projects
/
idzebra-moved-to-github.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Added test/codec/Makefile
[idzebra-moved-to-github.git]
/
recctrl
/
regxread.c
diff --git
a/recctrl/regxread.c
b/recctrl/regxread.c
index
9b8473b
..
4b26b8a
100644
(file)
--- a/
recctrl/regxread.c
+++ b/
recctrl/regxread.c
@@
-1,5
+1,5
@@
-/* $Id: regxread.c,v 1.47 2003-04-24 19:34:20 adam Exp $
- Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
+/* $Id: regxread.c,v 1.50 2004-05-25 12:13:15 adam Exp $
+ Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
Index Data Aps
This file is part of the Zebra server.
Index Data Aps
This file is part of the Zebra server.
@@
-473,6
+473,8
@@
static int actionListMk (struct lexSpec *spec, const char *s,
logf (LOG_WARN, "regular expression error '%.*s'", s-s0, s0);
return -1;
}
logf (LOG_WARN, "regular expression error '%.*s'", s-s0, s0);
return -1;
}
+ if (debug_dfa_tran)
+ printf ("pattern: %.*s\n", s-s0, s0);
dfa_mkstate ((*ap)->u.pattern.dfa);
s++;
break;
dfa_mkstate ((*ap)->u.pattern.dfa);
s++;
break;
@@
-590,6
+592,14
@@
int readFileSpec (struct lexSpec *spec)
if (spec->tcl_interp)
logf (LOG_LOG, "Tcl enabled");
#endif
if (spec->tcl_interp)
logf (LOG_LOG, "Tcl enabled");
#endif
+
+#if 0
+ debug_dfa_trav = 0;
+ debug_dfa_tran = 1;
+ debug_dfa_followpos = 0;
+ dfa_verbose = 1;
+#endif
+
lineBuf = wrbuf_alloc();
spec->lineNo = 0;
c = getc (spec_inf);
lineBuf = wrbuf_alloc();
spec->lineNo = 0;
c = getc (spec_inf);
@@
-632,12
+642,6
@@
int readFileSpec (struct lexSpec *spec)
fclose (spec_inf);
wrbuf_free(lineBuf, 1);
fclose (spec_inf);
wrbuf_free(lineBuf, 1);
-#if 0
- debug_dfa_trav = 1;
- debug_dfa_tran = 1;
- debug_dfa_followpos = 1;
- dfa_verbose = 1;
-#endif
for (lc = spec->context; lc; lc = lc->next)
{
struct lexRule *rp;
for (lc = spec->context; lc; lc = lc->next)
{
struct lexRule *rp;
@@
-668,13
+672,12
@@
static void execData (struct lexSpec *spec,
if (elen == 0) /* shouldn't happen, but it does! */
return ;
#if REGX_DEBUG
if (elen == 0) /* shouldn't happen, but it does! */
return ;
#if REGX_DEBUG
- if (elen > 40)
- logf (LOG_LOG, "data(%d bytes) %.15s ... %.*s", elen,
- ebuf, 15, ebuf + elen-15);
+ if (elen > 80)
+ logf (LOG_LOG, "data(%d bytes) %.40s ... %.*s", elen,
+ ebuf, 40, ebuf + elen-40);
else if (elen == 1 && ebuf[0] == '\n')
{
logf (LOG_LOG, "data(new line)");
else if (elen == 1 && ebuf[0] == '\n')
{
logf (LOG_LOG, "data(new line)");
- assert(0);
}
else if (elen > 0)
logf (LOG_LOG, "data(%d bytes) %.*s", elen, elen, ebuf);
}
else if (elen > 0)
logf (LOG_LOG, "data(%d bytes) %.*s", elen, elen, ebuf);
@@
-875,21
+878,33
@@
static void tagEnd (struct lexSpec *spec, int min_level,
static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
- struct DFA *dfa)
+ struct DFA *dfa, int greedy)
{
struct DFA_state *state = dfa->states[0];
struct DFA_tran *t;
{
struct DFA_state *state = dfa->states[0];
struct DFA_tran *t;
- unsigned char c;
+ unsigned char c = 0;
unsigned char c_prev = 0;
int ptr = *pptr; /* current pointer */
int start_ptr = *pptr; /* first char of match */
int last_ptr = 0; /* last char of match */
int last_rule = 0; /* rule number of current match */
unsigned char c_prev = 0;
int ptr = *pptr; /* current pointer */
int start_ptr = *pptr; /* first char of match */
int last_ptr = 0; /* last char of match */
int last_rule = 0; /* rule number of current match */
+ int restore_ptr = 0;
int i;
int i;
+ if (ptr)
+ {
+ --ptr;
+ c = f_win_advance (spec, &ptr);
+ }
while (1)
{
while (1)
{
+ if (dfa->states[0] == state)
+ {
+ c_prev = c;
+ restore_ptr = ptr;
+ }
c = f_win_advance (spec, &ptr);
c = f_win_advance (spec, &ptr);
+
if (ptr == F_WIN_EOF)
{
if (last_rule)
if (ptr == F_WIN_EOF)
{
if (last_rule)
@@
-900,10
+915,11
@@
static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
}
break;
}
}
break;
}
+
t = state->trans;
i = state->tran_no;
while (1)
t = state->trans;
i = state->tran_no;
while (1)
- if (--i < 0)
+ if (--i < 0) /* no transition for character c */
{
if (last_rule)
{
{
if (last_rule)
{
@@
-912,27
+928,28
@@
static int tryMatch (struct lexSpec *spec, int *pptr, int *mptr,
return 1;
}
state = dfa->states[0];
return 1;
}
state = dfa->states[0];
+
+ ptr = restore_ptr;
+ c = f_win_advance (spec, &ptr);
+
start_ptr = ptr;
start_ptr = ptr;
- c_prev = c;
+
break;
}
else if (c >= t->ch[0] && c <= t->ch[1])
{
state = dfa->states[t->to];
break;
}
else if (c >= t->ch[0] && c <= t->ch[1])
{
state = dfa->states[t->to];
- if (state->rule_no)
- {
- if (c_prev == '\n')
- {
- last_rule = state->rule_no;
- last_ptr = ptr;
- }
- else
- {
- last_rule = state->rule_nno;
- last_ptr = ptr;
- }
- }
- break;
+ if (state->rule_no && c_prev == '\n')
+ {
+ last_rule = state->rule_no;
+ last_ptr = ptr;
+ }
+ else if (state->rule_nno)
+ {
+ last_rule = state->rule_nno;
+ last_ptr = ptr;
+ }
+ break;
}
else
t++;
}
else
t++;
@@
-1031,6
+1048,8
@@
static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
#endif
res = data1_mk_root (spec->dh, spec->m, absynName);
#endif
res = data1_mk_root (spec->dh, spec->m, absynName);
+ spec->d1_level = 0;
+
spec->d1_stack[spec->d1_level++] = res;
res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
spec->d1_stack[spec->d1_level++] = res;
res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
@@
-1090,7
+1109,7
@@
static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
}
else if (!strcmp (argv[1], "element"))
{
}
else if (!strcmp (argv[1], "element"))
{
- int min_level = 1;
+ int min_level = 2;
char *element = 0;
if (argc >= 3 && !strcmp(argv[2], "-record"))
{
char *element = 0;
if (argc >= 3 && !strcmp(argv[2], "-record"))
{
@@
-1102,7
+1121,7
@@
static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
if (argc == 3)
element = argv[2];
tagEnd (spec, min_level, element, (element ? strlen(element) : 0));
if (argc == 3)
element = argv[2];
tagEnd (spec, min_level, element, (element ? strlen(element) : 0));
- if (spec->d1_level == 0)
+ if (spec->d1_level <= 1)
{
#if REGX_DEBUG
logf (LOG_LOG, "end element end records");
{
#if REGX_DEBUG
logf (LOG_LOG, "end element end records");
@@
-1163,7
+1182,7
@@
static int cmd_tcl_data (ClientData clientData, Tcl_Interp *interp,
argi++;
}
if (element)
argi++;
}
if (element)
- tagEnd (spec, 1, NULL, 0);
+ tagEnd (spec, 2, NULL, 0);
return TCL_OK;
}
return TCL_OK;
}
@@
-1266,7
+1285,7
@@
static void execCode (struct lexSpec *spec, struct regxCode *code)
r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
- if (spec->d1_level == 0)
+ if (spec->d1_level <= 1)
{
static char absynName[64];
data1_node *res;
{
static char absynName[64];
data1_node *res;
@@
-1280,6
+1299,8
@@
static void execCode (struct lexSpec *spec, struct regxCode *code)
#endif
res = data1_mk_root (spec->dh, spec->m, absynName);
#endif
res = data1_mk_root (spec->dh, spec->m, absynName);
+ spec->d1_level = 0;
+
spec->d1_stack[spec->d1_level++] = res;
res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
spec->d1_stack[spec->d1_level++] = res;
res = data1_mk_tag (spec->dh, spec->m, absynName, 0, res);
@@
-1378,7
+1399,7
@@
static void execCode (struct lexSpec *spec, struct regxCode *code)
}
else if (!strcmp (p, "element"))
{
}
else if (!strcmp (p, "element"))
{
- int min_level = 1;
+ int min_level = 2;
while ((r = execTok (spec, &s, &cmd_str, &cmd_len)) == 3)
{
if (cmd_len==7 && !memcmp ("-record", cmd_str, cmd_len))
while ((r = execTok (spec, &s, &cmd_str, &cmd_len)) == 3)
{
if (cmd_len==7 && !memcmp ("-record", cmd_str, cmd_len))
@@
-1391,7
+1412,7
@@
static void execCode (struct lexSpec *spec, struct regxCode *code)
}
else
tagEnd (spec, min_level, NULL, 0);
}
else
tagEnd (spec, min_level, NULL, 0);
- if (spec->d1_level == 0)
+ if (spec->d1_level <= 1)
{
#if REGX_DEBUG
logf (LOG_LOG, "end element end records");
{
#if REGX_DEBUG
logf (LOG_LOG, "end element end records");
@@
-1445,7
+1466,7
@@
static void execCode (struct lexSpec *spec, struct regxCode *code)
r = execTok (spec, &s, &cmd_str, &cmd_len);
} while (r > 1);
if (element_str)
r = execTok (spec, &s, &cmd_str, &cmd_len);
} while (r > 1);
if (element_str)
- tagEnd (spec, 1, NULL, 0);
+ tagEnd (spec, 2, NULL, 0);
}
else if (!strcmp (p, "unread"))
{
}
else if (!strcmp (p, "unread"))
{
@@
-1542,13
+1563,14
@@
static int execAction (struct lexSpec *spec, struct lexRuleAction *ap,
if (ap->u.pattern.body)
{
arg_start[arg_no] = *pptr;
if (ap->u.pattern.body)
{
arg_start[arg_no] = *pptr;
- if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa))
+ if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa, 0))
{
arg_end[arg_no] = F_WIN_EOF;
arg_no++;
arg_start[arg_no] = F_WIN_EOF;
arg_end[arg_no] = F_WIN_EOF;
{
arg_end[arg_no] = F_WIN_EOF;
arg_no++;
arg_start[arg_no] = F_WIN_EOF;
arg_end[arg_no] = F_WIN_EOF;
-/* return 1*/
+ yaz_log(LOG_DEBUG, "Pattern match rest of record");
+ *pptr = F_WIN_EOF;
}
else
{
}
else
{
@@
-1561,7
+1583,7
@@
static int execAction (struct lexSpec *spec, struct lexRuleAction *ap,
else
{
arg_start[arg_no] = *pptr;
else
{
arg_start[arg_no] = *pptr;
- if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa))
+ if (!tryMatch (spec, pptr, &sptr, ap->u.pattern.dfa, 1))
return 1;
if (sptr != arg_start[arg_no])
return 1;
return 1;
if (sptr != arg_start[arg_no])
return 1;