hyung-hwan/qse
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

enhanced record reading further

Browse Source
This commit is contained in:
hyung-hwan 2011-05-02 07:28:51 +00:00
parent d915a0fc7a
commit 42327ce6d1
8 changed files with 107 additions and 83 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
qse/cmd/awk/awk.c
View File

@ -1,5 +1,5 @@
/* /*
* $Id: awk.c 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: awk.c 447 2011-05-01 13:28:51Z 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.
@ -369,23 +369,23 @@ struct opttab_t
const qse_char_t* desc; const qse_char_t* desc;
} opttab[] = } opttab[] =
{ {
{ QSE_T("implicit"), QSE_AWK_IMPLICIT, QSE_T("allow undeclared variables") }, { QSE_T("implicit"), QSE_AWK_IMPLICIT, QSE_T("allow undeclared variables") },
{ QSE_T("explicit"), QSE_AWK_EXPLICIT, QSE_T("allow declared variables(local,global)") }, { QSE_T("explicit"), QSE_AWK_EXPLICIT, QSE_T("allow declared variables(local,global)") },
{ QSE_T("extraops"), QSE_AWK_EXTRAOPS, QSE_T("enable extra operators(<<,>>,^^,//)") }, { QSE_T("extraops"), QSE_AWK_EXTRAOPS, QSE_T("enable extra operators(<<,>>,^^,//)") },
{ QSE_T("rio"), QSE_AWK_RIO, QSE_T("enable builtin I/O including getline & print") }, { QSE_T("rio"), QSE_AWK_RIO, QSE_T("enable builtin I/O including getline & print") },
{ QSE_T("rwpipe"), QSE_AWK_RWPIPE, QSE_T("allow a dual-directional pipe") }, { QSE_T("rwpipe"), QSE_AWK_RWPIPE, QSE_T("allow a dual-directional pipe") },
{ QSE_T("newline"), QSE_AWK_NEWLINE, QSE_T("enable a newline to terminate a statement") }, { QSE_T("newline"), QSE_AWK_NEWLINE, QSE_T("enable a newline to terminate a statement") },
{ QSE_T("striprecspc"), QSE_AWK_STRIPRECSPC, QSE_T("strip spaces in splitting a record") }, { QSE_T("striprecspc"), QSE_AWK_STRIPRECSPC, QSE_T("strip spaces in splitting a record") },
{ QSE_T("stripstrspc"), QSE_AWK_STRIPSTRSPC, QSE_T("strip spaces in converting a string to a number") }, { QSE_T("stripstrspc"), QSE_AWK_STRIPSTRSPC, QSE_T("strip spaces in converting a string to a number") },
{ QSE_T("nextofile"), QSE_AWK_NEXTOFILE, QSE_T("enable 'nextofile'") }, { QSE_T("nextofile"), QSE_AWK_NEXTOFILE, QSE_T("enable 'nextofile'") },
{ QSE_T("reset"), QSE_AWK_RESET, QSE_T("enable 'reset'") }, { QSE_T("reset"), QSE_AWK_RESET, QSE_T("enable 'reset'") },
{ QSE_T("crlf"), QSE_AWK_CRLF, QSE_T("use CRLF for a newline") }, { QSE_T("crlf"), QSE_AWK_CRLF, QSE_T("use CRLF for a newline") },
{ QSE_T("maptovar"), QSE_AWK_MAPTOVAR, QSE_T("allow a map to be assigned or returned") }, { QSE_T("maptovar"), QSE_AWK_MAPTOVAR, QSE_T("allow a map to be assigned or returned") },
{ QSE_T("pablock"), QSE_AWK_PABLOCK, QSE_T("enable pattern-action loop") }, { QSE_T("pablock"), QSE_AWK_PABLOCK, QSE_T("enable pattern-action loop") },
{ QSE_T("rexbound"), QSE_AWK_REXBOUND, QSE_T("enable {n,m} in a regular expression") }, { QSE_T("rexbound"), QSE_AWK_REXBOUND, QSE_T("enable {n,m} in a regular expression") },
{ QSE_T("ncmponstr"), QSE_AWK_NCMPONSTR, QSE_T("perform numeric comparsion on numeric strings") }, { QSE_T("ncmponstr"), QSE_AWK_NCMPONSTR, QSE_T("perform numeric comparsion on numeric strings") },
{ QSE_T("strictnaming"), QSE_AWK_STRICTNAMING, QSE_T("enable the strict naming rule") }, { QSE_T("strictnaming"), QSE_AWK_STRICTNAMING, QSE_T("enable the strict naming rule") },
{ QSE_T("include"), QSE_AWK_INCLUDE, QSE_T("enable 'include'") }, { QSE_T("include"), QSE_AWK_INCLUDE, QSE_T("enable 'include'") },
{ QSE_NULL, 0 } { QSE_NULL, 0 }
}; };

4
qse/lib/awk/awk.c
View File

@ -1,5 +1,5 @@
/* /*
* $Id: awk.c 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: awk.c 447 2011-05-01 13:28:51Z 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.
@ -223,7 +223,7 @@ qse_awk_t* qse_awk_open (qse_mmgr_t* mmgr, qse_size_t xtn, qse_awk_prm_t* prm)
qse_lda_setcopier (awk->parse.params, QSE_LDA_COPIER_INLINE); qse_lda_setcopier (awk->parse.params, QSE_LDA_COPIER_INLINE);
awk->option = QSE_AWK_CLASSIC; awk->option = QSE_AWK_CLASSIC;
#if defined(_WIN32) || defined(__OS2__) #if defined(__OS2__) || defined(_WIN32) || defined(__DOS__)
awk->option |= QSE_AWK_CRLF; awk->option |= QSE_AWK_CRLF;
#endif #endif

91
qse/lib/awk/rio.c
View File

@ -1,5 +1,5 @@
/* /*
* $Id: rio.c 446 2011-04-30 15:24:38Z hyunghwan.chung $ * $Id: rio.c 447 2011-05-01 13:28:51Z 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.
@ -22,10 +22,10 @@
enum io_mask_t enum io_mask_t
{ {
MASK_READ = 0x0100, IO_MASK_READ = 0x0100,
MASK_WRITE = 0x0200, IO_MASK_WRITE = 0x0200,
MASK_RDWR = 0x0400, IO_MASK_RDWR = 0x0400,
MASK_CLEAR = 0x00FF IO_MASK_CLEAR = 0x00FF
}; };
static int in_type_map[] = static int in_type_map[] =
@ -50,10 +50,10 @@ static int in_mode_map[] =
static int in_mask_map[] = static int in_mask_map[] =
{ {
MASK_READ, IO_MASK_READ,
MASK_RDWR, IO_MASK_RDWR,
MASK_READ, IO_MASK_READ,
MASK_READ IO_MASK_READ
}; };
static int out_type_map[] = static int out_type_map[] =
@ -80,11 +80,11 @@ static int out_mode_map[] =
static int out_mask_map[] = static int out_mask_map[] =
{ {
MASK_WRITE, IO_MASK_WRITE,
MASK_RDWR, IO_MASK_RDWR,
MASK_WRITE, IO_MASK_WRITE,
MASK_WRITE, IO_MASK_WRITE,
MASK_WRITE IO_MASK_WRITE
}; };
static int find_rio_in ( static int find_rio_in (
@ -249,7 +249,8 @@ static QSE_INLINE int match_long_rs (
* A match in this case must end at the end of * A match in this case must end at the end of
* the current record buffer */ * the current record buffer */
QSE_ASSERT ( QSE_ASSERT (
QSE_STR_PTR(buf) + QSE_STR_LEN(buf) == match.ptr + match.len QSE_STR_PTR(buf) + QSE_STR_LEN(buf) ==
match.ptr + match.len
); );
/* drop the RS part. no extra character after RS to drop /* drop the RS part. no extra character after RS to drop
@ -462,28 +463,36 @@ int qse_awk_rtx_readio (
if (pc == QSE_T('\r') && if (pc == QSE_T('\r') &&
QSE_STR_LEN(buf) > 0) QSE_STR_LEN(buf) > 0)
{ {
/* shrink the line length and the record
* by dropping of CR before NL */
QSE_ASSERT (line_len > 0);
line_len--;
QSE_STR_LEN(buf) -= 1; QSE_STR_LEN(buf) -= 1;
} }
}
if (line_len == 0 && c == QSE_T('\n')) if (line_len == 0)
{
if (QSE_STR_LEN(buf) <= 0)
{ {
/* if the record is empty when a blank /* we got a blank line */
* line is encountered, the line if (QSE_STR_LEN(buf) <= 0)
* terminator should not be added to {
* the record */ /* if the record is empty when a blank
continue; * line is encountered, the line
* terminator should not be added to
* the record */
continue;
}
/* when a blank line is encountered,
* it needs to snip off the line
* terminator of the previous line */
QSE_STR_LEN(buf) -= 1;
done = 1;
break;
} }
/* when a blank line is encountered, line_len = 0;
* it needs to snip off the line
* terminator of the previous line */
QSE_STR_LEN(buf) -= 1;
done = 1;
break;
} }
else line_len++;
if (qse_str_ccat (buf, c) == (qse_size_t)-1) if (qse_str_ccat (buf, c) == (qse_size_t)-1)
{ {
@ -492,9 +501,6 @@ int qse_awk_rtx_readio (
done = 1; done = 1;
break; break;
} }
/* TODO: handle different line terminator */
if (c == QSE_T('\n')) line_len = 0;
else line_len = line_len + 1;
} }
while (p->in.pos < p->in.len); while (p->in.pos < p->in.len);
@ -561,7 +567,6 @@ int qse_awk_rtx_readio (
n = match_long_rs (run, buf, p); n = match_long_rs (run, buf, p);
if (n != 0) if (n != 0)
{ {
//p->in.pos--; /* unread the character in c */
if (n <= -1) ret = -1; if (n <= -1) ret = -1;
break; break;
} }
@ -983,7 +988,7 @@ int qse_awk_rtx_closio_read (
{ {
qse_awk_rio_fun_t handler; qse_awk_rio_fun_t handler;
handler = run->rio.handler[p->type & MASK_CLEAR]; handler = run->rio.handler[p->type & IO_MASK_CLEAR];
if (handler != QSE_NULL) if (handler != QSE_NULL)
{ {
if (handler (run, QSE_AWK_RIO_CLOSE, p, QSE_NULL, 0) <= -1) if (handler (run, QSE_AWK_RIO_CLOSE, p, QSE_NULL, 0) <= -1)
@ -1042,7 +1047,7 @@ int qse_awk_rtx_closio_write (
{ {
qse_awk_rio_fun_t handler; qse_awk_rio_fun_t handler;
handler = run->rio.handler[p->type & MASK_CLEAR]; handler = run->rio.handler[p->type & IO_MASK_CLEAR];
if (handler != QSE_NULL) if (handler != QSE_NULL)
{ {
qse_awk_rtx_seterrnum (run, QSE_AWK_ENOERR, QSE_NULL); qse_awk_rtx_seterrnum (run, QSE_AWK_ENOERR, QSE_NULL);
@ -1088,7 +1093,7 @@ int qse_awk_rtx_closeio (
{ {
if (opt[0] == QSE_T('r')) if (opt[0] == QSE_T('r'))
{ {
if (p->type & MASK_RDWR) if (p->type & IO_MASK_RDWR)
{ {
if (p->rwcstate != QSE_AWK_RIO_CLOSE_WRITE) if (p->rwcstate != QSE_AWK_RIO_CLOSE_WRITE)
{ {
@ -1098,12 +1103,12 @@ int qse_awk_rtx_closeio (
rwcmode = QSE_AWK_RIO_CLOSE_READ; rwcmode = QSE_AWK_RIO_CLOSE_READ;
} }
} }
else if (!(p->type & MASK_READ)) goto skip; else if (!(p->type & IO_MASK_READ)) goto skip;
} }
else else
{ {
QSE_ASSERT (opt[0] == QSE_T('w')); QSE_ASSERT (opt[0] == QSE_T('w'));
if (p->type & MASK_RDWR) if (p->type & IO_MASK_RDWR)
{ {
if (p->rwcstate != QSE_AWK_RIO_CLOSE_READ) if (p->rwcstate != QSE_AWK_RIO_CLOSE_READ)
{ {
@ -1113,11 +1118,11 @@ int qse_awk_rtx_closeio (
rwcmode = QSE_AWK_RIO_CLOSE_WRITE; rwcmode = QSE_AWK_RIO_CLOSE_WRITE;
} }
} }
else if (!(p->type & MASK_WRITE)) goto skip; else if (!(p->type & IO_MASK_WRITE)) goto skip;
} }
} }
handler = rtx->rio.handler[p->type & MASK_CLEAR]; handler = rtx->rio.handler[p->type & IO_MASK_CLEAR];
if (handler != QSE_NULL) if (handler != QSE_NULL)
{ {
qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOERR, QSE_NULL); qse_awk_rtx_seterrnum (rtx, QSE_AWK_ENOERR, QSE_NULL);
@ -1131,7 +1136,7 @@ int qse_awk_rtx_closeio (
} }
} }
if (p->type & MASK_RDWR) if (p->type & IO_MASK_RDWR)
{ {
p->rwcmode = rwcmode; p->rwcmode = rwcmode;
if (p->rwcstate == 0 && rwcmode != 0) if (p->rwcstate == 0 && rwcmode != 0)
@ -1172,7 +1177,7 @@ void qse_awk_rtx_cleario (qse_awk_rtx_t* run)
while (run->rio.chain != QSE_NULL) while (run->rio.chain != QSE_NULL)
{ {
handler = run->rio.handler[ handler = run->rio.handler[
run->rio.chain->type & MASK_CLEAR]; run->rio.chain->type & IO_MASK_CLEAR];
next = run->rio.chain->next; next = run->rio.chain->next;
if (handler != QSE_NULL) if (handler != QSE_NULL)

4
qse/lib/awk/std.c
View File

@ -1,5 +1,5 @@
/* /*
* $Id: std.c 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: std.c 447 2011-05-01 13:28:51Z 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.
@ -37,7 +37,7 @@
#endif #endif
#ifndef QSE_HAVE_CONFIG_H #ifndef QSE_HAVE_CONFIG_H
# if defined(__OS2__) || defined(_WIN32) # if defined(__OS2__) || defined(_WIN32) || defined(__DOS__)
# define HAVE_POW # define HAVE_POW
# define HAVE_SIN # define HAVE_SIN
# define HAVE_COS # define HAVE_COS

41
qse/lib/awk/tree.c
View File

@ -1,5 +1,5 @@
/* /*
* $Id: tree.c 441 2011-04-22 14:28:43Z hyunghwan.chung $ * $Id: tree.c 447 2011-05-01 13:28:51Z 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.
@ -103,29 +103,34 @@ static const qse_char_t* print_outop_str[] =
QSE_T("") QSE_T("")
}; };
#define PUT_SRCSTR(awk,str) \ #define PUT_SRCSTR(awk,str) QSE_BLOCK (\
do { if (qse_awk_putsrcstr (awk, str) == -1) return -1; } while (0) if (qse_awk_putsrcstr (awk, str) == -1) return -1; \
)
#define PUT_NL(awk) \ #define PUT_NL(awk) QSE_BLOCK (\
do { \ if (awk->option & QSE_AWK_CRLF) PUT_SRCSTR (awk, QSE_T("\r")); \
if (awk->option & QSE_AWK_CRLF) PUT_SRCSTR (awk, QSE_T("\r")); \ PUT_SRCSTR (awk, QSE_T("\n")); \
PUT_SRCSTR (awk, QSE_T("\n")); \ )
} while (0)
#define PUT_SRCSTRX(awk,str,len) \ #define PUT_SRCSTRX(awk,str,len) QSE_BLOCK (\
do { if (qse_awk_putsrcstrx (awk, str, len) == -1) return -1; } while (0) if (qse_awk_putsrcstrx (awk, str, len) == -1) return -1; \
)
#define PRINT_TABS(awk,depth) \ #define PRINT_TABS(awk,depth) QSE_BLOCK (\
do { if (print_tabs(awk,depth) == -1) return -1; } while (0) if (print_tabs(awk,depth) == -1) return -1; \
)
#define PRINT_EXPR(awk,nde) \ #define PRINT_EXPR(awk,nde) QSE_BLOCK (\
do { if (print_expr(awk,nde) == -1) return -1; } while (0) if (print_expr(awk,nde) == -1) return -1; \
)
#define PRINT_EXPR_LIST(awk,nde) \ #define PRINT_EXPR_LIST(awk,nde) QSE_BLOCK(\
do { if (print_expr_list(awk,nde) == -1) return -1; } while (0) if (print_expr_list(awk,nde) == -1) return -1; \
)
#define PRINT_STMTS(awk,nde,depth) \ #define PRINT_STMTS(awk,nde,depth) QSE_BLOCK(\
do { if (print_stmts(awk,nde,depth) == -1) return -1; } while (0) if (print_stmts(awk,nde,depth) == -1) return -1; \
)
static int print_tabs (qse_awk_t* awk, int depth); static int print_tabs (qse_awk_t* awk, int depth);
static int print_expr (qse_awk_t* awk, qse_awk_nde_t* nde); static int print_expr (qse_awk_t* awk, qse_awk_nde_t* nde);

2
qse/regress/awk/lang-044.awk Normal file
View File

@ -0,0 +1,2 @@
BEGIN { RS=""; }
{ print "[" $0 "]"; }

11
qse/regress/awk/lang-044.dat Normal file
View File

@ -0,0 +1,11 @@
abc
this is the second line
def
vwxyz
this is the second second line
ttttt
killer

1
qse/regress/awk/regress.sh
View File

@ -166,6 +166,7 @@ PROGS="
lang-041.awk!!!--newline=on -o- lang-041.awk!!!--newline=on -o-
lang-042.awk!!!--newline=on -o- lang-042.awk!!!--newline=on -o-
lang-043.awk!lang-043.dat!!--newline=on -o- lang-043.awk!lang-043.dat!!--newline=on -o-
lang-044.awk!lang-044.dat!!--newline=on -o-
columnate.awk!./passwd.dat!!--newline=on -F: columnate.awk!./passwd.dat!!--newline=on -F:
levenshtein-utests.awk!!!--newline=on --include=on levenshtein-utests.awk!!!--newline=on --include=on