diff --git a/qse/include/qse/cmn/Makefile.am b/qse/include/qse/cmn/Makefile.am
index 568d3c14..dfb1c27e 100644
--- a/qse/include/qse/cmn/Makefile.am
+++ b/qse/include/qse/cmn/Makefile.am
@@ -26,6 +26,7 @@ pkginclude_HEADERS = \
str.h \
time.h \
tio.h \
+ tre.h \
xma.h
if ENABLE_CXX
diff --git a/qse/include/qse/cmn/Makefile.in b/qse/include/qse/cmn/Makefile.in
index c47df1f5..0e8c7f95 100644
--- a/qse/include/qse/cmn/Makefile.in
+++ b/qse/include/qse/cmn/Makefile.in
@@ -53,8 +53,8 @@ SOURCES =
DIST_SOURCES =
am__pkginclude_HEADERS_DIST = alg.h chr.h dll.h env.h fio.h fma.h \
gdl.h htb.h lda.h main.h map.h mem.h misc.h oht.h opt.h pio.h \
- pma.h rbt.h rex.h sio.h sll.h stdio.h str.h time.h tio.h xma.h \
- Mmgr.hpp StdMmgr.hpp Mmged.hpp
+ pma.h rbt.h rex.h sio.h sll.h stdio.h str.h time.h tio.h tre.h \
+ xma.h Mmgr.hpp StdMmgr.hpp Mmged.hpp
am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
am__vpath_adj = case $$p in \
$(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
@@ -225,7 +225,7 @@ top_builddir = @top_builddir@
top_srcdir = @top_srcdir@
pkginclude_HEADERS = alg.h chr.h dll.h env.h fio.h fma.h gdl.h htb.h \
lda.h main.h map.h mem.h misc.h oht.h opt.h pio.h pma.h rbt.h \
- rex.h sio.h sll.h stdio.h str.h time.h tio.h xma.h \
+ rex.h sio.h sll.h stdio.h str.h time.h tio.h tre.h xma.h \
$(am__append_1)
all: all-am
diff --git a/qse/include/qse/cmn/dll.h b/qse/include/qse/cmn/dll.h
index 98e2eb7d..44413688 100644
--- a/qse/include/qse/cmn/dll.h
+++ b/qse/include/qse/cmn/dll.h
@@ -1,5 +1,5 @@
/*
- * $Id: dll.h 474 2011-05-23 16:52:37Z hyunghwan.chung $
+ * $Id: dll.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -257,7 +257,7 @@ void qse_dll_close (
/**
* The qse_dll_init() function initializes a statically declared list.
*/
-qse_dll_t* qse_dll_init (
+int qse_dll_init (
qse_dll_t* dll, /**< doubly linked list */
qse_mmgr_t* mmgr /**< memory manager */
);
diff --git a/qse/include/qse/cmn/env.h b/qse/include/qse/cmn/env.h
index 9ecb0a09..44a1e8e9 100644
--- a/qse/include/qse/cmn/env.h
+++ b/qse/include/qse/cmn/env.h
@@ -84,7 +84,7 @@ void qse_env_close (
qse_env_t* env
);
-qse_env_t* qse_env_init (
+int qse_env_init (
qse_env_t* env,
qse_mmgr_t* mmgr,
int fromcurenv
diff --git a/qse/include/qse/cmn/fio.h b/qse/include/qse/cmn/fio.h
index be5437ce..77318187 100644
--- a/qse/include/qse/cmn/fio.h
+++ b/qse/include/qse/cmn/fio.h
@@ -1,5 +1,5 @@
/*
- * $Id: fio.h 550 2011-08-14 15:59:55Z hyunghwan.chung $
+ * $Id: fio.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -152,7 +152,7 @@ void qse_fio_close (
/***
* The qse_fio_close() function opens a file into @a fio.
*/
-qse_fio_t* qse_fio_init (
+int qse_fio_init (
qse_fio_t* fio,
qse_mmgr_t* mmgr,
const qse_char_t* path,
diff --git a/qse/include/qse/cmn/fma.h b/qse/include/qse/cmn/fma.h
index 0f4fcabe..19b6a239 100644
--- a/qse/include/qse/cmn/fma.h
+++ b/qse/include/qse/cmn/fma.h
@@ -232,7 +232,7 @@ void qse_fma_close (
* The qse_fma_init() function initializes an memory allocator statically
* declared.
*/
-qse_fma_t* qse_fma_init (
+int qse_fma_init (
qse_fma_t* fma, /**< memory allocator */
qse_mmgr_t* mmgr, /**< outer memory manager */
qse_size_t blksize, /**< fixed block size in bytes */
diff --git a/qse/include/qse/cmn/htb.h b/qse/include/qse/cmn/htb.h
index 0ed56148..1fb8eb91 100644
--- a/qse/include/qse/cmn/htb.h
+++ b/qse/include/qse/cmn/htb.h
@@ -1,5 +1,5 @@
/*
- * $Id: htb.h 477 2011-05-24 04:22:40Z hyunghwan.chung $
+ * $Id: htb.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -341,7 +341,7 @@ void qse_htb_close (
/**
* The qse_htb_init() function initializes a hash table
*/
-qse_htb_t* qse_htb_init (
+int qse_htb_init (
qse_htb_t* htb, /**< hash table */
qse_mmgr_t* mmgr, /**< memory manager */
qse_size_t capa, /**< initial capacity */
diff --git a/qse/include/qse/cmn/lda.h b/qse/include/qse/cmn/lda.h
index 1bdbabef..9de8325b 100644
--- a/qse/include/qse/cmn/lda.h
+++ b/qse/include/qse/cmn/lda.h
@@ -1,5 +1,5 @@
/*
- * $Id: lda.h 479 2011-05-24 15:14:58Z hyunghwan.chung $
+ * $Id: lda.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -176,7 +176,7 @@ void qse_lda_close (
/**
* The qse_lda_init() function initializes a linear dynamic array.
*/
-qse_lda_t* qse_lda_init (
+int qse_lda_init (
qse_lda_t* lda,
qse_mmgr_t* mmgr,
qse_size_t capa
diff --git a/qse/include/qse/cmn/oht.h b/qse/include/qse/cmn/oht.h
index 9db47e0f..bf1be50e 100644
--- a/qse/include/qse/cmn/oht.h
+++ b/qse/include/qse/cmn/oht.h
@@ -165,10 +165,9 @@ void qse_oht_close (
);
/**
- * The qse_oht_open() function initializes a statically declared
- * open-addressed hash table.
+ * The qse_oht_open() function initializes an open-addressed hash table.
*/
-qse_oht_t* qse_oht_init (
+int qse_oht_init (
qse_oht_t* oht,
qse_mmgr_t* mmgr,
int scale,
diff --git a/qse/include/qse/cmn/pio.h b/qse/include/qse/cmn/pio.h
index 6e203462..bebbec97 100644
--- a/qse/include/qse/cmn/pio.h
+++ b/qse/include/qse/cmn/pio.h
@@ -1,5 +1,5 @@
/*
- * $Id: pio.h 539 2011-08-10 16:18:35Z hyunghwan.chung $
+ * $Id: pio.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -117,6 +117,7 @@ enum qse_pio_errnum_t
{
QSE_PIO_ENOERR = 0, /**< no error */
QSE_PIO_ENOMEM, /**< out of memory */
+ QSE_PIO_EINVAL, /**< out of memory */
QSE_PIO_ENOHND, /**< no handle available */
QSE_PIO_ECHILD, /**< the child is not valid */
QSE_PIO_EINTR, /**< interrupted */
@@ -223,9 +224,9 @@ void qse_pio_close (
* The qse_pio_init() functions performs the same task as the qse_pio_open()
* except that you need to allocate a #qse_pio_t structure and pass it to the
* function.
- * @return @a pio on success, #QSE_NULL on failure
+ * @return 0 on success, -1 on failure
*/
-qse_pio_t* qse_pio_init (
+int qse_pio_init (
qse_pio_t* pio, /**< pio object */
qse_mmgr_t* mmgr, /**< memory manager */
const qse_char_t* cmd, /**< command to execute */
diff --git a/qse/include/qse/cmn/pma.h b/qse/include/qse/cmn/pma.h
index 42dbcb37..ae9290da 100644
--- a/qse/include/qse/cmn/pma.h
+++ b/qse/include/qse/cmn/pma.h
@@ -71,7 +71,7 @@ void qse_pma_close (
qse_pma_t* pma /**< memory allocator */
);
-qse_pma_t* qse_pma_init (
+int qse_pma_init (
qse_pma_t* pma, /**< memory allocator */
qse_mmgr_t* mmgr /**< memory manager */
);
diff --git a/qse/include/qse/cmn/rbt.h b/qse/include/qse/cmn/rbt.h
index febd986c..05ccc6fa 100644
--- a/qse/include/qse/cmn/rbt.h
+++ b/qse/include/qse/cmn/rbt.h
@@ -301,7 +301,7 @@ void qse_rbt_close (
/**
* The qse_rbt_init() function initializes a red-black tree
*/
-qse_rbt_t* qse_rbt_init (
+int qse_rbt_init (
qse_rbt_t* rbt, /**< red-black tree */
qse_mmgr_t* mmgr, /**< memory manager */
int kscale, /**< key scale */
diff --git a/qse/include/qse/cmn/rex.h b/qse/include/qse/cmn/rex.h
index b8e94c83..e0fa5c42 100644
--- a/qse/include/qse/cmn/rex.h
+++ b/qse/include/qse/cmn/rex.h
@@ -1,5 +1,5 @@
/*
- * $Id: rex.h 462 2011-05-18 14:36:40Z hyunghwan.chung $
+ * $Id: rex.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -210,7 +210,7 @@ void qse_rex_close (
qse_rex_t* rex
);
-qse_rex_t* qse_rex_init (
+int qse_rex_init (
qse_rex_t* rex,
qse_mmgr_t* mmgr,
qse_rex_node_t* code
diff --git a/qse/include/qse/cmn/sio.h b/qse/include/qse/cmn/sio.h
index 617043d5..9529c136 100644
--- a/qse/include/qse/cmn/sio.h
+++ b/qse/include/qse/cmn/sio.h
@@ -1,5 +1,5 @@
/*
- * $Id: sio.h 547 2011-08-13 16:04:14Z hyunghwan.chung $
+ * $Id: sio.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -92,7 +92,7 @@ void qse_sio_close (
qse_sio_t* sio /**< stream */
);
-qse_sio_t* qse_sio_init (
+int qse_sio_init (
qse_sio_t* sio,
qse_mmgr_t* mmgr,
const qse_char_t* file,
diff --git a/qse/include/qse/cmn/sll.h b/qse/include/qse/cmn/sll.h
index 50ef43fd..c5fbf61c 100644
--- a/qse/include/qse/cmn/sll.h
+++ b/qse/include/qse/cmn/sll.h
@@ -1,5 +1,5 @@
/*
- * $Id: sll.h 474 2011-05-23 16:52:37Z hyunghwan.chung $
+ * $Id: sll.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -192,12 +192,11 @@ void qse_sll_close (
* to it. The caller may declare a static variable of the qse_sll_t type
* and pass its address. A memory manager still needs to be passed for
* node manipulation later.
- *
- * @return
+ * @return 0 on success, -1 on failure
* The qse_sll_init() function returns the first parameter on success and
* QSE_NULL on failure.
*/
-qse_sll_t* qse_sll_init (
+int qse_sll_init (
qse_sll_t* sll, /* singly linked list */
qse_mmgr_t* mmgr /* memory manager */
);
diff --git a/qse/include/qse/cmn/str.h b/qse/include/qse/cmn/str.h
index 805260d9..38484fd1 100644
--- a/qse/include/qse/cmn/str.h
+++ b/qse/include/qse/cmn/str.h
@@ -1,5 +1,5 @@
/*
- * $Id: str.h 550 2011-08-14 15:59:55Z hyunghwan.chung $
+ * $Id: str.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -877,6 +877,19 @@ int qse_wcsxncasecmp (
qse_size_t ln2 /**< length of the second string */
);
+int qse_mbszcmp (
+ const qse_mchar_t* s1,
+ const qse_mchar_t* s2,
+ qse_size_t n
+);
+
+int qse_wcszcmp (
+ const qse_wchar_t* s1,
+ const qse_wchar_t* s2,
+ qse_size_t n
+);
+
+
#ifdef QSE_CHAR_IS_MCHAR
# define qse_strcmp(s1,s2) qse_mbscmp(s1,s2)
# define qse_strxcmp(s1,ln1,s2) qse_mbsxcmp(s1,ln1,s2)
@@ -884,6 +897,7 @@ int qse_wcsxncasecmp (
# define qse_strcasecmp(s1,s2) qse_mbscasecmp(s1,s2)
# define qse_strxcasecmp(s1,ln1,s2) qse_mbsxcasecmp(s1,ln1,s2)
# define qse_strxncasecmp(s1,ln1,s2,ln2) qse_mbsxncasecmp(s1,ln1,s2,ln2)
+# define qse_strzcmp(s1,s2,n) qse_mbszcmp(s1,s2,n)
#else
# define qse_strcmp(s1,s2) qse_wcscmp(s1,s2)
# define qse_strxcmp(s1,ln1,s2) qse_wcsxcmp(s1,ln1,s2)
@@ -891,6 +905,7 @@ int qse_wcsxncasecmp (
# define qse_strcasecmp(s1,s2) qse_wcscasecmp(s1,s2)
# define qse_strxcasecmp(s1,ln1,s2) qse_wcsxcasecmp(s1,ln1,s2)
# define qse_strxncasecmp(s1,ln1,s2,ln2) qse_wcsxncasecmp(s1,ln1,s2,ln2)
+# define qse_strzcmp(s1,s2,n) qse_wcszcmp(s1,s2,n)
#endif
qse_mchar_t* qse_mbsdup (
@@ -2265,9 +2280,9 @@ void qse_mbs_close (
* The qse_mbs_init() function initializes a dynamically resizable string
* If the parameter capa is 0, it doesn't allocate the internal buffer
* in advance and always succeeds.
- * @return @a mbs on success, #QSE_NULL on failure.
+ * @return 0 on success, -1 on failure.
*/
-qse_mbs_t* qse_mbs_init (
+int qse_mbs_init (
qse_mbs_t* mbs,
qse_mmgr_t* mmgr,
qse_size_t capa
@@ -2439,9 +2454,9 @@ void qse_wcs_close (
* The qse_wcs_init() function initializes a dynamically resizable string
* If the parameter capa is 0, it doesn't allocate the internal buffer
* in advance and always succeeds.
- * @return @a wcs on success, #QSE_NULL on failure.
+ * @return 0 on success, -1 on failure.
*/
-qse_wcs_t* qse_wcs_init (
+int qse_wcs_init (
qse_wcs_t* wcs,
qse_mmgr_t* mmgr,
qse_size_t capa
diff --git a/qse/include/qse/cmn/tio.h b/qse/include/qse/cmn/tio.h
index edc6fb16..018cafd7 100644
--- a/qse/include/qse/cmn/tio.h
+++ b/qse/include/qse/cmn/tio.h
@@ -1,5 +1,5 @@
/*
- * $Id: tio.h 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: tio.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -126,7 +126,7 @@ QSE_DEFINE_COMMON_FUNCTIONS (tio)
*/
qse_tio_t* qse_tio_open (
qse_mmgr_t* mmgr,
- qse_size_t ext
+ qse_size_t xtnsize
);
/**
@@ -140,8 +140,8 @@ int qse_tio_close (
* The qse_tio_init() function initialize a statically declared
* text stream processor.
*/
-qse_tio_t* qse_tio_init (
- qse_tio_t* tip,
+int qse_tio_init (
+ qse_tio_t* tio,
qse_mmgr_t* mmgr
);
diff --git a/qse/include/qse/cmn/tre.h b/qse/include/qse/cmn/tre.h
new file mode 100644
index 00000000..abfb38d3
--- /dev/null
+++ b/qse/include/qse/cmn/tre.h
@@ -0,0 +1,134 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+#ifndef _QSE_TRE_TRE_H_
+#define _QSE_TRE_TRE_H_
+
+#include
+#include
+
+enum qse_tre_errnum_t
+{
+ QSE_TRE_ENOERR,
+ QSE_TRE_ENOMEM, /* Out of memory */
+ QSE_TRE_ENOMATCH, /* No match */
+ QSE_TRE_EBADPAT, /* Invalid regular expression */
+ QSE_TRE_ECOLLATE, /* Unknown collating element */
+ QSE_TRE_ECTYPE, /* Unknown character class name */
+ QSE_TRE_EESCAPE, /* Traling backslash */
+ QSE_TRE_ESUBREG, /* Invalid backreference */
+ QSE_TRE_EBRACK, /* "[]" imbalance */
+ QSE_TRE_EPAREN, /* "\(\)" or "()" imbalance */
+ QSE_TRE_EBRACE, /* "\{\}" or "{}" imbalance */
+ QSE_TRE_EBADBR, /* Invalid content of {} */
+ QSE_TRE_ERANGE, /* Invalid use of range operator */
+ QSE_TRE_EBADRPT /* Invalid use of repetition operators */
+};
+typedef enum qse_tre_errnum_t qse_tre_errnum_t;
+
+typedef struct qse_tre_t qse_tre_t;
+struct qse_tre_t
+{
+ qse_mmgr_t* mmgr;
+ qse_tre_errnum_t errnum;
+
+ qse_size_t re_nsub; /* Number of parenthesized subexpressions. */
+ void* value; /* For internal use only. */
+};
+
+#define QSE_TRE_ERRNUM(tre) ((const qse_tre_errnum_t)((tre)->errnum))
+
+typedef int qse_tre_off_t;
+
+typedef struct qse_tre_match_t qse_tre_match_t;
+struct qse_tre_match_t
+{
+ qse_tre_off_t rm_so;
+ qse_tre_off_t rm_eo;
+};
+
+enum qse_tre_cflag_t
+{
+ QSE_TRE_EXTENDED = (1 << 0),
+ QSE_TRE_IGNORECASE = (1 << 1),
+ QSE_TRE_NEWLINE = (1 << 2),
+ QSE_TRE_NOSUBREG = (1 << 3),
+ QSE_TRE_LITERAL = (1 << 4),
+ QSE_TRE_RIGHTASSOC = (1 << 5),
+ QSE_TRE_UNGREEDY = (1 << 6)
+};
+
+enum qse_tre_eflag_t
+{
+ QSE_TRE_NOTBOL = (1 << 0),
+ QSE_TRE_NOTEOL = (1 << 1),
+ QSE_TRE_BACKTRACKING = (1 << 2)
+};
+
+typedef struct qse_tre_strsrc_t qse_tre_strsrc_t;
+struct qse_tre_strsrc_t
+{
+ int (*get_next_char) (qse_char_t *c, unsigned int* pos_add, void* context);
+ void (*rewind)(qse_size_t pos, void *context);
+ int (*compare)(qse_size_t pos1, qse_size_t pos2, qse_size_t len, void* context);
+ void* context;
+};
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int qse_tre_init (
+ qse_tre_t* tre,
+ qse_mmgr_t* mmgr
+);
+
+void qse_tre_fini (
+ qse_tre_t* tre
+);
+
+
+int qse_tre_compx (
+ qse_tre_t* tre,
+ const qse_char_t* regex,
+ qse_size_t n,
+ int cflags
+);
+
+int qse_tre_comp (
+ qse_tre_t* tre,
+ const qse_char_t* regex,
+ int cflags
+);
+
+int qse_tre_execx (
+ qse_tre_t* tre,
+ const qse_char_t* str,
+ qse_size_t len,
+ qse_size_t nmatch,
+ qse_tre_match_t pmatch[],
+ int eflags
+);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/qse/include/qse/cmn/xma.h b/qse/include/qse/cmn/xma.h
index 615cb8bc..d404a58d 100644
--- a/qse/include/qse/cmn/xma.h
+++ b/qse/include/qse/cmn/xma.h
@@ -150,9 +150,9 @@ void qse_xma_close (
* to this function instead of calling qse_xma_open(). It obtains a memory zone
* of @a zonesize bytes with the memory manager @a mmgr. Unlike qse_xma_open(),
* it does not accept the extension size, thus not creating an extention area.
- * @return @a xma on success, #QSE_NULL on failure
+ * @return 0 on success, -1 on failure
*/
-qse_xma_t* qse_xma_init (
+int qse_xma_init (
qse_xma_t* xma, /**< memory allocator */
qse_mmgr_t* mmgr, /**< memory manager */
qse_size_t zonesize /**< zone size in bytes */
diff --git a/qse/include/qse/net/htrd.h b/qse/include/qse/net/htrd.h
index bbd75a93..67fc8c02 100644
--- a/qse/include/qse/net/htrd.h
+++ b/qse/include/qse/net/htrd.h
@@ -127,7 +127,7 @@ void qse_htrd_close (
qse_htrd_t* htrd
);
-qse_htrd_t* qse_htrd_init (
+int qse_htrd_init (
qse_htrd_t* htrd,
qse_mmgr_t* mmgr
);
diff --git a/qse/include/qse/net/htre.h b/qse/include/qse/net/htre.h
index 68ab2f14..78b4ed0b 100644
--- a/qse/include/qse/net/htre.h
+++ b/qse/include/qse/net/htre.h
@@ -129,7 +129,7 @@ typedef int (*qse_htre_header_walker_t) (
extern "C" {
#endif
-qse_htre_t* qse_htre_init (
+int qse_htre_init (
qse_htre_t* re,
qse_mmgr_t* mmgr
);
diff --git a/qse/lib/Makefile.am b/qse/lib/Makefile.am
index 8afb173d..b5254368 100644
--- a/qse/lib/Makefile.am
+++ b/qse/lib/Makefile.am
@@ -1,2 +1,2 @@
-SUBDIRS = cmn sed awk cut net stx
+SUBDIRS = cmn awk cut sed net stx
DIST_SUBDIRS = $(SUBDIRS)
diff --git a/qse/lib/Makefile.in b/qse/lib/Makefile.in
index c39ff054..52f51817 100644
--- a/qse/lib/Makefile.in
+++ b/qse/lib/Makefile.in
@@ -230,7 +230,7 @@ target_alias = @target_alias@
top_build_prefix = @top_build_prefix@
top_builddir = @top_builddir@
top_srcdir = @top_srcdir@
-SUBDIRS = cmn sed awk cut net stx
+SUBDIRS = cmn awk cut sed net stx
DIST_SUBDIRS = $(SUBDIRS)
all: all-recursive
diff --git a/qse/lib/awk/Awk.cpp b/qse/lib/awk/Awk.cpp
index ec460a99..49e9069f 100644
--- a/qse/lib/awk/Awk.cpp
+++ b/qse/lib/awk/Awk.cpp
@@ -1,5 +1,5 @@
/*
- * $Id: Awk.cpp 518 2011-07-24 14:24:13Z hyunghwan.chung $
+ * $Id: Awk.cpp 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -1066,7 +1066,8 @@ int Awk::open ()
QSE_ASSERT (awk == QSE_NULL && functionMap == QSE_NULL);
qse_awk_prm_t prm;
- memset (&prm, 0, QSE_SIZEOF(prm));
+
+ QSE_MEMSET (&prm, 0, QSE_SIZEOF(prm));
prm.sprintf = sprintf;
prm.math.pow = pow;
prm.math.mod = mod;
diff --git a/qse/lib/awk/fnc.c b/qse/lib/awk/fnc.c
index 2b0cd911..7a95aed8 100644
--- a/qse/lib/awk/fnc.c
+++ b/qse/lib/awk/fnc.c
@@ -1,5 +1,5 @@
/*
- * $Id: fnc.c 518 2011-07-24 14:24:13Z hyunghwan.chung $
+ * $Id: fnc.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -979,7 +979,7 @@ static int __substitute (qse_awk_rtx_t* run, qse_long_t max_count)
}
}
- if (qse_str_init (&new, run->awk->mmgr, s2.len) == QSE_NULL)
+ if (qse_str_init (&new, run->awk->mmgr, s2.len) <= -1)
{
qse_awk_rtx_seterrnum (run, QSE_AWK_ENOMEM, QSE_NULL);
goto oops;
@@ -1343,14 +1343,14 @@ static int fnc_sprintf (qse_awk_rtx_t* run, const qse_cstr_t* fnm)
nargs = qse_awk_rtx_getnargs (run);
QSE_ASSERT (nargs > 0);
- if (qse_str_init (&out, run->awk->mmgr, 256) == QSE_NULL)
+ if (qse_str_init (&out, run->awk->mmgr, 256) <= -1)
{
qse_awk_rtx_seterrnum (run, QSE_AWK_ENOMEM, QSE_NULL);
goto oops;
}
out_inited = 1;
- if (qse_str_init (&fbu, run->awk->mmgr, 256) == QSE_NULL)
+ if (qse_str_init (&fbu, run->awk->mmgr, 256) <= -1)
{
qse_awk_rtx_seterrnum (run, QSE_AWK_ENOMEM, QSE_NULL);
goto oops;
diff --git a/qse/lib/awk/misc.c b/qse/lib/awk/misc.c
index 078ae669..284a93c3 100644
--- a/qse/lib/awk/misc.c
+++ b/qse/lib/awk/misc.c
@@ -1,5 +1,5 @@
/*
- * $Id: misc.c 522 2011-07-25 13:08:07Z hyunghwan.chung $
+ * $Id: misc.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -1071,7 +1071,7 @@ void* qse_awk_buildrex (
p = qse_buildrex (
awk->mmgr, awk->rex.depth.max.build,
- ((awk->option&QSE_AWK_REXBOUND)? 0:QSE_REX_NOBOUND),
+ ((awk->option&QSE_AWK_REXBOUND)? 0: QSE_REX_NOBOUND),
ptn, len, &err
);
if (p == QSE_NULL) *errnum = QSE_AWK_REXERRTOERR(err);
diff --git a/qse/lib/awk/run.c b/qse/lib/awk/run.c
index c5fc1858..26c65817 100644
--- a/qse/lib/awk/run.c
+++ b/qse/lib/awk/run.c
@@ -1,5 +1,5 @@
/*
- * $Id: run.c 551 2011-08-15 13:52:48Z hyunghwan.chung $
+ * $Id: run.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -818,21 +818,21 @@ static int init_rtx (qse_awk_rtx_t* rtx, qse_awk_t* awk, qse_awk_rio_t* rio)
rtx->inrec.maxflds = 0;
rtx->inrec.d0 = qse_awk_val_nil;
if (qse_str_init (
- &rtx->inrec.line, MMGR(rtx), DEF_BUF_CAPA) == QSE_NULL)
+ &rtx->inrec.line, MMGR(rtx), DEF_BUF_CAPA) <= -1)
{
qse_awk_seterrnum (awk, QSE_AWK_ENOMEM, QSE_NULL);
return -1;
}
if (qse_str_init (
- &rtx->inrec.linew, MMGR(rtx), DEF_BUF_CAPA) == QSE_NULL)
+ &rtx->inrec.linew, MMGR(rtx), DEF_BUF_CAPA) <= -1)
{
qse_str_fini (&rtx->inrec.line);
qse_awk_seterrnum (awk, QSE_AWK_ENOMEM, QSE_NULL);
return -1;
}
- if (qse_str_init (&rtx->format.out, MMGR(rtx), 256) == QSE_NULL)
+ if (qse_str_init (&rtx->format.out, MMGR(rtx), 256) <= -1)
{
qse_str_fini (&rtx->inrec.linew);
qse_str_fini (&rtx->inrec.line);
@@ -840,7 +840,7 @@ static int init_rtx (qse_awk_rtx_t* rtx, qse_awk_t* awk, qse_awk_rio_t* rio)
return -1;
}
- if (qse_str_init (&rtx->format.fmt, MMGR(rtx), 256) == QSE_NULL)
+ if (qse_str_init (&rtx->format.fmt, MMGR(rtx), 256) <= -1)
{
qse_str_fini (&rtx->format.out);
qse_str_fini (&rtx->inrec.linew);
@@ -6311,7 +6311,7 @@ static qse_awk_val_t* eval_getline (qse_awk_rtx_t* run, qse_awk_nde_t* nde)
dst = (in == QSE_NULL)? QSE_T(""): in;
/* TODO: optimize the line buffer management */
- if (qse_str_init (&buf, MMGR(run), DEF_BUF_CAPA) == QSE_NULL)
+ if (qse_str_init (&buf, MMGR(run), DEF_BUF_CAPA) <= -1)
{
if (in != QSE_NULL) QSE_AWK_FREE (run->awk, in);
SETERR_LOC (run, QSE_AWK_ENOMEM, &nde->loc);
@@ -6483,7 +6483,7 @@ static int shorten_record (qse_awk_rtx_t* run, qse_size_t nflds)
}
if (qse_str_init (
- &tmp, MMGR(run), QSE_STR_LEN(&run->inrec.line)) == QSE_NULL)
+ &tmp, MMGR(run), QSE_STR_LEN(&run->inrec.line)) <= -1)
{
if (ofs_free != QSE_NULL)
QSE_AWK_FREE (run->awk, ofs_free);
@@ -6607,7 +6607,7 @@ static qse_char_t* idxnde_to_str (
out.type = QSE_AWK_RTX_VALTOSTR_STRPCAT;
out.u.strpcat = &idxstr;
- if (qse_str_init (&idxstr, MMGR(run), DEF_BUF_CAPA) == QSE_NULL)
+ if (qse_str_init (&idxstr, MMGR(run), DEF_BUF_CAPA) <= -1)
{
SETERR_LOC (run, QSE_AWK_ENOMEM, &nde->loc);
return QSE_NULL;
diff --git a/qse/lib/awk/val.c b/qse/lib/awk/val.c
index 469c02b4..a2221908 100644
--- a/qse/lib/awk/val.c
+++ b/qse/lib/awk/val.c
@@ -1,5 +1,5 @@
/*
- * $Id: val.c 518 2011-07-24 14:24:13Z hyunghwan.chung $
+ * $Id: val.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -448,10 +448,9 @@ qse_awk_val_t* qse_awk_rtx_makemapval (qse_awk_rtx_t* rtx)
val->nstr = 0;
val->map = (qse_htb_t*)(val + 1);
- val->map = qse_htb_init (
- val->map, rtx->awk->mmgr, 256, 70, QSE_SIZEOF(qse_char_t), 1
- );
- if (val->map == QSE_NULL)
+ if (qse_htb_init (
+ val->map, rtx->awk->mmgr,
+ 256, 70, QSE_SIZEOF(qse_char_t), 1) <= -1)
{
QSE_AWK_FREE (rtx->awk, val);
qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOMEM, QSE_NULL);
@@ -981,14 +980,14 @@ static int val_real_to_str (
tmp_len = rtx->gbl.convfmt.len;
}
- if (qse_str_init (&buf, rtx->awk->mmgr, 256) == QSE_NULL)
+ if (qse_str_init (&buf, rtx->awk->mmgr, 256) <= -1)
{
qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOMEM, QSE_NULL);
return -1;
}
buf_inited = 1;
- if (qse_str_init (&fbu, rtx->awk->mmgr, 256) == QSE_NULL)
+ if (qse_str_init (&fbu, rtx->awk->mmgr, 256) <= -1)
{
qse_str_fini (&buf);
qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOMEM, QSE_NULL);
diff --git a/qse/lib/cmn/Makefile.am b/qse/lib/cmn/Makefile.am
index 38e87d12..424f116c 100644
--- a/qse/lib/cmn/Makefile.am
+++ b/qse/lib/cmn/Makefile.am
@@ -5,26 +5,84 @@ AM_CPPFLAGS = \
-I$(top_srcdir)/include \
-I$(includedir)
+
lib_LTLIBRARIES = libqsecmn.la
+
libqsecmn_la_SOURCES = \
- syscall.h mem.h \
- mem.c xma.c fma.c pma.c chr.c chr_cnv.c rex.c \
- str_beg.c str_cat.c str_chr.c str_cnv.c str_cmp.c str_cpy.c \
- str_del.c str_dup.c str_dynm.c str_dynw.c str_end.c str_excl.c \
- str_fcpy.c str_incl.c str_len.c str_pac.c str_pbrk.c str_put.c \
- str_rev.c str_rot.c str_set.c str_spl.c str_spn.c str_str.c \
- str_subst.c str_tok.c str_trm.c str_word.c \
- lda.c oht.c htb.c rbt.c sll.c gdl.c dll.c opt.c \
- tio.c tio_get.c tio_put.c \
- fio.c pio.c sio.c \
- alg_search.c \
- alg_sort.c \
- env.c \
- time.c \
- misc.c \
+ tre.h \
+ alg-search.c \
+ alg-sort.c \
assert.c \
+ chr.c \
+ chr-cnv.c \
+ dll.c \
+ env.c \
+ gdl.c \
+ htb.c \
+ lda.c \
+ fio.c \
+ fma.c \
main.c \
- stdio.c
+ mem.c \
+ mem.h \
+ misc.c \
+ oht.c \
+ opt.c \
+ pma.c \
+ pio.c \
+ rbt.c \
+ rex.c \
+ sio.c \
+ sll.c \
+ str-beg.c \
+ str-cat.c \
+ str-chr.c \
+ str-cnv.c \
+ str-cmp.c \
+ str-cpy.c \
+ str-del.c \
+ str-dup.c \
+ str-dynm.c \
+ str-dynw.c \
+ str-end.c \
+ str-excl.c \
+ str-fcpy.c \
+ str-incl.c \
+ str-len.c \
+ str-pac.c \
+ str-pbrk.c \
+ str-put.c \
+ str-rev.c \
+ str-rot.c \
+ str-set.c \
+ str-spl.c \
+ str-spn.c \
+ str-str.c \
+ str-subst.c \
+ str-tok.c \
+ str-trm.c \
+ str-word.c \
+ syscall.h \
+ time.c \
+ tio.c \
+ tio-get.c \
+ tio-put.c \
+ tre.c \
+ tre.h \
+ tre-ast.c \
+ tre-ast.h \
+ tre-compile.c \
+ tre-compile.h \
+ tre-match-backtrack.c \
+ tre-match-parallel.c \
+ tre-match-utils.h \
+ tre-parse.c \
+ tre-parse.h \
+ tre-stack.c \
+ tre-stack.h \
+ stdio.c \
+ xma.c
+
libqsecmn_la_LDFLAGS = -L$(libdir) -version-info 1:0:0 -no-undefined
if WIN32
diff --git a/qse/lib/cmn/Makefile.in b/qse/lib/cmn/Makefile.in
index 22fd415e..2896e30d 100644
--- a/qse/lib/cmn/Makefile.in
+++ b/qse/lib/cmn/Makefile.in
@@ -73,16 +73,18 @@ am__base_list = \
am__installdirs = "$(DESTDIR)$(libdir)"
LTLIBRARIES = $(lib_LTLIBRARIES)
libqsecmn_la_DEPENDENCIES =
-am_libqsecmn_la_OBJECTS = mem.lo xma.lo fma.lo pma.lo chr.lo \
- chr_cnv.lo rex.lo str_beg.lo str_cat.lo str_chr.lo str_cnv.lo \
- str_cmp.lo str_cpy.lo str_del.lo str_dup.lo str_dynm.lo \
- str_dynw.lo str_end.lo str_excl.lo str_fcpy.lo str_incl.lo \
- str_len.lo str_pac.lo str_pbrk.lo str_put.lo str_rev.lo \
- str_rot.lo str_set.lo str_spl.lo str_spn.lo str_str.lo \
- str_subst.lo str_tok.lo str_trm.lo str_word.lo lda.lo oht.lo \
- htb.lo rbt.lo sll.lo gdl.lo dll.lo opt.lo tio.lo tio_get.lo \
- tio_put.lo fio.lo pio.lo sio.lo alg_search.lo alg_sort.lo \
- env.lo time.lo misc.lo assert.lo main.lo stdio.lo
+am_libqsecmn_la_OBJECTS = alg-search.lo alg-sort.lo assert.lo chr.lo \
+ chr-cnv.lo dll.lo env.lo gdl.lo htb.lo lda.lo fio.lo fma.lo \
+ main.lo mem.lo misc.lo oht.lo opt.lo pma.lo pio.lo rbt.lo \
+ rex.lo sio.lo sll.lo str-beg.lo str-cat.lo str-chr.lo \
+ str-cnv.lo str-cmp.lo str-cpy.lo str-del.lo str-dup.lo \
+ str-dynm.lo str-dynw.lo str-end.lo str-excl.lo str-fcpy.lo \
+ str-incl.lo str-len.lo str-pac.lo str-pbrk.lo str-put.lo \
+ str-rev.lo str-rot.lo str-set.lo str-spl.lo str-spn.lo \
+ str-str.lo str-subst.lo str-tok.lo str-trm.lo str-word.lo \
+ time.lo tio.lo tio-get.lo tio-put.lo tre.lo tre-ast.lo \
+ tre-compile.lo tre-match-backtrack.lo tre-match-parallel.lo \
+ tre-parse.lo tre-stack.lo stdio.lo xma.lo
libqsecmn_la_OBJECTS = $(am_libqsecmn_la_OBJECTS)
libqsecmn_la_LINK = $(LIBTOOL) --tag=CC $(AM_LIBTOOLFLAGS) \
$(LIBTOOLFLAGS) --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
@@ -272,24 +274,79 @@ AM_CPPFLAGS = \
lib_LTLIBRARIES = libqsecmn.la $(am__append_1)
libqsecmn_la_SOURCES = \
- syscall.h mem.h \
- mem.c xma.c fma.c pma.c chr.c chr_cnv.c rex.c \
- str_beg.c str_cat.c str_chr.c str_cnv.c str_cmp.c str_cpy.c \
- str_del.c str_dup.c str_dynm.c str_dynw.c str_end.c str_excl.c \
- str_fcpy.c str_incl.c str_len.c str_pac.c str_pbrk.c str_put.c \
- str_rev.c str_rot.c str_set.c str_spl.c str_spn.c str_str.c \
- str_subst.c str_tok.c str_trm.c str_word.c \
- lda.c oht.c htb.c rbt.c sll.c gdl.c dll.c opt.c \
- tio.c tio_get.c tio_put.c \
- fio.c pio.c sio.c \
- alg_search.c \
- alg_sort.c \
- env.c \
- time.c \
- misc.c \
+ tre.h \
+ alg-search.c \
+ alg-sort.c \
assert.c \
+ chr.c \
+ chr-cnv.c \
+ dll.c \
+ env.c \
+ gdl.c \
+ htb.c \
+ lda.c \
+ fio.c \
+ fma.c \
main.c \
- stdio.c
+ mem.c \
+ mem.h \
+ misc.c \
+ oht.c \
+ opt.c \
+ pma.c \
+ pio.c \
+ rbt.c \
+ rex.c \
+ sio.c \
+ sll.c \
+ str-beg.c \
+ str-cat.c \
+ str-chr.c \
+ str-cnv.c \
+ str-cmp.c \
+ str-cpy.c \
+ str-del.c \
+ str-dup.c \
+ str-dynm.c \
+ str-dynw.c \
+ str-end.c \
+ str-excl.c \
+ str-fcpy.c \
+ str-incl.c \
+ str-len.c \
+ str-pac.c \
+ str-pbrk.c \
+ str-put.c \
+ str-rev.c \
+ str-rot.c \
+ str-set.c \
+ str-spl.c \
+ str-spn.c \
+ str-str.c \
+ str-subst.c \
+ str-tok.c \
+ str-trm.c \
+ str-word.c \
+ syscall.h \
+ time.c \
+ tio.c \
+ tio-get.c \
+ tio-put.c \
+ tre.c \
+ tre.h \
+ tre-ast.c \
+ tre-ast.h \
+ tre-compile.c \
+ tre-compile.h \
+ tre-match-backtrack.c \
+ tre-match-parallel.c \
+ tre-match-utils.h \
+ tre-parse.c \
+ tre-parse.h \
+ tre-stack.c \
+ tre-stack.h \
+ stdio.c \
+ xma.c
libqsecmn_la_LDFLAGS = -L$(libdir) -version-info 1:0:0 -no-undefined
@WIN32_TRUE@libqsecmn_la_LIBADD = -lpsapi
@@ -375,11 +432,11 @@ distclean-compile:
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Mmgr.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/StdMmgr.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alg_search.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alg_sort.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alg-search.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/alg-sort.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/assert.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/chr-cnv.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/chr.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/chr_cnv.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/dll.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/env.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fio.Plo@am__quote@
@@ -399,38 +456,45 @@ distclean-compile:
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/sio.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/sll.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/stdio.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_beg.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_cat.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_chr.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_cmp.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_cnv.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_cpy.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_del.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_dup.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_dynm.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_dynw.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_end.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_excl.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_fcpy.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_incl.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_len.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_pac.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_pbrk.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_put.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_rev.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_rot.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_set.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_spl.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_spn.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_str.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_subst.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_tok.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_trm.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str_word.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-beg.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-cat.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-chr.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-cmp.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-cnv.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-cpy.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-del.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-dup.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-dynm.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-dynw.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-end.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-excl.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-fcpy.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-incl.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-len.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-pac.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-pbrk.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-put.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-rev.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-rot.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-set.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-spl.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-spn.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-str.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-subst.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-tok.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-trm.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str-word.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/time.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tio-get.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tio-put.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tio.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tio_get.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tio_put.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-ast.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-compile.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-match-backtrack.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-match-parallel.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-parse.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre-stack.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre.Plo@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/xma.Plo@am__quote@
.c.o:
diff --git a/qse/lib/cmn/alg_search.c b/qse/lib/cmn/alg-search.c
similarity index 100%
rename from qse/lib/cmn/alg_search.c
rename to qse/lib/cmn/alg-search.c
diff --git a/qse/lib/cmn/alg_sort.c b/qse/lib/cmn/alg-sort.c
similarity index 100%
rename from qse/lib/cmn/alg_sort.c
rename to qse/lib/cmn/alg-sort.c
diff --git a/qse/lib/cmn/chr_cnv.c b/qse/lib/cmn/chr-cnv.c
similarity index 98%
rename from qse/lib/cmn/chr_cnv.c
rename to qse/lib/cmn/chr-cnv.c
index 4508d736..90d92cb9 100644
--- a/qse/lib/cmn/chr_cnv.c
+++ b/qse/lib/cmn/chr-cnv.c
@@ -1,5 +1,5 @@
/*
- * $Id: chr_cnv.c 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: chr-cnv.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/chr.c b/qse/lib/cmn/chr.c
index a9706b35..9ee74564 100644
--- a/qse/lib/cmn/chr.c
+++ b/qse/lib/cmn/chr.c
@@ -1,5 +1,5 @@
/*
- * $Id: chr.c 555 2011-08-24 06:54:19Z hyunghwan.chung $
+ * $Id: chr.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -289,8 +289,8 @@ int qse_getwctypebyxname (const qse_wchar_t* name, qse_size_t len, qse_wctype_t*
mid = (left + right) / 2;
kwp = &wtab[mid];
- n = qse_wcsxcmp (name, len, wtab->name);
- if (n > 0)
+ n = qse_wcsxcmp (name, len, kwp->name);
+ if (n < 0)
{
/* if left, right, mid were of qse_size_t,
* you would need the following line.
@@ -298,7 +298,7 @@ int qse_getwctypebyxname (const qse_wchar_t* name, qse_size_t len, qse_wctype_t*
*/
right = mid - 1;
}
- else if (n < 0) left = mid + 1;
+ else if (n > 0) left = mid + 1;
else
{
*id = kwp->class;
@@ -332,8 +332,7 @@ static struct mtab_t
{ QSE_MT("punct"), QSE_MCTYPE_PUNCT },
{ QSE_MT("space"), QSE_MCTYPE_SPACE },
{ QSE_MT("upper"), QSE_MCTYPE_UPPER },
- { QSE_MT("xdigit"), QSE_MCTYPE_XDIGIT },
- { QSE_NULL, 0 }
+ { QSE_MT("xdigit"), QSE_MCTYPE_XDIGIT }
};
int qse_getmctypebyname (const qse_mchar_t* name, qse_mctype_t* id)
@@ -364,7 +363,7 @@ int qse_getmctypebyname (const qse_mchar_t* name, qse_mctype_t* id)
}
}
- return (qse_mctype_t)0;
+ return -1;
}
int qse_getmctypebyxname (const qse_mchar_t* name, qse_size_t len, qse_mctype_t* id)
@@ -378,8 +377,8 @@ int qse_getmctypebyxname (const qse_mchar_t* name, qse_size_t len, qse_mctype_t*
mid = (left + right) / 2;
kwp = &mtab[mid];
- n = qse_mbsxcmp (name, len, mtab->name);
- if (n > 0)
+ n = qse_mbsxcmp (name, len, kwp->name);
+ if (n < 0)
{
/* if left, right, mid were of qse_size_t,
* you would need the following line.
@@ -387,7 +386,7 @@ int qse_getmctypebyxname (const qse_mchar_t* name, qse_size_t len, qse_mctype_t*
*/
right = mid - 1;
}
- else if (n < 0) left = mid + 1;
+ else if (n > 0) left = mid + 1;
else
{
*id = kwp->class;
@@ -395,7 +394,7 @@ int qse_getmctypebyxname (const qse_mchar_t* name, qse_size_t len, qse_mctype_t*
}
}
- return (qse_mctype_t)0;
+ return -1;
}
qse_mctype_t qse_getmctype (const qse_mchar_t* name)
diff --git a/qse/lib/cmn/dll.c b/qse/lib/cmn/dll.c
index d1e92866..b7c43034 100644
--- a/qse/lib/cmn/dll.c
+++ b/qse/lib/cmn/dll.c
@@ -1,5 +1,5 @@
/*
- * $Id: dll.c 474 2011-05-23 16:52:37Z hyunghwan.chung $
+ * $Id: dll.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -29,8 +29,8 @@ QSE_IMPLEMENT_COMMON_FUNCTIONS (dll)
static int default_comper (
qse_dll_t* dll,
- const void* dptr1, size_t dlen1,
- const void* dptr2, size_t dlen2)
+ const void* dptr1, qse_size_t dlen1,
+ const void* dptr2, qse_size_t dlen2)
{
if (dlen1 == dlen2) return QSE_MEMCMP (dptr1, dptr2, TOB(dll,dlen1));
/* it just returns 1 to indicate that they are different. */
@@ -64,7 +64,7 @@ qse_dll_t* qse_dll_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
dll = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_dll_t) + xtnsize);
if (dll == QSE_NULL) return QSE_NULL;
- if (qse_dll_init (dll, mmgr) == QSE_NULL)
+ if (qse_dll_init (dll, mmgr) <= -1)
{
QSE_MMGR_FREE (mmgr, dll);
return QSE_NULL;
@@ -79,7 +79,7 @@ void qse_dll_close (qse_dll_t* dll)
QSE_MMGR_FREE (dll->mmgr, dll);
}
-qse_dll_t* qse_dll_init (qse_dll_t* dll, qse_mmgr_t* mmgr)
+int qse_dll_init (qse_dll_t* dll, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -93,7 +93,7 @@ qse_dll_t* qse_dll_init (qse_dll_t* dll, qse_mmgr_t* mmgr)
dll->copier = QSE_DLL_COPIER_SIMPLE;
QSE_DLL_INIT (dll);
- return dll;
+ return 0;
}
void qse_dll_fini (qse_dll_t* dll)
diff --git a/qse/lib/cmn/env.c b/qse/lib/cmn/env.c
index 0914fff8..10e1dc11 100644
--- a/qse/lib/cmn/env.c
+++ b/qse/lib/cmn/env.c
@@ -51,7 +51,7 @@ qse_env_t* qse_env_open (qse_mmgr_t* mmgr, qse_size_t xtnsize, int fromcurenv)
env = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_env_t) + xtnsize);
if (env == QSE_NULL) return QSE_NULL;
- if (qse_env_init (env, mmgr, fromcurenv) == QSE_NULL)
+ if (qse_env_init (env, mmgr, fromcurenv) <= -1)
{
QSE_MMGR_FREE (mmgr, env);
return QSE_NULL;
@@ -66,13 +66,13 @@ void qse_env_close (qse_env_t* env)
QSE_MMGR_FREE (env->mmgr, env);
}
-qse_env_t* qse_env_init (qse_env_t* env, qse_mmgr_t* mmgr, int fromcurenv)
+int qse_env_init (qse_env_t* env, qse_mmgr_t* mmgr, int fromcurenv)
{
QSE_MEMSET (env, 0, QSE_SIZEOF(*env));
env->mmgr = mmgr;
- if (fromcurenv && load_curenv (env) <= -1) return QSE_NULL;
- return env;
+ if (fromcurenv && load_curenv (env) <= -1) return -1;
+ return 0;
}
void qse_env_fini (qse_env_t* env)
diff --git a/qse/lib/cmn/fio.c b/qse/lib/cmn/fio.c
index 8f7a1113..1260d136 100644
--- a/qse/lib/cmn/fio.c
+++ b/qse/lib/cmn/fio.c
@@ -1,5 +1,5 @@
/*
- * $Id: fio.c 550 2011-08-14 15:59:55Z hyunghwan.chung $
+ * $Id: fio.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -24,7 +24,7 @@
#if defined(_WIN32)
# include
-# include
+# include /* for GetMappedFileName() */
# include
#elif defined(__OS2__)
# define INCL_DOSFILEMGR
@@ -67,7 +67,7 @@ qse_fio_t* qse_fio_open (
fio = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_fio_t) + ext);
if (fio == QSE_NULL) return QSE_NULL;
- if (qse_fio_init (fio, mmgr, path, flags, mode) == QSE_NULL)
+ if (qse_fio_init (fio, mmgr, path, flags, mode) <= -1)
{
QSE_MMGR_FREE (mmgr, fio);
return QSE_NULL;
@@ -82,7 +82,7 @@ void qse_fio_close (qse_fio_t* fio)
QSE_MMGR_FREE (fio->mmgr, fio);
}
-qse_fio_t* qse_fio_init (
+int qse_fio_init (
qse_fio_t* fio, qse_mmgr_t* mmgr,
const qse_char_t* path, int flags, int mode)
{
@@ -157,13 +157,13 @@ qse_fio_t* qse_fio_init (
creation_disposition, flag_and_attr, 0
);
}
- if (handle == INVALID_HANDLE_VALUE) return QSE_NULL;
+ if (handle == INVALID_HANDLE_VALUE) return -1;
/* some special check */
if (GetFileType(handle) == FILE_TYPE_UNKNOWN)
{
CloseHandle (handle);
- return QSE_NULL;
+ return -1;
}
/* TODO: support more features on WIN32 - TEMPORARY, DELETE_ON_CLOSE */
@@ -186,7 +186,7 @@ qse_fio_t* qse_fio_init (
#else
qse_mchar_t path_mb[CCHMAXPATH];
if (qse_wcstombs_strict (path,
- path_mb, QSE_COUNTOF(path_mb)) == -1) return QSE_NULL;
+ path_mb, QSE_COUNTOF(path_mb)) == -1) return -1;
#endif
zero.ulLo = 0;
@@ -253,7 +253,7 @@ qse_fio_t* qse_fio_init (
0L
);
- if (ret != NO_ERROR) return QSE_NULL;
+ if (ret != NO_ERROR) return -1;
}
#elif defined(__DOS__)
@@ -272,7 +272,7 @@ qse_fio_t* qse_fio_init (
#else
qse_mchar_t path_mb[_MAX_PATH];
if (qse_wcstombs_strict (path,
- path_mb, QSE_COUNTOF(path_mb)) == -1) return QSE_NULL;
+ path_mb, QSE_COUNTOF(path_mb)) == -1) return -1;
#endif
if (flags & QSE_FIO_APPEND)
@@ -303,7 +303,7 @@ qse_fio_t* qse_fio_init (
oflags,
permission
);
- if (handle <= -1) return QSE_NULL;
+ if (handle <= -1) return -1;
}
#else
@@ -321,7 +321,7 @@ qse_fio_t* qse_fio_init (
#else
qse_mchar_t path_mb[PATH_MAX + 1];
if (qse_wcstombs_strict (path,
- path_mb, QSE_COUNTOF(path_mb)) == -1) return QSE_NULL;
+ path_mb, QSE_COUNTOF(path_mb)) == -1) return -1;
#endif
/*
* rwa -> RDWR | APPEND
@@ -358,7 +358,7 @@ qse_fio_t* qse_fio_init (
handle = QSE_OPEN (path_mb, desired_access, mode);
}
- if (handle == -1) return QSE_NULL;
+ if (handle == -1) return -1;
#endif
@@ -369,8 +369,8 @@ qse_fio_t* qse_fio_init (
tio = qse_tio_open (fio->mmgr, 0);
if (tio == QSE_NULL) QSE_THROW_ERR (tio);
- if (qse_tio_attachin (tio, fio_input, fio) == -1 ||
- qse_tio_attachout (tio, fio_output, fio) == -1)
+ if (qse_tio_attachin (tio, fio_input, fio) <= -1 ||
+ qse_tio_attachout (tio, fio_output, fio) <= -1)
{
qse_tio_close (tio);
QSE_THROW_ERR (tio);
@@ -387,7 +387,7 @@ qse_fio_t* qse_fio_init (
#else
QSE_CLOSE (handle);
#endif
- return QSE_NULL;
+ return -1;
}
fio->tio = tio;
@@ -395,7 +395,7 @@ qse_fio_t* qse_fio_init (
fio->handle = handle;
- return fio;
+ return 0;
}
void qse_fio_fini (qse_fio_t* fio)
@@ -436,15 +436,17 @@ qse_fio_off_t qse_fio_seek (
QSE_ASSERT (QSE_SIZEOF(offset) <= QSE_SIZEOF(x.QuadPart));
+ /* SetFilePointerEx is not available on Windows NT 4.
+ * So let's use SetFilePointerEx */
+#if 0
x.QuadPart = offset;
if (SetFilePointerEx (fio->handle, x, &y, seek_map[origin]) == FALSE)
{
return (qse_fio_off_t)-1;
}
-
return (qse_fio_off_t)y.QuadPart;
+#endif
- /*
x.QuadPart = offset;
x.LowPart = SetFilePointer (
fio->handle, x.LowPart, &x.HighPart, seek_map[origin]);
@@ -452,9 +454,8 @@ qse_fio_off_t qse_fio_seek (
{
return (qse_fio_off_t)-1;
}
-
return (qse_fio_off_t)x.QuadPart;
- */
+
#elif defined(__OS2__)
static int seek_map[] =
{
@@ -518,11 +519,15 @@ qse_fio_off_t qse_fio_seek (
int qse_fio_truncate (qse_fio_t* fio, qse_fio_off_t size)
{
#if defined(_WIN32)
+#if 0
LARGE_INTEGER x;
x.QuadPart = size;
if (SetFilePointerEx(fio->handle,x,NULL,FILE_BEGIN) == FALSE ||
SetEndOfFile(fio->handle) == FALSE) return -1;
+#endif
+ if (qse_fio_seek (fio, size, QSE_FIO_BEGIN) == (qse_fio_off_t)-1) return -1;
+ if (SetEndOfFile(fio->handle) == FALSE) return -1;
return 0;
#elif defined(__OS2__)
@@ -637,7 +642,6 @@ qse_ssize_t qse_fio_flush (qse_fio_t* fio)
return qse_tio_flush (fio->tio);
}
-
#if defined(_WIN32)
static int get_devname_from_handle (
diff --git a/qse/lib/cmn/fma.c b/qse/lib/cmn/fma.c
index ee833c4d..ef222e62 100644
--- a/qse/lib/cmn/fma.c
+++ b/qse/lib/cmn/fma.c
@@ -42,7 +42,7 @@ qse_fma_t* qse_fma_open (
fma = (qse_fma_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(*fma) + xtnsize);
if (fma == QSE_NULL) return QSE_NULL;
- if (qse_fma_init (fma, mmgr, blksize, maxblks, maxcnks) == QSE_NULL)
+ if (qse_fma_init (fma, mmgr, blksize, maxblks, maxcnks) <= -1)
{
QSE_MMGR_FREE (mmgr, fma);
return QSE_NULL;
@@ -57,7 +57,7 @@ void qse_fma_close (qse_fma_t* fma)
QSE_MMGR_FREE (fma->mmgr, fma);
}
-qse_fma_t* qse_fma_init (
+int qse_fma_init (
qse_fma_t* fma, qse_mmgr_t* mmgr,
qse_size_t blksize, qse_size_t maxblks, qse_size_t maxcnks)
{
@@ -74,7 +74,7 @@ qse_fma_t* qse_fma_init (
fma->maxblks = maxblks;
fma->maxcnks = maxcnks;
- return fma;
+ return 0;
}
void qse_fma_fini (qse_fma_t* fma)
diff --git a/qse/lib/cmn/htb.c b/qse/lib/cmn/htb.c
index cbde752b..6fb3e8a1 100644
--- a/qse/lib/cmn/htb.c
+++ b/qse/lib/cmn/htb.c
@@ -1,5 +1,5 @@
/*
- * $Id: htb.c 492 2011-06-15 16:00:36Z hyunghwan.chung $
+ * $Id: htb.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -274,7 +274,7 @@ htb_t* qse_htb_open (
htb = (htb_t*) QSE_MMGR_ALLOC (mmgr, SIZEOF(htb_t) + xtnsize);
if (htb == QSE_NULL) return QSE_NULL;
- if (qse_htb_init (htb, mmgr, capa, factor, kscale, vscale) == QSE_NULL)
+ if (qse_htb_init (htb, mmgr, capa, factor, kscale, vscale) <= -1)
{
QSE_MMGR_FREE (mmgr, htb);
return QSE_NULL;
@@ -289,7 +289,7 @@ void qse_htb_close (htb_t* htb)
QSE_MMGR_FREE (htb->mmgr, htb);
}
-htb_t* qse_htb_init (
+int qse_htb_init (
htb_t* htb, mmgr_t* mmgr, size_t capa,
int factor, int kscale, int vscale)
{
@@ -310,10 +310,10 @@ htb_t* qse_htb_init (
htb->mmgr = mmgr;
htb->bucket = QSE_MMGR_ALLOC (mmgr, capa*SIZEOF(pair_t*));
- if (htb->bucket == QSE_NULL) return QSE_NULL;
+ if (htb->bucket == QSE_NULL) return -1;
/*for (i = 0; i < capa; i++) htb->bucket[i] = QSE_NULL;*/
- QSE_MEMSET (htb->bucket, 0, capa*SIZEOF(pair_t*));
+ QSE_MEMSET (htb->bucket, 0, capa * SIZEOF(pair_t*));
htb->factor = factor;
htb->scale[QSE_HTB_KEY] = (kscale < 1)? 1: kscale;
@@ -325,7 +325,7 @@ htb_t* qse_htb_init (
if (htb->capa > 0 && htb->threshold <= 0) htb->threshold = 1;
htb->mancbs = &mancbs[0];
- return htb;
+ return 0;
}
void qse_htb_fini (htb_t* htb)
diff --git a/qse/lib/cmn/lda.c b/qse/lib/cmn/lda.c
index b44b9007..c3b35cdf 100644
--- a/qse/lib/cmn/lda.c
+++ b/qse/lib/cmn/lda.c
@@ -1,5 +1,5 @@
/*
- * $Id: lda.c 479 2011-05-24 15:14:58Z hyunghwan.chung $
+ * $Id: lda.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -111,7 +111,7 @@ lda_t* qse_lda_open (mmgr_t* mmgr, size_t ext, size_t capa)
lda = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(lda_t) + ext);
if (lda == QSE_NULL) return QSE_NULL;
- if (qse_lda_init (lda, mmgr, capa) == QSE_NULL)
+ if (qse_lda_init (lda, mmgr, capa) <= -1)
{
QSE_MMGR_FREE (mmgr, lda);
return QSE_NULL;
@@ -126,7 +126,7 @@ void qse_lda_close (lda_t* lda)
QSE_MMGR_FREE (lda->mmgr, lda);
}
-lda_t* qse_lda_init (lda_t* lda, mmgr_t* mmgr, size_t capa)
+int qse_lda_init (lda_t* lda, mmgr_t* mmgr, size_t capa)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -140,8 +140,7 @@ lda_t* qse_lda_init (lda_t* lda, mmgr_t* mmgr, size_t capa)
lda->copier = QSE_LDA_COPIER_SIMPLE;
lda->comper = default_comparator;
- if (qse_lda_setcapa (lda, capa) == QSE_NULL) return QSE_NULL;
- return lda;
+ return (qse_lda_setcapa (lda, capa) == QSE_NULL)? -1: 0;
}
void qse_lda_fini (lda_t* lda)
diff --git a/qse/lib/cmn/main.c b/qse/lib/cmn/main.c
index 1b6120ad..27836be0 100644
--- a/qse/lib/cmn/main.c
+++ b/qse/lib/cmn/main.c
@@ -26,7 +26,8 @@
int qse_runmain (
int argc, qse_achar_t* argv[], qse_runmain_handler_t handler)
{
- setlocale (LC_ALL, ""); /* TODO: remove dependency on setlocale */
+ /* TODO: remove dependency on setlocale */
+ setlocale (LC_ALL, "");
#if (defined(QSE_ACHAR_IS_MCHAR) && defined(QSE_CHAR_IS_MCHAR)) || \
(defined(QSE_ACHAR_IS_WCHAR) && defined(QSE_CHAR_IS_WCHAR))
diff --git a/qse/lib/cmn/mem.h b/qse/lib/cmn/mem.h
index 5c67962f..551cdcaa 100644
--- a/qse/lib/cmn/mem.h
+++ b/qse/lib/cmn/mem.h
@@ -1,5 +1,5 @@
/*
- * $Id: mem.h 441 2011-04-22 14:28:43Z hyunghwan.chung $
+ * $Id: mem.h 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -24,22 +24,22 @@
#include
#ifdef MINIMIZE_PLATFORM_DEPENDENCY
-# define QSE_MEMCPY(dst,src,len) qse_memcpy(dst,src,len)
-# define QSE_MEMCMP(p1,p2,len) qse_memcmp(p1,p2,len)
-# define QSE_MEMSET(dst,val,len) qse_memset(dst,val,len)
-# define QSE_MEMBYTE(s,val,len) qse_membyte(s,val,len)
-# define QSE_MEMRBYTE(s,val,len) qse_memrbyte(s,val,len)
-# define QSE_MEMMEM(hs,hl,nd,nl) qse_memmem(hs,hl,nd,nl)
+# define QSE_MEMCPY(dst,src,len) qse_memcpy(dst,src,len)
+# define QSE_MEMCMP(p1,p2,len) qse_memcmp(p1,p2,len)
+# define QSE_MEMSET(dst,val,len) qse_memset(dst,val,len)
+# define QSE_MEMBYTE(s,val,len) qse_membyte(s,val,len)
+# define QSE_MEMRBYTE(s,val,len) qse_memrbyte(s,val,len)
+# define QSE_MEMMEM(hs,hl,nd,nl) qse_memmem(hs,hl,nd,nl)
# define QSE_MEMRMEM(hs,hl,nd,nl) qse_memrmem(hs,hl,nd,nl)
#else
# include
-# define QSE_MEMCPY(dst,src,len) memcpy(dst,src,len)
-# define QSE_MEMCMP(p1,p2,len) memcmp(p1,p2,len)
-# define QSE_MEMSET(dst,val,len) memset(dst,val,len)
-# define QSE_MEMBYTE(s,val,len) memchr(s,val,len)
-# define QSE_MEMRBYTE(s,val,len) memrchr(s,val,len)
-# define QSE_MEMMEM(hs,hl,nd,nl) memmem(hs,hl,nd,nl)
-# define QSE_MEMRMEM(hs,hl,nd,nl) memrmem(hs,hl,nd,nl)
+# define QSE_MEMCPY(dst,src,len) memcpy(dst,src,len)
+# define QSE_MEMCMP(p1,p2,len) memcmp(p1,p2,len)
+# define QSE_MEMSET(dst,val,len) memset(dst,val,len)
+# define QSE_MEMBYTE(s,val,len) memchr(s,val,len)
+# define QSE_MEMRBYTE(s,val,len) memrchr(s,val,len)
+# define QSE_MEMMEM(hs,hl,nd,nl) memmem(hs,hl,nd,nl)
+# define QSE_MEMRMEM(hs,hl,nd,nl) qse_memrmem(hs,hl,nd,nl)
#endif
#define QSE_MALLOC(mmgr,size) QSE_MMGR_ALLOC(mmgr,size)
diff --git a/qse/lib/cmn/oht.c b/qse/lib/cmn/oht.c
index 6aed9942..c8ac9e63 100644
--- a/qse/lib/cmn/oht.c
+++ b/qse/lib/cmn/oht.c
@@ -9,7 +9,7 @@ QSE_IMPLEMENT_COMMON_FUNCTIONS (oht)
static QSE_INLINE_ALWAYS qse_size_t default_hasher (
qse_oht_t* oht, const void* data)
{
- size_t h = 5381;
+ qse_size_t h = 5381;
const qse_byte_t* p = (const qse_byte_t*)data;
const qse_byte_t* bound = p + oht->scale;
while (p < bound) h = ((h << 5) + h) + *p++;
@@ -61,7 +61,7 @@ qse_oht_t* qse_oht_open (
oht = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_oht_t) + xtnsize);
if (oht == QSE_NULL) return QSE_NULL;
- if (qse_oht_init (oht, mmgr, scale, capa, limit) == QSE_NULL)
+ if (qse_oht_init (oht, mmgr, scale, capa, limit) <= -1)
{
QSE_MMGR_FREE (mmgr, oht);
return QSE_NULL;
@@ -76,7 +76,7 @@ void qse_oht_close (qse_oht_t* oht)
QSE_MMGR_FREE (oht->mmgr, oht);
}
-qse_oht_t* qse_oht_init (
+int qse_oht_init (
qse_oht_t* oht, qse_mmgr_t* mmgr,
int scale, qse_size_t capa, qse_size_t limit)
{
@@ -99,17 +99,17 @@ qse_oht_t* qse_oht_init (
oht->copier = default_copier;*/
oht->mark = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_oht_mark_t) * capa);
- if (!oht->mark) return QSE_NULL;
+ if (!oht->mark) return -1;
oht->data = QSE_MMGR_ALLOC (mmgr, scale * capa);
if (!oht->data)
{
QSE_MMGR_FREE (mmgr, oht->mark);
- return QSE_NULL;
+ return -1;
}
for (i = 0; i < capa; i++) oht->mark[i] = QSE_OHT_EMPTY;
- return oht;
+ return 0;
}
void qse_oht_fini (qse_oht_t* oht)
diff --git a/qse/lib/cmn/pio.c b/qse/lib/cmn/pio.c
index c88b4114..4819dc06 100644
--- a/qse/lib/cmn/pio.c
+++ b/qse/lib/cmn/pio.c
@@ -1,5 +1,5 @@
/*
- * $Id: pio.c 543 2011-08-12 16:35:34Z hyunghwan.chung $
+ * $Id: pio.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -62,7 +62,7 @@ qse_pio_t* qse_pio_open (
pio = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_pio_t) + ext);
if (pio == QSE_NULL) return QSE_NULL;
- if (qse_pio_init (pio, mmgr, cmd, env, oflags) == QSE_NULL)
+ if (qse_pio_init (pio, mmgr, cmd, env, oflags) <= -1)
{
QSE_MMGR_FREE (mmgr, pio);
return QSE_NULL;
@@ -77,7 +77,7 @@ void qse_pio_close (qse_pio_t* pio)
QSE_MMGR_FREE (pio->mmgr, pio);
}
-qse_pio_t* qse_pio_init (
+int qse_pio_init (
qse_pio_t* pio, qse_mmgr_t* mmgr, const qse_char_t* cmd,
qse_env_t* env, int oflags)
{
@@ -183,7 +183,11 @@ qse_pio_t* qse_pio_init (
maxidx = 5;
}
- if (maxidx == -1) goto oops;
+ if (maxidx == -1)
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops;
+ }
if ((oflags & QSE_PIO_INTONUL) ||
(oflags & QSE_PIO_OUTTONUL) ||
@@ -255,7 +259,11 @@ qse_pio_t* qse_pio_init (
{
const qse_mchar_t* mbs = (const qse_mchar_t*)cmd;
qse_size_t ll = qse_mbstowcslen (mbs, &reqlen);
- if (mbs[ll] != QSE_MT('\0')) goto oops; /* illegal sequence */
+ if (mbs[ll] != QSE_MT('\0'))
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops; /* illegal sequence */
+ }
}
else
{
@@ -270,7 +278,11 @@ qse_pio_t* qse_pio_init (
dupcmd = QSE_MMGR_ALLOC (
mmgr, (11 + reqlen) * QSE_SIZEOF(*dupcmd)
);
- if (dupcmd == QSE_NULL) goto oops;
+ if (dupcmd == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
qse_strcpy (dupcmd, QSE_T("cmd.exe /c "));
@@ -302,7 +314,11 @@ qse_pio_t* qse_pio_init (
#if defined(QSE_CHAR_IS_WCHAR)
}
#endif
- if (dupcmd == QSE_NULL) goto oops;
+ if (dupcmd == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
}
x = CreateProcess (
@@ -403,7 +419,11 @@ qse_pio_t* qse_pio_init (
maxidx = 5;
}
- if (maxidx == -1) goto oops;
+ if (maxidx == -1)
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops;
+ }
if ((oflags & QSE_PIO_INTONUL) ||
(oflags & QSE_PIO_OUTTONUL) ||
@@ -513,12 +533,20 @@ qse_pio_t* qse_pio_init (
else
{
n = qse_wcstombslen (cmd, &mn);
- if (cmd[n] != QSE_WT('\0')) goto oops; /* illegal sequence found */
+ if (cmd[n] != QSE_WT('\0'))
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops; /* illegal sequence found */
+ }
}
#endif
cmd_line = QSE_MMGR_ALLOC (
mmgr, ((11+mn+1+1) * QSE_SIZEOF(*cmd_line)));
- if (cmd_line == QSE_NULL) goto oops;
+ if (cmd_line == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
qse_mbscpy (cmd_line, QSE_MT("cmd.exe")); /* cmd.exe\0/c */
qse_mbscpy (&cmd_line[8], QSE_MT("/c "));
@@ -547,24 +575,40 @@ qse_pio_t* qse_pio_init (
#ifdef QSE_CHAR_IS_MCHAR
mn = qse_strlen(cmd);
cmd_line = qse_strdup2 (cmd, QSE_T(" "), pio->mmgr);
- if (cmd_line == QSE_NULL) goto oops;
+ if (cmd_line == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
#else
if (oflags & QSE_PIO_MBSCMD)
{
mn = qse_mbslen((const qse_mchar_t*)cmd);
cmd_line = qse_mbsdup2 ((const qse_mchar_t*)cmd, QSE_MT(" "), pio->mmgr);
- if (cmd_line == QSE_NULL) goto oops;
+ if (cmd_line == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
}
else
{
qse_size_t n;
n = qse_wcstombslen (cmd, &mn);
- if (cmd[n] != QSE_T('\0')) goto oops; /* illegal sequence in cmd */
+ if (cmd[n] != QSE_T('\0'))
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops; /* illegal sequence in cmd */
+ }
mn = mn + 1;
cmd_line = QSE_MMGR_ALLOC (pio->mmgr, mn * QSE_SIZEOF(qse_char_t));
- if (cmd_line == QSE_NULL) goto oops;
+ if (cmd_line == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
qse_wcstombs (cmd, cmd_line, &mn);
}
@@ -611,7 +655,7 @@ qse_pio_t* qse_pio_init (
#elif defined(__DOS__)
/* DOS not multi-processed. can't support pio */
- return QSE_NULL;
+ return -1;
#else
@@ -635,7 +679,11 @@ qse_pio_t* qse_pio_init (
maxidx = 5;
}
- if (maxidx == -1) goto oops;
+ if (maxidx == -1)
+ {
+ pio->errnum = QSE_PIO_EINVAL;
+ goto oops;
+ }
pid = QSE_FORK();
if (pid <= -1) goto oops;
@@ -956,7 +1004,11 @@ qse_pio_t* qse_pio_init (
int r;
tio[i] = qse_tio_open (pio->mmgr, 0);
- if (tio[i] == QSE_NULL) goto oops;
+ if (tio[i] == QSE_NULL)
+ {
+ pio->errnum = QSE_PIO_ENOMEM;
+ goto oops;
+ }
r = (i == QSE_PIO_IN)?
qse_tio_attachout (tio[i], pio_output, &pio->pin[i]):
@@ -969,9 +1021,11 @@ qse_pio_t* qse_pio_init (
}
pio->option = 0;
- return pio;
+ return 0;
oops:
+ if (pio->errnum == QSE_PIO_ENOERR)
+ pio->errnum = QSE_PIO_ESUBSYS;
#if defined(_WIN32)
if (windevnul != INVALID_HANDLE_VALUE) CloseHandle (windevnul);
@@ -1019,7 +1073,7 @@ oops:
}
#endif
- return QSE_NULL;
+ return -1;
}
void qse_pio_fini (qse_pio_t* pio)
@@ -1054,6 +1108,7 @@ const qse_char_t* qse_pio_geterrmsg (qse_pio_t* pio)
{
QSE_T("no error"),
QSE_T("out of memory"),
+ QSE_T("invalid parameter"),
QSE_T("no handle available"),
QSE_T("child process not valid"),
QSE_T("interruped"),
diff --git a/qse/lib/cmn/pma.c b/qse/lib/cmn/pma.c
index c5cc5686..87a22f70 100644
--- a/qse/lib/cmn/pma.c
+++ b/qse/lib/cmn/pma.c
@@ -56,7 +56,7 @@ qse_pma_t* qse_pma_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
pma = (qse_pma_t*)QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(*pma) + xtnsize);
if (pma == QSE_NULL) return QSE_NULL;
- if (qse_pma_init (pma, mmgr) == QSE_NULL)
+ if (qse_pma_init (pma, mmgr) <= -1)
{
QSE_MMGR_FREE (mmgr, pma);
return QSE_NULL;
@@ -71,14 +71,14 @@ void qse_pma_close (qse_pma_t* pma)
QSE_MMGR_FREE (pma->mmgr, pma);
}
-qse_pma_t* qse_pma_init (qse_pma_t* pma, qse_mmgr_t* mmgr)
+int qse_pma_init (qse_pma_t* pma, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
QSE_MEMSET (pma, 0, QSE_SIZEOF(*pma));
pma->mmgr = mmgr;
- return pma;
+ return 0;
}
/* Frees the memory allocator and all memory allocated with it. */
diff --git a/qse/lib/cmn/rbt.c b/qse/lib/cmn/rbt.c
index bc9fca1f..f5183dcb 100644
--- a/qse/lib/cmn/rbt.c
+++ b/qse/lib/cmn/rbt.c
@@ -214,7 +214,7 @@ rbt_t* qse_rbt_open (mmgr_t* mmgr, size_t xtnsize, int kscale, int vscale)
rbt = (rbt_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(rbt_t) + xtnsize);
if (rbt == QSE_NULL) return QSE_NULL;
- if (qse_rbt_init (rbt, mmgr, kscale, vscale) == QSE_NULL)
+ if (qse_rbt_init (rbt, mmgr, kscale, vscale) <= -1)
{
QSE_MMGR_FREE (mmgr, rbt);
return QSE_NULL;
@@ -229,7 +229,7 @@ void qse_rbt_close (rbt_t* rbt)
QSE_MMGR_FREE (rbt->mmgr, rbt);
}
-rbt_t* qse_rbt_init (rbt_t* rbt, mmgr_t* mmgr, int kscale, int vscale)
+int qse_rbt_init (rbt_t* rbt, mmgr_t* mmgr, int kscale, int vscale)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -252,7 +252,7 @@ rbt_t* qse_rbt_init (rbt_t* rbt, mmgr_t* mmgr, int kscale, int vscale)
/* root is set to nil initially */
rbt->root = &rbt->nil;
- return rbt;
+ return 0;
}
void qse_rbt_fini (rbt_t* rbt)
diff --git a/qse/lib/cmn/rex.c b/qse/lib/cmn/rex.c
index 58988bc4..eb2c63ef 100644
--- a/qse/lib/cmn/rex.c
+++ b/qse/lib/cmn/rex.c
@@ -1,5 +1,5 @@
/*
- * $Id: rex.c 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: rex.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -118,7 +118,7 @@ struct cand_t
QSE_IMPLEMENT_COMMON_FUNCTIONS (rex)
-qse_rex_t* qse_rex_init (qse_rex_t* rex, qse_mmgr_t* mmgr, qse_rex_node_t* code)
+int qse_rex_init (qse_rex_t* rex, qse_mmgr_t* mmgr, qse_rex_node_t* code)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -132,7 +132,7 @@ qse_rex_t* qse_rex_init (qse_rex_t* rex, qse_mmgr_t* mmgr, qse_rex_node_t* code)
* is closed, the code delegated is destroyed. */
rex->code = code;
- return rex;
+ return 0;
}
qse_rex_t* qse_rex_open (qse_mmgr_t* mmgr, qse_size_t xtn, qse_rex_node_t* code)
@@ -152,7 +152,7 @@ qse_rex_t* qse_rex_open (qse_mmgr_t* mmgr, qse_size_t xtn, qse_rex_node_t* code)
rex = (qse_rex_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_rex_t) + xtn);
if (rex == QSE_NULL) return QSE_NULL;
- if (qse_rex_init (rex, mmgr, code) == QSE_NULL)
+ if (qse_rex_init (rex, mmgr, code) <= -1)
{
QSE_MMGR_FREE (mmgr, rex);
return QSE_NULL;
@@ -1967,14 +1967,14 @@ static int comp_cand (qse_lda_t* lda,
static int init_exec_dds (exec_t* e, qse_mmgr_t* mmgr)
{
/* initializes dynamic data structures */
- if (qse_lda_init (&e->cand.set[0], mmgr, 100) == QSE_NULL)
+ if (qse_lda_init (&e->cand.set[0], mmgr, 100) <= -1)
{
- /* TOOD: set error */
+ e->rex->errnum = QSE_REX_ENOMEM;
return -1;
}
- if (qse_lda_init (&e->cand.set[1], mmgr, 100) == QSE_NULL)
+ if (qse_lda_init (&e->cand.set[1], mmgr, 100) <= -1)
{
- /* TOOD: set error */
+ e->rex->errnum = QSE_REX_ENOMEM;
qse_lda_fini (&e->cand.set[0]);
return -1;
}
diff --git a/qse/lib/cmn/sio.c b/qse/lib/cmn/sio.c
index 7c0b75bd..43de9c61 100644
--- a/qse/lib/cmn/sio.c
+++ b/qse/lib/cmn/sio.c
@@ -1,5 +1,5 @@
/*
- * $Id: sio.c 454 2011-05-06 15:28:27Z hyunghwan.chung $
+ * $Id: sio.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -176,7 +176,7 @@ qse_sio_t* qse_sio_open (
sio = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_sio_t) + xtnsize);
if (sio == QSE_NULL) return QSE_NULL;
- if (qse_sio_init (sio, mmgr, file, flags) == QSE_NULL)
+ if (qse_sio_init (sio, mmgr, file, flags) <= -1)
{
QSE_MMGR_FREE (mmgr, sio);
return QSE_NULL;
@@ -191,7 +191,7 @@ void qse_sio_close (qse_sio_t* sio)
QSE_MMGR_FREE (sio->mmgr, sio);
}
-qse_sio_t* qse_sio_init (
+int qse_sio_init (
qse_sio_t* sio, qse_mmgr_t* mmgr, const qse_char_t* file, int flags)
{
int mode;
@@ -204,26 +204,23 @@ qse_sio_t* qse_sio_init (
mode = QSE_FIO_RUSR | QSE_FIO_WUSR |
QSE_FIO_RGRP | QSE_FIO_ROTH;
- if (qse_fio_init (&sio->fio, mmgr, file, flags, mode) == QSE_NULL)
- {
- return QSE_NULL;
- }
+ if (qse_fio_init (&sio->fio, mmgr, file, flags, mode) <= -1) return -1;
- if (qse_tio_init(&sio->tio, mmgr) == QSE_NULL)
+ if (qse_tio_init(&sio->tio, mmgr) <= -1)
{
qse_fio_fini (&sio->fio);
- return QSE_NULL;
+ return -1;
}
- if (qse_tio_attachin(&sio->tio, __sio_input, sio) == -1 ||
- qse_tio_attachout(&sio->tio, __sio_output, sio) == -1)
+ if (qse_tio_attachin(&sio->tio, __sio_input, sio) <= -1 ||
+ qse_tio_attachout(&sio->tio, __sio_output, sio) <= -1)
{
qse_tio_fini (&sio->tio);
qse_fio_fini (&sio->fio);
- return QSE_NULL;
+ return -1;
}
- return sio;
+ return 0;
}
void qse_sio_fini (qse_sio_t* sio)
diff --git a/qse/lib/cmn/sll.c b/qse/lib/cmn/sll.c
index 897239b9..7706a27d 100644
--- a/qse/lib/cmn/sll.c
+++ b/qse/lib/cmn/sll.c
@@ -1,5 +1,5 @@
/*
- * $Id: sll.c 441 2011-04-22 14:28:43Z hyunghwan.chung $
+ * $Id: sll.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -117,7 +117,7 @@ sll_t* qse_sll_open (mmgr_t* mmgr, size_t ext)
sll = QSE_MMGR_ALLOC (mmgr, SIZEOF(sll_t) + ext);
if (sll == QSE_NULL) return QSE_NULL;
- if (qse_sll_init (sll, mmgr) == QSE_NULL)
+ if (qse_sll_init (sll, mmgr) <= -1)
{
QSE_MMGR_FREE (mmgr, sll);
return QSE_NULL;
@@ -132,7 +132,7 @@ void qse_sll_close (sll_t* sll)
QSE_MMGR_FREE (sll->mmgr, sll);
}
-sll_t* qse_sll_init (sll_t* sll, mmgr_t* mmgr)
+int qse_sll_init (sll_t* sll, mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -145,7 +145,7 @@ sll_t* qse_sll_init (sll_t* sll, mmgr_t* mmgr)
sll->comper = default_comper;
sll->copier = QSE_SLL_COPIER_SIMPLE;
- return sll;
+ return 0;
}
void qse_sll_fini (sll_t* sll)
diff --git a/qse/lib/cmn/str_beg.c b/qse/lib/cmn/str-beg.c
similarity index 98%
rename from qse/lib/cmn/str_beg.c
rename to qse/lib/cmn/str-beg.c
index 4b42ad40..0cd4f6c1 100644
--- a/qse/lib/cmn/str_beg.c
+++ b/qse/lib/cmn/str-beg.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_beg.c 550 2011-08-14 15:59:55Z hyunghwan.chung $
+ * $Id: str-beg.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_cat.c b/qse/lib/cmn/str-cat.c
similarity index 98%
rename from qse/lib/cmn/str_cat.c
rename to qse/lib/cmn/str-cat.c
index 7e803dec..5111e2cf 100644
--- a/qse/lib/cmn/str_cat.c
+++ b/qse/lib/cmn/str-cat.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_cat.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-cat.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_chr.c b/qse/lib/cmn/str-chr.c
similarity index 97%
rename from qse/lib/cmn/str_chr.c
rename to qse/lib/cmn/str-chr.c
index 6c96761d..09b337c7 100644
--- a/qse/lib/cmn/str_chr.c
+++ b/qse/lib/cmn/str-chr.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_chr.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-chr.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_cmp.c b/qse/lib/cmn/str-cmp.c
similarity index 90%
rename from qse/lib/cmn/str_cmp.c
rename to qse/lib/cmn/str-cmp.c
index 9f382b21..535c6d22 100644
--- a/qse/lib/cmn/str_cmp.c
+++ b/qse/lib/cmn/str-cmp.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_cmp.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-cmp.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -116,6 +116,20 @@ int qse_mbsxncasecmp (
return (s2 < end2)? -1: 0;
}
+
+int qse_mbszcmp (const qse_mchar_t* s1, const qse_mchar_t* s2, qse_size_t n)
+{
+ if (n == 0) return 0;
+
+ while (*s1 == *s2)
+ {
+ if (*s1 == QSE_MT('\0') || n == 1) return 0;
+ s1++, s2++, n--;
+ }
+
+ return (*s1 > *s2)? 1: -1;
+}
+
int qse_wcscmp (const qse_wchar_t* s1, const qse_wchar_t* s2)
{
while (*s1 == *s2)
@@ -210,3 +224,16 @@ int qse_wcsxncasecmp (
return (s2 < end2)? -1: 0;
}
+
+int qse_wcszcmp (const qse_wchar_t* s1, const qse_wchar_t* s2, qse_size_t n)
+{
+ if (n == 0) return 0;
+
+ while (*s1 == *s2)
+ {
+ if (*s1 == QSE_WT('\0') || n == 1) return 0;
+ s1++, s2++, n--;
+ }
+
+ return (*s1 > *s2)? 1: -1;
+}
diff --git a/qse/lib/cmn/str_cnv.c b/qse/lib/cmn/str-cnv.c
similarity index 98%
rename from qse/lib/cmn/str_cnv.c
rename to qse/lib/cmn/str-cnv.c
index aa82e10f..ab0abd29 100644
--- a/qse/lib/cmn/str_cnv.c
+++ b/qse/lib/cmn/str-cnv.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_cnv.c 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: str-cnv.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -197,12 +197,12 @@ qse_size_t qse_mbstowcs (
return 0; /* 0 byte processed */
}
- for (mp = mbs; *mp != '\0'; mp++);
+ for (mp = mbs; *mp != QSE_MT('\0'); mp++);
mlen = qse_mbsntowcsn (mbs, mp - mbs, wcs, &wlen);
if (wlen < *wcslen)
{
/* null-terminate wcs if it is large enough. */
- wcs[wlen] = L'\0';
+ wcs[wlen] = QSE_WT('\0');
}
/* if null-terminated properly, the input wcslen must be less than
diff --git a/qse/lib/cmn/str_cpy.c b/qse/lib/cmn/str-cpy.c
similarity index 98%
rename from qse/lib/cmn/str_cpy.c
rename to qse/lib/cmn/str-cpy.c
index e6644201..c475548d 100644
--- a/qse/lib/cmn/str_cpy.c
+++ b/qse/lib/cmn/str-cpy.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_cpy.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-cpy.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_del.c b/qse/lib/cmn/str-del.c
similarity index 96%
rename from qse/lib/cmn/str_del.c
rename to qse/lib/cmn/str-del.c
index 2040c822..176e9373 100644
--- a/qse/lib/cmn/str_del.c
+++ b/qse/lib/cmn/str-del.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_del.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-del.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_dup.c b/qse/lib/cmn/str-dup.c
similarity index 98%
rename from qse/lib/cmn/str_dup.c
rename to qse/lib/cmn/str-dup.c
index 5c6070fc..32263921 100644
--- a/qse/lib/cmn/str_dup.c
+++ b/qse/lib/cmn/str-dup.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_dup.c 499 2011-06-22 16:17:35Z hyunghwan.chung $
+ * $Id: str-dup.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_dynm.c b/qse/lib/cmn/str-dynm.c
similarity index 97%
rename from qse/lib/cmn/str_dynm.c
rename to qse/lib/cmn/str-dynm.c
index 0b32f40d..fdf3e304 100644
--- a/qse/lib/cmn/str_dynm.c
+++ b/qse/lib/cmn/str-dynm.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_dynm.c 502 2011-07-06 16:44:10Z hyunghwan.chung $
+ * $Id: str-dynm.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -40,7 +40,7 @@ qse_mbs_t* qse_mbs_open (qse_mmgr_t* mmgr, qse_size_t ext, qse_size_t capa)
str = (qse_mbs_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_mbs_t) + ext);
if (str == QSE_NULL) return QSE_NULL;
- if (qse_mbs_init (str, mmgr, capa) == QSE_NULL)
+ if (qse_mbs_init (str, mmgr, capa) <= -1)
{
QSE_MMGR_FREE (mmgr, str);
return QSE_NULL;
@@ -55,7 +55,7 @@ void qse_mbs_close (qse_mbs_t* str)
QSE_MMGR_FREE (str->mmgr, str);
}
-qse_mbs_t* qse_mbs_init (qse_mbs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
+int qse_mbs_init (qse_mbs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -69,14 +69,14 @@ qse_mbs_t* qse_mbs_init (qse_mbs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
{
str->val.ptr = (qse_mchar_t*) QSE_MMGR_ALLOC (
mmgr, QSE_SIZEOF(qse_mchar_t) * (capa + 1));
- if (str->val.ptr == QSE_NULL) return QSE_NULL;
+ if (str->val.ptr == QSE_NULL) return -1;
str->val.ptr[0] = QSE_MT('\0');
}
str->val.len = 0;
str->capa = capa;
- return str;
+ return 0;
}
void qse_mbs_fini (qse_mbs_t* str)
diff --git a/qse/lib/cmn/str_dynw.c b/qse/lib/cmn/str-dynw.c
similarity index 97%
rename from qse/lib/cmn/str_dynw.c
rename to qse/lib/cmn/str-dynw.c
index 10a14f71..068040b6 100644
--- a/qse/lib/cmn/str_dynw.c
+++ b/qse/lib/cmn/str-dynw.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_dynw.c 502 2011-07-06 16:44:10Z hyunghwan.chung $
+ * $Id: str-dynw.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -40,7 +40,7 @@ qse_wcs_t* qse_wcs_open (qse_mmgr_t* mmgr, qse_size_t ext, qse_size_t capa)
str = (qse_wcs_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_wcs_t) + ext);
if (str == QSE_NULL) return QSE_NULL;
- if (qse_wcs_init (str, mmgr, capa) == QSE_NULL)
+ if (qse_wcs_init (str, mmgr, capa) <= -1)
{
QSE_MMGR_FREE (mmgr, str);
return QSE_NULL;
@@ -55,7 +55,7 @@ void qse_wcs_close (qse_wcs_t* str)
QSE_MMGR_FREE (str->mmgr, str);
}
-qse_wcs_t* qse_wcs_init (qse_wcs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
+int qse_wcs_init (qse_wcs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -69,14 +69,14 @@ qse_wcs_t* qse_wcs_init (qse_wcs_t* str, qse_mmgr_t* mmgr, qse_size_t capa)
{
str->val.ptr = (qse_wchar_t*) QSE_MMGR_ALLOC (
mmgr, QSE_SIZEOF(qse_wchar_t) * (capa + 1));
- if (str->val.ptr == QSE_NULL) return QSE_NULL;
+ if (str->val.ptr == QSE_NULL) return -1;
str->val.ptr[0] = QSE_WT('\0');
}
str->val.len = 0;
str->capa = capa;
- return str;
+ return 0;
}
void qse_wcs_fini (qse_wcs_t* str)
diff --git a/qse/lib/cmn/str_end.c b/qse/lib/cmn/str-end.c
similarity index 97%
rename from qse/lib/cmn/str_end.c
rename to qse/lib/cmn/str-end.c
index defdcfd0..ac1455d2 100644
--- a/qse/lib/cmn/str_end.c
+++ b/qse/lib/cmn/str-end.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_end.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-end.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_excl.c b/qse/lib/cmn/str-excl.c
similarity index 96%
rename from qse/lib/cmn/str_excl.c
rename to qse/lib/cmn/str-excl.c
index b90096e7..5ea76e3f 100644
--- a/qse/lib/cmn/str_excl.c
+++ b/qse/lib/cmn/str-excl.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_excl.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-excl.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_fcpy.c b/qse/lib/cmn/str-fcpy.c
similarity index 99%
rename from qse/lib/cmn/str_fcpy.c
rename to qse/lib/cmn/str-fcpy.c
index 518d3147..08f37abf 100644
--- a/qse/lib/cmn/str_fcpy.c
+++ b/qse/lib/cmn/str-fcpy.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_fcpy.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-fcpy.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_incl.c b/qse/lib/cmn/str-incl.c
similarity index 96%
rename from qse/lib/cmn/str_incl.c
rename to qse/lib/cmn/str-incl.c
index dd644025..16bf9c73 100644
--- a/qse/lib/cmn/str_incl.c
+++ b/qse/lib/cmn/str-incl.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_incl.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-incl.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_len.c b/qse/lib/cmn/str-len.c
similarity index 95%
rename from qse/lib/cmn/str_len.c
rename to qse/lib/cmn/str-len.c
index 63fc2ad8..c3d078cb 100644
--- a/qse/lib/cmn/str_len.c
+++ b/qse/lib/cmn/str-len.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_len.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-len.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_pac.c b/qse/lib/cmn/str-pac.c
similarity index 97%
rename from qse/lib/cmn/str_pac.c
rename to qse/lib/cmn/str-pac.c
index c0953be5..5b6a19af 100644
--- a/qse/lib/cmn/str_pac.c
+++ b/qse/lib/cmn/str-pac.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_pac.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-pac.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_pbrk.c b/qse/lib/cmn/str-pbrk.c
similarity index 95%
rename from qse/lib/cmn/str_pbrk.c
rename to qse/lib/cmn/str-pbrk.c
index 0a9465a8..6f050467 100644
--- a/qse/lib/cmn/str_pbrk.c
+++ b/qse/lib/cmn/str-pbrk.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_pbrk.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-pbrk.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_put.c b/qse/lib/cmn/str-put.c
similarity index 97%
rename from qse/lib/cmn/str_put.c
rename to qse/lib/cmn/str-put.c
index b49a89c3..b7090dc8 100644
--- a/qse/lib/cmn/str_put.c
+++ b/qse/lib/cmn/str-put.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_put.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-put.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_rev.c b/qse/lib/cmn/str-rev.c
similarity index 95%
rename from qse/lib/cmn/str_rev.c
rename to qse/lib/cmn/str-rev.c
index 255ed750..1f804ea6 100644
--- a/qse/lib/cmn/str_rev.c
+++ b/qse/lib/cmn/str-rev.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_rev.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-rev.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_rot.c b/qse/lib/cmn/str-rot.c
similarity index 97%
rename from qse/lib/cmn/str_rot.c
rename to qse/lib/cmn/str-rot.c
index c767248f..e6e0c27a 100644
--- a/qse/lib/cmn/str_rot.c
+++ b/qse/lib/cmn/str-rot.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_rot.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-rot.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_set.c b/qse/lib/cmn/str-set.c
similarity index 96%
rename from qse/lib/cmn/str_set.c
rename to qse/lib/cmn/str-set.c
index e70d83fb..79459e6f 100644
--- a/qse/lib/cmn/str_set.c
+++ b/qse/lib/cmn/str-set.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_set.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-set.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_spl.c b/qse/lib/cmn/str-spl.c
similarity index 99%
rename from qse/lib/cmn/str_spl.c
rename to qse/lib/cmn/str-spl.c
index ef721296..54bc4843 100644
--- a/qse/lib/cmn/str_spl.c
+++ b/qse/lib/cmn/str-spl.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_spl.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-spl.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_spn.c b/qse/lib/cmn/str-spn.c
similarity index 97%
rename from qse/lib/cmn/str_spn.c
rename to qse/lib/cmn/str-spn.c
index c7f59771..46579946 100644
--- a/qse/lib/cmn/str_spn.c
+++ b/qse/lib/cmn/str-spn.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_spn.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-spn.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_str.c b/qse/lib/cmn/str-str.c
similarity index 99%
rename from qse/lib/cmn/str_str.c
rename to qse/lib/cmn/str-str.c
index 5b080c3e..fcc69044 100644
--- a/qse/lib/cmn/str_str.c
+++ b/qse/lib/cmn/str-str.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_str.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-str.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_subst.c b/qse/lib/cmn/str-subst.c
similarity index 98%
rename from qse/lib/cmn/str_subst.c
rename to qse/lib/cmn/str-subst.c
index 21085cf4..9462f220 100644
--- a/qse/lib/cmn/str_subst.c
+++ b/qse/lib/cmn/str-subst.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_subst.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-subst.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_tok.c b/qse/lib/cmn/str-tok.c
similarity index 99%
rename from qse/lib/cmn/str_tok.c
rename to qse/lib/cmn/str-tok.c
index 0a6c33c2..26d786e6 100644
--- a/qse/lib/cmn/str_tok.c
+++ b/qse/lib/cmn/str-tok.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_tok.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-tok.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_trm.c b/qse/lib/cmn/str-trm.c
similarity index 98%
rename from qse/lib/cmn/str_trm.c
rename to qse/lib/cmn/str-trm.c
index 8e2cb67a..b693cff2 100644
--- a/qse/lib/cmn/str_trm.c
+++ b/qse/lib/cmn/str-trm.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_trm.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-trm.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/str_word.c b/qse/lib/cmn/str-word.c
similarity index 98%
rename from qse/lib/cmn/str_word.c
rename to qse/lib/cmn/str-word.c
index 2476443f..ecd7aa8e 100644
--- a/qse/lib/cmn/str_word.c
+++ b/qse/lib/cmn/str-word.c
@@ -1,5 +1,5 @@
/*
- * $Id: str_word.c 443 2011-04-25 14:56:05Z hyunghwan.chung $
+ * $Id: str-word.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/tio_get.c b/qse/lib/cmn/tio-get.c
similarity index 97%
rename from qse/lib/cmn/tio_get.c
rename to qse/lib/cmn/tio-get.c
index 7bcd1940..6fc01d0c 100644
--- a/qse/lib/cmn/tio_get.c
+++ b/qse/lib/cmn/tio-get.c
@@ -1,5 +1,5 @@
/*
- * $Id: tio_get.c 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: tio-get.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/tio_put.c b/qse/lib/cmn/tio-put.c
similarity index 97%
rename from qse/lib/cmn/tio_put.c
rename to qse/lib/cmn/tio-put.c
index a1db0f88..c8cf98a8 100644
--- a/qse/lib/cmn/tio_put.c
+++ b/qse/lib/cmn/tio-put.c
@@ -1,5 +1,5 @@
/*
- * $Id: tio_put.c 554 2011-08-22 05:26:26Z hyunghwan.chung $
+ * $Id: tio-put.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
diff --git a/qse/lib/cmn/tio.c b/qse/lib/cmn/tio.c
index 4f83e423..14883c75 100644
--- a/qse/lib/cmn/tio.c
+++ b/qse/lib/cmn/tio.c
@@ -1,5 +1,5 @@
/*
- * $Id: tio.c 441 2011-04-22 14:28:43Z hyunghwan.chung $
+ * $Id: tio.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -23,7 +23,7 @@
QSE_IMPLEMENT_COMMON_FUNCTIONS (tio)
-qse_tio_t* qse_tio_open (qse_mmgr_t* mmgr, qse_size_t ext)
+qse_tio_t* qse_tio_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
{
qse_tio_t* tio;
@@ -37,10 +37,10 @@ qse_tio_t* qse_tio_open (qse_mmgr_t* mmgr, qse_size_t ext)
if (mmgr == QSE_NULL) return QSE_NULL;
}
- tio = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_tio_t) + ext);
+ tio = QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_tio_t) + xtnsize);
if (tio == QSE_NULL) return QSE_NULL;
- if (qse_tio_init (tio, mmgr) == QSE_NULL)
+ if (qse_tio_init (tio, mmgr) <= -1)
{
QSE_MMGR_FREE (mmgr, tio);
return QSE_NULL;
@@ -56,7 +56,7 @@ int qse_tio_close (qse_tio_t* tio)
return n;
}
-qse_tio_t* qse_tio_init (qse_tio_t* tio, qse_mmgr_t* mmgr)
+int qse_tio_init (qse_tio_t* tio, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -70,15 +70,15 @@ qse_tio_t* qse_tio_init (qse_tio_t* tio, qse_mmgr_t* mmgr)
tio->output_func = QSE_NULL;
tio->output_arg = QSE_NULL;
- tio->input_status = 0;
- tio->inbuf_curp = 0;
- tio->inbuf_len = 0;
- tio->outbuf_len = 0;
+ tio->input_status = 0;
+ tio->inbuf_curp = 0;
+ tio->inbuf_len = 0;
+ tio->outbuf_len = 0;
*/
tio->errnum = QSE_TIO_ENOERR;
- return tio;
+ return 0;
}
int qse_tio_fini (qse_tio_t* tio)
@@ -135,9 +135,9 @@ int qse_tio_attachin (qse_tio_t* tio, qse_tio_io_t input, void* arg)
tio->input_func = input;
tio->input_arg = arg;
- tio->input_status = 0;
- tio->inbuf_curp = 0;
- tio->inbuf_len = 0;
+ tio->input_status = 0;
+ tio->inbuf_curp = 0;
+ tio->inbuf_len = 0;
return 0;
}
diff --git a/qse/lib/cmn/tre-ast.c b/qse/lib/cmn/tre-ast.c
new file mode 100644
index 00000000..07025b71
--- /dev/null
+++ b/qse/lib/cmn/tre-ast.c
@@ -0,0 +1,267 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-ast.c - Abstract syntax tree (AST) routines
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "tre-ast.h"
+
+tre_ast_node_t *
+tre_ast_new_node(tre_mem_t mem, tre_ast_type_t type, size_t size)
+{
+ tre_ast_node_t *node;
+
+ node = tre_mem_calloc(mem, sizeof(*node));
+ if (!node)
+ return NULL;
+ node->obj = tre_mem_calloc(mem, size);
+ if (!node->obj)
+ return NULL;
+ node->type = type;
+ node->nullable = -1;
+ node->submatch_id = -1;
+
+ return node;
+}
+
+tre_ast_node_t *
+tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position)
+{
+ tre_ast_node_t *node;
+ tre_literal_t *lit;
+
+ node = tre_ast_new_node(mem, LITERAL, sizeof(tre_literal_t));
+ if (!node)
+ return NULL;
+ lit = node->obj;
+ lit->code_min = code_min;
+ lit->code_max = code_max;
+ lit->position = position;
+
+ return node;
+}
+
+tre_ast_node_t *
+tre_ast_new_iter(tre_mem_t mem, tre_ast_node_t *arg, int min, int max,
+ int minimal)
+{
+ tre_ast_node_t *node;
+ tre_iteration_t *iter;
+
+ node = tre_ast_new_node(mem, ITERATION, sizeof(tre_iteration_t));
+ if (!node)
+ return NULL;
+ iter = node->obj;
+ iter->arg = arg;
+ iter->min = min;
+ iter->max = max;
+ iter->minimal = minimal;
+ node->num_submatches = arg->num_submatches;
+
+ return node;
+}
+
+tre_ast_node_t *
+tre_ast_new_union(tre_mem_t mem, tre_ast_node_t *left, tre_ast_node_t *right)
+{
+ tre_ast_node_t *node;
+
+ node = tre_ast_new_node(mem, UNION, sizeof(tre_union_t));
+ if (node == NULL)
+ return NULL;
+ ((tre_union_t *)node->obj)->left = left;
+ ((tre_union_t *)node->obj)->right = right;
+ node->num_submatches = left->num_submatches + right->num_submatches;
+
+ return node;
+}
+
+tre_ast_node_t *
+tre_ast_new_catenation(tre_mem_t mem, tre_ast_node_t *left,
+ tre_ast_node_t *right)
+{
+ tre_ast_node_t *node;
+
+ node = tre_ast_new_node(mem, CATENATION, sizeof(tre_catenation_t));
+ if (node == NULL)
+ return NULL;
+ ((tre_catenation_t *)node->obj)->left = left;
+ ((tre_catenation_t *)node->obj)->right = right;
+ node->num_submatches = left->num_submatches + right->num_submatches;
+
+ return node;
+}
+
+#ifdef TRE_DEBUG
+
+static void
+tre_findent(FILE *stream, int i)
+{
+ while (i-- > 0)
+ fputc(' ', stream);
+}
+
+void
+tre_print_params(int *params)
+{
+ int i;
+ if (params)
+ {
+ DPRINT(("params ["));
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ {
+ if (params[i] == TRE_PARAM_UNSET)
+ DPRINT(("unset"));
+ else if (params[i] == TRE_PARAM_DEFAULT)
+ DPRINT(("default"));
+ else
+ DPRINT(("%d", params[i]));
+ if (i < TRE_PARAM_LAST - 1)
+ DPRINT((", "));
+ }
+ DPRINT(("]"));
+ }
+}
+
+static void
+tre_do_print(FILE *stream, tre_ast_node_t *ast, int indent)
+{
+ int code_min, code_max, pos;
+ int num_tags = ast->num_tags;
+ tre_literal_t *lit;
+ tre_iteration_t *iter;
+
+ tre_findent(stream, indent);
+ switch (ast->type)
+ {
+ case LITERAL:
+ lit = ast->obj;
+ code_min = lit->code_min;
+ code_max = lit->code_max;
+ pos = lit->position;
+ if (IS_EMPTY(lit))
+ {
+ fprintf(stream, "literal empty\n");
+ }
+ else if (IS_ASSERTION(lit))
+ {
+ int i;
+ char *assertions[] = { "bol", "eol", "ctype", "!ctype",
+ "bow", "eow", "wb", "!wb"
+ };
+ if (code_max >= ASSERT_LAST << 1)
+ assert(0);
+ fprintf(stream, "assertions: ");
+ for (i = 0; (1 << i) <= ASSERT_LAST; i++)
+ if (code_max & (1 << i))
+ fprintf(stream, "%s ", assertions[i]);
+ fprintf(stream, "\n");
+ }
+ else if (IS_TAG(lit))
+ {
+ fprintf(stream, "tag %d\n", code_max);
+ }
+ else if (IS_BACKREF(lit))
+ {
+ fprintf(stream, "backref %d, pos %d\n", code_max, pos);
+ }
+ else if (IS_PARAMETER(lit))
+ {
+ tre_print_params(lit->u.params);
+ fprintf(stream, "\n");
+ }
+ else
+ {
+ fprintf(stream, "literal (%c, %c) (%d, %d), pos %d, sub %d, "
+ "%d tags\n", code_min, code_max, code_min, code_max, pos,
+ ast->submatch_id, num_tags);
+ }
+ break;
+ case ITERATION:
+ iter = ast->obj;
+ fprintf(stream, "iteration {%d, %d}, sub %d, %d tags, %s\n",
+ iter->min, iter->max, ast->submatch_id, num_tags,
+ iter->minimal ? "minimal" : "greedy");
+ tre_do_print(stream, iter->arg, indent + 2);
+ break;
+ case UNION:
+ fprintf(stream, "union, sub %d, %d tags\n", ast->submatch_id, num_tags);
+ tre_do_print(stream, ((tre_union_t *)ast->obj)->left, indent + 2);
+ tre_do_print(stream, ((tre_union_t *)ast->obj)->right, indent + 2);
+ break;
+ case CATENATION:
+ fprintf(stream, "catenation, sub %d, %d tags\n", ast->submatch_id,
+ num_tags);
+ tre_do_print(stream, ((tre_catenation_t *)ast->obj)->left, indent + 2);
+ tre_do_print(stream, ((tre_catenation_t *)ast->obj)->right, indent + 2);
+ break;
+ default:
+ assert(0);
+ break;
+ }
+}
+
+static void
+tre_ast_fprint(FILE *stream, tre_ast_node_t *ast)
+{
+ tre_do_print(stream, ast, 0);
+}
+
+void
+tre_ast_print(tre_ast_node_t *tree)
+{
+ printf("AST:\n");
+ tre_ast_fprint(stdout, tree);
+}
+
+#endif /* TRE_DEBUG */
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-ast.h b/qse/lib/cmn/tre-ast.h
new file mode 100644
index 00000000..b848c113
--- /dev/null
+++ b/qse/lib/cmn/tre-ast.h
@@ -0,0 +1,170 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-ast.h - Abstract syntax tree (AST) definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+
+#ifndef _QSE_LIB_CMN_TRE_AST_H_
+#define _QSE_LIB_CMN_TRE_AST_H_
+
+#include "tre.h"
+#include "tre-compile.h"
+
+/* The different AST node types. */
+typedef enum
+{
+ LITERAL,
+ CATENATION,
+ ITERATION,
+ UNION
+} tre_ast_type_t;
+
+/* Special subtypes of TRE_LITERAL. */
+#define EMPTY -1 /* Empty leaf (denotes empty string). */
+#define ASSERTION -2 /* Assertion leaf. */
+#define TAG -3 /* Tag leaf. */
+#define BACKREF -4 /* Back reference leaf. */
+#define PARAMETER -5 /* Parameter. */
+
+#define IS_SPECIAL(x) ((x)->code_min < 0)
+#define IS_EMPTY(x) ((x)->code_min == EMPTY)
+#define IS_ASSERTION(x) ((x)->code_min == ASSERTION)
+#define IS_TAG(x) ((x)->code_min == TAG)
+#define IS_BACKREF(x) ((x)->code_min == BACKREF)
+#define IS_PARAMETER(x) ((x)->code_min == PARAMETER)
+
+
+/* A generic AST node. All AST nodes consist of this node on the top
+ level with `obj' pointing to the actual content. */
+typedef struct
+{
+ tre_ast_type_t type; /* Type of the node. */
+ void *obj; /* Pointer to actual node. */
+ int nullable;
+ int submatch_id;
+ int num_submatches;
+ int num_tags;
+ tre_pos_and_tags_t *firstpos;
+ tre_pos_and_tags_t *lastpos;
+} tre_ast_node_t;
+
+
+/* A "literal" node. These are created for assertions, back references,
+ tags, matching parameter settings, and all expressions that match one
+ character. */
+typedef struct
+{
+ long code_min;
+ long code_max;
+ int position;
+ union
+ {
+ tre_ctype_t class;
+ int *params;
+ } u;
+ tre_ctype_t *neg_classes;
+} tre_literal_t;
+
+/* A "catenation" node. These are created when two regexps are concatenated.
+ If there are more than one subexpressions in sequence, the `left' part
+ holds all but the last, and `right' part holds the last subexpression
+ (catenation is left associative). */
+typedef struct
+{
+ tre_ast_node_t *left;
+ tre_ast_node_t *right;
+} tre_catenation_t;
+
+/* An "iteration" node. These are created for the "*", "+", "?", and "{m,n}"
+ operators. */
+typedef struct
+{
+ /* Subexpression to match. */
+ tre_ast_node_t *arg;
+ /* Minimum number of consecutive matches. */
+ int min;
+ /* Maximum number of consecutive matches. */
+ int max;
+ /* If 0, match as many characters as possible, if 1 match as few as
+ possible. Note that this does not always mean the same thing as
+ matching as many/few repetitions as possible. */
+ unsigned int minimal:1;
+ /* Approximate matching parameters (or NULL). */
+ int *params;
+} tre_iteration_t;
+
+/* An "union" node. These are created for the "|" operator. */
+typedef struct
+{
+ tre_ast_node_t *left;
+ tre_ast_node_t *right;
+} tre_union_t;
+
+tre_ast_node_t* tre_ast_new_node(tre_mem_t mem, tre_ast_type_t type, size_t size);
+
+tre_ast_node_t* tre_ast_new_literal(tre_mem_t mem, int code_min, int code_max, int position);
+
+tre_ast_node_t* tre_ast_new_iter(tre_mem_t mem, tre_ast_node_t *arg, int min, int max, int minimal);
+
+tre_ast_node_t* tre_ast_new_union(tre_mem_t mem, tre_ast_node_t *left, tre_ast_node_t *right);
+
+tre_ast_node_t* tre_ast_new_catenation(tre_mem_t mem, tre_ast_node_t *left, tre_ast_node_t *right);
+
+#ifdef TRE_DEBUG
+void tre_ast_print(tre_ast_node_t *tree);
+/* XXX - rethink AST printing API */
+void tre_print_params(int *params);
+#endif /* TRE_DEBUG */
+
+#endif /* TRE_AST_H */
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-compile.c b/qse/lib/cmn/tre-compile.c
new file mode 100644
index 00000000..f287c6d0
--- /dev/null
+++ b/qse/lib/cmn/tre-compile.c
@@ -0,0 +1,2276 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-compile.c - TRE regex compiler
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+ TODO:
+ - Fix tre_ast_to_tnfa() to recurse using a stack instead of recursive
+ function calls.
+*/
+
+
+#include "tre.h"
+#include "tre-stack.h"
+#include "tre-ast.h"
+#include "tre-parse.h"
+#include "tre-compile.h"
+
+/*
+ Algorithms to setup tags so that submatch addressing can be done.
+*/
+
+static QSE_INLINE void* xcalloc (
+ qse_mmgr_t* mmgr, qse_size_t nmemb, qse_size_t size)
+{
+ void* ptr = QSE_MMGR_ALLOC (mmgr, nmemb * size);
+ if (ptr) QSE_MEMSET (ptr, 0, nmemb * size);
+ return ptr;
+}
+
+/* Inserts a catenation node to the root of the tree given in `node'.
+ As the left child a new tag with number `tag_id' to `node' is added,
+ and the right child is the old root. */
+static reg_errcode_t
+tre_add_tag_left(tre_mem_t mem, tre_ast_node_t *node, int tag_id)
+{
+ tre_catenation_t *c;
+
+ DPRINT(("add_tag_left: tag %d\n", tag_id));
+
+ c = tre_mem_alloc(mem, sizeof(*c));
+ if (c == NULL)
+ return REG_ESPACE;
+ c->left = tre_ast_new_literal(mem, TAG, tag_id, -1);
+ if (c->left == NULL)
+ return REG_ESPACE;
+ c->right = tre_mem_alloc(mem, sizeof(tre_ast_node_t));
+ if (c->right == NULL)
+ return REG_ESPACE;
+
+ c->right->obj = node->obj;
+ c->right->type = node->type;
+ c->right->nullable = -1;
+ c->right->submatch_id = -1;
+ c->right->firstpos = NULL;
+ c->right->lastpos = NULL;
+ c->right->num_tags = 0;
+ node->obj = c;
+ node->type = CATENATION;
+ return REG_OK;
+}
+
+/* Inserts a catenation node to the root of the tree given in `node'.
+ As the right child a new tag with number `tag_id' to `node' is added,
+ and the left child is the old root. */
+static reg_errcode_t
+tre_add_tag_right(tre_mem_t mem, tre_ast_node_t *node, int tag_id)
+{
+ tre_catenation_t *c;
+
+ DPRINT(("tre_add_tag_right: tag %d\n", tag_id));
+
+ c = tre_mem_alloc(mem, sizeof(*c));
+ if (c == NULL)
+ return REG_ESPACE;
+ c->right = tre_ast_new_literal(mem, TAG, tag_id, -1);
+ if (c->right == NULL)
+ return REG_ESPACE;
+ c->left = tre_mem_alloc(mem, sizeof(tre_ast_node_t));
+ if (c->left == NULL)
+ return REG_ESPACE;
+
+ c->left->obj = node->obj;
+ c->left->type = node->type;
+ c->left->nullable = -1;
+ c->left->submatch_id = -1;
+ c->left->firstpos = NULL;
+ c->left->lastpos = NULL;
+ c->left->num_tags = 0;
+ node->obj = c;
+ node->type = CATENATION;
+ return REG_OK;
+}
+
+typedef enum
+{
+ ADDTAGS_RECURSE,
+ ADDTAGS_AFTER_ITERATION,
+ ADDTAGS_AFTER_UNION_LEFT,
+ ADDTAGS_AFTER_UNION_RIGHT,
+ ADDTAGS_AFTER_CAT_LEFT,
+ ADDTAGS_AFTER_CAT_RIGHT,
+ ADDTAGS_SET_SUBMATCH_END
+} tre_addtags_symbol_t;
+
+
+typedef struct
+{
+ int tag;
+ int next_tag;
+} tre_tag_states_t;
+
+
+/* Go through `regset' and set submatch data for submatches that are
+ using this tag. */
+static void
+tre_purge_regset(int *regset, tre_tnfa_t *tnfa, int tag)
+{
+ int i;
+
+ for (i = 0; regset[i] >= 0; i++)
+ {
+ int id = regset[i] / 2;
+ int start = !(regset[i] % 2);
+ DPRINT((" Using tag %d for %s offset of "
+ "submatch %d\n", tag,
+ start ? "start" : "end", id));
+ if (start)
+ tnfa->submatch_data[id].so_tag = tag;
+ else
+ tnfa->submatch_data[id].eo_tag = tag;
+ }
+ regset[0] = -1;
+}
+
+
+/* Adds tags to appropriate locations in the parse tree in `tree', so that
+ subexpressions marked for submatch addressing can be traced. */
+static reg_errcode_t
+tre_add_tags(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree,
+ tre_tnfa_t *tnfa)
+{
+ reg_errcode_t status = REG_OK;
+ tre_addtags_symbol_t symbol;
+ tre_ast_node_t *node = tree; /* Tree node we are currently looking at. */
+ int bottom = tre_stack_num_objects(stack);
+ /* True for first pass (counting number of needed tags) */
+ int first_pass = (mem == NULL || tnfa == NULL);
+ int *regset, *orig_regset;
+ int num_tags = 0; /* Total number of tags. */
+ int num_minimals = 0; /* Number of special minimal tags. */
+ int tag = 0; /* The tag that is to be added next. */
+ int next_tag = 1; /* Next tag to use after this one. */
+ int *parents; /* Stack of submatches the current submatch is
+ contained in. */
+ int minimal_tag = -1; /* Tag that marks the beginning of a minimal match. */
+ tre_tag_states_t *saved_states;
+
+ tre_tag_direction_t direction = TRE_TAG_MINIMIZE;
+ if (!first_pass)
+ {
+ tnfa->end_tag = 0;
+ tnfa->minimal_tags[0] = -1;
+ }
+
+ regset = xmalloc(mem->mmgr, sizeof(*regset) * ((tnfa->num_submatches + 1) * 2));
+ if (regset == NULL)
+ return REG_ESPACE;
+ regset[0] = -1;
+ orig_regset = regset;
+
+ parents = xmalloc(mem->mmgr, sizeof(*parents) * (tnfa->num_submatches + 1));
+ if (parents == NULL)
+ {
+ xfree(mem->mmgr, regset);
+ return REG_ESPACE;
+ }
+ parents[0] = -1;
+
+ saved_states = xmalloc(mem->mmgr, sizeof(*saved_states) * (tnfa->num_submatches + 1));
+ if (saved_states == NULL)
+ {
+ xfree(mem->mmgr,regset);
+ xfree(mem->mmgr,parents);
+ return REG_ESPACE;
+ }
+ else
+ {
+ unsigned int i;
+ for (i = 0; i <= tnfa->num_submatches; i++)
+ saved_states[i].tag = -1;
+ }
+
+ STACK_PUSH(stack, voidptr, node);
+ STACK_PUSH(stack, int, ADDTAGS_RECURSE);
+
+ while (tre_stack_num_objects(stack) > bottom)
+ {
+ if (status != REG_OK)
+ break;
+
+ symbol = (tre_addtags_symbol_t)tre_stack_pop_int(stack);
+ switch (symbol)
+ {
+
+ case ADDTAGS_SET_SUBMATCH_END:
+ {
+ int id = tre_stack_pop_int(stack);
+ int i;
+
+ /* Add end of this submatch to regset. */
+ for (i = 0; regset[i] >= 0; i++);
+ regset[i] = id * 2 + 1;
+ regset[i + 1] = -1;
+
+ /* Pop this submatch from the parents stack. */
+ for (i = 0; parents[i] >= 0; i++);
+ parents[i - 1] = -1;
+ break;
+ }
+
+ case ADDTAGS_RECURSE:
+ node = tre_stack_pop_voidptr(stack);
+
+ if (node->submatch_id >= 0)
+ {
+ int id = node->submatch_id;
+ int i;
+
+
+ /* Add start of this submatch to regset. */
+ for (i = 0; regset[i] >= 0; i++);
+ regset[i] = id * 2;
+ regset[i + 1] = -1;
+
+ if (!first_pass)
+ {
+ for (i = 0; parents[i] >= 0; i++);
+ tnfa->submatch_data[id].parents = NULL;
+ if (i > 0)
+ {
+ int *p = xmalloc(mem->mmgr, sizeof(*p) * (i + 1));
+ if (p == NULL)
+ {
+ status = REG_ESPACE;
+ break;
+ }
+ assert(tnfa->submatch_data[id].parents == NULL);
+ tnfa->submatch_data[id].parents = p;
+ for (i = 0; parents[i] >= 0; i++)
+ p[i] = parents[i];
+ p[i] = -1;
+ }
+ }
+
+ /* Add end of this submatch to regset after processing this
+ node. */
+ STACK_PUSHX(stack, int, node->submatch_id);
+ STACK_PUSHX(stack, int, ADDTAGS_SET_SUBMATCH_END);
+ }
+
+ switch (node->type)
+ {
+ case LITERAL:
+ {
+ tre_literal_t *lit = node->obj;
+
+ if (!IS_SPECIAL(lit) || IS_BACKREF(lit))
+ {
+ int i;
+ DPRINT(("Literal %d-%d\n",
+ (int)lit->code_min, (int)lit->code_max));
+ if (regset[0] >= 0)
+ {
+ /* Regset is not empty, so add a tag before the
+ literal or backref. */
+ if (!first_pass)
+ {
+ status = tre_add_tag_left(mem, node, tag);
+ tnfa->tag_directions[tag] = direction;
+ if (minimal_tag >= 0)
+ {
+ DPRINT(("Minimal %d, %d\n", minimal_tag, tag));
+ for (i = 0; tnfa->minimal_tags[i] >= 0; i++);
+ tnfa->minimal_tags[i] = tag;
+ tnfa->minimal_tags[i + 1] = minimal_tag;
+ tnfa->minimal_tags[i + 2] = -1;
+ minimal_tag = -1;
+ num_minimals++;
+ }
+ tre_purge_regset(regset, tnfa, tag);
+ }
+ else
+ {
+ DPRINT((" num_tags = 1\n"));
+ node->num_tags = 1;
+ }
+
+ DPRINT((" num_tags++\n"));
+ regset[0] = -1;
+ tag = next_tag;
+ num_tags++;
+ next_tag++;
+ }
+ }
+ else
+ {
+ assert(!IS_TAG(lit));
+ }
+ break;
+ }
+ case CATENATION:
+ {
+ tre_catenation_t *cat = node->obj;
+ tre_ast_node_t *left = cat->left;
+ tre_ast_node_t *right = cat->right;
+ int reserved_tag = -1;
+ DPRINT(("Catenation, next_tag = %d\n", next_tag));
+
+
+ /* After processing right child. */
+ STACK_PUSHX(stack, voidptr, node);
+ STACK_PUSHX(stack, int, ADDTAGS_AFTER_CAT_RIGHT);
+
+ /* Process right child. */
+ STACK_PUSHX(stack, voidptr, right);
+ STACK_PUSHX(stack, int, ADDTAGS_RECURSE);
+
+ /* After processing left child. */
+ STACK_PUSHX(stack, int, next_tag + left->num_tags);
+ DPRINT((" Pushing %d for after left\n",
+ next_tag + left->num_tags));
+ if (left->num_tags > 0 && right->num_tags > 0)
+ {
+ /* Reserve the next tag to the right child. */
+ DPRINT((" Reserving next_tag %d to right child\n",
+ next_tag));
+ reserved_tag = next_tag;
+ next_tag++;
+ }
+ STACK_PUSHX(stack, int, reserved_tag);
+ STACK_PUSHX(stack, int, ADDTAGS_AFTER_CAT_LEFT);
+
+ /* Process left child. */
+ STACK_PUSHX(stack, voidptr, left);
+ STACK_PUSHX(stack, int, ADDTAGS_RECURSE);
+
+ }
+ break;
+ case ITERATION:
+ {
+ tre_iteration_t *iter = node->obj;
+ DPRINT(("Iteration\n"));
+
+ if (first_pass)
+ {
+ STACK_PUSHX(stack, int, regset[0] >= 0 || iter->minimal);
+ }
+ else
+ {
+ STACK_PUSHX(stack, int, tag);
+ STACK_PUSHX(stack, int, iter->minimal);
+ }
+ STACK_PUSHX(stack, voidptr, node);
+ STACK_PUSHX(stack, int, ADDTAGS_AFTER_ITERATION);
+
+ STACK_PUSHX(stack, voidptr, iter->arg);
+ STACK_PUSHX(stack, int, ADDTAGS_RECURSE);
+
+ /* Regset is not empty, so add a tag here. */
+ if (regset[0] >= 0 || iter->minimal)
+ {
+ if (!first_pass)
+ {
+ int i;
+ status = tre_add_tag_left(mem, node, tag);
+ if (iter->minimal)
+ tnfa->tag_directions[tag] = TRE_TAG_MAXIMIZE;
+ else
+ tnfa->tag_directions[tag] = direction;
+ if (minimal_tag >= 0)
+ {
+ DPRINT(("Minimal %d, %d\n", minimal_tag, tag));
+ for (i = 0; tnfa->minimal_tags[i] >= 0; i++);
+ tnfa->minimal_tags[i] = tag;
+ tnfa->minimal_tags[i + 1] = minimal_tag;
+ tnfa->minimal_tags[i + 2] = -1;
+ minimal_tag = -1;
+ num_minimals++;
+ }
+ tre_purge_regset(regset, tnfa, tag);
+ }
+
+ DPRINT((" num_tags++\n"));
+ regset[0] = -1;
+ tag = next_tag;
+ num_tags++;
+ next_tag++;
+ }
+ direction = TRE_TAG_MINIMIZE;
+ }
+ break;
+ case UNION:
+ {
+ tre_union_t *uni = node->obj;
+ tre_ast_node_t *left = uni->left;
+ tre_ast_node_t *right = uni->right;
+ int left_tag;
+ int right_tag;
+
+ if (regset[0] >= 0)
+ {
+ left_tag = next_tag;
+ right_tag = next_tag + 1;
+ }
+ else
+ {
+ left_tag = tag;
+ right_tag = next_tag;
+ }
+
+ DPRINT(("Union\n"));
+
+ /* After processing right child. */
+ STACK_PUSHX(stack, int, right_tag);
+ STACK_PUSHX(stack, int, left_tag);
+ STACK_PUSHX(stack, voidptr, regset);
+ STACK_PUSHX(stack, int, regset[0] >= 0);
+ STACK_PUSHX(stack, voidptr, node);
+ STACK_PUSHX(stack, voidptr, right);
+ STACK_PUSHX(stack, voidptr, left);
+ STACK_PUSHX(stack, int, ADDTAGS_AFTER_UNION_RIGHT);
+
+ /* Process right child. */
+ STACK_PUSHX(stack, voidptr, right);
+ STACK_PUSHX(stack, int, ADDTAGS_RECURSE);
+
+ /* After processing left child. */
+ STACK_PUSHX(stack, int, ADDTAGS_AFTER_UNION_LEFT);
+
+ /* Process left child. */
+ STACK_PUSHX(stack, voidptr, left);
+ STACK_PUSHX(stack, int, ADDTAGS_RECURSE);
+
+ /* Regset is not empty, so add a tag here. */
+ if (regset[0] >= 0)
+ {
+ if (!first_pass)
+ {
+ int i;
+ status = tre_add_tag_left(mem, node, tag);
+ tnfa->tag_directions[tag] = direction;
+ if (minimal_tag >= 0)
+ {
+ DPRINT(("Minimal %d, %d\n", minimal_tag, tag));
+ for (i = 0; tnfa->minimal_tags[i] >= 0; i++);
+ tnfa->minimal_tags[i] = tag;
+ tnfa->minimal_tags[i + 1] = minimal_tag;
+ tnfa->minimal_tags[i + 2] = -1;
+ minimal_tag = -1;
+ num_minimals++;
+ }
+ tre_purge_regset(regset, tnfa, tag);
+ }
+
+ DPRINT((" num_tags++\n"));
+ regset[0] = -1;
+ tag = next_tag;
+ num_tags++;
+ next_tag++;
+ }
+
+ if (node->num_submatches > 0)
+ {
+ /* The next two tags are reserved for markers. */
+ next_tag++;
+ tag = next_tag;
+ next_tag++;
+ }
+
+ break;
+ }
+ }
+
+ if (node->submatch_id >= 0)
+ {
+ int i;
+ /* Push this submatch on the parents stack. */
+ for (i = 0; parents[i] >= 0; i++);
+ parents[i] = node->submatch_id;
+ parents[i + 1] = -1;
+ }
+
+ break; /* end case: ADDTAGS_RECURSE */
+
+ case ADDTAGS_AFTER_ITERATION:
+ {
+ int minimal = 0;
+ int enter_tag;
+ node = tre_stack_pop_voidptr(stack);
+ if (first_pass)
+ {
+ node->num_tags = ((tre_iteration_t *)node->obj)->arg->num_tags
+ + tre_stack_pop_int(stack);
+ minimal_tag = -1;
+ }
+ else
+ {
+ minimal = tre_stack_pop_int(stack);
+ enter_tag = tre_stack_pop_int(stack);
+ if (minimal)
+ minimal_tag = enter_tag;
+ }
+
+ DPRINT(("After iteration\n"));
+ if (!first_pass)
+ {
+ DPRINT((" Setting direction to %s\n",
+ minimal ? "minimize" : "maximize"));
+ if (minimal)
+ direction = TRE_TAG_MINIMIZE;
+ else
+ direction = TRE_TAG_MAXIMIZE;
+ }
+ break;
+ }
+
+ case ADDTAGS_AFTER_CAT_LEFT:
+ {
+ int new_tag = tre_stack_pop_int(stack);
+ next_tag = tre_stack_pop_int(stack);
+ DPRINT(("After cat left, tag = %d, next_tag = %d\n",
+ tag, next_tag));
+ if (new_tag >= 0)
+ {
+ DPRINT((" Setting tag to %d\n", new_tag));
+ tag = new_tag;
+ }
+ break;
+ }
+
+ case ADDTAGS_AFTER_CAT_RIGHT:
+ DPRINT(("After cat right\n"));
+ node = tre_stack_pop_voidptr(stack);
+ if (first_pass)
+ node->num_tags = ((tre_catenation_t *)node->obj)->left->num_tags
+ + ((tre_catenation_t *)node->obj)->right->num_tags;
+ break;
+
+ case ADDTAGS_AFTER_UNION_LEFT:
+ DPRINT(("After union left\n"));
+ /* Lift the bottom of the `regset' array so that when processing
+ the right operand the items currently in the array are
+ invisible. The original bottom was saved at ADDTAGS_UNION and
+ will be restored at ADDTAGS_AFTER_UNION_RIGHT below. */
+ while (*regset >= 0)
+ regset++;
+ break;
+
+ case ADDTAGS_AFTER_UNION_RIGHT:
+ {
+ int added_tags, tag_left, tag_right;
+ tre_ast_node_t *left = tre_stack_pop_voidptr(stack);
+ tre_ast_node_t *right = tre_stack_pop_voidptr(stack);
+ DPRINT(("After union right\n"));
+ node = tre_stack_pop_voidptr(stack);
+ added_tags = tre_stack_pop_int(stack);
+ if (first_pass)
+ {
+ node->num_tags = ((tre_union_t *)node->obj)->left->num_tags
+ + ((tre_union_t *)node->obj)->right->num_tags + added_tags
+ + ((node->num_submatches > 0) ? 2 : 0);
+ }
+ regset = tre_stack_pop_voidptr(stack);
+ tag_left = tre_stack_pop_int(stack);
+ tag_right = tre_stack_pop_int(stack);
+
+ /* Add tags after both children, the left child gets a smaller
+ tag than the right child. This guarantees that we prefer
+ the left child over the right child. */
+ /* XXX - This is not always necessary (if the children have
+ tags which must be seen for every match of that child). */
+ /* XXX - Check if this is the only place where tre_add_tag_right
+ is used. If so, use tre_add_tag_left (putting the tag before
+ the child as opposed after the child) and throw away
+ tre_add_tag_right. */
+ if (node->num_submatches > 0)
+ {
+ if (!first_pass)
+ {
+ status = tre_add_tag_right(mem, left, tag_left);
+ tnfa->tag_directions[tag_left] = TRE_TAG_MAXIMIZE;
+ status = tre_add_tag_right(mem, right, tag_right);
+ tnfa->tag_directions[tag_right] = TRE_TAG_MAXIMIZE;
+ }
+ DPRINT((" num_tags += 2\n"));
+ num_tags += 2;
+ }
+ direction = TRE_TAG_MAXIMIZE;
+ break;
+ }
+
+ default:
+ assert(0);
+ break;
+
+ } /* end switch(symbol) */
+ } /* end while(tre_stack_num_objects(stack) > bottom) */
+
+ if (!first_pass)
+ tre_purge_regset(regset, tnfa, tag);
+
+ if (!first_pass && minimal_tag >= 0)
+ {
+ int i;
+ DPRINT(("Minimal %d, %d\n", minimal_tag, tag));
+ for (i = 0; tnfa->minimal_tags[i] >= 0; i++);
+ tnfa->minimal_tags[i] = tag;
+ tnfa->minimal_tags[i + 1] = minimal_tag;
+ tnfa->minimal_tags[i + 2] = -1;
+ minimal_tag = -1;
+ num_minimals++;
+ }
+
+ DPRINT(("tre_add_tags: %s complete. Number of tags %d.\n",
+ first_pass? "First pass" : "Second pass", num_tags));
+
+ assert(tree->num_tags == num_tags);
+ tnfa->end_tag = num_tags;
+ tnfa->num_tags = num_tags;
+ tnfa->num_minimals = num_minimals;
+ xfree(mem->mmgr,orig_regset);
+ xfree(mem->mmgr,parents);
+ xfree(mem->mmgr,saved_states);
+ return status;
+}
+
+
+
+/*
+ AST to TNFA compilation routines.
+*/
+
+typedef enum
+{
+ COPY_RECURSE,
+ COPY_SET_RESULT_PTR
+} tre_copyast_symbol_t;
+
+/* Flags for tre_copy_ast(). */
+#define COPY_REMOVE_TAGS 1
+#define COPY_MAXIMIZE_FIRST_TAG 2
+
+static reg_errcode_t
+tre_copy_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
+ int flags, int *pos_add, tre_tag_direction_t *tag_directions,
+ tre_ast_node_t **copy, int *max_pos)
+{
+ reg_errcode_t status = REG_OK;
+ int bottom = tre_stack_num_objects(stack);
+ int num_copied = 0;
+ int first_tag = 1;
+ tre_ast_node_t **result = copy;
+ tre_copyast_symbol_t symbol;
+
+ STACK_PUSH(stack, voidptr, ast);
+ STACK_PUSH(stack, int, COPY_RECURSE);
+
+ while (status == REG_OK && tre_stack_num_objects(stack) > bottom)
+ {
+ tre_ast_node_t *node;
+ if (status != REG_OK)
+ break;
+
+ symbol = (tre_copyast_symbol_t)tre_stack_pop_int(stack);
+ switch (symbol)
+ {
+ case COPY_SET_RESULT_PTR:
+ result = tre_stack_pop_voidptr(stack);
+ break;
+ case COPY_RECURSE:
+ node = tre_stack_pop_voidptr(stack);
+ switch (node->type)
+ {
+ case LITERAL:
+ {
+ tre_literal_t *lit = node->obj;
+ int pos = lit->position;
+ int min = lit->code_min;
+ int max = lit->code_max;
+ if (!IS_SPECIAL(lit) || IS_BACKREF(lit))
+ {
+ /* XXX - e.g. [ab] has only one position but two
+ nodes, so we are creating holes in the state space
+ here. Not fatal, just wastes memory. */
+ pos += *pos_add;
+ num_copied++;
+ }
+ else if (IS_TAG(lit) && (flags & COPY_REMOVE_TAGS))
+ {
+ /* Change this tag to empty. */
+ min = EMPTY;
+ max = pos = -1;
+ }
+ else if (IS_TAG(lit) && (flags & COPY_MAXIMIZE_FIRST_TAG)
+ && first_tag)
+ {
+ /* Maximize the first tag. */
+ tag_directions[max] = TRE_TAG_MAXIMIZE;
+ first_tag = 0;
+ }
+ *result = tre_ast_new_literal(mem, min, max, pos);
+ if (*result == NULL)
+ status = REG_ESPACE;
+
+ if (pos > *max_pos)
+ *max_pos = pos;
+ break;
+ }
+ case UNION:
+ {
+ tre_union_t *uni = node->obj;
+ tre_union_t *tmp;
+ *result = tre_ast_new_union(mem, uni->left, uni->right);
+ if (*result == NULL)
+ {
+ status = REG_ESPACE;
+ break;
+ }
+ tmp = (*result)->obj;
+ result = &tmp->left;
+ STACK_PUSHX(stack, voidptr, uni->right);
+ STACK_PUSHX(stack, int, COPY_RECURSE);
+ STACK_PUSHX(stack, voidptr, &tmp->right);
+ STACK_PUSHX(stack, int, COPY_SET_RESULT_PTR);
+ STACK_PUSHX(stack, voidptr, uni->left);
+ STACK_PUSHX(stack, int, COPY_RECURSE);
+ break;
+ }
+ case CATENATION:
+ {
+ tre_catenation_t *cat = node->obj;
+ tre_catenation_t *tmp;
+ *result = tre_ast_new_catenation(mem, cat->left, cat->right);
+ if (*result == NULL)
+ {
+ status = REG_ESPACE;
+ break;
+ }
+ tmp = (*result)->obj;
+ tmp->left = NULL;
+ tmp->right = NULL;
+ result = &tmp->left;
+
+ STACK_PUSHX(stack, voidptr, cat->right);
+ STACK_PUSHX(stack, int, COPY_RECURSE);
+ STACK_PUSHX(stack, voidptr, &tmp->right);
+ STACK_PUSHX(stack, int, COPY_SET_RESULT_PTR);
+ STACK_PUSHX(stack, voidptr, cat->left);
+ STACK_PUSHX(stack, int, COPY_RECURSE);
+ break;
+ }
+ case ITERATION:
+ {
+ tre_iteration_t *iter = node->obj;
+ STACK_PUSHX(stack, voidptr, iter->arg);
+ STACK_PUSHX(stack, int, COPY_RECURSE);
+ *result = tre_ast_new_iter(mem, iter->arg, iter->min,
+ iter->max, iter->minimal);
+ if (*result == NULL)
+ {
+ status = REG_ESPACE;
+ break;
+ }
+ iter = (*result)->obj;
+ result = &iter->arg;
+ break;
+ }
+ default:
+ assert(0);
+ break;
+ }
+ break;
+ }
+ }
+ *pos_add += num_copied;
+ return status;
+}
+
+typedef enum
+{
+ EXPAND_RECURSE,
+ EXPAND_AFTER_ITER
+} tre_expand_ast_symbol_t;
+
+/* Expands each iteration node that has a finite nonzero minimum or maximum
+ iteration count to a catenated sequence of copies of the node. */
+static reg_errcode_t
+tre_expand_ast(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *ast,
+ int *position, tre_tag_direction_t *tag_directions,
+ int *max_depth)
+{
+ reg_errcode_t status = REG_OK;
+ int bottom = tre_stack_num_objects(stack);
+ int pos_add = 0;
+ int pos_add_total = 0;
+ int max_pos = 0;
+ /* Current approximate matching parameters. */
+ int params[TRE_PARAM_LAST];
+ /* Approximate parameter nesting level. */
+ int params_depth = 0;
+ int iter_depth = 0;
+ int i;
+
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ params[i] = TRE_PARAM_DEFAULT;
+
+ STACK_PUSHR(stack, voidptr, ast);
+ STACK_PUSHR(stack, int, EXPAND_RECURSE);
+ while (status == REG_OK && tre_stack_num_objects(stack) > bottom)
+ {
+ tre_ast_node_t *node;
+ tre_expand_ast_symbol_t symbol;
+
+ if (status != REG_OK)
+ break;
+
+ DPRINT(("pos_add %d\n", pos_add));
+
+ symbol = (tre_expand_ast_symbol_t)tre_stack_pop_int(stack);
+ node = tre_stack_pop_voidptr(stack);
+ switch (symbol)
+ {
+ case EXPAND_RECURSE:
+ switch (node->type)
+ {
+ case LITERAL:
+ {
+ tre_literal_t *lit= node->obj;
+ if (!IS_SPECIAL(lit) || IS_BACKREF(lit))
+ {
+ lit->position += pos_add;
+ if (lit->position > max_pos)
+ max_pos = lit->position;
+ }
+ break;
+ }
+ case UNION:
+ {
+ tre_union_t *uni = node->obj;
+ STACK_PUSHX(stack, voidptr, uni->right);
+ STACK_PUSHX(stack, int, EXPAND_RECURSE);
+ STACK_PUSHX(stack, voidptr, uni->left);
+ STACK_PUSHX(stack, int, EXPAND_RECURSE);
+ break;
+ }
+ case CATENATION:
+ {
+ tre_catenation_t *cat = node->obj;
+ STACK_PUSHX(stack, voidptr, cat->right);
+ STACK_PUSHX(stack, int, EXPAND_RECURSE);
+ STACK_PUSHX(stack, voidptr, cat->left);
+ STACK_PUSHX(stack, int, EXPAND_RECURSE);
+ break;
+ }
+ case ITERATION:
+ {
+ tre_iteration_t *iter = node->obj;
+ STACK_PUSHX(stack, int, pos_add);
+ STACK_PUSHX(stack, voidptr, node);
+ STACK_PUSHX(stack, int, EXPAND_AFTER_ITER);
+ STACK_PUSHX(stack, voidptr, iter->arg);
+ STACK_PUSHX(stack, int, EXPAND_RECURSE);
+ /* If we are going to expand this node at EXPAND_AFTER_ITER
+ then don't increase the `pos' fields of the nodes now, it
+ will get done when expanding. */
+ if (iter->min > 1 || iter->max > 1)
+ pos_add = 0;
+ iter_depth++;
+ DPRINT(("iter\n"));
+ break;
+ }
+ default:
+ assert(0);
+ break;
+ }
+ break;
+ case EXPAND_AFTER_ITER:
+ {
+ tre_iteration_t *iter = node->obj;
+ int pos_add_last;
+ pos_add = tre_stack_pop_int(stack);
+ pos_add_last = pos_add;
+ if (iter->min > 1 || iter->max > 1)
+ {
+ tre_ast_node_t *seq1 = NULL, *seq2 = NULL;
+ int j;
+ int pos_add_save = pos_add;
+
+ /* Create a catenated sequence of copies of the node. */
+ for (j = 0; j < iter->min; j++)
+ {
+ tre_ast_node_t *copy;
+ /* Remove tags from all but the last copy. */
+ int flags = ((j + 1 < iter->min)
+ ? COPY_REMOVE_TAGS
+ : COPY_MAXIMIZE_FIRST_TAG);
+ DPRINT((" pos_add %d\n", pos_add));
+ pos_add_save = pos_add;
+ status = tre_copy_ast(mem, stack, iter->arg, flags,
+ &pos_add, tag_directions, ©,
+ &max_pos);
+ if (status != REG_OK)
+ return status;
+ if (seq1 != NULL)
+ seq1 = tre_ast_new_catenation(mem, seq1, copy);
+ else
+ seq1 = copy;
+ if (seq1 == NULL)
+ return REG_ESPACE;
+ }
+
+ if (iter->max == -1)
+ {
+ /* No upper limit. */
+ pos_add_save = pos_add;
+ status = tre_copy_ast(mem, stack, iter->arg, 0,
+ &pos_add, NULL, &seq2, &max_pos);
+ if (status != REG_OK)
+ return status;
+ seq2 = tre_ast_new_iter(mem, seq2, 0, -1, 0);
+ if (seq2 == NULL)
+ return REG_ESPACE;
+ }
+ else
+ {
+ for (j = iter->min; j < iter->max; j++)
+ {
+ tre_ast_node_t *tmp, *copy;
+ pos_add_save = pos_add;
+ status = tre_copy_ast(mem, stack, iter->arg, 0,
+ &pos_add, NULL, ©, &max_pos);
+ if (status != REG_OK)
+ return status;
+ if (seq2 != NULL)
+ seq2 = tre_ast_new_catenation(mem, copy, seq2);
+ else
+ seq2 = copy;
+ if (seq2 == NULL)
+ return REG_ESPACE;
+ tmp = tre_ast_new_literal(mem, EMPTY, -1, -1);
+ if (tmp == NULL)
+ return REG_ESPACE;
+ seq2 = tre_ast_new_union(mem, tmp, seq2);
+ if (seq2 == NULL)
+ return REG_ESPACE;
+ }
+ }
+
+ pos_add = pos_add_save;
+ if (seq1 == NULL)
+ seq1 = seq2;
+ else if (seq2 != NULL)
+ seq1 = tre_ast_new_catenation(mem, seq1, seq2);
+ if (seq1 == NULL)
+ return REG_ESPACE;
+ node->obj = seq1->obj;
+ node->type = seq1->type;
+ }
+
+ iter_depth--;
+ pos_add_total += pos_add - pos_add_last;
+ if (iter_depth == 0)
+ pos_add = pos_add_total;
+
+ /* If approximate parameters are specified, surround the result
+ with two parameter setting nodes. The one on the left sets
+ the specified parameters, and the one on the right restores
+ the old parameters. */
+ if (iter->params)
+ {
+ tre_ast_node_t *tmp_l, *tmp_r, *tmp_node, *node_copy;
+ int *old_params;
+
+ tmp_l = tre_ast_new_literal(mem, PARAMETER, 0, -1);
+ if (!tmp_l)
+ return REG_ESPACE;
+ ((tre_literal_t *)tmp_l->obj)->u.params = iter->params;
+ iter->params[TRE_PARAM_DEPTH] = params_depth + 1;
+ tmp_r = tre_ast_new_literal(mem, PARAMETER, 0, -1);
+ if (!tmp_r)
+ return REG_ESPACE;
+ old_params = tre_mem_alloc(mem, sizeof(*old_params)
+ * TRE_PARAM_LAST);
+ if (!old_params)
+ return REG_ESPACE;
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ old_params[i] = params[i];
+ ((tre_literal_t *)tmp_r->obj)->u.params = old_params;
+ old_params[TRE_PARAM_DEPTH] = params_depth;
+ /* XXX - this is the only place where ast_new_node is
+ needed -- should be moved inside AST module. */
+ node_copy = tre_ast_new_node(mem, ITERATION,
+ sizeof(tre_iteration_t));
+ if (!node_copy)
+ return REG_ESPACE;
+ node_copy->obj = node->obj;
+ tmp_node = tre_ast_new_catenation(mem, tmp_l, node_copy);
+ if (!tmp_node)
+ return REG_ESPACE;
+ tmp_node = tre_ast_new_catenation(mem, tmp_node, tmp_r);
+ if (!tmp_node)
+ return REG_ESPACE;
+ /* Replace the contents of `node' with `tmp_node'. */
+ QSE_MEMCPY (node, tmp_node, sizeof(*node));
+ node->obj = tmp_node->obj;
+ node->type = tmp_node->type;
+ params_depth++;
+ if (params_depth > *max_depth)
+ *max_depth = params_depth;
+ }
+ break;
+ }
+ default:
+ assert(0);
+ break;
+ }
+ }
+
+ *position += pos_add_total;
+
+ /* `max_pos' should never be larger than `*position' if the above
+ code works, but just an extra safeguard let's make sure
+ `*position' is set large enough so enough memory will be
+ allocated for the transition table. */
+ if (max_pos > *position)
+ *position = max_pos;
+
+#ifdef TRE_DEBUG
+ DPRINT(("Expanded AST:\n"));
+ tre_ast_print(ast);
+ DPRINT(("*position %d, max_pos %d\n", *position, max_pos));
+#endif
+
+ return status;
+}
+
+static tre_pos_and_tags_t *
+tre_set_empty(tre_mem_t mem)
+{
+ tre_pos_and_tags_t *new_set;
+
+ new_set = tre_mem_calloc(mem, sizeof(*new_set));
+ if (new_set == NULL)
+ return NULL;
+
+ new_set[0].position = -1;
+ new_set[0].code_min = -1;
+ new_set[0].code_max = -1;
+
+ return new_set;
+}
+
+static tre_pos_and_tags_t *
+tre_set_one(tre_mem_t mem, int position, int code_min, int code_max,
+ tre_ctype_t class, tre_ctype_t *neg_classes, int backref)
+{
+ tre_pos_and_tags_t *new_set;
+
+ new_set = tre_mem_calloc(mem, sizeof(*new_set) * 2);
+ if (new_set == NULL)
+ return NULL;
+
+ new_set[0].position = position;
+ new_set[0].code_min = code_min;
+ new_set[0].code_max = code_max;
+ new_set[0].class = class;
+ new_set[0].neg_classes = neg_classes;
+ new_set[0].backref = backref;
+ new_set[1].position = -1;
+ new_set[1].code_min = -1;
+ new_set[1].code_max = -1;
+
+ return new_set;
+}
+
+static tre_pos_and_tags_t *
+tre_set_union(tre_mem_t mem, tre_pos_and_tags_t *set1, tre_pos_and_tags_t *set2,
+ int *tags, int assertions, int *params)
+{
+ int s1, s2, i, j;
+ tre_pos_and_tags_t *new_set;
+ int *new_tags;
+ int num_tags;
+
+ for (num_tags = 0; tags != NULL && tags[num_tags] >= 0; num_tags++);
+ for (s1 = 0; set1[s1].position >= 0; s1++);
+ for (s2 = 0; set2[s2].position >= 0; s2++);
+ new_set = tre_mem_calloc(mem, sizeof(*new_set) * (s1 + s2 + 1));
+ if (!new_set )
+ return NULL;
+
+ for (s1 = 0; set1[s1].position >= 0; s1++)
+ {
+ new_set[s1].position = set1[s1].position;
+ new_set[s1].code_min = set1[s1].code_min;
+ new_set[s1].code_max = set1[s1].code_max;
+ new_set[s1].assertions = set1[s1].assertions | assertions;
+ new_set[s1].class = set1[s1].class;
+ new_set[s1].neg_classes = set1[s1].neg_classes;
+ new_set[s1].backref = set1[s1].backref;
+ if (set1[s1].tags == NULL && tags == NULL)
+ new_set[s1].tags = NULL;
+ else
+ {
+ for (i = 0; set1[s1].tags != NULL && set1[s1].tags[i] >= 0; i++);
+ new_tags = tre_mem_alloc(mem, (sizeof(*new_tags)
+ * (i + num_tags + 1)));
+ if (new_tags == NULL)
+ return NULL;
+ for (j = 0; j < i; j++)
+ new_tags[j] = set1[s1].tags[j];
+ for (i = 0; i < num_tags; i++)
+ new_tags[j + i] = tags[i];
+ new_tags[j + i] = -1;
+ new_set[s1].tags = new_tags;
+ }
+ if (set1[s1].params)
+ new_set[s1].params = set1[s1].params;
+ if (params)
+ {
+ if (!new_set[s1].params)
+ new_set[s1].params = params;
+ else
+ {
+ new_set[s1].params = tre_mem_alloc(mem, sizeof(*params) *
+ TRE_PARAM_LAST);
+ if (!new_set[s1].params)
+ return NULL;
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ if (params[i] != TRE_PARAM_UNSET)
+ new_set[s1].params[i] = params[i];
+ }
+ }
+ }
+
+ for (s2 = 0; set2[s2].position >= 0; s2++)
+ {
+ new_set[s1 + s2].position = set2[s2].position;
+ new_set[s1 + s2].code_min = set2[s2].code_min;
+ new_set[s1 + s2].code_max = set2[s2].code_max;
+ /* XXX - why not | assertions here as well? */
+ new_set[s1 + s2].assertions = set2[s2].assertions;
+ new_set[s1 + s2].class = set2[s2].class;
+ new_set[s1 + s2].neg_classes = set2[s2].neg_classes;
+ new_set[s1 + s2].backref = set2[s2].backref;
+ if (set2[s2].tags == NULL)
+ new_set[s1 + s2].tags = NULL;
+ else
+ {
+ for (i = 0; set2[s2].tags[i] >= 0; i++);
+ new_tags = tre_mem_alloc(mem, sizeof(*new_tags) * (i + 1));
+ if (new_tags == NULL)
+ return NULL;
+ for (j = 0; j < i; j++)
+ new_tags[j] = set2[s2].tags[j];
+ new_tags[j] = -1;
+ new_set[s1 + s2].tags = new_tags;
+ }
+ if (set2[s2].params)
+ new_set[s1 + s2].params = set2[s2].params;
+ if (params)
+ {
+ if (!new_set[s1 + s2].params)
+ new_set[s1 + s2].params = params;
+ else
+ {
+ new_set[s1 + s2].params = tre_mem_alloc(mem, sizeof(*params) *
+ TRE_PARAM_LAST);
+ if (!new_set[s1 + s2].params)
+ return NULL;
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ if (params[i] != TRE_PARAM_UNSET)
+ new_set[s1 + s2].params[i] = params[i];
+ }
+ }
+ }
+ new_set[s1 + s2].position = -1;
+ return new_set;
+}
+
+/* Finds the empty path through `node' which is the one that should be
+ taken according to POSIX.2 rules, and adds the tags on that path to
+ `tags'. `tags' may be NULL. If `num_tags_seen' is not NULL, it is
+ set to the number of tags seen on the path. */
+static reg_errcode_t
+tre_match_empty(tre_stack_t *stack, tre_ast_node_t *node, int *tags,
+ int *assertions, int *params, int *num_tags_seen,
+ int *params_seen)
+{
+ tre_literal_t *lit;
+ tre_union_t *uni;
+ tre_catenation_t *cat;
+ tre_iteration_t *iter;
+ int i;
+ int bottom = tre_stack_num_objects(stack);
+ reg_errcode_t status = REG_OK;
+ if (num_tags_seen)
+ *num_tags_seen = 0;
+ if (params_seen)
+ *params_seen = 0;
+
+ status = tre_stack_push_voidptr(stack, node);
+
+ /* Walk through the tree recursively. */
+ while (status == REG_OK && tre_stack_num_objects(stack) > bottom)
+ {
+ node = tre_stack_pop_voidptr(stack);
+
+ switch (node->type)
+ {
+ case LITERAL:
+ lit = (tre_literal_t *)node->obj;
+ switch (lit->code_min)
+ {
+ case TAG:
+ if (lit->code_max >= 0)
+ {
+ if (tags != NULL)
+ {
+ /* Add the tag to `tags'. */
+ for (i = 0; tags[i] >= 0; i++)
+ if (tags[i] == lit->code_max)
+ break;
+ if (tags[i] < 0)
+ {
+ tags[i] = lit->code_max;
+ tags[i + 1] = -1;
+ }
+ }
+ if (num_tags_seen)
+ (*num_tags_seen)++;
+ }
+ break;
+ case ASSERTION:
+ assert(lit->code_max >= 1
+ || lit->code_max <= ASSERT_LAST);
+ if (assertions != NULL)
+ *assertions |= lit->code_max;
+ break;
+ case PARAMETER:
+ if (params != NULL)
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ params[i] = lit->u.params[i];
+ if (params_seen != NULL)
+ *params_seen = 1;
+ break;
+ case EMPTY:
+ break;
+ default:
+ assert(0);
+ break;
+ }
+ break;
+
+ case UNION:
+ /* Subexpressions starting earlier take priority over ones
+ starting later, so we prefer the left subexpression over the
+ right subexpression. */
+ uni = (tre_union_t *)node->obj;
+ if (uni->left->nullable)
+ STACK_PUSHX(stack, voidptr, uni->left)
+ else if (uni->right->nullable)
+ STACK_PUSHX(stack, voidptr, uni->right)
+ else
+ assert(0);
+ break;
+
+ case CATENATION:
+ /* The path must go through both children. */
+ cat = (tre_catenation_t *)node->obj;
+ assert(cat->left->nullable);
+ assert(cat->right->nullable);
+ STACK_PUSHX(stack, voidptr, cat->left);
+ STACK_PUSHX(stack, voidptr, cat->right);
+ break;
+
+ case ITERATION:
+ /* A match with an empty string is preferred over no match at
+ all, so we go through the argument if possible. */
+ iter = (tre_iteration_t *)node->obj;
+ if (iter->arg->nullable)
+ STACK_PUSHX(stack, voidptr, iter->arg);
+ break;
+
+ default:
+ assert(0);
+ break;
+ }
+ }
+
+ return status;
+}
+
+
+typedef enum
+{
+ NFL_RECURSE,
+ NFL_POST_UNION,
+ NFL_POST_CATENATION,
+ NFL_POST_ITERATION
+} tre_nfl_stack_symbol_t;
+
+
+/* Computes and fills in the fields `nullable', `firstpos', and `lastpos' for
+ the nodes of the AST `tree'. */
+static reg_errcode_t
+tre_compute_nfl(tre_mem_t mem, tre_stack_t *stack, tre_ast_node_t *tree)
+{
+ int bottom = tre_stack_num_objects(stack);
+
+ STACK_PUSHR(stack, voidptr, tree);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+
+ while (tre_stack_num_objects(stack) > bottom)
+ {
+ tre_nfl_stack_symbol_t symbol;
+ tre_ast_node_t *node;
+
+ symbol = (tre_nfl_stack_symbol_t)tre_stack_pop_int(stack);
+ node = tre_stack_pop_voidptr(stack);
+ switch (symbol)
+ {
+ case NFL_RECURSE:
+ switch (node->type)
+ {
+ case LITERAL:
+ {
+ tre_literal_t *lit = (tre_literal_t *)node->obj;
+ if (IS_BACKREF(lit))
+ {
+ /* Back references: nullable = false, firstpos = {i},
+ lastpos = {i}. */
+ node->nullable = 0;
+ node->firstpos = tre_set_one(mem, lit->position, 0,
+ TRE_CHAR_MAX, 0, NULL, -1);
+ if (!node->firstpos)
+ return REG_ESPACE;
+ node->lastpos = tre_set_one(mem, lit->position, 0,
+ TRE_CHAR_MAX, 0, NULL,
+ (int)lit->code_max);
+ if (!node->lastpos)
+ return REG_ESPACE;
+ }
+ else if (lit->code_min < 0)
+ {
+ /* Tags, empty strings, params, and zero width assertions:
+ nullable = true, firstpos = {}, and lastpos = {}. */
+ node->nullable = 1;
+ node->firstpos = tre_set_empty(mem);
+ if (!node->firstpos)
+ return REG_ESPACE;
+ node->lastpos = tre_set_empty(mem);
+ if (!node->lastpos)
+ return REG_ESPACE;
+ }
+ else
+ {
+ /* Literal at position i: nullable = false, firstpos = {i},
+ lastpos = {i}. */
+ node->nullable = 0;
+ node->firstpos =
+ tre_set_one(mem, lit->position, (int)lit->code_min,
+ (int)lit->code_max, 0, NULL, -1);
+ if (!node->firstpos)
+ return REG_ESPACE;
+ node->lastpos = tre_set_one(mem, lit->position,
+ (int)lit->code_min,
+ (int)lit->code_max,
+ lit->u.class, lit->neg_classes,
+ -1);
+ if (!node->lastpos)
+ return REG_ESPACE;
+ }
+ break;
+ }
+
+ case UNION:
+ /* Compute the attributes for the two subtrees, and after that
+ for this node. */
+ STACK_PUSHR(stack, voidptr, node);
+ STACK_PUSHR(stack, int, NFL_POST_UNION);
+ STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->right);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+ STACK_PUSHR(stack, voidptr, ((tre_union_t *)node->obj)->left);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+ break;
+
+ case CATENATION:
+ /* Compute the attributes for the two subtrees, and after that
+ for this node. */
+ STACK_PUSHR(stack, voidptr, node);
+ STACK_PUSHR(stack, int, NFL_POST_CATENATION);
+ STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->right);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+ STACK_PUSHR(stack, voidptr, ((tre_catenation_t *)node->obj)->left);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+ break;
+
+ case ITERATION:
+ /* Compute the attributes for the subtree, and after that for
+ this node. */
+ STACK_PUSHR(stack, voidptr, node);
+ STACK_PUSHR(stack, int, NFL_POST_ITERATION);
+ STACK_PUSHR(stack, voidptr, ((tre_iteration_t *)node->obj)->arg);
+ STACK_PUSHR(stack, int, NFL_RECURSE);
+ break;
+ }
+ break; /* end case: NFL_RECURSE */
+
+ case NFL_POST_UNION:
+ {
+ tre_union_t *uni = (tre_union_t *)node->obj;
+ node->nullable = uni->left->nullable || uni->right->nullable;
+ node->firstpos = tre_set_union(mem, uni->left->firstpos,
+ uni->right->firstpos, NULL, 0, NULL);
+ if (!node->firstpos)
+ return REG_ESPACE;
+ node->lastpos = tre_set_union(mem, uni->left->lastpos,
+ uni->right->lastpos, NULL, 0, NULL);
+ if (!node->lastpos)
+ return REG_ESPACE;
+ break;
+ }
+
+ case NFL_POST_ITERATION:
+ {
+ tre_iteration_t *iter = (tre_iteration_t *)node->obj;
+
+ if (iter->min == 0 || iter->arg->nullable)
+ node->nullable = 1;
+ else
+ node->nullable = 0;
+ node->firstpos = iter->arg->firstpos;
+ node->lastpos = iter->arg->lastpos;
+ break;
+ }
+
+ case NFL_POST_CATENATION:
+ {
+ int num_tags, *tags, assertions, params_seen;
+ int *params;
+ reg_errcode_t status;
+ tre_catenation_t *cat = node->obj;
+ node->nullable = cat->left->nullable && cat->right->nullable;
+
+ /* Compute firstpos. */
+ if (cat->left->nullable)
+ {
+ /* The left side matches the empty string. Make a first pass
+ with tre_match_empty() to get the number of tags and
+ parameters. */
+ status = tre_match_empty(stack, cat->left,
+ NULL, NULL, NULL, &num_tags,
+ ¶ms_seen);
+ if (status != REG_OK)
+ return status;
+ /* Allocate arrays for the tags and parameters. */
+ tags = xmalloc(mem->mmgr, sizeof(*tags) * (num_tags + 1));
+ if (!tags)
+ return REG_ESPACE;
+ tags[0] = -1;
+ assertions = 0;
+ params = NULL;
+ if (params_seen)
+ {
+ params = tre_mem_alloc(mem, sizeof(*params)
+ * TRE_PARAM_LAST);
+ if (!params)
+ {
+ xfree(mem->mmgr,tags);
+ return REG_ESPACE;
+ }
+ }
+ /* Second pass with tre_mach_empty() to get the list of
+ tags and parameters. */
+ status = tre_match_empty(stack, cat->left, tags,
+ &assertions, params, NULL, NULL);
+ if (status != REG_OK)
+ {
+ xfree(mem->mmgr,tags);
+ return status;
+ }
+ node->firstpos =
+ tre_set_union(mem, cat->right->firstpos, cat->left->firstpos,
+ tags, assertions, params);
+ xfree(mem->mmgr,tags);
+ if (!node->firstpos)
+ return REG_ESPACE;
+ }
+ else
+ {
+ node->firstpos = cat->left->firstpos;
+ }
+
+ /* Compute lastpos. */
+ if (cat->right->nullable)
+ {
+ /* The right side matches the empty string. Make a first pass
+ with tre_match_empty() to get the number of tags and
+ parameters. */
+ status = tre_match_empty(stack, cat->right,
+ NULL, NULL, NULL, &num_tags,
+ ¶ms_seen);
+ if (status != REG_OK)
+ return status;
+ /* Allocate arrays for the tags and parameters. */
+ tags = xmalloc(mem->mmgr,sizeof(int) * (num_tags + 1));
+ if (!tags)
+ return REG_ESPACE;
+ tags[0] = -1;
+ assertions = 0;
+ params = NULL;
+ if (params_seen)
+ {
+ params = tre_mem_alloc(mem, sizeof(*params)
+ * TRE_PARAM_LAST);
+ if (!params)
+ {
+ xfree(mem->mmgr,tags);
+ return REG_ESPACE;
+ }
+ }
+ /* Second pass with tre_mach_empty() to get the list of
+ tags and parameters. */
+ status = tre_match_empty(stack, cat->right, tags,
+ &assertions, params, NULL, NULL);
+ if (status != REG_OK)
+ {
+ xfree(mem->mmgr,tags);
+ return status;
+ }
+ node->lastpos =
+ tre_set_union(mem, cat->left->lastpos, cat->right->lastpos,
+ tags, assertions, params);
+ xfree(mem->mmgr,tags);
+ if (!node->lastpos)
+ return REG_ESPACE;
+ }
+ else
+ {
+ node->lastpos = cat->right->lastpos;
+ }
+ break;
+ }
+
+ default:
+ assert(0);
+ break;
+ }
+ }
+
+ return REG_OK;
+}
+
+
+/* Adds a transition from each position in `p1' to each position in `p2'. */
+static reg_errcode_t
+tre_make_trans(qse_mmgr_t* mmgr, tre_pos_and_tags_t *p1, tre_pos_and_tags_t *p2,
+ tre_tnfa_transition_t *transitions,
+ int *counts, int *offs)
+{
+ tre_pos_and_tags_t *orig_p2 = p2;
+ tre_tnfa_transition_t *trans;
+ int i, j, k, l, dup, prev_p2_pos;
+
+ if (transitions != NULL)
+ while (p1->position >= 0)
+ {
+ p2 = orig_p2;
+ prev_p2_pos = -1;
+ while (p2->position >= 0)
+ {
+ /* Optimization: if this position was already handled, skip it. */
+ if (p2->position == prev_p2_pos)
+ {
+ p2++;
+ continue;
+ }
+ prev_p2_pos = p2->position;
+ /* Set `trans' to point to the next unused transition from
+ position `p1->position'. */
+ trans = transitions + offs[p1->position];
+ while (trans->state != NULL)
+ {
+#if 0
+ /* If we find a previous transition from `p1->position' to
+ `p2->position', it is overwritten. This can happen only
+ if there are nested loops in the regexp, like in "((a)*)*".
+ In POSIX.2 repetition using the outer loop is always
+ preferred over using the inner loop. Therefore the
+ transition for the inner loop is useless and can be thrown
+ away. */
+ /* XXX - The same position is used for all nodes in a bracket
+ expression, so this optimization cannot be used (it will
+ break bracket expressions) unless I figure out a way to
+ detect it here. */
+ if (trans->state_id == p2->position)
+ {
+ DPRINT(("*"));
+ break;
+ }
+#endif
+ trans++;
+ }
+
+ if (trans->state == NULL)
+ (trans + 1)->state = NULL;
+ /* Use the character ranges, assertions, etc. from `p1' for
+ the transition from `p1' to `p2'. */
+ trans->code_min = p1->code_min;
+ trans->code_max = p1->code_max;
+ trans->state = transitions + offs[p2->position];
+ trans->state_id = p2->position;
+ trans->assertions = p1->assertions | p2->assertions
+ | (p1->class ? ASSERT_CHAR_CLASS : 0)
+ | (p1->neg_classes != NULL ? ASSERT_CHAR_CLASS_NEG : 0);
+ if (p1->backref >= 0)
+ {
+ assert((trans->assertions & ASSERT_CHAR_CLASS) == 0);
+ assert(p2->backref < 0);
+ trans->u.backref = p1->backref;
+ trans->assertions |= ASSERT_BACKREF;
+ }
+ else
+ trans->u.class = p1->class;
+ if (p1->neg_classes != NULL)
+ {
+ for (i = 0; p1->neg_classes[i] != (tre_ctype_t)0; i++);
+ trans->neg_classes =
+ xmalloc(mmgr,sizeof(*trans->neg_classes) * (i + 1));
+ if (trans->neg_classes == NULL)
+ return REG_ESPACE;
+ for (i = 0; p1->neg_classes[i] != (tre_ctype_t)0; i++)
+ trans->neg_classes[i] = p1->neg_classes[i];
+ trans->neg_classes[i] = (tre_ctype_t)0;
+ }
+ else
+ trans->neg_classes = NULL;
+
+ /* Find out how many tags this transition has. */
+ i = 0;
+ if (p1->tags != NULL)
+ while(p1->tags[i] >= 0)
+ i++;
+ j = 0;
+ if (p2->tags != NULL)
+ while(p2->tags[j] >= 0)
+ j++;
+
+ /* If we are overwriting a transition, free the old tag array. */
+ if (trans->tags != NULL)
+ xfree(mmgr,trans->tags);
+ trans->tags = NULL;
+
+ /* If there were any tags, allocate an array and fill it. */
+ if (i + j > 0)
+ {
+ trans->tags = xmalloc(mmgr,sizeof(*trans->tags) * (i + j + 1));
+ if (!trans->tags)
+ return REG_ESPACE;
+ i = 0;
+ if (p1->tags != NULL)
+ while(p1->tags[i] >= 0)
+ {
+ trans->tags[i] = p1->tags[i];
+ i++;
+ }
+ l = i;
+ j = 0;
+ if (p2->tags != NULL)
+ while (p2->tags[j] >= 0)
+ {
+ /* Don't add duplicates. */
+ dup = 0;
+ for (k = 0; k < i; k++)
+ if (trans->tags[k] == p2->tags[j])
+ {
+ dup = 1;
+ break;
+ }
+ if (!dup)
+ trans->tags[l++] = p2->tags[j];
+ j++;
+ }
+ trans->tags[l] = -1;
+ }
+
+ /* Set the parameter array. If both `p2' and `p1' have same
+ parameters, the values in `p2' override those in `p1'. */
+ if (p1->params || p2->params)
+ {
+ if (!trans->params)
+ trans->params = xmalloc(mmgr,sizeof(*trans->params)
+ * TRE_PARAM_LAST);
+ if (!trans->params)
+ return REG_ESPACE;
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ {
+ trans->params[i] = TRE_PARAM_UNSET;
+ if (p1->params && p1->params[i] != TRE_PARAM_UNSET)
+ trans->params[i] = p1->params[i];
+ if (p2->params && p2->params[i] != TRE_PARAM_UNSET)
+ trans->params[i] = p2->params[i];
+ }
+ }
+ else
+ {
+ if (trans->params)
+ xfree(mmgr,trans->params);
+ trans->params = NULL;
+ }
+
+
+#ifdef TRE_DEBUG
+ {
+ int *tags;
+
+ DPRINT((" %2d -> %2d on %3d", p1->position, p2->position,
+ p1->code_min));
+ if (p1->code_max != p1->code_min)
+ DPRINT(("-%3d", p1->code_max));
+ tags = trans->tags;
+ if (tags)
+ {
+ DPRINT((", tags ["));
+ while (*tags >= 0)
+ {
+ DPRINT(("%d", *tags));
+ tags++;
+ if (*tags >= 0)
+ DPRINT((","));
+ }
+ DPRINT(("]"));
+ }
+ if (trans->assertions)
+ DPRINT((", assert %d", trans->assertions));
+ if (trans->assertions & ASSERT_BACKREF)
+ DPRINT((", backref %d", trans->u.backref));
+ else if (trans->u.class)
+ DPRINT((", class %ld", (long)trans->u.class));
+ if (trans->neg_classes)
+ DPRINT((", neg_classes %p", trans->neg_classes));
+ if (trans->params)
+ {
+ DPRINT((", "));
+ tre_print_params(trans->params);
+ }
+ DPRINT(("\n"));
+ }
+#endif /* TRE_DEBUG */
+ p2++;
+ }
+ p1++;
+ }
+ else
+ /* Compute a maximum limit for the number of transitions leaving
+ from each state. */
+ while (p1->position >= 0)
+ {
+ p2 = orig_p2;
+ while (p2->position >= 0)
+ {
+ counts[p1->position]++;
+ p2++;
+ }
+ p1++;
+ }
+ return REG_OK;
+}
+
+/* Converts the syntax tree to a TNFA. All the transitions in the TNFA are
+ labelled with one character range (there are no transitions on empty
+ strings). The TNFA takes O(n^2) space in the worst case, `n' is size of
+ the regexp. */
+static reg_errcode_t
+tre_ast_to_tnfa(qse_mmgr_t* mmgr, tre_ast_node_t *node, tre_tnfa_transition_t *transitions,
+ int *counts, int *offs)
+{
+ tre_union_t *uni;
+ tre_catenation_t *cat;
+ tre_iteration_t *iter;
+ reg_errcode_t errcode = REG_OK;
+
+ /* XXX - recurse using a stack!. */
+ switch (node->type)
+ {
+ case LITERAL:
+ break;
+ case UNION:
+ uni = (tre_union_t *)node->obj;
+ errcode = tre_ast_to_tnfa(mmgr, uni->left, transitions, counts, offs);
+ if (errcode != REG_OK)
+ return errcode;
+ errcode = tre_ast_to_tnfa(mmgr, uni->right, transitions, counts, offs);
+ break;
+
+ case CATENATION:
+ cat = (tre_catenation_t *)node->obj;
+ /* Add a transition from each position in cat->left->lastpos
+ to each position in cat->right->firstpos. */
+ errcode = tre_make_trans(mmgr, cat->left->lastpos, cat->right->firstpos,
+ transitions, counts, offs);
+ if (errcode != REG_OK)
+ return errcode;
+ errcode = tre_ast_to_tnfa(mmgr, cat->left, transitions, counts, offs);
+ if (errcode != REG_OK)
+ return errcode;
+ errcode = tre_ast_to_tnfa(mmgr, cat->right, transitions, counts, offs);
+ break;
+
+ case ITERATION:
+ iter = (tre_iteration_t *)node->obj;
+ assert(iter->max == -1 || iter->max == 1);
+
+ if (iter->max == -1)
+ {
+ assert(iter->min == 0 || iter->min == 1);
+ /* Add a transition from each last position in the iterated
+ expression to each first position. */
+ errcode = tre_make_trans(mmgr, iter->arg->lastpos, iter->arg->firstpos,
+ transitions, counts, offs);
+ if (errcode != REG_OK)
+ return errcode;
+ }
+ errcode = tre_ast_to_tnfa(mmgr, iter->arg, transitions, counts, offs);
+ break;
+ }
+ return errcode;
+}
+
+
+#define ERROR_EXIT(err) \
+ do \
+ { \
+ errcode = err; \
+ if (/*CONSTCOND*/1) \
+ goto error_exit; \
+ } \
+ while (/*CONSTCOND*/0)
+
+
+int tre_compile (regex_t *preg, const tre_char_t *regex, size_t n, int cflags)
+{
+ tre_stack_t *stack;
+ tre_ast_node_t *tree, *tmp_ast_l, *tmp_ast_r;
+ tre_pos_and_tags_t *p;
+ int *counts = NULL, *offs = NULL;
+ int i, add = 0;
+ tre_tnfa_transition_t *transitions, *initial;
+ tre_tnfa_t *tnfa = NULL;
+ tre_submatch_data_t *submatch_data;
+ tre_tag_direction_t *tag_directions = NULL;
+ reg_errcode_t errcode;
+ tre_mem_t mem;
+
+ /* Parse context. */
+ tre_parse_ctx_t parse_ctx;
+
+ /* Allocate a stack used throughout the compilation process for various
+ purposes. */
+ stack = tre_stack_new(preg->mmgr, 512, 10240, 128);
+ if (!stack)
+ return REG_ESPACE;
+ /* Allocate a fast memory allocator. */
+ mem = tre_mem_new(preg->mmgr);
+ if (!mem)
+ {
+ tre_stack_destroy(stack);
+ return REG_ESPACE;
+ }
+
+ /* Parse the regexp. */
+ QSE_MEMSET(&parse_ctx, 0, sizeof(parse_ctx));
+ parse_ctx.mem = mem;
+ parse_ctx.stack = stack;
+ parse_ctx.re = regex;
+ parse_ctx.len = n;
+ parse_ctx.cflags = cflags;
+ parse_ctx.max_backref = -1;
+ DPRINT(("tre_compile: parsing '%.*" STRF "'\n", (int)n, regex));
+ errcode = tre_parse(&parse_ctx);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+ preg->re_nsub = parse_ctx.submatch_id - 1;
+ tree = parse_ctx.result;
+
+ /* Back references and approximate matching cannot currently be used
+ in the same regexp. */
+ if (parse_ctx.max_backref >= 0 && parse_ctx.have_approx)
+ ERROR_EXIT(REG_BADPAT);
+
+#ifdef TRE_DEBUG
+ tre_ast_print(tree);
+#endif /* TRE_DEBUG */
+
+ /* Referring to nonexistent subexpressions is illegal. */
+ if (parse_ctx.max_backref > (int)preg->re_nsub)
+ ERROR_EXIT(REG_ESUBREG);
+
+ /* Allocate the TNFA struct. */
+ tnfa = xcalloc(preg->mmgr, 1, sizeof(tre_tnfa_t));
+ if (tnfa == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ tnfa->have_backrefs = parse_ctx.max_backref >= 0;
+ tnfa->have_approx = parse_ctx.have_approx;
+ tnfa->num_submatches = parse_ctx.submatch_id;
+
+ /* Set up tags for submatch addressing. If REG_NOSUB is set and the
+ regexp does not have back references, this can be skipped. */
+ if (tnfa->have_backrefs || !(cflags & REG_NOSUB))
+ {
+ DPRINT(("tre_compile: setting up tags\n"));
+
+ /* Figure out how many tags we will need. */
+ errcode = tre_add_tags(NULL, stack, tree, tnfa);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+#ifdef TRE_DEBUG
+ tre_ast_print(tree);
+#endif /* TRE_DEBUG */
+
+ if (tnfa->num_tags > 0)
+ {
+ tag_directions = xmalloc(preg->mmgr,sizeof(*tag_directions)
+ * (tnfa->num_tags + 1));
+ if (tag_directions == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ tnfa->tag_directions = tag_directions;
+ QSE_MEMSET(tag_directions, -1,
+ sizeof(*tag_directions) * (tnfa->num_tags + 1));
+ }
+ tnfa->minimal_tags = xcalloc(preg->mmgr, (unsigned)tnfa->num_tags * 2 + 1,
+ sizeof(tnfa->minimal_tags));
+ if (tnfa->minimal_tags == NULL)
+ ERROR_EXIT(REG_ESPACE);
+
+ submatch_data = xcalloc(preg->mmgr,(unsigned)parse_ctx.submatch_id,
+ sizeof(*submatch_data));
+ if (submatch_data == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ tnfa->submatch_data = submatch_data;
+
+ errcode = tre_add_tags(mem, stack, tree, tnfa);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+
+#ifdef TRE_DEBUG
+ for (i = 0; i < parse_ctx.submatch_id; i++)
+ DPRINT(("pmatch[%d] = {t%d, t%d}\n",
+ i, submatch_data[i].so_tag, submatch_data[i].eo_tag));
+ for (i = 0; i < tnfa->num_tags; i++)
+ DPRINT(("t%d is %s\n", i,
+ tag_directions[i] == TRE_TAG_MINIMIZE ?
+ "minimized" : "maximized"));
+#endif /* TRE_DEBUG */
+ }
+
+ /* Expand iteration nodes. */
+ errcode = tre_expand_ast(mem, stack, tree, &parse_ctx.position,
+ tag_directions, &tnfa->params_depth);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+
+ /* Add a dummy node for the final state.
+ XXX - For certain patterns this dummy node can be optimized away,
+ for example "a*" or "ab*". Figure out a simple way to detect
+ this possibility. */
+ tmp_ast_l = tree;
+ tmp_ast_r = tre_ast_new_literal(mem, 0, 0, parse_ctx.position++);
+ if (tmp_ast_r == NULL)
+ ERROR_EXIT(REG_ESPACE);
+
+ tree = tre_ast_new_catenation(mem, tmp_ast_l, tmp_ast_r);
+ if (tree == NULL)
+ ERROR_EXIT(REG_ESPACE);
+
+#ifdef TRE_DEBUG
+ tre_ast_print(tree);
+ DPRINT(("Number of states: %d\n", parse_ctx.position));
+#endif /* TRE_DEBUG */
+
+ errcode = tre_compute_nfl(mem, stack, tree);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+
+ counts = xmalloc(preg->mmgr,sizeof(int) * parse_ctx.position);
+ if (counts == NULL)
+ ERROR_EXIT(REG_ESPACE);
+
+ offs = xmalloc(preg->mmgr,sizeof(int) * parse_ctx.position);
+ if (offs == NULL)
+ ERROR_EXIT(REG_ESPACE);
+
+ for (i = 0; i < parse_ctx.position; i++)
+ counts[i] = 0;
+ tre_ast_to_tnfa(preg->mmgr, tree, NULL, counts, NULL);
+
+ add = 0;
+ for (i = 0; i < parse_ctx.position; i++)
+ {
+ offs[i] = add;
+ add += counts[i] + 1;
+ counts[i] = 0;
+ }
+ transitions = xcalloc(preg->mmgr, (unsigned)add + 1, sizeof(*transitions));
+ if (transitions == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ tnfa->transitions = transitions;
+ tnfa->num_transitions = add;
+
+ DPRINT(("Converting to TNFA:\n"));
+ errcode = tre_ast_to_tnfa(preg->mmgr, tree, transitions, counts, offs);
+ if (errcode != REG_OK)
+ ERROR_EXIT(errcode);
+
+ /* If in eight bit mode, compute a table of characters that can be the
+ first character of a match. */
+ tnfa->first_char = -1;
+ if (TRE_MB_CUR_MAX == 1 && !tmp_ast_l->nullable)
+ {
+ int count = 0;
+ tre_cint_t k;
+ DPRINT(("Characters that can start a match:"));
+ tnfa->firstpos_chars = xcalloc(preg->mmgr, 256, sizeof(char));
+ if (tnfa->firstpos_chars == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ for (p = tree->firstpos; p->position >= 0; p++)
+ {
+ tre_tnfa_transition_t *j = transitions + offs[p->position];
+ while (j->state != NULL)
+ {
+ for (k = j->code_min; k <= j->code_max && k < 256; k++)
+ {
+ DPRINT((" %d", k));
+ tnfa->firstpos_chars[k] = 1;
+ count++;
+ }
+ j++;
+ }
+ }
+ DPRINT(("\n"));
+#define TRE_OPTIMIZE_FIRST_CHAR 1
+#if TRE_OPTIMIZE_FIRST_CHAR
+ if (count == 1)
+ {
+ for (k = 0; k < 256; k++)
+ if (tnfa->firstpos_chars[k])
+ {
+ DPRINT(("first char must be %d\n", k));
+ tnfa->first_char = k;
+ xfree(preg->mmgr,tnfa->firstpos_chars);
+ tnfa->firstpos_chars = NULL;
+ break;
+ }
+ }
+#endif
+
+ }
+ else
+ tnfa->firstpos_chars = NULL;
+
+
+ p = tree->firstpos;
+ i = 0;
+ while (p->position >= 0)
+ {
+ i++;
+
+#ifdef TRE_DEBUG
+ {
+ int *tags;
+ DPRINT(("initial: %d", p->position));
+ tags = p->tags;
+ if (tags != NULL)
+ {
+ if (*tags >= 0)
+ DPRINT(("/"));
+ while (*tags >= 0)
+ {
+ DPRINT(("%d", *tags));
+ tags++;
+ if (*tags >= 0)
+ DPRINT((","));
+ }
+ }
+ DPRINT((", assert %d", p->assertions));
+ if (p->params)
+ {
+ DPRINT((", "));
+ tre_print_params(p->params);
+ }
+ DPRINT(("\n"));
+ }
+#endif /* TRE_DEBUG */
+
+ p++;
+ }
+
+ initial = xcalloc(preg->mmgr, (unsigned)i + 1, sizeof(tre_tnfa_transition_t));
+ if (initial == NULL)
+ ERROR_EXIT(REG_ESPACE);
+ tnfa->initial = initial;
+
+ i = 0;
+ for (p = tree->firstpos; p->position >= 0; p++)
+ {
+ initial[i].state = transitions + offs[p->position];
+ initial[i].state_id = p->position;
+ initial[i].tags = NULL;
+ /* Copy the arrays p->tags, and p->params, they are allocated
+ from a tre_mem object. */
+ if (p->tags)
+ {
+ int j;
+ for (j = 0; p->tags[j] >= 0; j++);
+ initial[i].tags = xmalloc(preg->mmgr,sizeof(*p->tags) * (j + 1));
+ if (!initial[i].tags)
+ ERROR_EXIT(REG_ESPACE);
+ QSE_MEMCPY (initial[i].tags, p->tags, sizeof(*p->tags) * (j + 1));
+ }
+ initial[i].params = NULL;
+ if (p->params)
+ {
+ initial[i].params = xmalloc(preg->mmgr,sizeof(*p->params) * TRE_PARAM_LAST);
+ if (!initial[i].params)
+ ERROR_EXIT(REG_ESPACE);
+ QSE_MEMCPY (initial[i].params, p->params,
+ sizeof(*p->params) * TRE_PARAM_LAST);
+ }
+ initial[i].assertions = p->assertions;
+ i++;
+ }
+ initial[i].state = NULL;
+
+ tnfa->num_transitions = add;
+ tnfa->final = transitions + offs[tree->lastpos[0].position];
+ tnfa->num_states = parse_ctx.position;
+ tnfa->cflags = cflags;
+
+ DPRINT(("final state %p\n", (void *)tnfa->final));
+
+ tre_mem_destroy(mem);
+ tre_stack_destroy(stack);
+ xfree(preg->mmgr,counts);
+ xfree(preg->mmgr,offs);
+
+ preg->TRE_REGEX_T_FIELD = (void *)tnfa;
+ return REG_OK;
+
+error_exit:
+ /* Free everything that was allocated and return the error code. */
+ tre_mem_destroy(mem);
+ if (stack != NULL)
+ tre_stack_destroy(stack);
+ if (counts != NULL)
+ xfree(preg->mmgr,counts);
+ if (offs != NULL)
+ xfree(preg->mmgr,offs);
+ preg->TRE_REGEX_T_FIELD = (void *)tnfa;
+ tre_free(preg);
+ return errcode;
+}
+
+void tre_free (regex_t *preg)
+{
+ tre_tnfa_t *tnfa;
+ unsigned int i;
+ tre_tnfa_transition_t *trans;
+
+ tnfa = (void *)preg->TRE_REGEX_T_FIELD;
+ if (!tnfa)
+ return;
+
+ for (i = 0; i < tnfa->num_transitions; i++)
+ if (tnfa->transitions[i].state)
+ {
+ if (tnfa->transitions[i].tags)
+ xfree(preg->mmgr,tnfa->transitions[i].tags);
+ if (tnfa->transitions[i].neg_classes)
+ xfree(preg->mmgr,tnfa->transitions[i].neg_classes);
+ if (tnfa->transitions[i].params)
+ xfree(preg->mmgr,tnfa->transitions[i].params);
+ }
+ if (tnfa->transitions)
+ xfree(preg->mmgr,tnfa->transitions);
+
+ if (tnfa->initial)
+ {
+ for (trans = tnfa->initial; trans->state; trans++)
+ {
+ if (trans->tags)
+ xfree(preg->mmgr,trans->tags);
+ if (trans->params)
+ xfree(preg->mmgr,trans->params);
+ }
+ xfree(preg->mmgr,tnfa->initial);
+ }
+
+ if (tnfa->submatch_data)
+ {
+ for (i = 0; i < tnfa->num_submatches; i++)
+ if (tnfa->submatch_data[i].parents)
+ xfree(preg->mmgr,tnfa->submatch_data[i].parents);
+ xfree(preg->mmgr,tnfa->submatch_data);
+ }
+
+ if (tnfa->tag_directions)
+ xfree(preg->mmgr,tnfa->tag_directions);
+ if (tnfa->firstpos_chars)
+ xfree(preg->mmgr,tnfa->firstpos_chars);
+ if (tnfa->minimal_tags)
+ xfree(preg->mmgr,tnfa->minimal_tags);
+ xfree(preg->mmgr,tnfa);
+}
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-compile.h b/qse/lib/cmn/tre-compile.h
new file mode 100644
index 00000000..f918e422
--- /dev/null
+++ b/qse/lib/cmn/tre-compile.h
@@ -0,0 +1,75 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-compile.h: Regex compilation definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+
+#ifndef _QSE_LIB_CMN_TRE_COMPILE_H_
+#define _QSE_LIB_CMN_TRE_COMPILE_H_
+
+typedef struct
+{
+ int position;
+ int code_min;
+ int code_max;
+ int *tags;
+ int assertions;
+ tre_ctype_t class;
+ tre_ctype_t *neg_classes;
+ int backref;
+ int *params;
+} tre_pos_and_tags_t;
+
+
+#endif /* TRE_COMPILE_H */
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-match-backtrack.c b/qse/lib/cmn/tre-match-backtrack.c
new file mode 100644
index 00000000..c1ce2293
--- /dev/null
+++ b/qse/lib/cmn/tre-match-backtrack.c
@@ -0,0 +1,658 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-match-backtrack.c - TRE backtracking regex matching engine
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+ This matcher is for regexps that use back referencing. Regexp matching
+ with back referencing is an NP-complete problem on the number of back
+ references. The easiest way to match them is to use a backtracking
+ routine which basically goes through all possible paths in the TNFA
+ and chooses the one which results in the best (leftmost and longest)
+ match. This can be spectacularly expensive and may run out of stack
+ space, but there really is no better known generic algorithm. Quoting
+ Henry Spencer from comp.compilers:
+
+
+ POSIX.2 REs require longest match, which is really exciting to
+ implement since the obsolete ("basic") variant also includes
+ \. I haven't found a better way of tackling this than doing
+ a preliminary match using a DFA (or simulation) on a modified RE
+ that just replicates subREs for \, and then doing a
+ backtracking match to determine whether the subRE matches were
+ right. This can be rather slow, but I console myself with the
+ thought that people who use \ deserve very slow execution.
+ (Pun unintentional but very appropriate.)
+
+*/
+
+#include "tre.h"
+#include "tre-match-utils.h"
+
+typedef struct
+{
+ int pos;
+ const char *str_byte;
+#ifdef TRE_WCHAR
+ const qse_wchar_t *str_wide;
+#endif /* TRE_WCHAR */
+ tre_tnfa_transition_t *state;
+ int state_id;
+ int next_c;
+ int *tags;
+#ifdef TRE_MBSTATE
+ qse_mbstate_t mbstate;
+#endif /* TRE_MBSTATE */
+} tre_backtrack_item_t;
+
+typedef struct tre_backtrack_struct
+{
+ tre_backtrack_item_t item;
+ struct tre_backtrack_struct *prev;
+ struct tre_backtrack_struct *next;
+} *tre_backtrack_t;
+
+#ifdef TRE_WCHAR
+#define BT_STACK_WIDE_IN(_str_wide) stack->item.str_wide = (_str_wide)
+#define BT_STACK_WIDE_OUT (str_wide) = stack->item.str_wide
+#else /* !TRE_WCHAR */
+#define BT_STACK_WIDE_IN(_str_wide)
+#define BT_STACK_WIDE_OUT
+#endif /* !TRE_WCHAR */
+
+#ifdef TRE_MBSTATE
+#define BT_STACK_MBSTATE_IN stack->item.mbstate = (mbstate)
+#define BT_STACK_MBSTATE_OUT (mbstate) = stack->item.mbstate
+#else /* !TRE_MBSTATE */
+#define BT_STACK_MBSTATE_IN
+#define BT_STACK_MBSTATE_OUT
+#endif /* !TRE_MBSTATE */
+
+#define tre_bt_mem_new tre_mem_new
+#define tre_bt_mem_alloc tre_mem_alloc
+#define tre_bt_mem_destroy tre_mem_destroy
+
+
+#define BT_STACK_PUSH(_mmgr, _pos, _str_byte, _str_wide, _state, _state_id, _next_c, _tags, _mbstate) \
+ do \
+ { \
+ int i; \
+ if (!stack->next) \
+ { \
+ tre_backtrack_t s; \
+ s = tre_bt_mem_alloc(mem, sizeof(*s)); \
+ if (!s) \
+ { \
+ tre_bt_mem_destroy(mem); \
+ if (tags) \
+ xfree(_mmgr,tags); \
+ if (pmatch) \
+ xfree(_mmgr,pmatch); \
+ if (states_seen) \
+ xfree(_mmgr,states_seen); \
+ return REG_ESPACE; \
+ } \
+ s->prev = stack; \
+ s->next = NULL; \
+ s->item.tags = tre_bt_mem_alloc(mem, \
+ sizeof(*tags) * tnfa->num_tags); \
+ if (!s->item.tags) \
+ { \
+ tre_bt_mem_destroy(mem); \
+ if (tags) \
+ xfree(_mmgr,tags); \
+ if (pmatch) \
+ xfree(_mmgr,pmatch); \
+ if (states_seen) \
+ xfree(_mmgr,states_seen); \
+ return REG_ESPACE; \
+ } \
+ stack->next = s; \
+ stack = s; \
+ } \
+ else \
+ stack = stack->next; \
+ stack->item.pos = (_pos); \
+ stack->item.str_byte = (_str_byte); \
+ BT_STACK_WIDE_IN(_str_wide); \
+ stack->item.state = (_state); \
+ stack->item.state_id = (_state_id); \
+ stack->item.next_c = (_next_c); \
+ for (i = 0; i < tnfa->num_tags; i++) \
+ stack->item.tags[i] = (_tags)[i]; \
+ BT_STACK_MBSTATE_IN; \
+ } \
+ while (/*CONSTCOND*/0)
+
+#define BT_STACK_POP() \
+ do \
+ { \
+ int i; \
+ assert(stack->prev); \
+ pos = stack->item.pos; \
+ if (type == STR_USER) \
+ str_source->rewind(pos + pos_add_next, str_source->context); \
+ str_byte = stack->item.str_byte; \
+ BT_STACK_WIDE_OUT; \
+ state = stack->item.state; \
+ next_c = stack->item.next_c; \
+ for (i = 0; i < tnfa->num_tags; i++) \
+ tags[i] = stack->item.tags[i]; \
+ BT_STACK_MBSTATE_OUT; \
+ stack = stack->prev; \
+ } \
+ while (/*CONSTCOND*/0)
+
+#undef MIN
+#define MIN(a, b) ((a) <= (b) ? (a) : (b))
+
+reg_errcode_t
+tre_tnfa_run_backtrack(qse_mmgr_t* mmgr, const tre_tnfa_t *tnfa, const void *string,
+ int len, tre_str_type_t type, int *match_tags,
+ int eflags, int *match_end_ofs)
+{
+ /* State variables required by GET_NEXT_WCHAR. */
+ tre_char_t prev_c = 0, next_c = 0;
+ const char *str_byte = string;
+ int pos = 0;
+ unsigned int pos_add_next = 1;
+#ifdef TRE_WCHAR
+ const qse_wchar_t *str_wide = string;
+#ifdef TRE_MBSTATE
+ qse_mbstate_t mbstate;
+#endif /* TRE_MBSTATE */
+#endif /* TRE_WCHAR */
+ int reg_notbol = eflags & REG_NOTBOL;
+ int reg_noteol = eflags & REG_NOTEOL;
+ int reg_newline = tnfa->cflags & REG_NEWLINE;
+ int str_user_end = 0;
+
+ /* These are used to remember the necessary values of the above
+ variables to return to the position where the current search
+ started from. */
+ int next_c_start;
+ const char *str_byte_start;
+ int pos_start = -1;
+#ifdef TRE_WCHAR
+ const qse_wchar_t *str_wide_start;
+#endif /* TRE_WCHAR */
+#ifdef TRE_MBSTATE
+ qse_mbstate_t mbstate_start;
+#endif /* TRE_MBSTATE */
+
+ /* End offset of best match so far, or -1 if no match found yet. */
+ int match_eo = -1;
+ /* Tag arrays. */
+ int *next_tags, *tags = NULL;
+ /* Current TNFA state. */
+ tre_tnfa_transition_t *state;
+ int *states_seen = NULL;
+
+ /* Memory allocator to for allocating the backtracking stack. */
+ tre_mem_t mem = tre_bt_mem_new(mmgr);
+
+ /* The backtracking stack. */
+ tre_backtrack_t stack;
+
+ tre_tnfa_transition_t *trans_i;
+ regmatch_t *pmatch = NULL;
+ int ret;
+
+#ifdef TRE_MBSTATE
+ QSE_MEMSET(&mbstate, '\0', sizeof(mbstate));
+#endif /* TRE_MBSTATE */
+
+ if (!mem)
+ return REG_ESPACE;
+ stack = tre_bt_mem_alloc(mem, sizeof(*stack));
+ if (!stack)
+ {
+ ret = REG_ESPACE;
+ goto error_exit;
+ }
+ stack->prev = NULL;
+ stack->next = NULL;
+
+ DPRINT(("tnfa_execute_backtrack, input type %d\n", type));
+ DPRINT(("len = %d\n", len));
+
+ if (tnfa->num_tags)
+ {
+ tags = xmalloc(mmgr,sizeof(*tags) * tnfa->num_tags);
+ if (!tags)
+ {
+ ret = REG_ESPACE;
+ goto error_exit;
+ }
+ }
+ if (tnfa->num_submatches)
+ {
+ pmatch = xmalloc(mmgr,sizeof(*pmatch) * tnfa->num_submatches);
+ if (!pmatch)
+ {
+ ret = REG_ESPACE;
+ goto error_exit;
+ }
+ }
+ if (tnfa->num_states)
+ {
+ states_seen = xmalloc(mmgr,sizeof(*states_seen) * tnfa->num_states);
+ if (!states_seen)
+ {
+ ret = REG_ESPACE;
+ goto error_exit;
+ }
+ }
+
+retry:
+ {
+ int i;
+ for (i = 0; i < tnfa->num_tags; i++)
+ {
+ tags[i] = -1;
+ if (match_tags)
+ match_tags[i] = -1;
+ }
+ for (i = 0; i < tnfa->num_states; i++)
+ states_seen[i] = 0;
+ }
+
+ state = NULL;
+ pos = pos_start;
+ if (type == STR_USER)
+ str_source->rewind(pos + pos_add_next, str_source->context);
+ GET_NEXT_WCHAR();
+ pos_start = pos;
+ next_c_start = next_c;
+ str_byte_start = str_byte;
+#ifdef TRE_WCHAR
+ str_wide_start = str_wide;
+#endif /* TRE_WCHAR */
+#ifdef TRE_MBSTATE
+ mbstate_start = mbstate;
+#endif /* TRE_MBSTATE */
+
+ /* Handle initial states. */
+ next_tags = NULL;
+ for (trans_i = tnfa->initial; trans_i->state; trans_i++)
+ {
+ DPRINT(("> init %p, prev_c %lc\n", trans_i->state, (tre_cint_t)prev_c));
+ if (trans_i->assertions && CHECK_ASSERTIONS(trans_i->assertions))
+ {
+ DPRINT(("assert failed\n"));
+ continue;
+ }
+ if (state == NULL)
+ {
+ /* Start from this state. */
+ state = trans_i->state;
+ next_tags = trans_i->tags;
+ }
+ else
+ {
+ /* Backtrack to this state. */
+ DPRINT(("saving state %d for backtracking\n", trans_i->state_id));
+ BT_STACK_PUSH(mmgr, pos, str_byte, str_wide, trans_i->state,
+ trans_i->state_id, next_c, tags, mbstate);
+ {
+ int *tmp = trans_i->tags;
+ if (tmp)
+ while (*tmp >= 0)
+ stack->item.tags[*tmp++] = pos;
+ }
+ }
+ }
+
+ if (next_tags)
+ for (; *next_tags >= 0; next_tags++)
+ tags[*next_tags] = pos;
+
+
+ DPRINT(("entering match loop, pos %d, str_byte %p\n", pos, str_byte));
+ DPRINT(("pos:chr/code | state and tags\n"));
+ DPRINT(("-------------+------------------------------------------------\n"));
+
+ if (state == NULL)
+ goto backtrack;
+
+ while (/*CONSTCOND*/1)
+ {
+ tre_tnfa_transition_t *next_state;
+ int empty_br_match;
+
+ DPRINT(("start loop\n"));
+ if (state == tnfa->final)
+ {
+ DPRINT((" match found, %d %d\n", match_eo, pos));
+ if (match_eo < pos
+ || (match_eo == pos
+ && match_tags
+ && tre_tag_order(tnfa->num_tags, tnfa->tag_directions,
+ tags, match_tags)))
+ {
+ int i;
+ /* This match wins the previous match. */
+ DPRINT((" win previous\n"));
+ match_eo = pos;
+ if (match_tags)
+ for (i = 0; i < tnfa->num_tags; i++)
+ match_tags[i] = tags[i];
+ }
+ /* Our TNFAs never have transitions leaving from the final state,
+ so we jump right to backtracking. */
+ goto backtrack;
+ }
+
+#ifdef TRE_DEBUG
+ DPRINT(("%3d:%2lc/%05d | %p ", pos, (tre_cint_t)next_c, (int)next_c,
+ state));
+ {
+ int i;
+ for (i = 0; i < tnfa->num_tags; i++)
+ DPRINT(("%d%s", tags[i], i < tnfa->num_tags - 1 ? ", " : ""));
+ DPRINT(("\n"));
+ }
+#endif /* TRE_DEBUG */
+
+ /* Go to the next character in the input string. */
+ empty_br_match = 0;
+ trans_i = state;
+ if (trans_i->state && trans_i->assertions & ASSERT_BACKREF)
+ {
+ /* This is a back reference state. All transitions leaving from
+ this state have the same back reference "assertion". Instead
+ of reading the next character, we match the back reference. */
+ int so, eo, bt = trans_i->u.backref;
+ int bt_len;
+ int result;
+
+ DPRINT((" should match back reference %d\n", bt));
+ /* Get the substring we need to match against. Remember to
+ turn off REG_NOSUB temporarily. */
+ tre_fill_pmatch(bt + 1, pmatch, tnfa->cflags & /*LINTED*/!REG_NOSUB,
+ tnfa, tags, pos);
+ so = pmatch[bt].rm_so;
+ eo = pmatch[bt].rm_eo;
+ bt_len = eo - so;
+
+#ifdef TRE_DEBUG
+ {
+ int slen;
+ if (len < 0)
+ slen = bt_len;
+ else
+ slen = MIN(bt_len, len - pos);
+
+ if (type == STR_BYTE)
+ {
+ DPRINT((" substring (len %d) is [%d, %d[: '%.*s'\n",
+ bt_len, so, eo, bt_len, (char*)string + so));
+ DPRINT((" current string is '%.*s'\n", slen, str_byte - 1));
+ }
+#ifdef TRE_WCHAR
+ else if (type == STR_WIDE)
+ {
+ DPRINT((" substring (len %d) is [%d, %d[: '%.*" STRF "'\n",
+ bt_len, so, eo, bt_len, (qse_wchar_t*)string + so));
+ DPRINT((" current string is '%.*" STRF "'\n",
+ slen, str_wide - 1));
+ }
+#endif /* TRE_WCHAR */
+ }
+#endif
+
+ if (len < 0)
+ {
+ if (type == STR_USER)
+ result = str_source->compare((unsigned)so, (unsigned)pos,
+ (unsigned)bt_len,
+ str_source->context);
+#ifdef TRE_WCHAR
+ else if (type == STR_WIDE)
+ result = qse_wcszcmp((const qse_wchar_t*)string + so, str_wide - 1, (size_t)bt_len);
+#endif /* TRE_WCHAR */
+ else
+ result = qse_mbszcmp((const char*)string + so, str_byte - 1, (size_t)bt_len);
+ }
+ else if (len - pos < bt_len)
+ result = 1;
+#ifdef TRE_WCHAR
+ else if (type == STR_WIDE)
+ {
+ /*result = wmemcmp((const qse_wchar_t*)string + so, str_wide - 1, (size_t)bt_len);*/
+ result = qse_wcsxncmp((const qse_wchar_t*)string + so, (size_t)bt_len, str_wide - 1, (size_t)bt_len);
+ }
+#endif /* TRE_WCHAR */
+ else
+ {
+ /*result = memcmp((const char*)string + so, str_byte - 1, (size_t)bt_len); */
+ result = qse_mbsxncmp((const char*)string + so, (size_t)bt_len, str_byte - 1, (size_t)bt_len);
+ }
+
+
+ if (result == 0)
+ {
+ /* Back reference matched. Check for infinite loop. */
+ if (bt_len == 0)
+ empty_br_match = 1;
+ if (empty_br_match && states_seen[trans_i->state_id])
+ {
+ DPRINT((" avoid loop\n"));
+ goto backtrack;
+ }
+
+ states_seen[trans_i->state_id] = empty_br_match;
+
+ /* Advance in input string and resync `prev_c', `next_c'
+ and pos. */
+ DPRINT((" back reference matched\n"));
+ str_byte += bt_len - 1;
+#ifdef TRE_WCHAR
+ str_wide += bt_len - 1;
+#endif /* TRE_WCHAR */
+ pos += bt_len - 1;
+ GET_NEXT_WCHAR();
+ DPRINT((" pos now %d\n", pos));
+ }
+ else
+ {
+ DPRINT((" back reference did not match\n"));
+ goto backtrack;
+ }
+ }
+ else
+ {
+ /* Check for end of string. */
+ if (len < 0)
+ {
+ if (type == STR_USER)
+ {
+ if (str_user_end)
+ goto backtrack;
+ }
+ else if (next_c == QSE_T('\0'))
+ goto backtrack;
+ }
+ else
+ {
+ if (pos >= len)
+ goto backtrack;
+ }
+
+ /* Read the next character. */
+ GET_NEXT_WCHAR();
+ }
+
+ next_state = NULL;
+ for (trans_i = state; trans_i->state; trans_i++)
+ {
+ DPRINT((" transition %d-%d (%c-%c) %d to %d\n",
+ trans_i->code_min, trans_i->code_max,
+ trans_i->code_min, trans_i->code_max,
+ trans_i->assertions, trans_i->state_id));
+ if (trans_i->code_min <= (tre_cint_t)prev_c
+ && trans_i->code_max >= (tre_cint_t)prev_c)
+ {
+ if (trans_i->assertions
+ && (CHECK_ASSERTIONS(trans_i->assertions)
+ || CHECK_CHAR_CLASSES(trans_i, tnfa, eflags)))
+ {
+ DPRINT((" assertion failed\n"));
+ continue;
+ }
+
+ if (next_state == NULL)
+ {
+ /* First matching transition. */
+ DPRINT((" Next state is %d\n", trans_i->state_id));
+ next_state = trans_i->state;
+ next_tags = trans_i->tags;
+ }
+ else
+ {
+ /* Second matching transition. We may need to backtrack here
+ to take this transition instead of the first one, so we
+ push this transition in the backtracking stack so we can
+ jump back here if needed. */
+ DPRINT((" saving state %d for backtracking\n",
+ trans_i->state_id));
+ BT_STACK_PUSH(mmgr, pos, str_byte, str_wide, trans_i->state,
+ trans_i->state_id, next_c, tags, mbstate);
+ {
+ int *tmp;
+ for (tmp = trans_i->tags; tmp && *tmp >= 0; tmp++)
+ stack->item.tags[*tmp] = pos;
+ }
+#if 0 /* XXX - it's important not to look at all transitions here to keep
+ the stack small! */
+ break;
+#endif
+ }
+ }
+ }
+
+ if (next_state != NULL)
+ {
+ /* Matching transitions were found. Take the first one. */
+ state = next_state;
+
+ /* Update the tag values. */
+ if (next_tags)
+ while (*next_tags >= 0)
+ tags[*next_tags++] = pos;
+ }
+ else
+ {
+backtrack:
+ /* A matching transition was not found. Try to backtrack. */
+ if (stack->prev)
+ {
+ DPRINT((" backtracking\n"));
+ if (stack->item.state->assertions && ASSERT_BACKREF)
+ {
+ DPRINT((" states_seen[%d] = 0\n",
+ stack->item.state_id));
+ states_seen[stack->item.state_id] = 0;
+ }
+
+ BT_STACK_POP();
+ }
+ else if (match_eo < 0)
+ {
+ /* Try starting from a later position in the input string. */
+ /* Check for end of string. */
+ if (len < 0)
+ {
+ if (next_c == QSE_T('\0'))
+ {
+ DPRINT(("end of string.\n"));
+ break;
+ }
+ }
+ else
+ {
+ if (pos >= len)
+ {
+ DPRINT(("end of string.\n"));
+ break;
+ }
+ }
+ DPRINT(("restarting from next start position\n"));
+ next_c = next_c_start;
+#ifdef TRE_MBSTATE
+ mbstate = mbstate_start;
+#endif /* TRE_MBSTATE */
+ str_byte = str_byte_start;
+#ifdef TRE_WCHAR
+ str_wide = str_wide_start;
+#endif /* TRE_WCHAR */
+ goto retry;
+ }
+ else
+ {
+ DPRINT(("finished\n"));
+ break;
+ }
+ }
+ }
+
+ ret = match_eo >= 0 ? REG_OK : REG_NOMATCH;
+ *match_end_ofs = match_eo;
+
+error_exit:
+ tre_bt_mem_destroy(mem);
+ if (tags) xfree(mmgr,tags);
+ if (pmatch) xfree(mmgr,pmatch);
+ if (states_seen) xfree(mmgr,states_seen);
+
+ return ret;
+}
diff --git a/qse/lib/cmn/tre-match-parallel.c b/qse/lib/cmn/tre-match-parallel.c
new file mode 100644
index 00000000..321bf748
--- /dev/null
+++ b/qse/lib/cmn/tre-match-parallel.c
@@ -0,0 +1,496 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-match-parallel.c - TRE parallel regex matching engine
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+ This algorithm searches for matches basically by reading characters
+ in the searched string one by one, starting at the beginning. All
+ matching paths in the TNFA are traversed in parallel. When two or
+ more paths reach the same state, exactly one is chosen according to
+ tag ordering rules; if returning submatches is not required it does
+ not matter which path is chosen.
+
+ The worst case time required for finding the leftmost and longest
+ match, or determining that there is no match, is always linearly
+ dependent on the length of the text being searched.
+
+ This algorithm cannot handle TNFAs with back referencing nodes.
+ See `tre-match-backtrack.c'.
+*/
+
+#include "tre.h"
+#include "tre-match-utils.h"
+
+typedef struct
+{
+ tre_tnfa_transition_t *state;
+ int *tags;
+} tre_tnfa_reach_t;
+
+typedef struct
+{
+ int pos;
+ int **tags;
+} tre_reach_pos_t;
+
+
+#ifdef TRE_DEBUG
+static void
+tre_print_reach(const tre_tnfa_t *tnfa, tre_tnfa_reach_t *reach, int num_tags)
+{
+ int i;
+
+ while (reach->state != NULL)
+ {
+ DPRINT((" %p", (void *)reach->state));
+ if (num_tags > 0)
+ {
+ DPRINT(("/"));
+ for (i = 0; i < num_tags; i++)
+ {
+ DPRINT(("%d:%d", i, reach->tags[i]));
+ if (i < (num_tags-1))
+ DPRINT((","));
+ }
+ }
+ reach++;
+ }
+ DPRINT(("\n"));
+
+}
+#endif /* TRE_DEBUG */
+
+reg_errcode_t
+tre_tnfa_run_parallel(qse_mmgr_t* mmgr, const tre_tnfa_t *tnfa, const void *string, int len,
+ tre_str_type_t type, int *match_tags, int eflags,
+ int *match_end_ofs)
+{
+ /* State variables required by GET_NEXT_WCHAR. */
+ tre_char_t prev_c = 0, next_c = 0;
+ const char *str_byte = string;
+ int pos = -1;
+ unsigned int pos_add_next = 1;
+#ifdef TRE_WCHAR
+ const qse_wchar_t *str_wide = string;
+#ifdef TRE_MBSTATE
+ qse_mbstate_t mbstate;
+#endif
+#endif /* TRE_WCHAR */
+ int reg_notbol = eflags & REG_NOTBOL;
+ int reg_noteol = eflags & REG_NOTEOL;
+ int reg_newline = tnfa->cflags & REG_NEWLINE;
+ int str_user_end = 0;
+
+ char *buf;
+ tre_tnfa_transition_t *trans_i;
+ tre_tnfa_reach_t *reach, *reach_next, *reach_i, *reach_next_i;
+ tre_reach_pos_t *reach_pos;
+ int *tag_i;
+ int num_tags, i;
+
+ int match_eo = -1; /* end offset of match (-1 if no match found yet) */
+ int new_match = 0;
+ int *tmp_tags = NULL;
+ int *tmp_iptr;
+
+#ifdef TRE_MBSTATE
+ QSE_MEMSET(&mbstate, '\0', sizeof(mbstate));
+#endif /* TRE_MBSTATE */
+
+ DPRINT(("tre_tnfa_run_parallel, input type %d\n", type));
+
+ if (!match_tags)
+ num_tags = 0;
+ else
+ num_tags = tnfa->num_tags;
+
+ /* Allocate memory for temporary data required for matching. This needs to
+ be done for every matching operation to be thread safe. This allocates
+ everything in a single large block from the stack frame using alloca()
+ or with malloc() if alloca is unavailable. */
+ {
+ int tbytes, rbytes, pbytes, xbytes, total_bytes;
+ char *tmp_buf;
+ /* Compute the length of the block we need. */
+ tbytes = sizeof(*tmp_tags) * num_tags;
+ rbytes = sizeof(*reach_next) * (tnfa->num_states + 1);
+ pbytes = sizeof(*reach_pos) * tnfa->num_states;
+ xbytes = sizeof(int) * num_tags;
+ total_bytes =
+ (sizeof(long) - 1) * 4 /* for alignment paddings */
+ + (rbytes + xbytes * tnfa->num_states) * 2 + tbytes + pbytes;
+
+ /* Allocate the memory. */
+ buf = xmalloc(mmgr, (unsigned)total_bytes);
+ if (buf == NULL) return REG_ESPACE;
+ QSE_MEMSET(buf, 0, (size_t)total_bytes);
+
+ /* Get the various pointers within tmp_buf (properly aligned). */
+ tmp_tags = (void *)buf;
+ tmp_buf = buf + tbytes;
+ tmp_buf += ALIGN(tmp_buf, long);
+ reach_next = (void *)tmp_buf;
+ tmp_buf += rbytes;
+ tmp_buf += ALIGN(tmp_buf, long);
+ reach = (void *)tmp_buf;
+ tmp_buf += rbytes;
+ tmp_buf += ALIGN(tmp_buf, long);
+ reach_pos = (void *)tmp_buf;
+ tmp_buf += pbytes;
+ tmp_buf += ALIGN(tmp_buf, long);
+ for (i = 0; i < tnfa->num_states; i++)
+ {
+ reach[i].tags = (void *)tmp_buf;
+ tmp_buf += xbytes;
+ reach_next[i].tags = (void *)tmp_buf;
+ tmp_buf += xbytes;
+ }
+ }
+
+ for (i = 0; i < tnfa->num_states; i++)
+ reach_pos[i].pos = -1;
+
+ /* If only one character can start a match, find it first. */
+ if (tnfa->first_char >= 0 && type == STR_BYTE && str_byte)
+ {
+ const char *orig_str = str_byte;
+ int first = tnfa->first_char;
+
+ if (len >= 0)
+ str_byte = QSE_MEMBYTE(orig_str, first, (size_t)len);
+ else
+ str_byte = qse_mbschr(orig_str, first);
+ if (str_byte == NULL)
+ {
+ if (buf) xfree(mmgr, buf);
+ return REG_NOMATCH;
+ }
+ DPRINT(("skipped %lu chars\n", (unsigned long)(str_byte - orig_str)));
+ if (str_byte >= orig_str + 1)
+ prev_c = (unsigned char)*(str_byte - 1);
+ next_c = (unsigned char)*str_byte;
+ pos = str_byte - orig_str;
+ if (len < 0 || pos < len)
+ str_byte++;
+ }
+ else
+ {
+ GET_NEXT_WCHAR();
+ pos = 0;
+ }
+
+#if 0
+ /* Skip over characters that cannot possibly be the first character
+ of a match. */
+ if (tnfa->firstpos_chars != NULL)
+ {
+ char *chars = tnfa->firstpos_chars;
+
+ if (len < 0)
+ {
+ const char *orig_str = str_byte;
+ /* XXX - use strpbrk() and wcspbrk() because they might be
+ optimized for the target architecture. Try also strcspn()
+ and wcscspn() and compare the speeds. */
+ while (next_c != QSE_T('\0') && !chars[next_c])
+ {
+ next_c = *str_byte++;
+ }
+ prev_c = *(str_byte - 2);
+ pos += str_byte - orig_str;
+ DPRINT(("skipped %d chars\n", str_byte - orig_str));
+ }
+ else
+ {
+ while (pos <= len && !chars[next_c])
+ {
+ prev_c = next_c;
+ next_c = (unsigned char)(*str_byte++);
+ pos++;
+ }
+ }
+ }
+#endif
+
+ DPRINT(("length: %d\n", len));
+ DPRINT(("pos:chr/code | states and tags\n"));
+ DPRINT(("-------------+------------------------------------------------\n"));
+
+ reach_next_i = reach_next;
+ while (/*CONSTCOND*/1)
+ {
+ /* If no match found yet, add the initial states to `reach_next'. */
+ if (match_eo < 0)
+ {
+ DPRINT((" init >"));
+ trans_i = tnfa->initial;
+ while (trans_i->state != NULL)
+ {
+ if (reach_pos[trans_i->state_id].pos < pos)
+ {
+ if (trans_i->assertions
+ && CHECK_ASSERTIONS(trans_i->assertions))
+ {
+ DPRINT(("assertion failed\n"));
+ trans_i++;
+ continue;
+ }
+
+ DPRINT((" %p", (void *)trans_i->state));
+ reach_next_i->state = trans_i->state;
+ for (i = 0; i < num_tags; i++)
+ reach_next_i->tags[i] = -1;
+ tag_i = trans_i->tags;
+ if (tag_i)
+ while (*tag_i >= 0)
+ {
+ if (*tag_i < num_tags)
+ reach_next_i->tags[*tag_i] = pos;
+ tag_i++;
+ }
+ if (reach_next_i->state == tnfa->final)
+ {
+ DPRINT((" found empty match\n"));
+ match_eo = pos;
+ new_match = 1;
+ for (i = 0; i < num_tags; i++)
+ match_tags[i] = reach_next_i->tags[i];
+ }
+ reach_pos[trans_i->state_id].pos = pos;
+ reach_pos[trans_i->state_id].tags = &reach_next_i->tags;
+ reach_next_i++;
+ }
+ trans_i++;
+ }
+ DPRINT(("\n"));
+ reach_next_i->state = NULL;
+ }
+ else
+ {
+ if (num_tags == 0 || reach_next_i == reach_next)
+ /* We have found a match. */
+ break;
+ }
+
+ /* Check for end of string. */
+ if (len < 0)
+ {
+ if (type == STR_USER)
+ {
+ if (str_user_end)
+ break;
+ }
+ else if (next_c == QSE_T('\0'))
+ break;
+ }
+ else
+ {
+ if (pos >= len)
+ break;
+ }
+
+ GET_NEXT_WCHAR();
+
+#ifdef TRE_DEBUG
+ DPRINT(("%3d:%2lc/%05d |", pos - 1, (tre_cint_t)prev_c, (int)prev_c));
+ tre_print_reach(tnfa, reach_next, num_tags);
+ DPRINT(("%3d:%2lc/%05d |", pos, (tre_cint_t)next_c, (int)next_c));
+ tre_print_reach(tnfa, reach_next, num_tags);
+#endif /* TRE_DEBUG */
+
+ /* Swap `reach' and `reach_next'. */
+ reach_i = reach;
+ reach = reach_next;
+ reach_next = reach_i;
+
+ /* For each state in `reach', weed out states that don't fulfill the
+ minimal matching conditions. */
+ if (tnfa->num_minimals && new_match)
+ {
+ new_match = 0;
+ reach_next_i = reach_next;
+ for (reach_i = reach; reach_i->state; reach_i++)
+ {
+ int skip = 0;
+ for (i = 0; tnfa->minimal_tags[i] >= 0; i += 2)
+ {
+ int end = tnfa->minimal_tags[i];
+ int start = tnfa->minimal_tags[i + 1];
+ DPRINT((" Minimal start %d, end %d\n", start, end));
+ if (end >= num_tags)
+ {
+ DPRINT((" Throwing %p out.\n", reach_i->state));
+ skip = 1;
+ break;
+ }
+ else if (reach_i->tags[start] == match_tags[start]
+ && reach_i->tags[end] < match_tags[end])
+ {
+ DPRINT((" Throwing %p out because t%d < %d\n",
+ reach_i->state, end, match_tags[end]));
+ skip = 1;
+ break;
+ }
+ }
+ if (!skip)
+ {
+ reach_next_i->state = reach_i->state;
+ tmp_iptr = reach_next_i->tags;
+ reach_next_i->tags = reach_i->tags;
+ reach_i->tags = tmp_iptr;
+ reach_next_i++;
+ }
+ }
+ reach_next_i->state = NULL;
+
+ /* Swap `reach' and `reach_next'. */
+ reach_i = reach;
+ reach = reach_next;
+ reach_next = reach_i;
+ }
+
+ /* For each state in `reach' see if there is a transition leaving with
+ the current input symbol to a state not yet in `reach_next', and
+ add the destination states to `reach_next'. */
+ reach_next_i = reach_next;
+ for (reach_i = reach; reach_i->state; reach_i++)
+ {
+ for (trans_i = reach_i->state; trans_i->state; trans_i++)
+ {
+ /* Does this transition match the input symbol? */
+ if (trans_i->code_min <= (tre_cint_t)prev_c &&
+ trans_i->code_max >= (tre_cint_t)prev_c)
+ {
+ if (trans_i->assertions
+ && (CHECK_ASSERTIONS(trans_i->assertions)
+ || CHECK_CHAR_CLASSES(trans_i, tnfa, eflags)))
+ {
+ DPRINT(("assertion failed\n"));
+ continue;
+ }
+
+ /* Compute the tags after this transition. */
+ for (i = 0; i < num_tags; i++)
+ tmp_tags[i] = reach_i->tags[i];
+ tag_i = trans_i->tags;
+ if (tag_i != NULL)
+ while (*tag_i >= 0)
+ {
+ if (*tag_i < num_tags)
+ tmp_tags[*tag_i] = pos;
+ tag_i++;
+ }
+
+ if (reach_pos[trans_i->state_id].pos < pos)
+ {
+ /* Found an unvisited node. */
+ reach_next_i->state = trans_i->state;
+ tmp_iptr = reach_next_i->tags;
+ reach_next_i->tags = tmp_tags;
+ tmp_tags = tmp_iptr;
+ reach_pos[trans_i->state_id].pos = pos;
+ reach_pos[trans_i->state_id].tags = &reach_next_i->tags;
+
+ if (reach_next_i->state == tnfa->final
+ && (match_eo == -1
+ || (num_tags > 0
+ && reach_next_i->tags[0] <= match_tags[0])))
+ {
+ DPRINT((" found match %p\n", trans_i->state));
+ match_eo = pos;
+ new_match = 1;
+ for (i = 0; i < num_tags; i++)
+ match_tags[i] = reach_next_i->tags[i];
+ }
+ reach_next_i++;
+
+ }
+ else
+ {
+ assert(reach_pos[trans_i->state_id].pos == pos);
+ /* Another path has also reached this state. We choose
+ the winner by examining the tag values for both
+ paths. */
+ if (tre_tag_order(num_tags, tnfa->tag_directions,
+ tmp_tags,
+ *reach_pos[trans_i->state_id].tags))
+ {
+ /* The new path wins. */
+ tmp_iptr = *reach_pos[trans_i->state_id].tags;
+ *reach_pos[trans_i->state_id].tags = tmp_tags;
+ if (trans_i->state == tnfa->final)
+ {
+ DPRINT((" found better match\n"));
+ match_eo = pos;
+ new_match = 1;
+ for (i = 0; i < num_tags; i++)
+ match_tags[i] = tmp_tags[i];
+ }
+ tmp_tags = tmp_iptr;
+ }
+ }
+ }
+ }
+ }
+ reach_next_i->state = NULL;
+ }
+
+ DPRINT(("match end offset = %d\n", match_eo));
+
+ if (buf) xfree(mmgr, buf);
+
+ *match_end_ofs = match_eo;
+ return match_eo >= 0 ? REG_OK : REG_NOMATCH;
+}
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-match-utils.h b/qse/lib/cmn/tre-match-utils.h
new file mode 100644
index 00000000..fe3c7f74
--- /dev/null
+++ b/qse/lib/cmn/tre-match-utils.h
@@ -0,0 +1,261 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-match-utils.h - TRE matcher helper definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define str_source ((const tre_str_source*)string)
+
+#ifdef TRE_WCHAR
+
+#ifdef TRE_MULTIBYTE
+
+/* Wide character and multibyte support. */
+
+#define GET_NEXT_WCHAR() \
+ do { \
+ prev_c = next_c; \
+ if (type == STR_BYTE) \
+ { \
+ pos++; \
+ if (len >= 0 && pos >= len) \
+ next_c = '\0'; \
+ else \
+ next_c = (unsigned char)(*str_byte++); \
+ } \
+ else if (type == STR_WIDE) \
+ { \
+ pos++; \
+ if (len >= 0 && pos >= len) \
+ next_c = QSE_T('\0'); \
+ else \
+ next_c = *str_wide++; \
+ } \
+ else if (type == STR_MBS) \
+ { \
+ pos += pos_add_next; \
+ if (str_byte == NULL) \
+ next_c = QSE_T('\0'); \
+ else \
+ { \
+ size_t w; \
+ int max; \
+ if (len >= 0) \
+ max = len - pos; \
+ else \
+ max = 32; \
+ if (max <= 0) \
+ { \
+ next_c = QSE_T('\0'); \
+ pos_add_next = 1; \
+ } \
+ else \
+ { \
+ w = qse_mbrtowc(str_byte, (size_t)max, &next_c, &mbstate); \
+ if (w <= 0 || w > max) \
+ return REG_NOMATCH; \
+ if (next_c == QSE_T('\0') && len >= 0) \
+ { \
+ pos_add_next = 1; \
+ next_c = 0; \
+ str_byte++; \
+ } \
+ else \
+ { \
+ pos_add_next = w; \
+ str_byte += w; \
+ } \
+ } \
+ } \
+ } \
+ else if (type == STR_USER) \
+ { \
+ pos += pos_add_next; \
+ str_user_end = str_source->get_next_char(&next_c, &pos_add_next, \
+ str_source->context); \
+ } \
+ } while(/*CONSTCOND*/0)
+
+#else /* !TRE_MULTIBYTE */
+
+/* Wide character support, no multibyte support. */
+
+#define GET_NEXT_WCHAR() \
+ do { \
+ prev_c = next_c; \
+ if (type == STR_BYTE) \
+ { \
+ pos++; \
+ if (len >= 0 && pos >= len) \
+ next_c = '\0'; \
+ else \
+ next_c = (unsigned char)(*str_byte++); \
+ } \
+ else if (type == STR_WIDE) \
+ { \
+ pos++; \
+ if (len >= 0 && pos >= len) \
+ next_c = QSE_T('\0'); \
+ else \
+ next_c = *str_wide++; \
+ } \
+ else if (type == STR_USER) \
+ { \
+ pos += pos_add_next; \
+ str_user_end = str_source->get_next_char(&next_c, &pos_add_next, \
+ str_source->context); \
+ } \
+ } while(/*CONSTCOND*/0)
+
+#endif /* !TRE_MULTIBYTE */
+
+#else /* !TRE_WCHAR */
+
+/* No wide character or multibyte support. */
+
+#define GET_NEXT_WCHAR() \
+ do { \
+ prev_c = next_c; \
+ if (type == STR_BYTE) \
+ { \
+ pos++; \
+ if (len >= 0 && pos >= len) \
+ next_c = '\0'; \
+ else \
+ next_c = (unsigned char)(*str_byte++); \
+ } \
+ else if (type == STR_USER) \
+ { \
+ pos += pos_add_next; \
+ str_user_end = str_source->get_next_char(&next_c, &pos_add_next, \
+ str_source->context); \
+ } \
+ } while(/*CONSTCOND*/0)
+
+#endif /* !TRE_WCHAR */
+
+
+
+#define IS_WORD_CHAR(c) ((c) == QSE_T('_') || tre_isalnum(c))
+
+#define CHECK_ASSERTIONS(assertions) \
+ (((assertions & ASSERT_AT_BOL) \
+ && (pos > 0 || reg_notbol) \
+ && (prev_c != QSE_T('\n') || !reg_newline)) \
+ || ((assertions & ASSERT_AT_EOL) \
+ && (next_c != QSE_T('\0') || reg_noteol) \
+ && (next_c != QSE_T('\n') || !reg_newline)) \
+ || ((assertions & ASSERT_AT_BOW) \
+ && (IS_WORD_CHAR(prev_c) || !IS_WORD_CHAR(next_c))) \
+ || ((assertions & ASSERT_AT_EOW) \
+ && (!IS_WORD_CHAR(prev_c) || IS_WORD_CHAR(next_c))) \
+ || ((assertions & ASSERT_AT_WB) \
+ && (pos != 0 && next_c != QSE_T('\0') \
+ && IS_WORD_CHAR(prev_c) == IS_WORD_CHAR(next_c))) \
+ || ((assertions & ASSERT_AT_WB_NEG) \
+ && (pos == 0 || next_c == QSE_T('\0') \
+ || IS_WORD_CHAR(prev_c) != IS_WORD_CHAR(next_c))))
+
+#define CHECK_CHAR_CLASSES(trans_i, tnfa, eflags) \
+ (((trans_i->assertions & ASSERT_CHAR_CLASS) \
+ && !(tnfa->cflags & REG_ICASE) \
+ && !tre_isctype((tre_cint_t)prev_c, trans_i->u.class)) \
+ || ((trans_i->assertions & ASSERT_CHAR_CLASS) \
+ && (tnfa->cflags & REG_ICASE) \
+ && !tre_isctype(tre_tolower((tre_cint_t)prev_c),trans_i->u.class) \
+ && !tre_isctype(tre_toupper((tre_cint_t)prev_c),trans_i->u.class)) \
+ || ((trans_i->assertions & ASSERT_CHAR_CLASS_NEG) \
+ && tre_neg_char_classes_match(trans_i->neg_classes,(tre_cint_t)prev_c,\
+ tnfa->cflags & REG_ICASE)))
+
+
+
+
+/* Returns 1 if `t1' wins `t2', 0 otherwise. */
+QSE_INLINE static int
+tre_tag_order(int num_tags, tre_tag_direction_t *tag_directions,
+ int *t1, int *t2)
+{
+ int i;
+ for (i = 0; i < num_tags; i++)
+ {
+ if (tag_directions[i] == TRE_TAG_MINIMIZE)
+ {
+ if (t1[i] < t2[i])
+ return 1;
+ if (t1[i] > t2[i])
+ return 0;
+ }
+ else
+ {
+ if (t1[i] > t2[i])
+ return 1;
+ if (t1[i] < t2[i])
+ return 0;
+ }
+ }
+ /* assert(0);*/
+ return 0;
+}
+
+QSE_INLINE static int
+tre_neg_char_classes_match(tre_ctype_t *classes, tre_cint_t wc, int icase)
+{
+ DPRINT(("neg_char_classes_test: %p, %d, %d\n", classes, wc, icase));
+ while (*classes != (tre_ctype_t)0)
+ if ((!icase && tre_isctype(wc, *classes))
+ || (icase && (tre_isctype(tre_toupper(wc), *classes)
+ || tre_isctype(tre_tolower(wc), *classes))))
+ return 1; /* Match. */
+ else
+ classes++;
+ return 0; /* No match. */
+}
diff --git a/qse/lib/cmn/tre-parse.c b/qse/lib/cmn/tre-parse.c
new file mode 100644
index 00000000..d46bb72c
--- /dev/null
+++ b/qse/lib/cmn/tre-parse.c
@@ -0,0 +1,1837 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-parse.c - Regexp parser
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+ This parser is just a simple recursive descent parser for POSIX.2
+ regexps. The parser supports both the obsolete default syntax and
+ the "extended" syntax, and some nonstandard extensions.
+*/
+
+
+#include "tre.h"
+#include
+#include
+
+#include "tre-ast.h"
+#include "tre-stack.h"
+#include "tre-parse.h"
+
+/* Characters with special meanings in regexp syntax. */
+#define CHAR_PIPE QSE_T('|')
+#define CHAR_LPAREN QSE_T('(')
+#define CHAR_RPAREN QSE_T(')')
+#define CHAR_LBRACE QSE_T('{')
+#define CHAR_RBRACE QSE_T('}')
+#define CHAR_LBRACKET QSE_T('[')
+#define CHAR_RBRACKET QSE_T(']')
+#define CHAR_MINUS QSE_T('-')
+#define CHAR_STAR QSE_T('*')
+#define CHAR_QUESTIONMARK QSE_T('?')
+#define CHAR_PLUS QSE_T('+')
+#define CHAR_PERIOD QSE_T('.')
+#define CHAR_COLON QSE_T(':')
+#define CHAR_EQUAL QSE_T('=')
+#define CHAR_COMMA QSE_T(',')
+#define CHAR_CARET QSE_T('^')
+#define CHAR_DOLLAR QSE_T('$')
+#define CHAR_BACKSLASH QSE_T('\\')
+#define CHAR_HASH QSE_T('#')
+#define CHAR_TILDE QSE_T('~')
+
+
+/* Some macros for expanding \w, \s, etc. */
+static const struct tre_macro_struct
+{
+ const char c;
+ const char *expansion;
+} tre_macros[] =
+{
+ {'t', "\t"}, {'n', "\n"}, {'r', "\r"},
+ {'f', "\f"}, {'a', "\a"}, {'e', "\033"},
+ {'w', "[[:alnum:]_]"}, {'W', "[^[:alnum:]_]"}, {'s', "[[:space:]]"},
+ {'S', "[^[:space:]]"}, {'d', "[[:digit:]]"}, {'D', "[^[:digit:]]"},
+ { 0, NULL }
+};
+
+static QSE_INLINE int xdigit_to_num (qse_char_t c)
+{
+ return (c >= QSE_T('0') && c <= QSE_T('9'))? (c - QSE_T('0')):
+ (c >= QSE_T('A') && c <= QSE_T('F'))? (c - QSE_T('A') + 10):
+ (c >= QSE_T('a') && c <= QSE_T('f'))? (c - QSE_T('a') + 10): -1;
+}
+
+/* Expands a macro delimited by `regex' and `regex_end' to `buf', which
+ must have at least `len' items. Sets buf[0] to zero if the there
+ is no match in `tre_macros'. */
+static void
+tre_expand_macro(const tre_char_t *regex, const tre_char_t *regex_end,
+ tre_char_t *buf, size_t buf_len)
+{
+ int i;
+
+ buf[0] = 0;
+ if (regex >= regex_end)
+ return;
+
+ for (i = 0; tre_macros[i].expansion; i++)
+ {
+ if (tre_macros[i].c == *regex)
+ {
+ unsigned int j;
+ DPRINT(("Expanding macro '%c' => '%s'\n",
+ tre_macros[i].c, tre_macros[i].expansion));
+ for (j = 0; tre_macros[i].expansion[j] && j < buf_len; j++)
+ buf[j] = tre_macros[i].expansion[j];
+ buf[j] = 0;
+ break;
+ }
+ }
+}
+
+static reg_errcode_t
+tre_new_item(tre_mem_t mem, int min, int max, int *i, int *max_i,
+ tre_ast_node_t ***items)
+{
+ reg_errcode_t status;
+ tre_ast_node_t **array = *items;
+ /* Allocate more space if necessary. */
+ if (*i >= *max_i)
+ {
+ tre_ast_node_t **new_items;
+ DPRINT(("out of array space, i = %d\n", *i));
+ /* If the array is already 1024 items large, give up -- there's
+ probably an error in the regexp (e.g. not a '\0' terminated
+ string and missing ']') */
+ if (*max_i > 1024)
+ return REG_ESPACE;
+ *max_i *= 2;
+ new_items = xrealloc(mem->mmgr, array, sizeof(*items) * *max_i);
+ if (new_items == NULL)
+ return REG_ESPACE;
+ *items = array = new_items;
+ }
+ array[*i] = tre_ast_new_literal(mem, min, max, -1);
+ status = array[*i] == NULL ? REG_ESPACE : REG_OK;
+ (*i)++;
+ return status;
+}
+
+
+/* Expands a character class to character ranges. */
+static reg_errcode_t
+tre_expand_ctype(tre_mem_t mem, tre_ctype_t class, tre_ast_node_t ***items,
+ int *i, int *max_i, int cflags)
+{
+ reg_errcode_t status = REG_OK;
+ tre_cint_t c;
+ int j, min = -1, max = 0;
+ assert(TRE_MB_CUR_MAX == 1);
+
+ DPRINT((" expanding class to character ranges\n"));
+ for (j = 0; (j < 256) && (status == REG_OK); j++)
+ {
+ c = j;
+ if (tre_isctype(c, class) ||
+ ((cflags & REG_ICASE) && (tre_isctype(tre_tolower(c), class) ||
+ tre_isctype(tre_toupper(c), class))))
+ {
+ if (min < 0) min = c;
+ max = c;
+ }
+ else if (min >= 0)
+ {
+ DPRINT((" range %c (%d) to %c (%d)\n", min, min, max, max));
+ status = tre_new_item(mem, min, max, i, max_i, items);
+ min = -1;
+ }
+ }
+ if (min >= 0 && status == REG_OK)
+ status = tre_new_item(mem, min, max, i, max_i, items);
+ return status;
+}
+
+
+static int
+tre_compare_items(const void *a, const void *b, void* ctx)
+{
+ const tre_ast_node_t *node_a = *(tre_ast_node_t * const *)a;
+ const tre_ast_node_t *node_b = *(tre_ast_node_t * const *)b;
+ tre_literal_t *l_a = node_a->obj, *l_b = node_b->obj;
+ int a_min = l_a->code_min, b_min = l_b->code_min;
+
+ if (a_min < b_min)
+ return -1;
+ else if (a_min > b_min)
+ return 1;
+ else
+ return 0;
+}
+
+#if 0
+#ifndef TRE_USE_SYSTEM_WCTYPE
+/* isalnum() and the rest may be macros, so wrap them to functions. */
+int tre_isalnum_func(tre_cint_t c)
+{
+ return tre_isalnum(c);
+}
+int tre_isalpha_func(tre_cint_t c)
+{
+ return tre_isalpha(c);
+}
+int tre_isascii_func(tre_cint_t c)
+{
+ return !(c >> 7);
+}
+int tre_isblank_func(tre_cint_t c)
+{
+ return tre_isblank(c);
+}
+int tre_iscntrl_func(tre_cint_t c)
+{
+ return tre_iscntrl(c);
+}
+int tre_isdigit_func(tre_cint_t c)
+{
+ return tre_isdigit(c);
+}
+int tre_isgraph_func(tre_cint_t c)
+{
+ return tre_isgraph(c);
+}
+int tre_islower_func(tre_cint_t c)
+{
+ return tre_islower(c);
+}
+int tre_isprint_func(tre_cint_t c)
+{
+ return tre_isprint(c);
+}
+int tre_ispunct_func(tre_cint_t c)
+{
+ return tre_ispunct(c);
+}
+int tre_isspace_func(tre_cint_t c)
+{
+ return tre_isspace(c);
+}
+int tre_isupper_func(tre_cint_t c)
+{
+ return tre_isupper(c);
+}
+int tre_isxdigit_func(tre_cint_t c)
+{
+ return tre_isxdigit(c);
+}
+
+struct
+{
+ char *name;
+ int (*func)(tre_cint_t);
+} tre_ctype_map[] =
+{
+ { "alnum", &tre_isalnum_func },
+ { "alpha", &tre_isalpha_func },
+ { "ascii", &tre_isascii_func },
+ { "blank", &tre_isblank_func },
+ { "cntrl", &tre_iscntrl_func },
+ { "digit", &tre_isdigit_func },
+ { "graph", &tre_isgraph_func },
+ { "lower", &tre_islower_func },
+ { "print", &tre_isprint_func },
+ { "punct", &tre_ispunct_func },
+ { "space", &tre_isspace_func },
+ { "upper", &tre_isupper_func },
+ { "xdigit", &tre_isxdigit_func },
+ { NULL, NULL}
+};
+
+tre_ctype_t tre_ctype(const char *name)
+{
+ int i;
+ for (i = 0; tre_ctype_map[i].name != NULL; i++)
+ {
+ if (qse_mbscmp(name, tre_ctype_map[i].name) == 0)
+ return tre_ctype_map[i].func;
+ }
+ return (tre_ctype_t)0;
+}
+
+#endif /* !TRE_USE_SYSTEM_WCTYPE */
+#endif
+
+/* Maximum number of character classes that can occur in a negated bracket
+ expression. */
+#define MAX_NEG_CLASSES 64
+
+/* Maximum length of character class names. */
+#define MAX_CLASS_NAME
+
+#define REST(re) (int)(ctx->re_end - (re)), (re)
+
+static reg_errcode_t
+tre_parse_bracket_items(tre_parse_ctx_t *ctx, int negate,
+ tre_ctype_t neg_classes[], int *num_neg_classes,
+ tre_ast_node_t ***items, int *num_items,
+ int *items_size)
+{
+ const tre_char_t *re = ctx->re;
+ reg_errcode_t status = REG_OK;
+ tre_ctype_t class = (tre_ctype_t)0;
+ int i = *num_items;
+ int max_i = *items_size;
+ int skip;
+
+ /* Build an array of the items in the bracket expression. */
+ while (status == REG_OK)
+ {
+ skip = 0;
+ if (re == ctx->re_end)
+ {
+ status = REG_EBRACK;
+ }
+ else if (*re == CHAR_RBRACKET && re > ctx->re)
+ {
+ DPRINT(("tre_parse_bracket: done: '%.*" STRF "'\n", REST(re)));
+ re++;
+ break;
+ }
+ else
+ {
+ tre_cint_t min = 0, max = 0;
+
+ class = (tre_ctype_t)0;
+ if (re + 2 < ctx->re_end
+ && *(re + 1) == CHAR_MINUS && *(re + 2) != CHAR_RBRACKET)
+ {
+ DPRINT(("tre_parse_bracket: range: '%.*" STRF "'\n", REST(re)));
+ min = *re;
+ max = *(re + 2);
+ re += 3;
+ /* XXX - Should use collation order instead of encoding values
+ in character ranges. */
+ if (min > max)
+ status = REG_ERANGE;
+ }
+ else if (re + 1 < ctx->re_end
+ && *re == CHAR_LBRACKET && *(re + 1) == CHAR_PERIOD)
+ status = REG_ECOLLATE;
+ else if (re + 1 < ctx->re_end
+ && *re == CHAR_LBRACKET && *(re + 1) == CHAR_EQUAL)
+ status = REG_ECOLLATE;
+ else if (re + 1 < ctx->re_end
+ && *re == CHAR_LBRACKET && *(re + 1) == CHAR_COLON)
+ {
+#if 0
+ char tmp_str[64];
+#endif
+ const tre_char_t *endptr = re + 2;
+ int len;
+ DPRINT(("tre_parse_bracket: class: '%.*" STRF "'\n", REST(re)));
+ while (endptr < ctx->re_end && *endptr != CHAR_COLON)
+ endptr++;
+ if (endptr != ctx->re_end)
+ {
+ len = MIN(endptr - re - 2, 63);
+
+ if (qse_getctypebyxname (re + 2, len, &class) <= -1) status = REG_ECTYPE;
+
+ /* Optimize character classes for 8 bit character sets. */
+ if (status == REG_OK && TRE_MB_CUR_MAX == 1)
+ {
+ status = tre_expand_ctype(ctx->mem, class, items,
+ &i, &max_i, ctx->cflags);
+ class = (tre_ctype_t)0;
+ skip = 1;
+ }
+ re = endptr + 2;
+ }
+ else
+ status = REG_ECTYPE;
+ min = 0;
+ max = TRE_CHAR_MAX;
+ }
+ else
+ {
+ DPRINT(("tre_parse_bracket: char: '%.*" STRF "'\n", REST(re)));
+ if (*re == CHAR_MINUS && *(re + 1) != CHAR_RBRACKET
+ && ctx->re != re)
+ /* Two ranges are not allowed to share and endpoint. */
+ status = REG_ERANGE;
+ min = max = *re++;
+ }
+
+ if (status != REG_OK)
+ break;
+
+ if (class && negate)
+ if (*num_neg_classes >= MAX_NEG_CLASSES)
+ status = REG_ESPACE;
+ else
+ neg_classes[(*num_neg_classes)++] = class;
+ else if (!skip)
+ {
+ status = tre_new_item(ctx->mem, min, max, &i, &max_i, items);
+ if (status != REG_OK)
+ break;
+ ((tre_literal_t*)((*items)[i-1])->obj)->u.class = class;
+ }
+
+ /* Add opposite-case counterpoints if REG_ICASE is present.
+ This is broken if there are more than two "same" characters. */
+ if (ctx->cflags & REG_ICASE && !class && status == REG_OK && !skip)
+ {
+ tre_cint_t cmin, ccurr;
+
+ DPRINT(("adding opposite-case counterpoints\n"));
+ while (min <= max)
+ {
+ if (tre_islower(min))
+ {
+ cmin = ccurr = tre_toupper(min++);
+ while (tre_islower(min) && tre_toupper(min) == ccurr + 1
+ && min <= max)
+ ccurr = tre_toupper(min++);
+ status = tre_new_item(ctx->mem, cmin, ccurr,
+ &i, &max_i, items);
+ }
+ else if (tre_isupper(min))
+ {
+ cmin = ccurr = tre_tolower(min++);
+ while (tre_isupper(min) && tre_tolower(min) == ccurr + 1
+ && min <= max)
+ ccurr = tre_tolower(min++);
+ status = tre_new_item(ctx->mem, cmin, ccurr,
+ &i, &max_i, items);
+ }
+ else min++;
+ if (status != REG_OK)
+ break;
+ }
+ if (status != REG_OK)
+ break;
+ }
+ }
+ }
+ *num_items = i;
+ *items_size = max_i;
+ ctx->re = re;
+ return status;
+}
+
+static reg_errcode_t
+tre_parse_bracket(tre_parse_ctx_t *ctx, tre_ast_node_t **result)
+{
+ tre_ast_node_t *node = NULL;
+ int negate = 0;
+ reg_errcode_t status = REG_OK;
+ tre_ast_node_t **items, *u, *n;
+ int i = 0, j, max_i = 32, curr_max, curr_min;
+ tre_ctype_t neg_classes[MAX_NEG_CLASSES];
+ int num_neg_classes = 0;
+
+ /* Start off with an array of `max_i' elements. */
+ items = xmalloc(ctx->mem->mmgr, sizeof(*items) * max_i);
+ if (items == NULL)
+ return REG_ESPACE;
+
+ if (*ctx->re == CHAR_CARET)
+ {
+ DPRINT(("tre_parse_bracket: negate: '%.*" STRF "'\n", REST(ctx->re)));
+ negate = 1;
+ ctx->re++;
+ }
+
+ status = tre_parse_bracket_items(ctx, negate, neg_classes, &num_neg_classes,
+ &items, &i, &max_i);
+
+ if (status != REG_OK)
+ goto parse_bracket_done;
+
+ /* Sort the array if we need to negate it. */
+ if (negate)
+ qse_qsort(items, (unsigned)i, sizeof(*items), tre_compare_items, QSE_NULL);
+
+ curr_max = curr_min = 0;
+ /* Build a union of the items in the array, negated if necessary. */
+ for (j = 0; j < i && status == REG_OK; j++)
+ {
+ int min, max;
+ tre_literal_t *l = items[j]->obj;
+ min = l->code_min;
+ max = l->code_max;
+
+ DPRINT(("item: %d - %d, class %ld, curr_max = %d\n",
+ (int)l->code_min, (int)l->code_max, (long)l->u.class, curr_max));
+
+ if (negate)
+ {
+ if (min < curr_max)
+ {
+ /* Overlap. */
+ curr_max = MAX(max + 1, curr_max);
+ DPRINT(("overlap, curr_max = %d\n", curr_max));
+ l = NULL;
+ }
+ else
+ {
+ /* No overlap. */
+ curr_max = min - 1;
+ if (curr_max >= curr_min)
+ {
+ DPRINT(("no overlap\n"));
+ l->code_min = curr_min;
+ l->code_max = curr_max;
+ }
+ else
+ {
+ DPRINT(("no overlap, zero room\n"));
+ l = NULL;
+ }
+ curr_min = curr_max = max + 1;
+ }
+ }
+
+ if (l != NULL)
+ {
+ int k;
+ DPRINT(("creating %d - %d\n", (int)l->code_min, (int)l->code_max));
+ l->position = ctx->position;
+ if (num_neg_classes > 0)
+ {
+ l->neg_classes = tre_mem_alloc(ctx->mem,
+ (sizeof(l->neg_classes)
+ * (num_neg_classes + 1)));
+ if (l->neg_classes == NULL)
+ {
+ status = REG_ESPACE;
+ break;
+ }
+ for (k = 0; k < num_neg_classes; k++)
+ l->neg_classes[k] = neg_classes[k];
+ l->neg_classes[k] = (tre_ctype_t)0;
+ }
+ else
+ l->neg_classes = NULL;
+ if (node == NULL)
+ node = items[j];
+ else
+ {
+ u = tre_ast_new_union(ctx->mem, node, items[j]);
+ if (u == NULL)
+ status = REG_ESPACE;
+ node = u;
+ }
+ }
+ }
+
+ if (status != REG_OK)
+ goto parse_bracket_done;
+
+ if (negate)
+ {
+ int k;
+ DPRINT(("final: creating %d - %d\n", curr_min, (int)TRE_CHAR_MAX));
+ n = tre_ast_new_literal(ctx->mem, curr_min, TRE_CHAR_MAX, ctx->position);
+ if (n == NULL)
+ status = REG_ESPACE;
+ else
+ {
+ tre_literal_t *l = n->obj;
+ if (num_neg_classes > 0)
+ {
+ l->neg_classes = tre_mem_alloc(ctx->mem,
+ (sizeof(l->neg_classes)
+ * (num_neg_classes + 1)));
+ if (l->neg_classes == NULL)
+ {
+ status = REG_ESPACE;
+ goto parse_bracket_done;
+ }
+ for (k = 0; k < num_neg_classes; k++)
+ l->neg_classes[k] = neg_classes[k];
+ l->neg_classes[k] = (tre_ctype_t)0;
+ }
+ else
+ l->neg_classes = NULL;
+ if (node == NULL)
+ node = n;
+ else
+ {
+ u = tre_ast_new_union(ctx->mem, node, n);
+ if (u == NULL)
+ status = REG_ESPACE;
+ node = u;
+ }
+ }
+ }
+
+ if (status != REG_OK)
+ goto parse_bracket_done;
+
+#ifdef TRE_DEBUG
+ tre_ast_print(node);
+#endif /* TRE_DEBUG */
+
+parse_bracket_done:
+ xfree(ctx->mem->mmgr, items);
+ ctx->position++;
+ *result = node;
+ return status;
+}
+
+
+/* Parses a positive decimal integer. Returns -1 if the string does not
+ contain a valid number. */
+static int
+tre_parse_int(const tre_char_t **regex, const tre_char_t *regex_end)
+{
+ int num = -1;
+ const tre_char_t *r = *regex;
+ while (r < regex_end && *r >= QSE_T('0') && *r <= QSE_T('9'))
+ {
+ if (num < 0)
+ num = 0;
+ num = num * 10 + *r - QSE_T('0');
+ r++;
+ }
+ *regex = r;
+ return num;
+}
+
+
+static reg_errcode_t
+tre_parse_bound(tre_parse_ctx_t *ctx, tre_ast_node_t **result)
+{
+ int min, max, i;
+ int cost_ins, cost_del, cost_subst, cost_max;
+ int limit_ins, limit_del, limit_subst, limit_err;
+ const tre_char_t *r = ctx->re;
+ const tre_char_t *start;
+ int minimal = (ctx->cflags & REG_UNGREEDY) ? 1 : 0;
+ int approx = 0;
+ int costs_set = 0;
+ int counts_set = 0;
+
+ cost_ins = cost_del = cost_subst = cost_max = TRE_PARAM_UNSET;
+ limit_ins = limit_del = limit_subst = limit_err = TRE_PARAM_UNSET;
+
+ /* Parse number (minimum repetition count). */
+ min = -1;
+ if (r < ctx->re_end && *r >= QSE_T('0') && *r <= QSE_T('9'))
+ {
+ DPRINT(("tre_parse: min count: '%.*" STRF "'\n", REST(r)));
+ min = tre_parse_int(&r, ctx->re_end);
+ }
+
+ /* Parse comma and second number (maximum repetition count). */
+ max = min;
+ if (r < ctx->re_end && *r == CHAR_COMMA)
+ {
+ r++;
+ DPRINT(("tre_parse: max count: '%.*" STRF "'\n", REST(r)));
+ max = tre_parse_int(&r, ctx->re_end);
+ }
+
+ /* Check that the repeat counts are sane. */
+ /*if ((max >= 0 && min > max) || max > RE_DUP_MAX) return REG_BADBR;
+
+ hyunghwan.chung:
+ this original check still allows something like {100000,}
+ while it does not allow {1,256}. Why is RE_DUP_MAX necessary?
+ */
+ if ((max >= 0 && min > max)) return REG_BADBR;
+
+
+ /*
+ '{'
+ optionally followed immediately by a number == minimum repcount
+ optionally followed by , then a number == maximum repcount
+ + then a number == maximum insertion count
+ - then a number == maximum deletion count
+ # then a number == maximum substitution count
+ ~ then a number == maximum number of errors
+ Any of +, -, # or ~ without followed by a number means that
+ the maximum count/number of errors is infinite.
+
+ An equation of the form
+ Xi + Yd + Zs < C
+ can be specified to set costs and the cost limit to a value
+ different from the default value:
+ - X is the cost of an insertion
+ - Y is the cost of a deletion
+ - Z is the cost of a substitution
+ - C is the maximum cost
+
+ If no count limit or cost is set for an operation, the operation
+ is not allowed at all.
+ */
+
+
+ do
+ {
+ int done;
+ start = r;
+
+ /* Parse count limit settings */
+ done = 0;
+ if (!counts_set)
+ while (r + 1 < ctx->re_end && !done)
+ {
+ switch (*r)
+ {
+ case CHAR_PLUS: /* Insert limit */
+ DPRINT(("tre_parse: ins limit: '%.*" STRF "'\n", REST(r)));
+ r++;
+ limit_ins = tre_parse_int(&r, ctx->re_end);
+ if (limit_ins < 0)
+ limit_ins = QSE_TYPE_MAX(int);
+ counts_set = 1;
+ break;
+ case CHAR_MINUS: /* Delete limit */
+ DPRINT(("tre_parse: del limit: '%.*" STRF "'\n", REST(r)));
+ r++;
+ limit_del = tre_parse_int(&r, ctx->re_end);
+ if (limit_del < 0)
+ limit_del = QSE_TYPE_MAX(int);
+ counts_set = 1;
+ break;
+ case CHAR_HASH: /* Substitute limit */
+ DPRINT(("tre_parse: subst limit: '%.*" STRF "'\n", REST(r)));
+ r++;
+ limit_subst = tre_parse_int(&r, ctx->re_end);
+ if (limit_subst < 0)
+ limit_subst = QSE_TYPE_MAX(int);
+ counts_set = 1;
+ break;
+ case CHAR_TILDE: /* Maximum number of changes */
+ DPRINT(("tre_parse: count limit: '%.*" STRF "'\n", REST(r)));
+ r++;
+ limit_err = tre_parse_int(&r, ctx->re_end);
+ if (limit_err < 0)
+ limit_err = QSE_TYPE_MAX(int);
+ approx = 1;
+ break;
+ case CHAR_COMMA:
+ r++;
+ break;
+ case QSE_T(' '):
+ r++;
+ break;
+ case QSE_T('}'):
+ done = 1;
+ break;
+ default:
+ done = 1;
+ break;
+ }
+ }
+
+ /* Parse cost restriction equation. */
+ done = 0;
+ if (!costs_set)
+ while (r + 1 < ctx->re_end && !done)
+ {
+ switch (*r)
+ {
+ case CHAR_PLUS:
+ case QSE_T(' '):
+ r++;
+ break;
+ case QSE_T('<'):
+ DPRINT(("tre_parse: max cost: '%.*" STRF "'\n", REST(r)));
+ r++;
+ while (*r == QSE_T(' '))
+ r++;
+ cost_max = tre_parse_int(&r, ctx->re_end);
+ if (cost_max < 0)
+ cost_max = QSE_TYPE_MAX(int);
+ else
+ cost_max--;
+ approx = 1;
+ break;
+ case CHAR_COMMA:
+ r++;
+ done = 1;
+ break;
+ default:
+ if (*r >= QSE_T('0') && *r <= QSE_T('9'))
+ {
+#ifdef TRE_DEBUG
+ const tre_char_t *sr = r;
+#endif /* TRE_DEBUG */
+ int cost = tre_parse_int(&r, ctx->re_end);
+ /* XXX - make sure r is not past end. */
+ switch (*r)
+ {
+ case QSE_T('i'): /* Insert cost */
+ DPRINT(("tre_parse: ins cost: '%.*" STRF "'\n",
+ REST(sr)));
+ r++;
+ cost_ins = cost;
+ costs_set = 1;
+ break;
+ case QSE_T('d'): /* Delete cost */
+ DPRINT(("tre_parse: del cost: '%.*" STRF "'\n",
+ REST(sr)));
+ r++;
+ cost_del = cost;
+ costs_set = 1;
+ break;
+ case QSE_T('s'): /* Substitute cost */
+ DPRINT(("tre_parse: subst cost: '%.*" STRF "'\n",
+ REST(sr)));
+ r++;
+ cost_subst = cost;
+ costs_set = 1;
+ break;
+ default:
+ return REG_BADBR;
+ }
+ }
+ else
+ {
+ done = 1;
+ break;
+ }
+ }
+ }
+ }
+ while (start != r);
+
+ /* Missing }. */
+ if (r >= ctx->re_end)
+ return REG_EBRACE;
+
+ /* Empty contents of {}. */
+ if (r == ctx->re)
+ return REG_BADBR;
+
+ /* Parse the ending '}' or '\}'.*/
+ if (ctx->cflags & REG_EXTENDED)
+ {
+ if (r >= ctx->re_end || *r != CHAR_RBRACE)
+ return REG_BADBR;
+ r++;
+ }
+ else
+ {
+ if (r + 1 >= ctx->re_end
+ || *r != CHAR_BACKSLASH
+ || *(r + 1) != CHAR_RBRACE)
+ return REG_BADBR;
+ r += 2;
+ }
+
+
+ /* Parse trailing '?' marking minimal repetition. */
+ if (r < ctx->re_end)
+ {
+ if (*r == CHAR_QUESTIONMARK)
+ {
+ minimal = !(ctx->cflags & REG_UNGREEDY);
+ r++;
+ }
+ else if (*r == CHAR_STAR || *r == CHAR_PLUS)
+ {
+ /* These are reserved for future extensions. */
+ return REG_BADRPT;
+ }
+ }
+
+ /* Create the AST node(s). */
+ if (min == 0 && max == 0)
+ {
+ *result = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1);
+ if (*result == NULL)
+ return REG_ESPACE;
+ }
+ else
+ {
+ if (min < 0 && max < 0)
+ /* Only approximate parameters set, no repetitions. */
+ min = max = 1;
+
+ *result = tre_ast_new_iter(ctx->mem, *result, min, max, minimal);
+ if (!*result)
+ return REG_ESPACE;
+
+ /* If approximate matching parameters are set, add them to the
+ iteration node. */
+ if (approx || costs_set || counts_set)
+ {
+ int *params;
+ tre_iteration_t *iter = (*result)->obj;
+
+ if (costs_set || counts_set)
+ {
+ if (limit_ins == TRE_PARAM_UNSET)
+ {
+ if (cost_ins == TRE_PARAM_UNSET)
+ limit_ins = 0;
+ else
+ limit_ins = QSE_TYPE_MAX(int);
+ }
+
+ if (limit_del == TRE_PARAM_UNSET)
+ {
+ if (cost_del == TRE_PARAM_UNSET)
+ limit_del = 0;
+ else
+ limit_del = QSE_TYPE_MAX(int);
+ }
+
+ if (limit_subst == TRE_PARAM_UNSET)
+ {
+ if (cost_subst == TRE_PARAM_UNSET)
+ limit_subst = 0;
+ else
+ limit_subst = QSE_TYPE_MAX(int);
+ }
+ }
+
+ if (cost_max == TRE_PARAM_UNSET)
+ cost_max = QSE_TYPE_MAX(int);
+ if (limit_err == TRE_PARAM_UNSET)
+ limit_err = QSE_TYPE_MAX(int);
+
+ ctx->have_approx = 1;
+ params = tre_mem_alloc(ctx->mem, sizeof(*params) * TRE_PARAM_LAST);
+ if (!params)
+ return REG_ESPACE;
+ for (i = 0; i < TRE_PARAM_LAST; i++)
+ params[i] = TRE_PARAM_UNSET;
+ params[TRE_PARAM_COST_INS] = cost_ins;
+ params[TRE_PARAM_COST_DEL] = cost_del;
+ params[TRE_PARAM_COST_SUBST] = cost_subst;
+ params[TRE_PARAM_COST_MAX] = cost_max;
+ params[TRE_PARAM_MAX_INS] = limit_ins;
+ params[TRE_PARAM_MAX_DEL] = limit_del;
+ params[TRE_PARAM_MAX_SUBST] = limit_subst;
+ params[TRE_PARAM_MAX_ERR] = limit_err;
+ iter->params = params;
+ }
+ }
+
+ DPRINT(("tre_parse_bound: min %d, max %d, costs [%d,%d,%d, total %d], "
+ "limits [%d,%d,%d, total %d]\n",
+ min, max, cost_ins, cost_del, cost_subst, cost_max,
+ limit_ins, limit_del, limit_subst, limit_err));
+
+
+ ctx->re = r;
+ return REG_OK;
+}
+
+typedef enum
+{
+ PARSE_RE = 0,
+ PARSE_ATOM,
+ PARSE_MARK_FOR_SUBMATCH,
+ PARSE_BRANCH,
+ PARSE_PIECE,
+ PARSE_CATENATION,
+ PARSE_POST_CATENATION,
+ PARSE_UNION,
+ PARSE_POST_UNION,
+ PARSE_POSTFIX,
+ PARSE_RESTORE_CFLAGS
+} tre_parse_re_stack_symbol_t;
+
+
+reg_errcode_t
+tre_parse(tre_parse_ctx_t *ctx)
+{
+ tre_ast_node_t *result = NULL;
+ tre_parse_re_stack_symbol_t symbol;
+ reg_errcode_t status = REG_OK;
+ tre_stack_t *stack = ctx->stack;
+ int bottom = tre_stack_num_objects(stack);
+ int depth = 0;
+ int temporary_cflags = 0;
+
+ DPRINT(("tre_parse: parsing '%.*" STRF "', len = %d\n",
+ ctx->len, ctx->re, ctx->len));
+
+ if (!ctx->nofirstsub)
+ {
+ STACK_PUSH(stack, int, ctx->submatch_id);
+ STACK_PUSH(stack, int, PARSE_MARK_FOR_SUBMATCH);
+ ctx->submatch_id++;
+ }
+ STACK_PUSH(stack, int, PARSE_RE);
+ ctx->re_start = ctx->re;
+ ctx->re_end = ctx->re + ctx->len;
+
+
+ /* The following is basically just a recursive descent parser. I use
+ an explicit stack instead of recursive functions mostly because of
+ two reasons: compatibility with systems which have an overflowable
+ call stack, and efficiency (both in lines of code and speed). */
+ while (tre_stack_num_objects(stack) > bottom && status == REG_OK)
+ {
+ if (status != REG_OK)
+ break;
+ symbol = tre_stack_pop_int(stack);
+ switch (symbol)
+ {
+ case PARSE_RE:
+ /* Parse a full regexp. A regexp is one or more branches,
+ separated by the union operator `|'. */
+#ifdef REG_LITERAL
+ if (!(ctx->cflags & REG_LITERAL)
+ && ctx->cflags & REG_EXTENDED)
+#endif /* REG_LITERAL */
+ STACK_PUSHX(stack, int, PARSE_UNION);
+ STACK_PUSHX(stack, int, PARSE_BRANCH);
+ break;
+
+ case PARSE_BRANCH:
+ /* Parse a branch. A branch is one or more pieces, concatenated.
+ A piece is an atom possibly followed by a postfix operator. */
+ STACK_PUSHX(stack, int, PARSE_CATENATION);
+ STACK_PUSHX(stack, int, PARSE_PIECE);
+ break;
+
+ case PARSE_PIECE:
+ /* Parse a piece. A piece is an atom possibly followed by one
+ or more postfix operators. */
+#ifdef REG_LITERAL
+ if (!(ctx->cflags & REG_LITERAL))
+#endif /* REG_LITERAL */
+ STACK_PUSHX(stack, int, PARSE_POSTFIX);
+ STACK_PUSHX(stack, int, PARSE_ATOM);
+ break;
+
+ case PARSE_CATENATION:
+ /* If the expression has not ended, parse another piece. */
+ {
+ tre_char_t c;
+ if (ctx->re >= ctx->re_end)
+ break;
+ c = *ctx->re;
+#ifdef REG_LITERAL
+ if (!(ctx->cflags & REG_LITERAL))
+ {
+#endif /* REG_LITERAL */
+ if (ctx->cflags & REG_EXTENDED && c == CHAR_PIPE)
+ break;
+ if ((ctx->cflags & REG_EXTENDED
+ && c == CHAR_RPAREN && depth > 0)
+ || (!(ctx->cflags & REG_EXTENDED)
+ && (c == CHAR_BACKSLASH
+ && *(ctx->re + 1) == CHAR_RPAREN)))
+ {
+ if (!(ctx->cflags & REG_EXTENDED) && depth == 0)
+ status = REG_EPAREN;
+ DPRINT(("tre_parse: group end: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ depth--;
+ if (!(ctx->cflags & REG_EXTENDED))
+ ctx->re += 2;
+ break;
+ }
+#ifdef REG_LITERAL
+ }
+#endif /* REG_LITERAL */
+
+#ifdef REG_RIGHT_ASSOC
+ if (ctx->cflags & REG_RIGHT_ASSOC)
+ {
+ /* Right associative concatenation. */
+ STACK_PUSHX(stack, voidptr, result);
+ STACK_PUSHX(stack, int, PARSE_POST_CATENATION);
+ STACK_PUSHX(stack, int, PARSE_CATENATION);
+ STACK_PUSHX(stack, int, PARSE_PIECE);
+ }
+ else
+#endif
+ { /* REG_RIGHT_ASSOC */
+ /* Default case, left associative concatenation. */
+ STACK_PUSHX(stack, int, PARSE_CATENATION);
+ STACK_PUSHX(stack, voidptr, result);
+ STACK_PUSHX(stack, int, PARSE_POST_CATENATION);
+ STACK_PUSHX(stack, int, PARSE_PIECE);
+ }
+ break;
+ }
+
+ case PARSE_POST_CATENATION:
+ {
+ tre_ast_node_t *tree = tre_stack_pop_voidptr(stack);
+ tre_ast_node_t *tmp_node;
+ tmp_node = tre_ast_new_catenation(ctx->mem, tree, result);
+ if (!tmp_node)
+ return REG_ESPACE;
+ result = tmp_node;
+ break;
+ }
+
+ case PARSE_UNION:
+ if (ctx->re >= ctx->re_end)
+ break;
+#ifdef REG_LITERAL
+ if (ctx->cflags & REG_LITERAL)
+ break;
+#endif /* REG_LITERAL */
+ switch (*ctx->re)
+ {
+ case CHAR_PIPE:
+ DPRINT(("tre_parse: union: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ STACK_PUSHX(stack, int, PARSE_UNION);
+ STACK_PUSHX(stack, voidptr, result);
+ STACK_PUSHX(stack, int, PARSE_POST_UNION);
+ STACK_PUSHX(stack, int, PARSE_BRANCH);
+ ctx->re++;
+ break;
+
+ case CHAR_RPAREN:
+ ctx->re++;
+ break;
+
+ default:
+ break;
+ }
+ break;
+
+ case PARSE_POST_UNION:
+ {
+ tre_ast_node_t *tmp_node;
+ tre_ast_node_t *tree = tre_stack_pop_voidptr(stack);
+ tmp_node = tre_ast_new_union(ctx->mem, tree, result);
+ if (!tmp_node)
+ return REG_ESPACE;
+ result = tmp_node;
+ break;
+ }
+
+ case PARSE_POSTFIX:
+ /* Parse postfix operators. */
+ if (ctx->re >= ctx->re_end)
+ break;
+#ifdef REG_LITERAL
+ if (ctx->cflags & REG_LITERAL)
+ break;
+#endif /* REG_LITERAL */
+ switch (*ctx->re)
+ {
+ case CHAR_PLUS:
+ case CHAR_QUESTIONMARK:
+ if (!(ctx->cflags & REG_EXTENDED))
+ break;
+ /*FALLTHROUGH*/
+ case CHAR_STAR:
+ {
+ tre_ast_node_t *tmp_node;
+ int minimal = (ctx->cflags & REG_UNGREEDY) ? 1 : 0;
+ int rep_min = 0;
+ int rep_max = -1;
+#ifdef TRE_DEBUG
+ const tre_char_t *tmp_re;
+#endif
+
+ if (*ctx->re == CHAR_PLUS)
+ rep_min = 1;
+ if (*ctx->re == CHAR_QUESTIONMARK)
+ rep_max = 1;
+#ifdef TRE_DEBUG
+ tmp_re = ctx->re;
+#endif
+
+ if (ctx->re + 1 < ctx->re_end)
+ {
+ if (*(ctx->re + 1) == CHAR_QUESTIONMARK)
+ {
+ minimal = !(ctx->cflags & REG_UNGREEDY);
+ ctx->re++;
+ }
+ else if (*(ctx->re + 1) == CHAR_STAR
+ || *(ctx->re + 1) == CHAR_PLUS)
+ {
+ /* These are reserved for future extensions. */
+ return REG_BADRPT;
+ }
+ }
+
+ DPRINT(("tre_parse: %s star: '%.*" STRF "'\n",
+ minimal ? " minimal" : "greedy", REST(tmp_re)));
+ ctx->re++;
+ tmp_node = tre_ast_new_iter(ctx->mem, result, rep_min, rep_max,
+ minimal);
+ if (tmp_node == NULL)
+ return REG_ESPACE;
+ result = tmp_node;
+ STACK_PUSHX(stack, int, PARSE_POSTFIX);
+ }
+ break;
+
+ case CHAR_BACKSLASH:
+ /* "\{" is special without REG_EXTENDED */
+ if (!(ctx->cflags & REG_EXTENDED)
+ && ctx->re + 1 < ctx->re_end
+ && *(ctx->re + 1) == CHAR_LBRACE)
+ {
+ ctx->re++;
+ goto parse_brace;
+ }
+ else
+ break;
+
+ case CHAR_LBRACE:
+ /* "{" is literal without REG_EXTENDED */
+ if (!(ctx->cflags & REG_EXTENDED))
+ break;
+
+parse_brace:
+ DPRINT(("tre_parse: bound: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ ctx->re++;
+
+ status = tre_parse_bound(ctx, &result);
+ if (status != REG_OK)
+ return status;
+ STACK_PUSHX(stack, int, PARSE_POSTFIX);
+ break;
+ }
+ break;
+
+ case PARSE_ATOM:
+ /* Parse an atom. An atom is a regular expression enclosed in `()',
+ an empty set of `()', a bracket expression, `.', `^', `$',
+ a `\' followed by a character, or a single character. */
+
+ /* End of regexp? (empty string). */
+ if (ctx->re >= ctx->re_end)
+ goto parse_literal;
+
+#ifdef REG_LITERAL
+ if (ctx->cflags & REG_LITERAL)
+ goto parse_literal;
+#endif /* REG_LITERAL */
+
+ switch (*ctx->re)
+ {
+ case CHAR_LPAREN: /* parenthesized subexpression */
+
+ /* Handle "(?...)" extensions. They work in a way similar
+ to Perls corresponding extensions. */
+ if (ctx->cflags & REG_EXTENDED
+ && *(ctx->re + 1) == CHAR_QUESTIONMARK)
+ {
+ int new_cflags = ctx->cflags;
+ int bit = 1;
+ DPRINT(("tre_parse: extension: '%.*" STRF "\n",
+ REST(ctx->re)));
+ ctx->re += 2;
+ while (/*CONSTCOND*/1)
+ {
+ if (*ctx->re == QSE_T('i'))
+ {
+ DPRINT(("tre_parse: icase: '%.*" STRF "\n",
+ REST(ctx->re)));
+ if (bit)
+ new_cflags |= REG_ICASE;
+ else
+ new_cflags &= ~REG_ICASE;
+ ctx->re++;
+ }
+ else if (*ctx->re == QSE_T('n'))
+ {
+ DPRINT(("tre_parse: newline: '%.*" STRF "\n",
+ REST(ctx->re)));
+ if (bit)
+ new_cflags |= REG_NEWLINE;
+ else
+ new_cflags &= ~REG_NEWLINE;
+ ctx->re++;
+ }
+#ifdef REG_RIGHT_ASSOC
+ else if (*ctx->re == QSE_T('r'))
+ {
+ DPRINT(("tre_parse: right assoc: '%.*" STRF "\n",
+ REST(ctx->re)));
+ if (bit)
+ new_cflags |= REG_RIGHT_ASSOC;
+ else
+ new_cflags &= ~REG_RIGHT_ASSOC;
+ ctx->re++;
+ }
+#endif /* REG_RIGHT_ASSOC */
+#ifdef REG_UNGREEDY
+ else if (*ctx->re == QSE_T('U'))
+ {
+ DPRINT(("tre_parse: ungreedy: '%.*" STRF "\n",
+ REST(ctx->re)));
+ if (bit)
+ new_cflags |= REG_UNGREEDY;
+ else
+ new_cflags &= ~REG_UNGREEDY;
+ ctx->re++;
+ }
+#endif /* REG_UNGREEDY */
+ else if (*ctx->re == CHAR_MINUS)
+ {
+ DPRINT(("tre_parse: turn off: '%.*" STRF "\n",
+ REST(ctx->re)));
+ ctx->re++;
+ bit = 0;
+ }
+ else if (*ctx->re == CHAR_COLON)
+ {
+ DPRINT(("tre_parse: no group: '%.*" STRF "\n",
+ REST(ctx->re)));
+ ctx->re++;
+ depth++;
+ break;
+ }
+ else if (*ctx->re == CHAR_HASH)
+ {
+ DPRINT(("tre_parse: comment: '%.*" STRF "\n",
+ REST(ctx->re)));
+ /* A comment can contain any character except a
+ right parenthesis */
+ while (*ctx->re != CHAR_RPAREN
+ && ctx->re < ctx->re_end)
+ ctx->re++;
+ if (*ctx->re == CHAR_RPAREN && ctx->re < ctx->re_end)
+ {
+ ctx->re++;
+ break;
+ }
+ else
+ return REG_BADPAT;
+ }
+ else if (*ctx->re == CHAR_RPAREN)
+ {
+ ctx->re++;
+ break;
+ }
+ else
+ return REG_BADPAT;
+ }
+
+ /* Turn on the cflags changes for the rest of the
+ enclosing group. */
+ STACK_PUSHX(stack, int, ctx->cflags);
+ STACK_PUSHX(stack, int, PARSE_RESTORE_CFLAGS);
+ STACK_PUSHX(stack, int, PARSE_RE);
+ ctx->cflags = new_cflags;
+ break;
+ }
+
+ if (ctx->cflags & REG_EXTENDED
+ || (ctx->re > ctx->re_start
+ && *(ctx->re - 1) == CHAR_BACKSLASH))
+ {
+ depth++;
+ if (ctx->re + 2 < ctx->re_end
+ && *(ctx->re + 1) == CHAR_QUESTIONMARK
+ && *(ctx->re + 2) == CHAR_COLON)
+ {
+ DPRINT(("tre_parse: group begin: '%.*" STRF
+ "', no submatch\n", REST(ctx->re)));
+ /* Don't mark for submatching. */
+ ctx->re += 3;
+ STACK_PUSHX(stack, int, PARSE_RE);
+ }
+ else
+ {
+ DPRINT(("tre_parse: group begin: '%.*" STRF
+ "', submatch %d\n", REST(ctx->re),
+ ctx->submatch_id));
+ ctx->re++;
+ /* First parse a whole RE, then mark the resulting tree
+ for submatching. */
+ STACK_PUSHX(stack, int, ctx->submatch_id);
+ STACK_PUSHX(stack, int, PARSE_MARK_FOR_SUBMATCH);
+ STACK_PUSHX(stack, int, PARSE_RE);
+ ctx->submatch_id++;
+ }
+ }
+ else
+ goto parse_literal;
+ break;
+
+ case CHAR_RPAREN: /* end of current subexpression */
+ if ((ctx->cflags & REG_EXTENDED && depth > 0)
+ || (ctx->re > ctx->re_start
+ && *(ctx->re - 1) == CHAR_BACKSLASH))
+ {
+ DPRINT(("tre_parse: empty: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ /* We were expecting an atom, but instead the current
+ subexpression was closed. POSIX leaves the meaning of
+ this to be implementation-defined. We interpret this as
+ an empty expression (which matches an empty string). */
+ result = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1);
+ if (result == NULL)
+ return REG_ESPACE;
+ if (!(ctx->cflags & REG_EXTENDED))
+ ctx->re--;
+ }
+ else
+ goto parse_literal;
+ break;
+
+ case CHAR_LBRACKET: /* bracket expression */
+ DPRINT(("tre_parse: bracket: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ ctx->re++;
+ status = tre_parse_bracket(ctx, &result);
+ if (status != REG_OK)
+ return status;
+ break;
+
+ case CHAR_BACKSLASH:
+ /* If this is "\(" or "\)" chew off the backslash and
+ try again. */
+ if (!(ctx->cflags & REG_EXTENDED)
+ && ctx->re + 1 < ctx->re_end
+ && (*(ctx->re + 1) == CHAR_LPAREN
+ || *(ctx->re + 1) == CHAR_RPAREN))
+ {
+ ctx->re++;
+ STACK_PUSHX(stack, int, PARSE_ATOM);
+ break;
+ }
+
+ /* If a macro is used, parse the expanded macro recursively. */
+ {
+ tre_char_t buf[64];
+ tre_expand_macro(ctx->re + 1, ctx->re_end,
+ buf, QSE_COUNTOF(buf));
+ if (buf[0] != 0)
+ {
+ tre_parse_ctx_t subctx;
+ QSE_MEMCPY (&subctx, ctx, sizeof(subctx));
+ subctx.re = buf;
+ subctx.len = tre_strlen(buf);
+ subctx.nofirstsub = 1;
+ status = tre_parse(&subctx);
+ if (status != REG_OK) return status;
+ ctx->re += 2;
+ ctx->position = subctx.position;
+ result = subctx.result;
+ break;
+ }
+ }
+
+ if (ctx->re + 1 >= ctx->re_end)
+ /* Trailing backslash. */
+ return REG_EESCAPE;
+
+#ifdef REG_LITERAL
+ if (*(ctx->re + 1) == QSE_T('Q'))
+ {
+ DPRINT(("tre_parse: tmp literal: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ ctx->cflags |= REG_LITERAL;
+ temporary_cflags |= REG_LITERAL;
+ ctx->re += 2;
+ STACK_PUSHX(stack, int, PARSE_ATOM);
+ break;
+ }
+#endif /* REG_LITERAL */
+
+ DPRINT(("tre_parse: bleep: '%.*" STRF "'\n", REST(ctx->re)));
+ ctx->re++;
+ switch (*ctx->re)
+ {
+ case QSE_T('b'):
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_WB, -1);
+ ctx->re++;
+ break;
+ case QSE_T('B'):
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_WB_NEG, -1);
+ ctx->re++;
+ break;
+ case QSE_T('<'):
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_BOW, -1);
+ ctx->re++;
+ break;
+ case QSE_T('>'):
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_EOW, -1);
+ ctx->re++;
+ break;
+ case QSE_T('x'):
+ ctx->re++;
+ if (ctx->re[0] != CHAR_LBRACE && ctx->re < ctx->re_end)
+ {
+ #if 0
+ /* 8 bit hex char. */
+ char tmp[3] = {0, 0, 0};
+ long val;
+ DPRINT(("tre_parse: 8 bit hex: '%.*" STRF "'\n",
+ REST(ctx->re - 2)));
+
+ if (tre_isxdigit(ctx->re[0]) && ctx->re < ctx->re_end)
+ {
+ tmp[0] = (char)ctx->re[0];
+ ctx->re++;
+ }
+ if (tre_isxdigit(ctx->re[0]) && ctx->re < ctx->re_end)
+ {
+ tmp[1] = (char)ctx->re[0];
+ ctx->re++;
+ }
+ val = strtol(tmp, NULL, 16);
+ #endif
+ long val = 0;
+ int tmp;
+ if ((tmp = xdigit_to_num(ctx->re[0])) >= 0 && ctx->re < ctx->re_end)
+ {
+ val = val * 16 + tmp;
+ ctx->re++;
+ }
+ if ((tmp = xdigit_to_num(ctx->re[1])) >= 0 && ctx->re < ctx->re_end)
+ {
+ val = val * 16 + tmp;
+ ctx->re++;
+ }
+
+ result = tre_ast_new_literal(ctx->mem, (int)val,
+ (int)val, ctx->position);
+ ctx->position++;
+ break;
+ }
+ else if (ctx->re < ctx->re_end)
+ {
+ /* Wide char. */
+ #if 0
+ char tmp[32];
+ long val;
+ int i = 0;
+ ctx->re++;
+ while (ctx->re_end - ctx->re >= 0)
+ {
+ if (ctx->re[0] == CHAR_RBRACE)
+ break;
+ if (tre_isxdigit(ctx->re[0]))
+ {
+ tmp[i] = (char)ctx->re[0];
+ i++;
+ ctx->re++;
+ continue;
+ }
+ return REG_EBRACE;
+ }
+ ctx->re++;
+ tmp[i] = 0;
+ val = strtol(tmp, NULL, 16);
+ #endif
+ long val = 0;
+ int tmp;
+
+ ctx->re++;
+ while (ctx->re_end - ctx->re >= 0)
+ {
+ if (ctx->re[0] == CHAR_RBRACE)
+ break;
+ tmp = xdigit_to_num(ctx->re[0]);
+ if (tmp >= 0)
+ {
+ val = val * 16 + tmp;
+ ctx->re++;
+ continue;
+ }
+ return REG_EBRACE;
+ }
+
+ result = tre_ast_new_literal(ctx->mem, (int)val, (int)val,
+ ctx->position);
+ ctx->position++;
+ break;
+ }
+ /*FALLTHROUGH*/
+
+ default:
+ if (tre_isdigit(*ctx->re))
+ {
+ /* Back reference. */
+ int val = *ctx->re - QSE_T('0');
+ DPRINT(("tre_parse: backref: '%.*" STRF "'\n",
+ REST(ctx->re - 1)));
+ result = tre_ast_new_literal(ctx->mem, BACKREF, val,
+ ctx->position);
+ if (result == NULL)
+ return REG_ESPACE;
+ ctx->position++;
+ ctx->max_backref = MAX(val, ctx->max_backref);
+ ctx->re++;
+ }
+ else
+ {
+ /* Escaped character. */
+ DPRINT(("tre_parse: escaped: '%.*" STRF "'\n",
+ REST(ctx->re - 1)));
+ result = tre_ast_new_literal(ctx->mem, *ctx->re, *ctx->re,
+ ctx->position);
+ ctx->position++;
+ ctx->re++;
+ }
+ break;
+ }
+ if (result == NULL)
+ return REG_ESPACE;
+ break;
+
+ case CHAR_PERIOD: /* the any-symbol */
+ DPRINT(("tre_parse: any: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ if (ctx->cflags & REG_NEWLINE)
+ {
+ tre_ast_node_t *tmp1;
+ tre_ast_node_t *tmp2;
+ tmp1 = tre_ast_new_literal(ctx->mem, 0, QSE_T('\n') - 1,
+ ctx->position);
+ if (!tmp1)
+ return REG_ESPACE;
+ tmp2 = tre_ast_new_literal(ctx->mem, QSE_T('\n') + 1, TRE_CHAR_MAX,
+ ctx->position + 1);
+ if (!tmp2)
+ return REG_ESPACE;
+ result = tre_ast_new_union(ctx->mem, tmp1, tmp2);
+ if (!result)
+ return REG_ESPACE;
+ ctx->position += 2;
+ }
+ else
+ {
+ result = tre_ast_new_literal(ctx->mem, 0, TRE_CHAR_MAX,
+ ctx->position);
+ if (!result)
+ return REG_ESPACE;
+ ctx->position++;
+ }
+ ctx->re++;
+ break;
+
+ case CHAR_CARET: /* beginning of line assertion */
+ /* '^' has a special meaning everywhere in EREs, and in the
+ beginning of the RE and after \( is BREs. */
+ if (ctx->cflags & REG_EXTENDED
+ || (ctx->re - 2 >= ctx->re_start
+ && *(ctx->re - 2) == CHAR_BACKSLASH
+ && *(ctx->re - 1) == CHAR_LPAREN)
+ || ctx->re == ctx->re_start)
+ {
+ DPRINT(("tre_parse: BOL: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_BOL, -1);
+ if (result == NULL)
+ return REG_ESPACE;
+ ctx->re++;
+ }
+ else
+ goto parse_literal;
+ break;
+
+ case CHAR_DOLLAR: /* end of line assertion. */
+ /* '$' is special everywhere in EREs, and in the end of the
+ string and before \) is BREs. */
+ if (ctx->cflags & REG_EXTENDED
+ || (ctx->re + 2 < ctx->re_end
+ && *(ctx->re + 1) == CHAR_BACKSLASH
+ && *(ctx->re + 2) == CHAR_RPAREN)
+ || ctx->re + 1 == ctx->re_end)
+ {
+ DPRINT(("tre_parse: EOL: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ result = tre_ast_new_literal(ctx->mem, ASSERTION,
+ ASSERT_AT_EOL, -1);
+ if (result == NULL)
+ return REG_ESPACE;
+ ctx->re++;
+ }
+ else
+ goto parse_literal;
+ break;
+
+ default:
+parse_literal:
+
+ if (temporary_cflags && ctx->re + 1 < ctx->re_end
+ && *ctx->re == CHAR_BACKSLASH && *(ctx->re + 1) == QSE_T('E'))
+ {
+ DPRINT(("tre_parse: end tmps: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ ctx->cflags &= ~temporary_cflags;
+ temporary_cflags = 0;
+ ctx->re += 2;
+ STACK_PUSHX(stack, int, PARSE_PIECE);
+ break;
+ }
+
+
+ /* We are expecting an atom. If the subexpression (or the whole
+ regexp ends here, we interpret it as an empty expression
+ (which matches an empty string). */
+ if (
+#ifdef REG_LITERAL
+ !(ctx->cflags & REG_LITERAL) &&
+#endif /* REG_LITERAL */
+ (ctx->re >= ctx->re_end
+ || *ctx->re == CHAR_STAR
+ || (ctx->cflags & REG_EXTENDED
+ && (*ctx->re == CHAR_PIPE
+ || *ctx->re == CHAR_LBRACE
+ || *ctx->re == CHAR_PLUS
+ || *ctx->re == CHAR_QUESTIONMARK))
+ /* Test for "\)" in BRE mode. */
+ || (!(ctx->cflags & REG_EXTENDED)
+ && ctx->re + 1 < ctx->re_end
+ && *ctx->re == CHAR_BACKSLASH
+ && *(ctx->re + 1) == CHAR_LBRACE)))
+ {
+ DPRINT(("tre_parse: empty: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ result = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1);
+ if (!result)
+ return REG_ESPACE;
+ break;
+ }
+
+ DPRINT(("tre_parse: literal: '%.*" STRF "'\n",
+ REST(ctx->re)));
+ /* Note that we can't use an tre_isalpha() test here, since there
+ may be characters which are alphabetic but neither upper or
+ lower case. */
+ if (ctx->cflags & REG_ICASE
+ && (tre_isupper(*ctx->re) || tre_islower(*ctx->re)))
+ {
+ tre_ast_node_t *tmp1;
+ tre_ast_node_t *tmp2;
+
+ /* XXX - Can there be more than one opposite-case
+ counterpoints for some character in some locale? Or
+ more than two characters which all should be regarded
+ the same character if case is ignored? If yes, there
+ does not seem to be a portable way to detect it. I guess
+ that at least for multi-character collating elements there
+ could be several opposite-case counterpoints, but they
+ cannot be supported portably anyway. */
+ tmp1 = tre_ast_new_literal(ctx->mem, tre_toupper(*ctx->re),
+ tre_toupper(*ctx->re),
+ ctx->position);
+ if (!tmp1)
+ return REG_ESPACE;
+ tmp2 = tre_ast_new_literal(ctx->mem, tre_tolower(*ctx->re),
+ tre_tolower(*ctx->re),
+ ctx->position);
+ if (!tmp2)
+ return REG_ESPACE;
+ result = tre_ast_new_union(ctx->mem, tmp1, tmp2);
+ if (!result)
+ return REG_ESPACE;
+ }
+ else
+ {
+ result = tre_ast_new_literal(ctx->mem, *ctx->re, *ctx->re,
+ ctx->position);
+ if (!result)
+ return REG_ESPACE;
+ }
+ ctx->position++;
+ ctx->re++;
+ break;
+ }
+ break;
+
+ case PARSE_MARK_FOR_SUBMATCH:
+ {
+ int submatch_id = tre_stack_pop_int(stack);
+
+ if (result->submatch_id >= 0)
+ {
+ tre_ast_node_t *n, *tmp_node;
+ n = tre_ast_new_literal(ctx->mem, EMPTY, -1, -1);
+ if (n == NULL)
+ return REG_ESPACE;
+ tmp_node = tre_ast_new_catenation(ctx->mem, n, result);
+ if (tmp_node == NULL)
+ return REG_ESPACE;
+ tmp_node->num_submatches = result->num_submatches;
+ result = tmp_node;
+ }
+ result->submatch_id = submatch_id;
+ result->num_submatches++;
+ break;
+ }
+
+ case PARSE_RESTORE_CFLAGS:
+ ctx->cflags = tre_stack_pop_int(stack);
+ break;
+
+ default:
+ assert(0);
+ break;
+ }
+}
+
+/* Check for missing closing parentheses. */
+if (depth > 0)
+ return REG_EPAREN;
+
+if (status == REG_OK)
+ ctx->result = result;
+
+return status;
+}
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-parse.h b/qse/lib/cmn/tre-parse.h
new file mode 100644
index 00000000..976950e8
--- /dev/null
+++ b/qse/lib/cmn/tre-parse.h
@@ -0,0 +1,96 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-parse.c - Regexp parser definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef _QSE_LIB_CMN_TRE_PARSE_H_
+#define _QSE_LIB_CMN_TRE_PARSE_H_
+
+/* Parse context. */
+typedef struct
+{
+ /* Memory allocator. The AST is allocated using this. */
+ tre_mem_t mem;
+ /* Stack used for keeping track of regexp syntax. */
+ tre_stack_t *stack;
+ /* The parse result. */
+ tre_ast_node_t *result;
+ /* The regexp to parse and its length. */
+ const tre_char_t *re;
+ /* The first character of the entire regexp. */
+ const tre_char_t *re_start;
+ /* The first character after the end of the regexp. */
+ const tre_char_t *re_end;
+ int len;
+ /* Current submatch ID. */
+ int submatch_id;
+ /* Current position (number of literal). */
+ int position;
+ /* The highest back reference or -1 if none seen so far. */
+ int max_backref;
+ /* This flag is set if the regexp uses approximate matching. */
+ int have_approx;
+ /* Compilation flags. */
+ int cflags;
+ /* If this flag is set the top-level submatch is not captured. */
+ int nofirstsub;
+ /* The currently set approximate matching parameters. */
+ int params[TRE_PARAM_LAST];
+} tre_parse_ctx_t;
+
+/* Parses a wide character regexp pattern into a syntax tree. This parser
+ handles both syntaxes (BRE and ERE), including the TRE extensions. */
+reg_errcode_t tre_parse(tre_parse_ctx_t *ctx);
+
+#endif /* TRE_PARSE_H */
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-stack.c b/qse/lib/cmn/tre-stack.c
new file mode 100644
index 00000000..a7d61394
--- /dev/null
+++ b/qse/lib/cmn/tre-stack.c
@@ -0,0 +1,167 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-stack.c - Simple stack implementation
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "tre.h"
+#include "tre-stack.h"
+
+union tre_stack_item
+{
+ void *voidptr_value;
+ int int_value;
+};
+
+struct tre_stack_rec
+{
+ qse_mmgr_t* mmgr;
+ int size;
+ int max_size;
+ int increment;
+ int ptr;
+ union tre_stack_item *stack;
+};
+
+
+tre_stack_t* tre_stack_new(qse_mmgr_t* mmgr, int size, int max_size, int increment)
+{
+ tre_stack_t *s;
+
+ s = xmalloc(mmgr, sizeof(*s));
+ if (s != NULL)
+ {
+ s->stack = xmalloc(mmgr, sizeof(*s->stack) * size);
+ if (s->stack == NULL)
+ {
+ xfree(mmgr, s);
+ return NULL;
+ }
+ s->size = size;
+ s->max_size = max_size;
+ s->increment = increment;
+ s->ptr = 0;
+ s->mmgr = mmgr;
+ }
+ return s;
+}
+
+void
+tre_stack_destroy(tre_stack_t *s)
+{
+ xfree(s->mmgr,s->stack);
+ xfree(s->mmgr,s);
+}
+
+int
+tre_stack_num_objects(tre_stack_t *s)
+{
+ return s->ptr;
+}
+
+static reg_errcode_t
+tre_stack_push(tre_stack_t *s, union tre_stack_item value)
+{
+ if (s->ptr < s->size)
+ {
+ s->stack[s->ptr] = value;
+ s->ptr++;
+ }
+ else
+ {
+ if (s->size >= s->max_size)
+ {
+ DPRINT(("tre_stack_push: stack full\n"));
+ return REG_ESPACE;
+ }
+ else
+ {
+ union tre_stack_item *new_buffer;
+ int new_size;
+ DPRINT(("tre_stack_push: trying to realloc more space\n"));
+ new_size = s->size + s->increment;
+ if (new_size > s->max_size)
+ new_size = s->max_size;
+ new_buffer = xrealloc(s->mmgr, s->stack, sizeof(*new_buffer) * new_size);
+ if (new_buffer == NULL)
+ {
+ DPRINT(("tre_stack_push: realloc failed.\n"));
+ return REG_ESPACE;
+ }
+ DPRINT(("tre_stack_push: realloc succeeded.\n"));
+ assert(new_size > s->size);
+ s->size = new_size;
+ s->stack = new_buffer;
+ tre_stack_push(s, value);
+ }
+ }
+ return REG_OK;
+}
+
+#define define_pushf(typetag, type) \
+ declare_pushf(typetag, type) { \
+ union tre_stack_item item; \
+ item.typetag ## _value = value; \
+ return tre_stack_push(s, item); \
+}
+
+define_pushf(int, int)
+define_pushf(voidptr, void *)
+
+#define define_popf(typetag, type) \
+ declare_popf(typetag, type) { \
+ return s->stack[--s->ptr].typetag ## _value; \
+ }
+
+define_popf(int, int)
+define_popf(voidptr, void *)
+
+/* EOF */
diff --git a/qse/lib/cmn/tre-stack.h b/qse/lib/cmn/tre-stack.h
new file mode 100644
index 00000000..301ce16e
--- /dev/null
+++ b/qse/lib/cmn/tre-stack.h
@@ -0,0 +1,122 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-stack.h: Stack definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+
+#ifndef _QSE_LIB_CMN_TRE_STACK_H_
+#define _QSE_LIB_CMN_TRE_STACK_H_
+
+#include "tre.h"
+
+typedef struct tre_stack_rec tre_stack_t;
+
+/* Creates a new stack object. `size' is initial size in bytes, `max_size'
+ is maximum size, and `increment' specifies how much more space will be
+ allocated with realloc() if all space gets used up. Returns the stack
+ object or NULL if out of memory. */
+tre_stack_t *
+tre_stack_new(qse_mmgr_t* mmgr, int size, int max_size, int increment);
+
+/* Frees the stack object. */
+void
+tre_stack_destroy(tre_stack_t *s);
+
+/* Returns the current number of objects in the stack. */
+int
+tre_stack_num_objects(tre_stack_t *s);
+
+/* Each tre_stack_push_*(tre_stack_t *s, value) function pushes
+ `value' on top of stack `s'. Returns REG_ESPACE if out of memory.
+ This tries to realloc() more space before failing if maximum size
+ has not yet been reached. Returns REG_OK if successful. */
+#define declare_pushf(typetag, type) \
+ reg_errcode_t tre_stack_push_ ## typetag(tre_stack_t *s, type value)
+
+declare_pushf(voidptr, void *);
+declare_pushf(int, int);
+
+/* Each tre_stack_pop_*(tre_stack_t *s) function pops the topmost
+ element off of stack `s' and returns it. The stack must not be
+ empty. */
+#define declare_popf(typetag, type) \
+ type tre_stack_pop_ ## typetag(tre_stack_t *s)
+
+declare_popf(voidptr, void *);
+declare_popf(int, int);
+
+/* Just to save some typing. */
+#define STACK_PUSH(s, typetag, value) \
+ do \
+ { \
+ status = tre_stack_push_ ## typetag(s, value); \
+ } \
+ while (/*CONSTCOND*/0)
+
+#define STACK_PUSHX(s, typetag, value) \
+ { \
+ status = tre_stack_push_ ## typetag(s, value); \
+ if (status != REG_OK) \
+ break; \
+ }
+
+#define STACK_PUSHR(s, typetag, value) \
+ { \
+ reg_errcode_t _status; \
+ _status = tre_stack_push_ ## typetag(s, value); \
+ if (_status != REG_OK) \
+ return _status; \
+ }
+
+#endif /* TRE_STACK_H */
+
+/* EOF */
diff --git a/qse/lib/cmn/tre.c b/qse/lib/cmn/tre.c
new file mode 100644
index 00000000..64420ecb
--- /dev/null
+++ b/qse/lib/cmn/tre.c
@@ -0,0 +1,279 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+#include "tre.h"
+#include "tre-compile.h"
+#include
+
+#if 0
+QSE_IMPLEMENT_COMMON_FUNCTIONS (tre)
+
+qse_tre_t* qse_tre_open (qse_mmgr_t* mmgr, qse_size_t xtn, qse_tre_code_t* code)
+{
+ qse_tre_t* tre;
+
+ if (mmgr == QSE_NULL)
+ {
+ mmgr = QSE_MMGR_GETDFL();
+
+ QSE_ASSERTX (mmgr != QSE_NULL,
+ "Set the memory manager with QSE_MMGR_SETDFL()");
+
+ if (mmgr == QSE_NULL) return QSE_NULL;
+ }
+
+ tre = (qse_tre_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_tre_t) + xtn);
+ if (tre == QSE_NULL) return QSE_NULL;
+
+ if (qse_tre_init (tre, mmgr, code) <= -1)
+ {
+ QSE_MMGR_FREE (mmgr, tre);
+ return QSE_NULL;
+ }
+
+ return QSE_NULL;
+}
+
+void qse_tre_close (qse_tre_t* tre)
+{
+ qse_tre_fini (tre);
+ QSE_MMGR_FREE (tre->mmgr, tre);
+}
+#endif
+
+/*
+ tre_regcomp.c - TRE POSIX compatible regex compilation functions.
+
+ This software is released under a BSD-style license.
+ See the file LICENSE for details and copyright.
+
+*/
+
+int qse_tre_init (qse_tre_t* tre, qse_mmgr_t* mmgr)
+{
+ if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
+
+ QSE_MEMSET (tre, 0, QSE_SIZEOF(*tre));
+ tre->mmgr = mmgr;
+
+ return 0;
+}
+
+void qse_tre_fini (qse_tre_t* tre)
+{
+ if (tre->value)
+ {
+ tre_free (tre);
+ tre->value = QSE_NULL;
+ }
+}
+
+
+int qse_tre_compx (qse_tre_t* tre, const qse_char_t* regex, qse_size_t n, int cflags)
+{
+ int ret;
+
+ if (tre->value)
+ {
+ tre_free (tre);
+ tre->value = QSE_NULL;
+ }
+
+ ret = tre_compile (tre, regex, n, cflags);
+ if (ret > 0)
+ {
+ tre->value = QSE_NULL; /* just to make sure */
+ tre->errnum = ret;
+ return -1;
+ }
+
+ return 0;
+}
+
+int qse_tre_comp (qse_tre_t* tre, const qse_char_t* regex, int cflags)
+{
+ return qse_tre_compx (tre, regex, (regex? qse_strlen(regex):0), cflags);
+}
+
+/* Fills the POSIX.2 regmatch_t array according to the TNFA tag and match
+ endpoint values. */
+void tre_fill_pmatch(size_t nmatch, regmatch_t pmatch[], int cflags,
+ const tre_tnfa_t *tnfa, int *tags, int match_eo)
+{
+ tre_submatch_data_t *submatch_data;
+ unsigned int i, j;
+ int *parents;
+
+ i = 0;
+ if (match_eo >= 0 && !(cflags & REG_NOSUB))
+ {
+ /* Construct submatch offsets from the tags. */
+ DPRINT(("end tag = t%d = %d\n", tnfa->end_tag, match_eo));
+ submatch_data = tnfa->submatch_data;
+ while (i < tnfa->num_submatches && i < nmatch)
+ {
+ if (submatch_data[i].so_tag == tnfa->end_tag)
+ pmatch[i].rm_so = match_eo;
+ else
+ pmatch[i].rm_so = tags[submatch_data[i].so_tag];
+
+ if (submatch_data[i].eo_tag == tnfa->end_tag)
+ pmatch[i].rm_eo = match_eo;
+ else
+ pmatch[i].rm_eo = tags[submatch_data[i].eo_tag];
+
+ /* If either of the endpoints were not used, this submatch
+ was not part of the match. */
+ if (pmatch[i].rm_so == -1 || pmatch[i].rm_eo == -1)
+ pmatch[i].rm_so = pmatch[i].rm_eo = -1;
+
+ DPRINT(("pmatch[%d] = {t%d = %d, t%d = %d}\n", i,
+ submatch_data[i].so_tag, pmatch[i].rm_so,
+ submatch_data[i].eo_tag, pmatch[i].rm_eo));
+ i++;
+ }
+ /* Reset all submatches that are not within all of their parent
+ submatches. */
+ i = 0;
+ while (i < tnfa->num_submatches && i < nmatch)
+ {
+ if (pmatch[i].rm_eo == -1)
+ assert(pmatch[i].rm_so == -1);
+ assert(pmatch[i].rm_so <= pmatch[i].rm_eo);
+
+ parents = submatch_data[i].parents;
+ if (parents != QSE_NULL)
+ for (j = 0; parents[j] >= 0; j++)
+ {
+ DPRINT(("pmatch[%d] parent %d\n", i, parents[j]));
+ if (pmatch[i].rm_so < pmatch[parents[j]].rm_so
+ || pmatch[i].rm_eo > pmatch[parents[j]].rm_eo)
+ pmatch[i].rm_so = pmatch[i].rm_eo = -1;
+ }
+ i++;
+ }
+ }
+
+ while (i < nmatch)
+ {
+ pmatch[i].rm_so = -1;
+ pmatch[i].rm_eo = -1;
+ i++;
+ }
+}
+
+
+/*
+ Wrapper functions for POSIX compatible regexp matching.
+*/
+
+int tre_have_backrefs(const regex_t *preg)
+{
+ tre_tnfa_t *tnfa = (void *)preg->TRE_REGEX_T_FIELD;
+ return tnfa->have_backrefs;
+}
+
+static int tre_match(
+ const regex_t* preg, const void *string, qse_size_t len,
+ tre_str_type_t type, qse_size_t nmatch, regmatch_t pmatch[],
+ int eflags)
+{
+ tre_tnfa_t *tnfa = (void *)preg->TRE_REGEX_T_FIELD;
+ reg_errcode_t status;
+ int *tags = QSE_NULL, eo;
+ if (tnfa->num_tags > 0 && nmatch > 0)
+ {
+ tags = xmalloc (preg->mmgr, sizeof(*tags) * tnfa->num_tags);
+ if (tags == QSE_NULL) return REG_ESPACE;
+ }
+
+ /* Dispatch to the appropriate matcher. */
+ if (tnfa->have_backrefs || eflags & REG_BACKTRACKING_MATCHER)
+ {
+ /* The regex has back references, use the backtracking matcher. */
+ if (type == STR_USER)
+ {
+ const tre_str_source *source = string;
+ if (source->rewind == QSE_NULL || source->compare == QSE_NULL)
+ /* The backtracking matcher requires rewind and compare
+ capabilities from the input stream. */
+ return REG_BADPAT;
+ }
+
+ status = tre_tnfa_run_backtrack (
+ preg->mmgr, tnfa, string, (int)len, type,
+ tags, eflags, &eo);
+ }
+ else
+ {
+ /* Exact matching, no back references, use the parallel matcher. */
+ status = tre_tnfa_run_parallel (
+ preg->mmgr, tnfa, string, (int)len, type,
+ tags, eflags, &eo);
+ }
+
+ if (status == REG_OK)
+ /* A match was found, so fill the submatch registers. */
+ tre_fill_pmatch(nmatch, pmatch, tnfa->cflags, tnfa, tags, eo);
+ if (tags) xfree (preg->mmgr, tags);
+ return status;
+}
+
+int qse_tre_execx (
+ qse_tre_t* tre, const qse_char_t *str, qse_size_t len,
+ qse_size_t nmatch, regmatch_t pmatch[], int eflags)
+{
+ int ret;
+
+ if (tre->value == QSE_NULL)
+ {
+ /* regular expression is bad as none is compiled yet */
+ tre->errnum = QSE_TRE_EBADPAT;
+ return -1;
+ }
+#ifdef QSE_CHAR_IS_WCHAR
+ ret = tre_match (tre, str, len, STR_WIDE, nmatch, pmatch, eflags);
+#else
+ ret = tre_match (tre, str, len, STR_BYTE, nmatch, pmatch, eflags);
+#endif
+ if (ret > 0)
+ {
+ tre->errnum = ret;
+ return -1;
+ }
+
+ return 0;
+}
+
+int qse_tre_exec (
+ qse_tre_t* tre, const qse_char_t* str,
+ qse_size_t nmatch, regmatch_t pmatch[], int eflags)
+{
+ return qse_tre_execx (tre, str, (unsigned)-1, nmatch, pmatch, eflags);
+}
+
+#if 0
+int qse_tre_execsrc (
+ const regex_t *preg, const tre_str_source *str,
+ qse_size_t nmatch, regmatch_t pmatch[], int eflags)
+{
+ return tre_match (preg, str, (unsigned)-1, STR_USER, nmatch, pmatch, eflags);
+}
+#endif
diff --git a/qse/lib/cmn/tre.h b/qse/lib/cmn/tre.h
new file mode 100644
index 00000000..dbd630df
--- /dev/null
+++ b/qse/lib/cmn/tre.h
@@ -0,0 +1,354 @@
+/*
+ * $Id$
+ *
+ Copyright 2006-2011 Chung, Hyung-Hwan.
+ This file is part of QSE.
+
+ QSE is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Lesser General Public License as
+ published by the Free Software Foundation, either version 3 of
+ the License, or (at your option) any later version.
+
+ QSE is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with QSE. If not, see .
+ */
+
+/*
+ tre-internal.h - TRE internal definitions
+
+This is the license, copyright notice, and disclaimer for TRE, a regex
+matching package (library and tools) with support for approximate
+matching.
+
+Copyright (c) 2001-2009 Ville Laurikari
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions
+are met:
+
+ 1. Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ 2. Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in the
+ documentation and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS
+``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef _QSE_LIB_CMN_TRE_H_
+#define _QSE_LIB_CMN_TRE_H_
+
+#include
+
+#ifdef QSE_CHAR_IS_WCHAR
+# define TRE_WCHAR
+/*
+# define TRE_MULTIBYTE
+# define TRE_MBSTATE
+*/
+#endif
+
+#define TRE_REGEX_T_FIELD value
+#define assert QSE_ASSERT
+#define NULL QSE_NULL
+
+#include
+#include
+#include
+#include "mem.h"
+
+#define tre_islower(c) QSE_ISLOWER(c)
+#define tre_isupper(c) QSE_ISUPPER(c)
+#define tre_isalpha(c) QSE_ISALPHA(c)
+#define tre_isdigit(c) QSE_ISDIGIT(c)
+#define tre_isxdigit(c) QSE_ISXDIGIT(c)
+#define tre_isalnum(c) QSE_ISALNUM(c)
+
+#define tre_isspace(c) QSE_ISSPACE(c)
+#define tre_isprint(c) QSE_ISPRINT(c)
+#define tre_isgraph(c) QSE_ISGRAPH(c)
+#define tre_iscntrl(c) QSE_ISCNTRL(c)
+#define tre_ispunct(c) QSE_ISPUNCT(c)
+#define tre_isblank(c) QSE_ISBLANK(c)
+
+#define tre_tolower(c) QSE_TOLOWER(c)
+#define tre_toupper(c) QSE_TOUPPER(c)
+
+typedef qse_char_t tre_char_t;
+typedef qse_cint_t tre_cint_t;
+
+#define size_t qse_size_t
+#define regex_t qse_tre_t
+#define regmatch_t qse_tre_match_t
+#define reg_errcode_t qse_tre_errnum_t
+#define tre_str_source qse_tre_strsrc_t
+
+
+#define REG_OK QSE_TRE_ENOERR
+#define REG_ESPACE QSE_TRE_ENOMEM
+#define REG_NOMATCH QSE_TRE_ENOMATCH
+#define REG_BADPAT QSE_TRE_EBADPAT
+#define REG_ECOLLATE QSE_TRE_ECOLLATE
+#define REG_ECTYPE QSE_TRE_ECTYPE
+#define REG_EESCAPE QSE_TRE_EESCAPE
+#define REG_ESUBREG QSE_TRE_ESUBREG
+#define REG_EBRACK QSE_TRE_EBRACK
+#define REG_EPAREN QSE_TRE_EPAREN
+#define REG_EBRACE QSE_TRE_EBRACE
+#define REG_BADBR QSE_TRE_EBADBR
+#define REG_ERANGE QSE_TRE_ERANGE
+#define REG_BADRPT QSE_TRE_EBADRPT
+
+/* The maximum number of iterations in a bound expression. */
+#undef RE_DUP_MAX
+#define RE_DUP_MAX 255
+
+/* POSIX tre_regcomp() flags. */
+#define REG_EXTENDED QSE_TRE_EXTENDED
+#define REG_ICASE QSE_TRE_IGNORECASE
+#define REG_NEWLINE QSE_TRE_NEWLINE
+#define REG_NOSUB QSE_TRE_NOSUBREG
+/* Extra tre_regcomp() flags. */
+#define REG_LITERAL QSE_TRE_LITERAL
+#define REG_RIGHT_ASSOC QSE_TRE_RIGHTASSOC
+#define REG_UNGREEDY QSE_TRE_UNGREEDY
+
+/* POSIX tre_regexec() flags. */
+#define REG_NOTBOL QSE_TRE_NOTBOL
+#define REG_NOTEOL QSE_TRE_NOTEOL
+#define REG_BACKTRACKING_MATCHER QSE_TRE_BACKTRACKING
+
+
+#define tre_strlen(c) qse_strlen(c)
+
+typedef qse_pma_t* tre_mem_t;
+
+#define tre_mem_new(mmgr) qse_pma_open(mmgr,0)
+#define tre_mem_destroy(mem) qse_pma_close(mem)
+#define tre_mem_alloc(mem,size) qse_pma_alloc(mem,size)
+#define tre_mem_calloc(mem,size) qse_pma_calloc(mem,size)
+
+#define xmalloc(mmgr,size) QSE_MMGR_ALLOC(mmgr,size)
+#define xfree(mmgr,ptr) QSE_MMGR_FREE(mmgr,ptr)
+#define xrealloc(mmgr,ptr,new_size) QSE_MMGR_REALLOC(mmgr, ptr, new_size)
+
+
+/* tre-ast.h */
+#define tre_ast_new_node qse_tre_astnewnode
+#define tre_ast_new_literal qse_tre_astnewliteral
+#define tre_ast_new_iter qse_tre_astnewiter
+#define tre_ast_new_union qse_tre_astnewunion
+#define tre_ast_new_catenation qse_tre_astnewcatenation
+
+/* tre-parse.h */
+#define tre_parse qse_tre_parse
+
+/* tre-stack.h */
+#define tre_stack_destroy qse_tre_stackfree
+#define tre_stack_new qse_tre_stacknew
+#define tre_stack_num_objects qse_tre_stacknumobjs
+#define tre_stack_pop_int qse_tre_stackpopint
+#define tre_stack_pop_voidptr qse_tre_stackpopvoidptr
+#define tre_stack_push_int qse_tre_stackpushint
+#define tre_stack_push_voidptr qse_tre_stackpushvoidptr
+
+/* this tre.h */
+#define tre_compile qse_tre_compile
+#define tre_free qse_tre_free
+#define tre_fill_pmatch qse_tre_fillpmatch
+#define tre_tnfa_run_backtrack qse_tre_runbacktrack
+#define tre_tnfa_run_parallel qse_tre_runparallel
+#define tre_have_backrefs qse_tre_havebackrefs
+
+/* Define the character types and functions. */
+#ifdef TRE_WCHAR
+# define TRE_CHAR_MAX QSE_TYPE_MAX(qse_wchar_t)
+# ifdef TRE_MULTIBYTE
+# define TRE_MB_CUR_MAX (qse_getmbcurmax())
+# else /* !TRE_MULTIBYTE */
+# define TRE_MB_CUR_MAX 1
+# endif /* !TRE_MULTIBYTE */
+#else /* !TRE_WCHAR */
+# define TRE_CHAR_MAX 255
+# define TRE_MB_CUR_MAX 1
+#endif /* !TRE_WCHAR */
+
+#define DPRINT(msg)
+
+typedef qse_ctype_t tre_ctype_t;
+#define tre_isctype(c,t) QSE_ISCTYPE(c,t)
+
+typedef enum { STR_WIDE, STR_BYTE, STR_MBS, STR_USER } tre_str_type_t;
+
+/* Returns number of bytes to add to (char *)ptr to make it
+ properly aligned for the type. */
+#define ALIGN(ptr, type) \
+ ((((long)ptr) % sizeof(type)) \
+ ? (sizeof(type) - (((long)ptr) % sizeof(type))) \
+ : 0)
+
+#undef MAX
+#undef MIN
+#define MAX(a, b) (((a) >= (b)) ? (a) : (b))
+#define MIN(a, b) (((a) <= (b)) ? (a) : (b))
+
+/* Define STRF to the correct printf formatter for strings. */
+#ifdef TRE_WCHAR
+#define STRF "ls"
+#else /* !TRE_WCHAR */
+#define STRF "s"
+#endif /* !TRE_WCHAR */
+
+/* TNFA transition type. A TNFA state is an array of transitions,
+ the terminator is a transition with NULL `state'. */
+typedef struct tnfa_transition tre_tnfa_transition_t;
+
+struct tnfa_transition
+{
+ /* Range of accepted characters. */
+ tre_cint_t code_min;
+ tre_cint_t code_max;
+ /* Pointer to the destination state. */
+ tre_tnfa_transition_t *state;
+ /* ID number of the destination state. */
+ int state_id;
+ /* -1 terminated array of tags (or NULL). */
+ int *tags;
+ /* Matching parameters settings (or NULL). */
+ int *params;
+ /* Assertion bitmap. */
+ int assertions;
+ /* Assertion parameters. */
+ union
+ {
+ /* Character class assertion. */
+ tre_ctype_t class;
+ /* Back reference assertion. */
+ int backref;
+ } u;
+ /* Negative character class assertions. */
+ tre_ctype_t *neg_classes;
+};
+
+
+/* Assertions. */
+#define ASSERT_AT_BOL 1 /* Beginning of line. */
+#define ASSERT_AT_EOL 2 /* End of line. */
+#define ASSERT_CHAR_CLASS 4 /* Character class in `class'. */
+#define ASSERT_CHAR_CLASS_NEG 8 /* Character classes in `neg_classes'. */
+#define ASSERT_AT_BOW 16 /* Beginning of word. */
+#define ASSERT_AT_EOW 32 /* End of word. */
+#define ASSERT_AT_WB 64 /* Word boundary. */
+#define ASSERT_AT_WB_NEG 128 /* Not a word boundary. */
+#define ASSERT_BACKREF 256 /* A back reference in `backref'. */
+#define ASSERT_LAST 256
+
+/* Tag directions. */
+typedef enum
+{
+ TRE_TAG_MINIMIZE = 0,
+ TRE_TAG_MAXIMIZE = 1
+} tre_tag_direction_t;
+
+/* Parameters that can be changed dynamically while matching. */
+typedef enum
+{
+ TRE_PARAM_COST_INS = 0,
+ TRE_PARAM_COST_DEL = 1,
+ TRE_PARAM_COST_SUBST = 2,
+ TRE_PARAM_COST_MAX = 3,
+ TRE_PARAM_MAX_INS = 4,
+ TRE_PARAM_MAX_DEL = 5,
+ TRE_PARAM_MAX_SUBST = 6,
+ TRE_PARAM_MAX_ERR = 7,
+ TRE_PARAM_DEPTH = 8,
+ TRE_PARAM_LAST = 9
+} tre_param_t;
+
+/* Unset matching parameter */
+#define TRE_PARAM_UNSET -1
+
+/* Signifies the default matching parameter value. */
+#define TRE_PARAM_DEFAULT -2
+
+/* Instructions to compute submatch register values from tag values
+ after a successful match. */
+struct tre_submatch_data
+{
+ /* Tag that gives the value for rm_so (submatch start offset). */
+ int so_tag;
+ /* Tag that gives the value for rm_eo (submatch end offset). */
+ int eo_tag;
+ /* List of submatches this submatch is contained in. */
+ int *parents;
+};
+
+typedef struct tre_submatch_data tre_submatch_data_t;
+
+
+/* TNFA definition. */
+typedef struct tnfa tre_tnfa_t;
+
+struct tnfa
+{
+ tre_tnfa_transition_t *transitions;
+ unsigned int num_transitions;
+ tre_tnfa_transition_t *initial;
+ tre_tnfa_transition_t *final;
+ tre_submatch_data_t *submatch_data;
+ char *firstpos_chars;
+ int first_char;
+ unsigned int num_submatches;
+ tre_tag_direction_t *tag_directions;
+ int *minimal_tags;
+ int num_tags;
+ int num_minimals;
+ int end_tag;
+ int num_states;
+ int cflags;
+ int have_backrefs;
+ int have_approx;
+ int params_depth;
+};
+
+
+int tre_compile (regex_t *preg, const tre_char_t *regex, size_t n, int cflags);
+
+void tre_free (regex_t *preg);
+
+void tre_fill_pmatch(
+ size_t nmatch, regmatch_t pmatch[], int cflags,
+ const tre_tnfa_t *tnfa, int *tags, int match_eo);
+
+reg_errcode_t tre_tnfa_run_backtrack(
+ qse_mmgr_t* mmgr, const tre_tnfa_t *tnfa, const void *string,
+ int len, tre_str_type_t type, int *match_tags,
+ int eflags, int *match_end_ofs);
+
+
+reg_errcode_t tre_tnfa_run_parallel(
+ qse_mmgr_t* mmgr, const tre_tnfa_t *tnfa, const void *string, int len,
+ tre_str_type_t type, int *match_tags, int eflags,
+ int *match_end_ofs);
+
+
+#endif
+
+/* EOF */
diff --git a/qse/lib/cmn/xma.c b/qse/lib/cmn/xma.c
index 8e9c9946..13e2e7d7 100644
--- a/qse/lib/cmn/xma.c
+++ b/qse/lib/cmn/xma.c
@@ -122,7 +122,7 @@ qse_xma_t* qse_xma_open (
xma = (qse_xma_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(*xma) + xtnsize);
if (xma == QSE_NULL) return QSE_NULL;
- if (qse_xma_init (xma, mmgr, zonesize) == QSE_NULL)
+ if (qse_xma_init (xma, mmgr, zonesize) <= -1)
{
QSE_MMGR_FREE (mmgr, xma);
return QSE_NULL;
@@ -137,7 +137,7 @@ void qse_xma_close (qse_xma_t* xma)
QSE_MMGR_FREE (xma->mmgr, xma);
}
-qse_xma_t* qse_xma_init (qse_xma_t* xma, qse_mmgr_t* mmgr, qse_size_t zonesize)
+int qse_xma_init (qse_xma_t* xma, qse_mmgr_t* mmgr, qse_size_t zonesize)
{
qse_xma_blk_t* free;
qse_size_t xfi;
@@ -152,7 +152,7 @@ qse_xma_t* qse_xma_init (qse_xma_t* xma, qse_mmgr_t* mmgr, qse_size_t zonesize)
/* allocate a memory chunk to use for actual memory allocation */
free = QSE_MMGR_ALLOC (mmgr, zonesize);
- if (free == QSE_NULL) return QSE_NULL;
+ if (free == QSE_NULL) return -1;
/* initialize the header part of the free chunk */
free->avail = 1;
@@ -184,7 +184,7 @@ qse_xma_t* qse_xma_init (qse_xma_t* xma, qse_mmgr_t* mmgr, qse_size_t zonesize)
xma->stat.nused = 0;
#endif
- return xma;
+ return 0;
}
void qse_xma_fini (qse_xma_t* xma)
diff --git a/qse/lib/cut/StdCut.cpp b/qse/lib/cut/StdCut.cpp
index d75c7f45..9b807b29 100644
--- a/qse/lib/cut/StdCut.cpp
+++ b/qse/lib/cut/StdCut.cpp
@@ -173,7 +173,7 @@ int StdCut::StringStream::open (Data& io)
{
if (!out.inited)
{
- if (qse_str_init (&out.buf, ((Cut*)io)->getMmgr(), 256) == QSE_NULL)
+ if (qse_str_init (&out.buf, ((Cut*)io)->getMmgr(), 256) <= -1)
{
((Cut*)io)->setError (QSE_CUT_ENOMEM);
return -1;
diff --git a/qse/lib/cut/cut.c b/qse/lib/cut/cut.c
index 240d3e26..fc3e95ed 100644
--- a/qse/lib/cut/cut.c
+++ b/qse/lib/cut/cut.c
@@ -24,7 +24,7 @@
QSE_IMPLEMENT_COMMON_FUNCTIONS (cut)
-static qse_cut_t* qse_cut_init (qse_cut_t* cut, qse_mmgr_t* mmgr);
+static int qse_cut_init (qse_cut_t* cut, qse_mmgr_t* mmgr);
static void qse_cut_fini (qse_cut_t* cut);
#define SETERR0(cut,num) \
@@ -92,7 +92,7 @@ qse_cut_t* qse_cut_open (qse_mmgr_t* mmgr, qse_size_t xtn)
cut = (qse_cut_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_cut_t) + xtn);
if (cut == QSE_NULL) return QSE_NULL;
- if (qse_cut_init (cut, mmgr) == QSE_NULL)
+ if (qse_cut_init (cut, mmgr) <= -1)
{
QSE_MMGR_FREE (cut->mmgr, cut);
return QSE_NULL;
@@ -107,7 +107,7 @@ void qse_cut_close (qse_cut_t* cut)
QSE_MMGR_FREE (cut->mmgr, cut);
}
-static qse_cut_t* qse_cut_init (qse_cut_t* cut, qse_mmgr_t* mmgr)
+static int qse_cut_init (qse_cut_t* cut, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -125,13 +125,13 @@ static qse_cut_t* qse_cut_init (qse_cut_t* cut, qse_mmgr_t* mmgr)
cut->e.in.flds = cut->e.in.sflds;
if (qse_str_init (
- &cut->e.in.line, QSE_MMGR(cut), DFL_LINE_CAPA) == QSE_NULL)
+ &cut->e.in.line, QSE_MMGR(cut), DFL_LINE_CAPA) <= -1)
{
SETERR0 (cut, QSE_CUT_ENOMEM);
- return QSE_NULL;
+ return -1;
}
- return cut;
+ return 0;
}
static void qse_cut_fini (qse_cut_t* cut)
diff --git a/qse/lib/net/htrd.c b/qse/lib/net/htrd.c
index 7602a726..46250909 100644
--- a/qse/lib/net/htrd.c
+++ b/qse/lib/net/htrd.c
@@ -151,7 +151,7 @@ qse_htrd_t* qse_htrd_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
);
if (htrd == QSE_NULL) return QSE_NULL;
- if (qse_htrd_init (htrd, mmgr) == QSE_NULL)
+ if (qse_htrd_init (htrd, mmgr) <= -1)
{
QSE_MMGR_FREE (htrd->mmgr, htrd);
return QSE_NULL;
@@ -166,7 +166,7 @@ void qse_htrd_close (qse_htrd_t* htrd)
QSE_MMGR_FREE (htrd->mmgr, htrd);
}
-qse_htrd_t* qse_htrd_init (qse_htrd_t* htrd, qse_mmgr_t* mmgr)
+int qse_htrd_init (qse_htrd_t* htrd, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -180,17 +180,17 @@ qse_htrd_t* qse_htrd_init (qse_htrd_t* htrd, qse_mmgr_t* mmgr)
qse_mbs_init (&htrd->fed.b.raw, htrd->mmgr, 0);
qse_mbs_init (&htrd->fed.b.tra, htrd->mmgr, 0);
- if (qse_htre_init (&htrd->re, mmgr) == QSE_NULL)
+ if (qse_htre_init (&htrd->re, mmgr) <= -1)
{
qse_mbs_fini (&htrd->fed.b.tra);
qse_mbs_fini (&htrd->fed.b.raw);
#if 0
qse_mbs_fini (&htrd->tmp.qparam);
#endif
- return QSE_NULL;
+ return -1;
}
- return htrd;
+ return 0;
}
void qse_htrd_fini (qse_htrd_t* htrd)
diff --git a/qse/lib/net/htre.c b/qse/lib/net/htre.c
index 3410325a..dce3a54d 100644
--- a/qse/lib/net/htre.c
+++ b/qse/lib/net/htre.c
@@ -21,21 +21,18 @@
#include
#include "../cmn/mem.h"
-qse_htre_t* qse_htre_init (qse_htre_t* re, qse_mmgr_t* mmgr)
+int qse_htre_init (qse_htre_t* re, qse_mmgr_t* mmgr)
{
QSE_MEMSET (re, 0, QSE_SIZEOF(*re));
re->mmgr = mmgr;
- if (qse_htb_init (&re->hdrtab, mmgr, 60, 70, 1, 1) == QSE_NULL)
- {
- return QSE_NULL;
- }
+ if (qse_htb_init (&re->hdrtab, mmgr, 60, 70, 1, 1) <= -1) return -1;
qse_mbs_init (&re->content, mmgr, 0);
qse_mbs_init (&re->qpath_or_smesg, mmgr, 0);
qse_mbs_init (&re->qparam, mmgr, 0);
- return re;
+ return 0;
}
void qse_htre_fini (qse_htre_t* re)
diff --git a/qse/lib/net/httpd.c b/qse/lib/net/httpd.c
index 9c3f589c..ace83686 100644
--- a/qse/lib/net/httpd.c
+++ b/qse/lib/net/httpd.c
@@ -72,7 +72,7 @@ qse_httpd_t* qse_httpd_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
);
if (httpd == QSE_NULL) return QSE_NULL;
- if (qse_httpd_init (httpd, mmgr) == QSE_NULL)
+ if (qse_httpd_init (httpd, mmgr) <= -1)
{
QSE_MMGR_FREE (httpd->mmgr, httpd);
return QSE_NULL;
@@ -87,7 +87,7 @@ void qse_httpd_close (qse_httpd_t* httpd)
QSE_MMGR_FREE (httpd->mmgr, httpd);
}
-qse_httpd_t* qse_httpd_init (qse_httpd_t* httpd, qse_mmgr_t* mmgr)
+int qse_httpd_init (qse_httpd_t* httpd, qse_mmgr_t* mmgr)
{
QSE_MEMSET (httpd, 0, QSE_SIZEOF(*httpd));
httpd->mmgr = mmgr;
@@ -97,7 +97,7 @@ qse_httpd_t* qse_httpd_init (qse_httpd_t* httpd, qse_mmgr_t* mmgr)
pthread_mutex_init (&httpd->listener.mutex, QSE_NULL);
#endif
- return httpd;
+ return 0;
}
void qse_httpd_fini (qse_httpd_t* httpd)
diff --git a/qse/lib/net/httpd.h b/qse/lib/net/httpd.h
index bfbab0f8..2d2f8fa7 100644
--- a/qse/lib/net/httpd.h
+++ b/qse/lib/net/httpd.h
@@ -139,7 +139,7 @@ struct qse_httpd_t
extern "C" {
#endif
-qse_httpd_t* qse_httpd_init (
+int qse_httpd_init (
qse_httpd_t* httpd,
qse_mmgr_t* mmgr
);
diff --git a/qse/lib/scm/scm.c b/qse/lib/scm/scm.c
index 35c34740..56927f37 100644
--- a/qse/lib/scm/scm.c
+++ b/qse/lib/scm/scm.c
@@ -22,7 +22,7 @@
QSE_IMPLEMENT_COMMON_FUNCTIONS (scm)
-static qse_scm_t* qse_scm_init (
+static int qse_scm_init (
qse_scm_t* scm,
qse_mmgr_t* mmgr,
qse_size_t mem_ubound,
@@ -54,7 +54,7 @@ qse_scm_t* qse_scm_open (
);
if (scm == QSE_NULL) return QSE_NULL;
- if (qse_scm_init (scm, mmgr, mem_ubound, mem_ubound_inc) == QSE_NULL)
+ if (qse_scm_init (scm, mmgr, mem_ubound, mem_ubound_inc) <= -1)
{
QSE_MMGR_FREE (scm->mmgr, scm);
return QSE_NULL;
@@ -183,7 +183,7 @@ static qse_scm_t* qse_scm_init (
scm->r.curc = QSE_CHAR_EOF;
scm->r.curloc.line = 1;
scm->r.curloc.colm = 0;
- if (qse_str_init(&scm->r.t.name, mmgr, 256) == QSE_NULL) return QSE_NULL;
+ if (qse_str_init(&scm->r.t.name, mmgr, 256) <= -1) return -1;
/* initialize common values */
scm->nil = &static_values[0];
@@ -220,12 +220,12 @@ static qse_scm_t* qse_scm_init (
scm->e.env = scm->gloenv;
if (build_syntax_entities (scm) <= -1) goto oops;
- return scm;
+ return 0;
oops:
delete_all_entity_blocks (scm);
qse_str_fini (&scm->r.t.name);
- return QSE_NULL;
+ return -1;
}
static void qse_scm_fini (qse_scm_t* scm)
diff --git a/qse/lib/sed/StdSed.cpp b/qse/lib/sed/StdSed.cpp
index e52efe2a..5c784a44 100644
--- a/qse/lib/sed/StdSed.cpp
+++ b/qse/lib/sed/StdSed.cpp
@@ -1,5 +1,5 @@
/*
- * $Id: StdSed.cpp 441 2011-04-22 14:28:43Z hyunghwan.chung $
+ * $Id: StdSed.cpp 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -202,7 +202,7 @@ int StdSed::StringStream::open (Data& io)
{
if (!out.inited)
{
- if (qse_str_init (&out.buf, ((Sed*)io)->getMmgr(), 256) == QSE_NULL)
+ if (qse_str_init (&out.buf, ((Sed*)io)->getMmgr(), 256) <= -1)
{
((Sed*)io)->setError (QSE_SED_ENOMEM);
return -1;
diff --git a/qse/lib/sed/sed.c b/qse/lib/sed/sed.c
index 63f478e3..95dcf42f 100644
--- a/qse/lib/sed/sed.c
+++ b/qse/lib/sed/sed.c
@@ -1,5 +1,5 @@
/*
- * $Id: sed.c 462 2011-05-18 14:36:40Z hyunghwan.chung $
+ * $Id: sed.c 556 2011-08-31 15:43:46Z hyunghwan.chung $
*
Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE.
@@ -28,7 +28,7 @@ QSE_IMPLEMENT_COMMON_FUNCTIONS (sed)
static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd);
static void free_all_command_blocks (qse_sed_t* sed);
-static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr);
+static int qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr);
static void qse_sed_fini (qse_sed_t* sed);
#define SETERR0(sed,num,loc) \
@@ -58,7 +58,7 @@ qse_sed_t* qse_sed_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
sed = (qse_sed_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(qse_sed_t) + xtnsize);
if (sed == QSE_NULL) return QSE_NULL;
- if (qse_sed_init (sed, mmgr) == QSE_NULL)
+ if (qse_sed_init (sed, mmgr) <= -1)
{
QSE_MMGR_FREE (sed->mmgr, sed);
return QSE_NULL;
@@ -73,7 +73,7 @@ void qse_sed_close (qse_sed_t* sed)
QSE_MMGR_FREE (sed->mmgr, sed);
}
-static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
+static int qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
{
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@@ -81,75 +81,44 @@ static qse_sed_t* qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
sed->mmgr = mmgr;
sed->errstr = qse_sed_dflerrstr;
- if (qse_str_init (&sed->tmp.rex, mmgr, 0) == QSE_NULL)
- {
- SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
- return QSE_NULL;
- }
-
- if (qse_str_init (&sed->tmp.lab, mmgr, 0) == QSE_NULL)
- {
- qse_str_fini (&sed->tmp.lab);
- SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
- return QSE_NULL;
- }
+ if (qse_str_init (&sed->tmp.rex, mmgr, 0) <= -1) goto oops_1;
+ if (qse_str_init (&sed->tmp.lab, mmgr, 0) <= -1) goto oops_2;
if (qse_map_init (&sed->tmp.labs, mmgr,
- 128, 70, QSE_SIZEOF(qse_char_t), 1) == QSE_NULL)
- {
- qse_str_fini (&sed->tmp.lab);
- qse_str_fini (&sed->tmp.rex);
- SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
- return QSE_NULL;
- }
- qse_map_setmancbs (&sed->tmp.labs,
+ 128, 70, QSE_SIZEOF(qse_char_t), 1) <= -1) goto oops_3;
+ qse_map_setmancbs (
+ &sed->tmp.labs,
qse_map_mancbs(QSE_MAP_MANCBS_INLINE_KEY_COPIER)
);
- if (qse_lda_init (&sed->e.txt.appended, mmgr, 32) == QSE_NULL)
- {
- qse_map_fini (&sed->tmp.labs);
- qse_str_fini (&sed->tmp.lab);
- qse_str_fini (&sed->tmp.rex);
- return QSE_NULL;
- }
-
- if (qse_str_init (&sed->e.txt.read, mmgr, 256) == QSE_NULL)
- {
- qse_lda_fini (&sed->e.txt.appended);
- qse_map_fini (&sed->tmp.labs);
- qse_str_fini (&sed->tmp.lab);
- qse_str_fini (&sed->tmp.rex);
- return QSE_NULL;
- }
-
- if (qse_str_init (&sed->e.txt.held, mmgr, 256) == QSE_NULL)
- {
- qse_str_fini (&sed->e.txt.read);
- qse_lda_fini (&sed->e.txt.appended);
- qse_map_fini (&sed->tmp.labs);
- qse_str_fini (&sed->tmp.lab);
- qse_str_fini (&sed->tmp.rex);
- return QSE_NULL;
- }
-
- if (qse_str_init (&sed->e.txt.subst, mmgr, 256) == QSE_NULL)
- {
- qse_str_fini (&sed->e.txt.held);
- qse_str_fini (&sed->e.txt.read);
- qse_lda_fini (&sed->e.txt.appended);
- qse_map_fini (&sed->tmp.labs);
- qse_str_fini (&sed->tmp.lab);
- qse_str_fini (&sed->tmp.rex);
- return QSE_NULL;
- }
+ if (qse_lda_init (&sed->e.txt.appended, mmgr, 32) <= -1) goto oops_4;
+ if (qse_str_init (&sed->e.txt.read, mmgr, 256) <= -1) goto oops_5;
+ if (qse_str_init (&sed->e.txt.held, mmgr, 256) <= -1) goto oops_6;
+ if (qse_str_init (&sed->e.txt.subst, mmgr, 256) <= -1) goto oops_7;
/* on init, the last points to the first */
sed->cmd.lb = &sed->cmd.fb;
/* the block has no data yet */
sed->cmd.fb.len = 0;
- return sed;
+ return 0;
+
+
+oops_7:
+ qse_str_fini (&sed->e.txt.held);
+oops_6:
+ qse_str_fini (&sed->e.txt.read);
+oops_5:
+ qse_lda_fini (&sed->e.txt.appended);
+oops_4:
+ qse_map_fini (&sed->tmp.labs);
+oops_3:
+ qse_str_fini (&sed->tmp.lab);
+oops_2:
+ qse_str_fini (&sed->tmp.rex);
+oops_1:
+ SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
+ return -1;
}
static void qse_sed_fini (qse_sed_t* sed)
@@ -2659,8 +2628,9 @@ int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
sed->e.out.fun = outf;
sed->e.out.eof = 0;
sed->e.out.len = 0;
- if (qse_map_init (&sed->e.out.files, sed->mmgr,
- 128, 70, QSE_SIZEOF(qse_char_t), 1) == QSE_NULL)
+ if (qse_map_init (
+ &sed->e.out.files, sed->mmgr,
+ 128, 70, QSE_SIZEOF(qse_char_t), 1) <= -1)
{
SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
return -1;
@@ -2673,7 +2643,7 @@ int qse_sed_exec (qse_sed_t* sed, qse_sed_io_fun_t inf, qse_sed_io_fun_t outf)
sed->e.in.len = 0;
sed->e.in.pos = 0;
sed->e.in.num = 0;
- if (qse_str_init (&sed->e.in.line, QSE_MMGR(sed), 256) == QSE_NULL)
+ if (qse_str_init (&sed->e.in.line, QSE_MMGR(sed), 256) <= -1)
{
qse_map_fini (&sed->e.out.files);
SETERR0 (sed, QSE_SED_ENOMEM, QSE_NULL);
diff --git a/qse/lib/stx/par.c b/qse/lib/stx/par.c
index bf3ba061..6e27435a 100644
--- a/qse/lib/stx/par.c
+++ b/qse/lib/stx/par.c
@@ -85,7 +85,7 @@ qse_stc_t* qse_stc_open (qse_mmgr_t* mmgr, qse_size_t xtnsize, qse_stx_t* stx)
stc = (qse_stc_t*) QSE_MMGR_ALLOC (mmgr, QSE_SIZEOF(*stc) + xtnsize);
if (stc == QSE_NULL) return QSE_NULL;
- if (qse_stc_init (stc, mmgr, stx) == QSE_NULL)
+ if (qse_stc_init (stc, mmgr, stx) <= -1)
{
QSE_MMGR_FREE (mmgr, stc);
return QSE_NULL;
@@ -100,21 +100,21 @@ void qse_stc_close (qse_stc_t* stc)
QSE_MMGR_FREE (stc->mmgr, stc);
}
-qse_stc_t* qse_stc_init (qse_stc_t* stc, qse_mmgr_t* mmgr, qse_stx_t* stx)
+int qse_stc_init (qse_stc_t* stc, qse_mmgr_t* mmgr, qse_stx_t* stx)
{
QSE_MEMSET (stc, 0, QSE_SIZEOF(*stc));
stc->mmgr = mmgr;
stc->stx = stx;
- if (qse_str_init (&stc->method_name, mmgr, 0) == QSE_NULL)
+ if (qse_str_init (&stc->method_name, mmgr, 0) <= -1)
{
- return QSE_NULL;
+ return -1;
}
- if (qse_str_init (&stc->token.name, mmgr, 0) == QSE_NULL)
+ if (qse_str_init (&stc->token.name, mmgr, 0) <= -1)
{
qse_str_fini (&stc->method_name);
- return QSE_NULL;
+ return -1;
}
stc->token.type = TOKEN_END;
@@ -125,7 +125,7 @@ qse_lba_t linear byte array
*/
if (qse_lda_init (
&stc->bytecode, mmgr, 256,
- QSE_SIZEOF(qse_byte_t), QSE_NULL) == QSE_NULL)
+ QSE_SIZEOF(qse_byte_t), QSE_NULL) <= -1)
{
qse_str_fini (&stc->method_name);
qse_str_fini (&stc->token.name);
@@ -1238,7 +1238,7 @@ static int __parse_keyword_message (qse_stc_t* stc, qse_bool_t is_super)
if (__parse_binary_message (stc, is_super) == -1) return -1;
if (stc->token.type != TOKEN_KEYWORD) return 0;
- if (qse_str_init (&name, stc->mmgr, 0) == QSE_NULL)
+ if (qse_str_init (&name, stc->mmgr, 0) <= -1)
{
stc->error_code = QSE_STC_ERROR_MEMORY;
return -1;
diff --git a/qse/lib/stx/stx.c b/qse/lib/stx/stx.c
index 6dcdfb1c..074211a3 100644
--- a/qse/lib/stx/stx.c
+++ b/qse/lib/stx/stx.c
@@ -7,15 +7,15 @@
#include "boot.h"
#include "../cmn/mem.h"
-qse_stx_t* qse_stx_init (qse_stx_t* stx, qse_mmgr_t* mmgr, qse_size_t memcapa)
+int qse_stx_init (qse_stx_t* stx, qse_mmgr_t* mmgr, qse_size_t memcapa)
{
QSE_MEMSET (stx, 0, QSE_SIZEOF(*stx));
stx->mmgr = mmgr;
/* initialize object memory subsystem */
- if (qse_stx_initmem (stx, memcapa) <= -1) return QSE_NULL;
+ if (qse_stx_initmem (stx, memcapa) <= -1) return -1;
- return stx;
+ return 0;
}
void qse_stx_fini (qse_stx_t* stx)
@@ -41,7 +41,7 @@ qse_stx_t* qse_stx_open (
);
if (stx == QSE_NULL) return QSE_NULL;
- if (qse_stx_init (stx, mmgr, memcapa) == QSE_NULL)
+ if (qse_stx_init (stx, mmgr, memcapa) <= -1)
{
QSE_MMGR_FREE (stx->mmgr, stx);
return QSE_NULL;
diff --git a/qse/samples/cmn/Makefile.am b/qse/samples/cmn/Makefile.am
index edb2f046..88ec407c 100644
--- a/qse/samples/cmn/Makefile.am
+++ b/qse/samples/cmn/Makefile.am
@@ -6,7 +6,7 @@ AM_CPPFLAGS = \
-I$(includedir)
-bin_PROGRAMS = xma fma pma chr str sll dll lda oht htb rbt fio pio sio time main main2 rex01 env
+bin_PROGRAMS = xma fma pma chr str sll dll lda oht htb rbt fio pio sio time main main2 rex01 env tre01
LDFLAGS = -L../../lib/cmn
LDADD = -lqsecmn
@@ -30,6 +30,7 @@ main_SOURCES = main.c
main2_SOURCES = main2.c
rex01_SOURCES = rex01.c
env_SOURCES = env.c
+tre01_SOURCES = tre01.c
if ENABLE_CXX
diff --git a/qse/samples/cmn/Makefile.in b/qse/samples/cmn/Makefile.in
index 1ce349a7..cca5820f 100644
--- a/qse/samples/cmn/Makefile.in
+++ b/qse/samples/cmn/Makefile.in
@@ -38,7 +38,7 @@ bin_PROGRAMS = xma$(EXEEXT) fma$(EXEEXT) pma$(EXEEXT) chr$(EXEEXT) \
str$(EXEEXT) sll$(EXEEXT) dll$(EXEEXT) lda$(EXEEXT) \
oht$(EXEEXT) htb$(EXEEXT) rbt$(EXEEXT) fio$(EXEEXT) \
pio$(EXEEXT) sio$(EXEEXT) time$(EXEEXT) main$(EXEEXT) \
- main2$(EXEEXT) rex01$(EXEEXT) env$(EXEEXT)
+ main2$(EXEEXT) rex01$(EXEEXT) env$(EXEEXT) tre01$(EXEEXT)
subdir = samples/cmn
DIST_COMMON = $(srcdir)/Makefile.am $(srcdir)/Makefile.in
ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
@@ -127,6 +127,10 @@ am_time_OBJECTS = time.$(OBJEXT)
time_OBJECTS = $(am_time_OBJECTS)
time_LDADD = $(LDADD)
time_DEPENDENCIES =
+am_tre01_OBJECTS = tre01.$(OBJEXT)
+tre01_OBJECTS = $(am_tre01_OBJECTS)
+tre01_LDADD = $(LDADD)
+tre01_DEPENDENCIES =
am_xma_OBJECTS = xma.$(OBJEXT)
xma_OBJECTS = $(am_xma_OBJECTS)
xma_LDADD = $(LDADD)
@@ -148,12 +152,13 @@ SOURCES = $(chr_SOURCES) $(dll_SOURCES) $(env_SOURCES) $(fio_SOURCES) \
$(fma_SOURCES) $(htb_SOURCES) $(lda_SOURCES) $(main_SOURCES) \
$(main2_SOURCES) $(oht_SOURCES) $(pio_SOURCES) $(pma_SOURCES) \
$(rbt_SOURCES) $(rex01_SOURCES) $(sio_SOURCES) $(sll_SOURCES) \
- $(str_SOURCES) $(time_SOURCES) $(xma_SOURCES)
+ $(str_SOURCES) $(time_SOURCES) $(tre01_SOURCES) $(xma_SOURCES)
DIST_SOURCES = $(chr_SOURCES) $(dll_SOURCES) $(env_SOURCES) \
$(fio_SOURCES) $(fma_SOURCES) $(htb_SOURCES) $(lda_SOURCES) \
$(main_SOURCES) $(main2_SOURCES) $(oht_SOURCES) $(pio_SOURCES) \
$(pma_SOURCES) $(rbt_SOURCES) $(rex01_SOURCES) $(sio_SOURCES) \
- $(sll_SOURCES) $(str_SOURCES) $(time_SOURCES) $(xma_SOURCES)
+ $(sll_SOURCES) $(str_SOURCES) $(time_SOURCES) $(tre01_SOURCES) \
+ $(xma_SOURCES)
ETAGS = etags
CTAGS = ctags
DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
@@ -324,6 +329,7 @@ main_SOURCES = main.c
main2_SOURCES = main2.c
rex01_SOURCES = rex01.c
env_SOURCES = env.c
+tre01_SOURCES = tre01.c
all: all-am
.SUFFIXES:
@@ -455,6 +461,9 @@ str$(EXEEXT): $(str_OBJECTS) $(str_DEPENDENCIES)
time$(EXEEXT): $(time_OBJECTS) $(time_DEPENDENCIES)
@rm -f time$(EXEEXT)
$(LINK) $(time_OBJECTS) $(time_LDADD) $(LIBS)
+tre01$(EXEEXT): $(tre01_OBJECTS) $(tre01_DEPENDENCIES)
+ @rm -f tre01$(EXEEXT)
+ $(LINK) $(tre01_OBJECTS) $(tre01_LDADD) $(LIBS)
xma$(EXEEXT): $(xma_OBJECTS) $(xma_DEPENDENCIES)
@rm -f xma$(EXEEXT)
$(LINK) $(xma_OBJECTS) $(xma_LDADD) $(LIBS)
@@ -483,6 +492,7 @@ distclean-compile:
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/sll.Po@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/str.Po@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/time.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/tre01.Po@am__quote@
@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/xma.Po@am__quote@
.c.o:
diff --git a/qse/samples/cmn/tre01.c b/qse/samples/cmn/tre01.c
new file mode 100644
index 00000000..6171e0cb
--- /dev/null
+++ b/qse/samples/cmn/tre01.c
@@ -0,0 +1,41 @@
+
+#include
+#include
+#include
+
+static int test_main (int argc, qse_char_t* argv[], qse_char_t* envp[])
+{
+ qse_tre_t tre;
+
+ qse_tre_init (&tre, QSE_NULL);
+
+ if (qse_tre_comp (&tre, argv[1], QSE_TRE_EXTENDED|QSE_TRE_NOSUBREG) <= -1)
+ {
+ qse_printf (QSE_T("Cannot compile pattern [%s] - %d\n"), argv[1], QSE_TRE_ERRNUM(&tre));
+ goto oops;
+ }
+
+ if (qse_tre_exec(&tre, argv[2], (size_t) 0, NULL, 0) <= -1)
+ {
+ if (QSE_TRE_ERRNUM(&tre) == QSE_TRE_ENOMATCH) qse_printf (QSE_T("no match\n"));
+ else qse_printf (QSE_T("ERROR %d\n"), QSE_TRE_ERRNUM(&tre));
+ goto oops;
+ }
+ else
+ {
+ qse_printf (QSE_T("match...\n"));
+ }
+
+ qse_tre_fini (&tre);
+ return 0;
+
+oops:
+ qse_tre_fini (&tre);
+ return -1;
+}
+
+int qse_main (int argc, qse_achar_t* argv[], qse_achar_t* envp[])
+{
+ return qse_runmainwithenv (argc, argv, envp, test_main);
+}
+