* fixed a rounding bug of reallocation size in _realloc_merge of qse_xma_t.

* replaced rex with tre in qse_sed_t
* added submatch reference for the 's' command into qse_sed_t
* changed some options and functions for tre integration in qse_sed_t
This commit is contained in:
hyung-hwan 2011-09-03 09:27:44 +00:00
parent 1efa41052e
commit 3fe32033cb
9 changed files with 265 additions and 113 deletions

View File

@ -137,7 +137,7 @@ static void print_usage (QSE_FILE* out, int argc, qse_char_t* argv[])
qse_fprintf (out, QSE_T(" -h show this message\n")); qse_fprintf (out, QSE_T(" -h show this message\n"));
qse_fprintf (out, QSE_T(" -n disable auto-print\n")); qse_fprintf (out, QSE_T(" -n disable auto-print\n"));
qse_fprintf (out, QSE_T(" -a perform strict address check\n")); qse_fprintf (out, QSE_T(" -a perform strict address check\n"));
qse_fprintf (out, QSE_T(" -r allow {n,m} in a regular expression\n")); qse_fprintf (out, QSE_T(" -r use the extended regular expression\n"));
qse_fprintf (out, QSE_T(" -s allow text on the same line as c, a, i\n")); qse_fprintf (out, QSE_T(" -s allow text on the same line as c, a, i\n"));
qse_fprintf (out, QSE_T(" -l ensure a newline at text end\n")); qse_fprintf (out, QSE_T(" -l ensure a newline at text end\n"));
qse_fprintf (out, QSE_T(" -f file specify a script file\n")); qse_fprintf (out, QSE_T(" -f file specify a script file\n"));
@ -190,7 +190,7 @@ static int handle_args (int argc, qse_char_t* argv[])
break; break;
case QSE_T('r'): case QSE_T('r'):
g_option |= QSE_SED_REXBOUND; g_option |= QSE_SED_EXTENDEDREX;
break; break;
case QSE_T('s'): case QSE_T('s'):
@ -237,26 +237,25 @@ qse_char_t* load_script_file (const qse_char_t* file)
fp = qse_fopen (file, QSE_T("r")); fp = qse_fopen (file, QSE_T("r"));
if (fp == QSE_NULL) return QSE_NULL; if (fp == QSE_NULL) return QSE_NULL;
if (qse_str_init (&script, QSE_MMGR_GETDFL(), 1024) == QSE_NULL) if (qse_str_init (&script, QSE_MMGR_GETDFL(), 1024) <= -1)
{ {
qse_fclose (fp); qse_fclose (fp);
return QSE_NULL; return QSE_NULL;
} }
while ((c = qse_fgetc (fp)) != QSE_CHAR_EOF) while ((c = qse_fgetc (fp)) != QSE_CHAR_EOF)
{ {
if (qse_str_ccat (&script, c) == (qse_size_t)-1) if (qse_str_ccat (&script, c) == (qse_size_t)-1)
{ {
qse_fclose (fp);
qse_str_fini (&script); qse_str_fini (&script);
qse_fclose (fp);
return QSE_NULL; return QSE_NULL;
} }
} }
qse_fclose (fp);
qse_str_yield (&script, &xstr, 0); qse_str_yield (&script, &xstr, 0);
qse_str_fini (&script); qse_str_fini (&script);
qse_fclose (fp);
return xstr.ptr; return xstr.ptr;
} }
@ -293,7 +292,7 @@ int sed_main (int argc, qse_char_t* argv[])
qse_sed_setoption (sed, g_option); qse_sed_setoption (sed, g_option);
if (g_script_file != QSE_NULL) if (g_script_file)
{ {
QSE_ASSERT (g_script == QSE_NULL); QSE_ASSERT (g_script == QSE_NULL);

View File

@ -143,7 +143,7 @@ int qse_tre_execx (
const qse_char_t* str, const qse_char_t* str,
qse_size_t len, qse_size_t len,
qse_tre_match_t* pmatch, qse_tre_match_t* pmatch,
qse_size_t nmatch, qse_size_t nmatch,
int eflags int eflags
); );
@ -151,7 +151,7 @@ int qse_tre_exec (
qse_tre_t* tre, qse_tre_t* tre,
const qse_char_t* str, const qse_char_t* str,
qse_tre_match_t* pmatch, qse_tre_match_t* pmatch,
qse_size_t nmatch, qse_size_t nmatch,
int eflags int eflags
); );

View File

@ -1,5 +1,5 @@
/* /*
* $Id: Sed.hpp 507 2011-07-15 15:53:49Z hyunghwan.chung $ * $Id: Sed.hpp 558 2011-09-02 15:27:44Z hyunghwan.chung $
* *
Copyright 2006-2011 Chung, Hyung-Hwan. Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -54,8 +54,10 @@ public:
typedef qse_sed_io_arg_t io_arg_t; typedef qse_sed_io_arg_t io_arg_t;
/// The option_t type redefines an option type /// The option_t type redefines an option type
typedef qse_sed_option_t option_t; typedef qse_sed_option_t option_t;
#if 0
/// The depth_t type redefines an depth IDs /// The depth_t type redefines an depth IDs
typedef qse_sed_depth_t depth_t; typedef qse_sed_depth_t depth_t;
#endif
/// ///
/// The Stream class is a base class for I/O operation during /// The Stream class is a base class for I/O operation during
@ -214,6 +216,7 @@ public:
int opt ///< option code int opt ///< option code
); );
#if 0
/// ///
/// The getMaxDepth() function gets the maximum processing depth for /// The getMaxDepth() function gets the maximum processing depth for
/// an operation type identified by @a id. /// an operation type identified by @a id.
@ -229,6 +232,7 @@ public:
int ids, ///< 0 or a number OR'ed of depth_t values int ids, ///< 0 or a number OR'ed of depth_t values
size_t depth ///< 0 maximum depth size_t depth ///< 0 maximum depth
); );
#endif
/// ///
/// The getErrorMessage() function gets the description of the last /// The getErrorMessage() function gets the description of the last

View File

@ -1,5 +1,5 @@
/* /*
* $Id: sed.h 507 2011-07-15 15:53:49Z hyunghwan.chung $ * $Id: sed.h 558 2011-09-02 15:27:44Z hyunghwan.chung $
* *
Copyright 2006-2011 Chung, Hyung-Hwan. Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -131,29 +131,17 @@ typedef const qse_char_t* (*qse_sed_errstr_t) (
*/ */
enum qse_sed_option_t enum qse_sed_option_t
{ {
QSE_SED_STRIPLS = (1 << 0), /**< strip leading spaces from text */ QSE_SED_STRIPLS = (1 << 0), /**< strip leading spaces from text */
QSE_SED_KEEPTBS = (1 << 1), /**< keep an trailing backslash */ QSE_SED_KEEPTBS = (1 << 1), /**< keep an trailing backslash */
QSE_SED_ENSURENL = (1 << 2), /**< ensure NL at the text end */ QSE_SED_ENSURENL = (1 << 2), /**< ensure NL at the text end */
QSE_SED_QUIET = (1 << 3), /**< do not print pattern space */ QSE_SED_QUIET = (1 << 3), /**< do not print pattern space */
QSE_SED_STRICT = (1 << 4), /**< do strict address check */ QSE_SED_STRICT = (1 << 4), /**< do strict address check */
QSE_SED_STARTSTEP = (1 << 5), /**< allow start~step */ QSE_SED_STARTSTEP = (1 << 5), /**< allow start~step */
QSE_SED_REXBOUND = (1 << 6), /**< allow {n,m} in regular expression */ QSE_SED_EXTENDEDREX = (1 << 6), /**< allow {n,m} in regular expression */
QSE_SED_SAMELINE = (1 << 7), /**< allow text on the same line as c, a, i */ QSE_SED_SAMELINE = (1 << 7), /**< allow text on the same line as c, a, i */
}; };
typedef enum qse_sed_option_t qse_sed_option_t; typedef enum qse_sed_option_t qse_sed_option_t;
/**
* The qse_sed_depth_t type defines IDs for qse_sed_getmaxdepth() and
* qse_sed_setmaxdepth().
*/
enum qse_sed_depth_t
{
QSE_SED_DEPTH_REX_BUILD = (1 << 0),
QSE_SED_DEPTH_REX_MATCH = (1 << 1)
};
typedef enum qse_sed_depth_t qse_sed_depth_t;
/** /**
* The qse_sed_io_cmd_t type defines I/O command codes. The code indicates * The qse_sed_io_cmd_t type defines I/O command codes. The code indicates
* the action to take in an I/O handler. * the action to take in an I/O handler.
@ -246,23 +234,6 @@ void qse_sed_setoption (
int opt /**< 0 or a number OR'ed of #qse_sed_option_t values */ int opt /**< 0 or a number OR'ed of #qse_sed_option_t values */
); );
/**
* The qse_sed_getmaxdepth() gets the maximum processing depth.
*/
qse_size_t qse_sed_getmaxdepth (
qse_sed_t* sed, /**< stream editor */
qse_sed_depth_t id /**< one of qse_sed_depth_t values */
);
/**
* The qse_sed_setmaxdepth() sets the maximum processing depth.
*/
void qse_sed_setmaxdepth (
qse_sed_t* sed, /**< stream editor */
int ids, /**< 0 or a number OR'ed of #qse_sed_depth_t values */
qse_size_t depth /**< maximum depth level */
);
/** /**
* The qse_sed_geterrstr() gets an error string getter. * The qse_sed_geterrstr() gets an error string getter.
*/ */

View File

@ -47,7 +47,7 @@ qse_tre_t* qse_tre_open (qse_mmgr_t* mmgr, qse_size_t xtnsize)
return QSE_NULL; return QSE_NULL;
} }
return QSE_NULL; return tre;
} }
void qse_tre_close (qse_tre_t* tre) void qse_tre_close (qse_tre_t* tre)

View File

@ -49,8 +49,8 @@ struct qse_xma_blk_t
struct struct
{ {
qse_xma_blk_t* prev; /**< link to the previous block */ qse_xma_blk_t* prev; /**< link to the previous adjacent block */
qse_xma_blk_t* next; /**< link to the next block */ qse_xma_blk_t* next; /**< link to the next adjacent block */
} b; } b;
}; };
@ -379,7 +379,7 @@ void* qse_xma_alloc (qse_xma_t* xma, qse_size_t size)
for (++xfi; xfi < XFIMAX(xma) - 1; xfi++) for (++xfi; xfi < XFIMAX(xma) - 1; xfi++)
{ {
free = alloc_from_freelist (xma, xfi, size); free = alloc_from_freelist (xma, xfi, size);
if (free != QSE_NULL) break; if (free) break;
} }
if (free == QSE_NULL) return QSE_NULL; if (free == QSE_NULL) return QSE_NULL;
} }
@ -392,16 +392,20 @@ static void* _realloc_merge (qse_xma_t* xma, void* b, qse_size_t size)
{ {
qse_xma_blk_t* blk = USR_TO_SYS(b); qse_xma_blk_t* blk = USR_TO_SYS(b);
/* rounds up 'size' to be multiples of ALIGN */
size = ((size + ALIGN - 1) / ALIGN) * ALIGN;
if (size > blk->size) if (size > blk->size)
{ {
/* /*
* grow the current block * grow the current block
*/ */
qse_size_t req;
qse_size_t req = size - blk->size;
qse_xma_blk_t* n; qse_xma_blk_t* n;
qse_size_t rem; qse_size_t rem;
req = size - blk->size;
n = blk->b.next; n = blk->b.next;
/* check if the next adjacent block is available */ /* check if the next adjacent block is available */

View File

@ -1,5 +1,5 @@
/* /*
* $Id: Sed.cpp 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: Sed.cpp 558 2011-09-02 15:27:44Z hyunghwan.chung $
* *
Copyright 2006-2011 Chung, Hyung-Hwan. Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -80,6 +80,7 @@ void Sed::setOption (int opt)
qse_sed_setoption (sed, opt); qse_sed_setoption (sed, opt);
} }
#if 0
Sed::size_t Sed::getMaxDepth (depth_t id) const Sed::size_t Sed::getMaxDepth (depth_t id) const
{ {
QSE_ASSERT (sed != QSE_NULL); QSE_ASSERT (sed != QSE_NULL);
@ -91,6 +92,7 @@ void Sed::setMaxDepth (int ids, size_t depth)
QSE_ASSERT (sed != QSE_NULL); QSE_ASSERT (sed != QSE_NULL);
qse_sed_setmaxdepth (sed, ids, depth); qse_sed_setmaxdepth (sed, ids, depth);
} }
#endif
const Sed::char_t* Sed::getErrorMessage () const const Sed::char_t* Sed::getErrorMessage () const
{ {

View File

@ -1,5 +1,5 @@
/* /*
* $Id: sed.c 557 2011-09-01 14:45:06Z hyunghwan.chung $ * $Id: sed.c 558 2011-09-02 15:27:44Z hyunghwan.chung $
* *
Copyright 2006-2011 Chung, Hyung-Hwan. Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -20,17 +20,19 @@
#include "sed.h" #include "sed.h"
#include "../cmn/mem.h" #include "../cmn/mem.h"
#include <qse/cmn/rex.h>
#include <qse/cmn/chr.h> #include <qse/cmn/chr.h>
#ifdef USE_REX
# include <qse/cmn/rex.h>
#else
# include <qse/cmn/tre.h>
#endif
QSE_IMPLEMENT_COMMON_FUNCTIONS (sed) QSE_IMPLEMENT_COMMON_FUNCTIONS (sed)
static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd); static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd);
static void free_all_command_blocks (qse_sed_t* sed); static void free_all_command_blocks (qse_sed_t* sed);
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) \ #define SETERR0(sed,num,loc) \
do { qse_sed_seterror (sed, num, QSE_NULL, loc); } while (0) do { qse_sed_seterror (sed, num, QSE_NULL, loc); } while (0)
@ -73,7 +75,7 @@ void qse_sed_close (qse_sed_t* sed)
QSE_MMGR_FREE (sed->mmgr, sed); QSE_MMGR_FREE (sed->mmgr, sed);
} }
static int qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr) int qse_sed_init (qse_sed_t* sed, qse_mmgr_t* mmgr)
{ {
if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL(); if (mmgr == QSE_NULL) mmgr = QSE_MMGR_GETDFL();
@ -121,7 +123,7 @@ oops_1:
return -1; return -1;
} }
static void qse_sed_fini (qse_sed_t* sed) void qse_sed_fini (qse_sed_t* sed)
{ {
free_all_command_blocks (sed); free_all_command_blocks (sed);
@ -145,6 +147,8 @@ int qse_sed_getoption (qse_sed_t* sed)
return sed->option; return sed->option;
} }
#ifdef USE_REX
qse_size_t qse_sed_getmaxdepth (qse_sed_t* sed, qse_sed_depth_t id) qse_size_t qse_sed_getmaxdepth (qse_sed_t* sed, qse_sed_depth_t id)
{ {
return (id & QSE_SED_DEPTH_REX_BUILD)? sed->depth.rex.build: return (id & QSE_SED_DEPTH_REX_BUILD)? sed->depth.rex.build:
@ -156,6 +160,75 @@ void qse_sed_setmaxdepth (qse_sed_t* sed, int ids, qse_size_t depth)
if (ids & QSE_SED_DEPTH_REX_BUILD) sed->depth.rex.build = depth; if (ids & QSE_SED_DEPTH_REX_BUILD) sed->depth.rex.build = depth;
if (ids & QSE_SED_DEPTH_REX_MATCH) sed->depth.rex.match = depth; if (ids & QSE_SED_DEPTH_REX_MATCH) sed->depth.rex.match = depth;
} }
#endif
static qse_tre_t* maketre (
qse_sed_t* sed, const qse_cstr_t* str, const qse_sed_loc_t* loc)
{
qse_tre_t* tre;
tre = qse_tre_open (sed->mmgr, 0);
if (tre == QSE_NULL)
{
SETERR0 (sed, QSE_SED_ENOMEM, loc);
return QSE_NULL;
}
if (qse_tre_compx (tre, str->ptr, str->len, QSE_NULL,
((sed->option & QSE_SED_EXTENDEDREX)? QSE_TRE_EXTENDED: 0)) <= -1)
{
SETERR1 (sed, QSE_SED_EREXBL, str->ptr, str->len, loc);
qse_tre_close (tre);
return QSE_NULL;
}
return tre;
}
static QSE_INLINE void freetre (qse_sed_t* sed, qse_tre_t* tre)
{
qse_tre_close (tre);
}
static int matchtre (
qse_sed_t* sed, qse_tre_t* tre, int opt,
const qse_cstr_t* str, qse_cstr_t* mat,
qse_cstr_t submat[9], const qse_sed_loc_t* loc)
{
int n;
qse_tre_match_t match[10];
n = qse_tre_execx (tre, str->ptr, str->len, match, 10, opt);
if (n <= -1)
{
if (QSE_TRE_ERRNUM(tre) == QSE_TRE_ENOMATCH) return 0;
SETERR0 (sed, QSE_SED_EREXMA, loc);
return -1;
}
QSE_ASSERT (match[0].rm_so != -1);
if (mat)
{
mat->ptr = &str->ptr[match[0].rm_so];
mat->len = match[0].rm_eo - match[0].rm_so;
}
if (submat)
{
int i;
/* you must intialize submat before you pass into this
* function because it can abort filling */
for (i = 1; i <= 10; i++)
{
if (match[i].rm_so == -1) break;
submat[i-1].ptr = &str->ptr[match[i].rm_so];
submat[i-1].len = match[i].rm_eo - match[i].rm_so;
}
}
return 1;
}
/* check if c is a space character */ /* check if c is a space character */
#define IS_SPACE(c) ((c) == QSE_T(' ') || (c) == QSE_T('\t') || (c) == QSE_T('\r')) #define IS_SPACE(c) ((c) == QSE_T(' ') || (c) == QSE_T('\t') || (c) == QSE_T('\r'))
@ -200,13 +273,15 @@ static void free_address (qse_sed_t* sed, qse_sed_cmd_t* cmd)
if (cmd->a2.type == QSE_SED_ADR_REX) if (cmd->a2.type == QSE_SED_ADR_REX)
{ {
QSE_ASSERT (cmd->a2.u.rex != QSE_NULL); QSE_ASSERT (cmd->a2.u.rex != QSE_NULL);
qse_freerex (sed->mmgr, cmd->a2.u.rex); /*qse_freerex (sed->mmgr, cmd->a2.u.rex);*/
freetre (sed, cmd->a2.u.rex);
cmd->a2.type = QSE_SED_ADR_NONE; cmd->a2.type = QSE_SED_ADR_NONE;
} }
if (cmd->a1.type == QSE_SED_ADR_REX) if (cmd->a1.type == QSE_SED_ADR_REX)
{ {
QSE_ASSERT (cmd->a1.u.rex != QSE_NULL); QSE_ASSERT (cmd->a1.u.rex != QSE_NULL);
qse_freerex (sed->mmgr, cmd->a1.u.rex); /*qse_freerex (sed->mmgr, cmd->a1.u.rex);*/
freetre (sed, cmd->a1.u.rex);
cmd->a1.type = QSE_SED_ADR_NONE; cmd->a1.type = QSE_SED_ADR_NONE;
} }
} }
@ -285,7 +360,8 @@ static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd)
if (cmd->u.subst.rpl.ptr) if (cmd->u.subst.rpl.ptr)
QSE_MMGR_FREE (sed->mmgr, cmd->u.subst.rpl.ptr); QSE_MMGR_FREE (sed->mmgr, cmd->u.subst.rpl.ptr);
if (cmd->u.subst.rex) if (cmd->u.subst.rex)
qse_freerex (sed->mmgr, cmd->u.subst.rex); /*qse_freerex (sed->mmgr, cmd->u.subst.rex);*/
freetre (sed, cmd->u.subst.rex);
break; break;
case QSE_SED_CMD_TRANSLATE: case QSE_SED_CMD_TRANSLATE:
@ -300,7 +376,9 @@ static void free_command (qse_sed_t* sed, qse_sed_cmd_t* cmd)
static void* compile_rex (qse_sed_t* sed, qse_char_t rxend) static void* compile_rex (qse_sed_t* sed, qse_char_t rxend)
{ {
#ifdef USE_REX
void* code; void* code;
#endif
qse_cint_t c; qse_cint_t c;
qse_str_clear (&sed->tmp.rex); qse_str_clear (&sed->tmp.rex);
@ -346,43 +424,11 @@ static void* compile_rex (qse_sed_t* sed, qse_char_t rxend)
} }
} }
#ifdef USE_REX
#if 0
{
qse_tre_t* tre;
tre = qse_tre_open (sed->mmgr, 0);
if (tre)
{
if (qse_tre_comp (tre,
QSE_STR_PTR(&sed->tmp.rex),
QSE_STR_LEN(&sed->tmp.rex),
QSE_NULL,
QSE_TRE_EXTENDED) <= -1)
{
qse_tre_close (tre);
goto fail:
}
return tre;
}
else
{
SETERR1 (
sed, QSE_SED_EREXBL,
QSE_STR_PTR(&sed->tmp.rex),
QSE_STR_LEN(&sed->tmp.rex),
&sed->src.loc
);
return QSE_NULL;
}
}
#endif
code = qse_buildrex ( code = qse_buildrex (
sed->mmgr, sed->mmgr,
sed->depth.rex.build, sed->depth.rex.build,
((sed->option&QSE_SED_REXBOUND)? 0:QSE_REX_NOBOUND), ((sed->option&QSE_SED_EXTENDEDREX)? 0:QSE_REX_NOBOUND),
QSE_STR_PTR(&sed->tmp.rex), QSE_STR_PTR(&sed->tmp.rex),
QSE_STR_LEN(&sed->tmp.rex), QSE_STR_LEN(&sed->tmp.rex),
QSE_NULL QSE_NULL
@ -399,6 +445,10 @@ static void* compile_rex (qse_sed_t* sed, qse_char_t rxend)
} }
return code; return code;
#else
return maketre (sed, QSE_STR_CSTR(&sed->tmp.rex), &sed->src.loc);
#endif
} }
static qse_sed_adr_t* get_address (qse_sed_t* sed, qse_sed_adr_t* a) static qse_sed_adr_t* get_address (qse_sed_t* sed, qse_sed_adr_t* a)
@ -809,12 +859,14 @@ static int get_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
{ {
CHECK_CMDIC (sed, cmd, c, goto oops); CHECK_CMDIC (sed, cmd, c, goto oops);
#if 0
if (c == QSE_T('\\')) if (c == QSE_T('\\'))
{ {
c = NXTSC (sed); c = NXTSC (sed);
CHECK_CMDIC (sed, cmd, c, goto oops); CHECK_CMDIC (sed, cmd, c, goto oops);
if (c == QSE_T('n')) c = QSE_T('\n'); if (c == QSE_T('n')) c = QSE_T('\n');
} }
#endif
if (qse_str_ccat (t[i], c) == (qse_size_t)-1) if (qse_str_ccat (t[i], c) == (qse_size_t)-1)
{ {
@ -895,10 +947,12 @@ static int get_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
terminate_command (sed) <= -1) goto oops; terminate_command (sed) <= -1) goto oops;
QSE_ASSERT (cmd->u.subst.rex == QSE_NULL); QSE_ASSERT (cmd->u.subst.rex == QSE_NULL);
#ifdef USE_REX
cmd->u.subst.rex = qse_buildrex ( cmd->u.subst.rex = qse_buildrex (
sed->mmgr, sed->mmgr,
sed->depth.rex.build, sed->depth.rex.build,
((sed->option&QSE_SED_REXBOUND)? 0:QSE_REX_NOBOUND), ((sed->option&QSE_SED_EXTENDEDREX)? 0:QSE_REX_NOBOUND),
QSE_STR_PTR(t[0]), QSE_STR_PTR(t[0]),
QSE_STR_LEN(t[0]), QSE_STR_LEN(t[0]),
QSE_NULL QSE_NULL
@ -913,6 +967,10 @@ static int get_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
); );
goto oops; goto oops;
} }
#else
cmd->u.subst.rex = maketre (sed, QSE_STR_CSTR(t[0]), &sed->src.loc);
if (cmd->u.subst.rex == QSE_NULL) goto oops;
#endif
qse_str_yield (t[1], &cmd->u.subst.rpl, 0); qse_str_yield (t[1], &cmd->u.subst.rpl, 0);
if (cmd->u.subst.g == 0 && cmd->u.subst.occ == 0) cmd->u.subst.occ = 1; if (cmd->u.subst.g == 0 && cmd->u.subst.occ == 0) cmd->u.subst.occ = 1;
@ -1833,7 +1891,9 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
{ {
qse_cstr_t mat, pmat; qse_cstr_t mat, pmat;
int opt = 0, repl = 0, n; int opt = 0, repl = 0, n;
#ifdef USE_REX
qse_rex_errnum_t errnum; qse_rex_errnum_t errnum;
#endif
qse_cstr_t str, cur; qse_cstr_t str, cur;
const qse_char_t* str_end; const qse_char_t* str_end;
@ -1842,7 +1902,11 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
QSE_ASSERT (cmd->type == QSE_SED_CMD_SUBSTITUTE); QSE_ASSERT (cmd->type == QSE_SED_CMD_SUBSTITUTE);
qse_str_clear (&sed->e.txt.subst); qse_str_clear (&sed->e.txt.subst);
#ifdef USE_REX
if (cmd->u.subst.i) opt = QSE_REX_IGNORECASE; if (cmd->u.subst.i) opt = QSE_REX_IGNORECASE;
#else
if (cmd->u.subst.i) opt = QSE_TRE_IGNORECASE;
#endif
str.ptr = QSE_STR_PTR(&sed->e.in.line); str.ptr = QSE_STR_PTR(&sed->e.in.line);
str.len = QSE_STR_LEN(&sed->e.in.line); str.len = QSE_STR_LEN(&sed->e.in.line);
@ -1863,6 +1927,7 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
* end of string($) needs to be tested */ * end of string($) needs to be tested */
while (cur.ptr <= str_end) while (cur.ptr <= str_end)
{ {
#ifdef USE_REX
if (max_count == 0 || sub_count < max_count) if (max_count == 0 || sub_count < max_count)
{ {
n = qse_matchrex ( n = qse_matchrex (
@ -1871,14 +1936,29 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
cmd->u.subst.rex, opt, cmd->u.subst.rex, opt,
&str, &cur, &mat, &errnum &str, &cur, &mat, &errnum
); );
if (n <= -1)
{
SETERR0 (sed, QSE_SED_EREXMA, &cmd->loc);
return -1;
}
} }
else n = 0; else n = 0;
#else
qse_cstr_t submat[9];
if (n <= -1) QSE_MEMSET (submat, 0, QSE_SIZEOF(submat));
if (max_count == 0 || sub_count < max_count)
{ {
SETERR0 (sed, QSE_SED_EREXMA, &cmd->loc); n = matchtre (
return -1; sed,
cmd->u.subst.rex,
((str.ptr == cur.ptr)? opt: (opt | QSE_TRE_NOTBOL)),
&cur, &mat, submat, &cmd->loc
);
if (n <= -1) return -1;
} }
else n = 0;
#endif
if (n == 0) if (n == 0)
{ {
@ -1930,11 +2010,49 @@ static int do_subst (qse_sed_t* sed, qse_sed_cmd_t* cmd)
for (i = 0; i < cmd->u.subst.rpl.len; i++) for (i = 0; i < cmd->u.subst.rpl.len; i++)
{ {
if ((i+1) < cmd->u.subst.rpl.len && if ((i+1) < cmd->u.subst.rpl.len &&
cmd->u.subst.rpl.ptr[i] == QSE_T('\\') && cmd->u.subst.rpl.ptr[i] == QSE_T('\\'))
cmd->u.subst.rpl.ptr[i+1] == QSE_T('&'))
{ {
m = qse_str_ccat ( qse_char_t nc = cmd->u.subst.rpl.ptr[i+1];
&sed->e.txt.subst, QSE_T('&'));
#ifndef USE_REX
if (nc >= QSE_T('1') && nc <= QSE_T('9'))
{
int smi = nc - QSE_T('1');
m = qse_str_ncat (&sed->e.txt.subst, submat[smi].ptr, submat[smi].len);
}
else
{
#endif
switch (nc)
{
case QSE_T('n'):
nc = QSE_T('\n');
break;
case QSE_T('r'):
nc = QSE_T('\r');
break;
case QSE_T('t'):
nc = QSE_T('\t');
break;
case QSE_T('f'):
nc = QSE_T('\f');
break;
case QSE_T('b'):
nc = QSE_T('\b');
break;
case QSE_T('v'):
nc = QSE_T('\v');
break;
case QSE_T('a'):
nc = QSE_T('\a');
break;
#ifndef USE_REX
}
#endif
m = qse_str_ccat (&sed->e.txt.subst, nc);
}
i++; i++;
} }
else if (cmd->u.subst.rpl.ptr[i] == QSE_T('&')) else if (cmd->u.subst.rpl.ptr[i] == QSE_T('&'))
@ -2032,10 +2150,12 @@ static int match_a (qse_sed_t* sed, qse_sed_cmd_t* cmd, qse_sed_adr_t* a)
case QSE_SED_ADR_REX: case QSE_SED_ADR_REX:
{ {
#ifdef USE_REX
int n; int n;
qse_cstr_t match;
qse_cstr_t line;
qse_rex_errnum_t errnum; qse_rex_errnum_t errnum;
qse_cstr_t match;
#endif
qse_cstr_t line;
QSE_ASSERT (a->u.rex != QSE_NULL); QSE_ASSERT (a->u.rex != QSE_NULL);
@ -2045,6 +2165,7 @@ static int match_a (qse_sed_t* sed, qse_sed_cmd_t* cmd, qse_sed_adr_t* a)
if (line.len > 0 && if (line.len > 0 &&
line.ptr[line.len-1] == QSE_T('\n')) line.len--; line.ptr[line.len-1] == QSE_T('\n')) line.len--;
#ifdef USE_REX
n = qse_matchrex ( n = qse_matchrex (
sed->mmgr, sed->mmgr,
sed->depth.rex.match, sed->depth.rex.match,
@ -2058,6 +2179,10 @@ static int match_a (qse_sed_t* sed, qse_sed_cmd_t* cmd, qse_sed_adr_t* a)
} }
return n; return n;
#else
return matchtre (sed, a->u.rex, 0, &line, QSE_NULL, QSE_NULL, &cmd->loc);
#endif
} }
case QSE_SED_ADR_DOL: case QSE_SED_ADR_DOL:
{ {

View File

@ -1,5 +1,5 @@
/* /*
* $Id: sed.h 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: sed.h 558 2011-09-02 15:27:44Z hyunghwan.chung $
* *
Copyright 2006-2011 Chung, Hyung-Hwan. Copyright 2006-2011 Chung, Hyung-Hwan.
This file is part of QSE. This file is part of QSE.
@ -28,6 +28,23 @@
#define QSE_MAP_AS_RBT #define QSE_MAP_AS_RBT
#include <qse/cmn/map.h> #include <qse/cmn/map.h>
/*
* Define USE_REX to use rex.h on behalf of tre.h
* rex.h currently does not support backreference.
*/
#ifdef USE_REX
enum qse_sed_depth_t
{
QSE_SED_DEPTH_REX_BUILD = (1 << 0),
QSE_SED_DEPTH_REX_MATCH = (1 << 1)
};
typedef enum qse_sed_depth_t qse_sed_depth_t
#endif
#define QSE_SED_CMD_NOOP QSE_T('\0')
#define QSE_SED_CMD_QUIT QSE_T('q')
typedef struct qse_sed_adr_t qse_sed_adr_t; typedef struct qse_sed_adr_t qse_sed_adr_t;
typedef struct qse_sed_cmd_t qse_sed_cmd_t; typedef struct qse_sed_cmd_t qse_sed_cmd_t;
typedef struct qse_sed_cmd_blk_t qse_sed_cmd_blk_t; typedef struct qse_sed_cmd_blk_t qse_sed_cmd_blk_t;
@ -50,8 +67,6 @@ struct qse_sed_adr_t
} u; } u;
}; };
#define QSE_SED_CMD_NOOP QSE_T('\0')
#define QSE_SED_CMD_QUIT QSE_T('q')
#define QSE_SED_CMD_QUIT_QUIET QSE_T('Q') #define QSE_SED_CMD_QUIT_QUIET QSE_T('Q')
#define QSE_SED_CMD_APPEND QSE_T('a') #define QSE_SED_CMD_APPEND QSE_T('a')
#define QSE_SED_CMD_INSERT QSE_T('i') #define QSE_SED_CMD_INSERT QSE_T('i')
@ -262,7 +277,39 @@ struct qse_sed_t
extern "C" { extern "C" {
#endif #endif
const qse_char_t* qse_sed_dflerrstr (qse_sed_t* sed, qse_sed_errnum_t errnum); int qse_sed_init (
qse_sed_t* sed,
qse_mmgr_t* mmgr
);
void qse_sed_fini (
qse_sed_t* sed
);
const qse_char_t* qse_sed_dflerrstr (
qse_sed_t* sed,
qse_sed_errnum_t errnum
);
#ifdef USE_REX
/**
* The qse_sed_getmaxdepth() gets the maximum processing depth.
*/
qse_size_t qse_sed_getmaxdepth (
qse_sed_t* sed, /**< stream editor */
qse_sed_depth_t id /**< one of qse_sed_depth_t values */
);
/**
* The qse_sed_setmaxdepth() sets the maximum processing depth.
*/
void qse_sed_setmaxdepth (
qse_sed_t* sed, /**< stream editor */
int ids, /**< 0 or a number OR'ed of #qse_sed_depth_t values */
qse_size_t depth /**< maximum depth level */
);
#endif
#ifdef __cplusplus #ifdef __cplusplus
} }