/*
- * Copyright (C) 1994-1998, Index Data I/S
+ * Copyright (C) 1994-1999, Index Data
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: regxread.c,v $
- * Revision 1.16 1998-06-30 15:15:09 adam
+ * Revision 1.26 1999-05-26 07:49:14 adam
+ * C++ compilation.
+ *
+ * Revision 1.25 1999/05/25 12:33:32 adam
+ * Fixed bug in Tcl filter.
+ *
+ * Revision 1.24 1999/05/21 11:08:46 adam
+ * Tcl filter attempts to read <filt>.tflt. Improvements to configure
+ * script so that it reads uninstalled Tcl source.
+ *
+ * Revision 1.23 1999/05/20 12:57:18 adam
+ * Implemented TCL filter. Updated recctrl system.
+ *
+ * Revision 1.22 1998/11/03 16:07:13 adam
+ * Yet another fix.
+ *
+ * Revision 1.21 1998/11/03 15:43:39 adam
+ * Fixed bug introduced by previous commit.
+ *
+ * Revision 1.20 1998/11/03 14:51:28 adam
+ * Changed code so that it creates as few data1 nodes as possible.
+ *
+ * Revision 1.19 1998/11/03 10:22:39 adam
+ * Fixed memory leak that could occur for when large data1 node were
+ * concatenated. Data-type data1_nodes may have multiple nodes.
+ *
+ * Revision 1.18 1998/10/15 13:11:47 adam
+ * Added support for option -record for "end element". When specified
+ * end element will mark end-of-record when at outer-level.
+ *
+ * Revision 1.17 1998/07/01 10:13:51 adam
+ * Minor fix.
+ *
+ * Revision 1.16 1998/06/30 15:15:09 adam
* Tags are trimmed: white space removed before- and after the tag.
*
* Revision 1.15 1998/06/30 12:55:45 adam
#include <dfa.h>
#include "grsread.h"
+#if HAVE_TCL_H
+#include <tcl.h>
+#endif
+
#define REGX_DEBUG 0
#define F_WIN_EOF 2000000000
#define REGX_END 4
#define REGX_CODE 5
#define REGX_CONTEXT 6
+#define REGX_INIT 7
struct regxCode {
char *str;
struct lexRule *rules;
struct lexRuleInfo **fastRule;
int ruleNo;
+ int initFlag;
struct lexRuleAction *beginActionList;
struct lexRuleAction *endActionList;
+ struct lexRuleAction *initActionList;
struct lexContext *next;
};
+struct lexConcatBuf {
+ int len;
+ int max;
+ char *buf;
+};
+
struct lexSpec {
char *name;
struct lexContext *context;
int lineNo;
NMEM m;
data1_handle dh;
+#if HAVE_TCL_H
+ Tcl_Interp *tcl_interp;
+#endif
void *f_win_fh;
void (*f_win_ef)(void *, off_t);
int (*f_win_rf)(void *, char *, size_t);
off_t (*f_win_sf)(void *, off_t);
+ struct lexConcatBuf **concatBuf;
+ int maxLevel;
+ data1_node **d1_stack;
+ int d1_level;
+ int stop_flag;
+
+ int *arg_start;
+ int *arg_end;
+ int arg_no;
+ int ptr;
};
+struct lexSpecs {
+ struct lexSpec *spec;
+};
static char *f_win_get (struct lexSpec *spec, off_t start_pos, off_t end_pos,
int *size)
spec->f_win_start = start_pos;
if (!spec->f_win_buf)
- spec->f_win_buf = xmalloc (spec->f_win_size);
+ spec->f_win_buf = (char *) xmalloc (spec->f_win_size);
*size = (*spec->f_win_rf)(spec->f_win_fh, spec->f_win_buf,
spec->f_win_size);
spec->f_win_end = spec->f_win_start + *size;
{
struct regxCode *p;
- p = xmalloc (sizeof(*p));
- p->str = xmalloc (len+1);
+ p = (struct regxCode *) xmalloc (sizeof(*p));
+ p->str = (char *) xmalloc (len+1);
memcpy (p->str, buf, len);
p->str[len] = '\0';
*pp = p;
static struct lexContext *lexContextCreate (const char *name)
{
- struct lexContext *p = xmalloc (sizeof(*p));
+ struct lexContext *p = (struct lexContext *) xmalloc (sizeof(*p));
p->name = xstrdup (name);
p->ruleNo = 1;
+ p->initFlag = 0;
p->dfa = lexSpecDFA ();
p->rules = NULL;
p->fastRule = NULL;
p->beginActionList = NULL;
p->endActionList = NULL;
+ p->initActionList = NULL;
p->next = NULL;
return p;
}
xfree (p);
}
-static struct lexSpec *lexSpecCreate (const char *name)
+static struct lexSpec *lexSpecCreate (const char *name, data1_handle dh)
{
struct lexSpec *p;
-
- p = xmalloc (sizeof(*p));
- p->name = xmalloc (strlen(name)+1);
+ int i;
+
+ p = (struct lexSpec *) xmalloc (sizeof(*p));
+ p->name = (char *) xmalloc (strlen(name)+1);
strcpy (p->name, name);
+#if HAVE_TCL_H
+ p->tcl_interp = 0;
+#endif
+ p->dh = dh;
p->context = NULL;
p->context_stack_size = 100;
- p->context_stack = xmalloc (sizeof(*p->context_stack) *
- p->context_stack_size);
+ p->context_stack = (struct lexContext **)
+ xmalloc (sizeof(*p->context_stack) * p->context_stack_size);
p->f_win_buf = NULL;
+
+ p->maxLevel = 128;
+ p->concatBuf = (struct lexConcatBuf **)
+ xmalloc (sizeof(*p->concatBuf) * p->maxLevel);
+ for (i = 0; i < p->maxLevel; i++)
+ {
+ p->concatBuf[i] = (struct lexConcatBuf *)
+ xmalloc (sizeof(**p->concatBuf));
+ p->concatBuf[i]->len = p->concatBuf[i]->max = 0;
+ p->concatBuf[i]->buf = 0;
+ }
+ p->d1_stack = (data1_node **) xmalloc (sizeof(*p->d1_stack) * p->maxLevel);
+ p->d1_level = 0;
return p;
}
{
struct lexSpec *p;
struct lexContext *lt;
+ int i;
assert (pp);
p = *pp;
if (!p)
return ;
+
+ for (i = 0; i < p->maxLevel; i++)
+ xfree (p->concatBuf[i]);
+ xfree (p->concatBuf);
+
lt = p->context;
while (lt)
{
lexContextDestroy (lt);
lt = lt_next;
}
+#if HAVE_TCL_H
+ if (p->tcl_interp)
+ Tcl_DeleteInterp (p->tcl_interp);
+#endif
xfree (p->name);
xfree (p->f_win_buf);
xfree (p->context_stack);
+ xfree (p->d1_stack);
xfree (p);
*pp = NULL;
}
cmd[i] = *cp + 'a' - 'A';
else
break;
- if (i < sizeof(cmd)-2)
+ if (i < (int) sizeof(cmd)-2)
i++;
cp++;
}
return REGX_BODY;
else if (!strcmp (cmd, "context"))
return REGX_CONTEXT;
+ else if (!strcmp (cmd, "init"))
+ return REGX_INIT;
else
{
logf (LOG_WARN, "bad command %s", cmd);
bodyMark = 1;
continue;
case REGX_CODE:
- *ap = xmalloc (sizeof(**ap));
+ *ap = (struct lexRuleAction *) xmalloc (sizeof(**ap));
(*ap)->which = tok;
regxCodeMk (&(*ap)->u.code, s, len);
s += len+1;
break;
case REGX_PATTERN:
- *ap = xmalloc (sizeof(**ap));
+ *ap = (struct lexRuleAction *) xmalloc (sizeof(**ap));
(*ap)->which = tok;
(*ap)->u.pattern.body = bodyMark;
bodyMark = 0;
s++;
break;
case REGX_BEGIN:
- logf (LOG_WARN, "cannot use begin here");
+ logf (LOG_WARN, "cannot use BEGIN here");
+ continue;
+ case REGX_INIT:
+ logf (LOG_WARN, "cannot use INIT here");
continue;
case REGX_END:
- *ap = xmalloc (sizeof(**ap));
+ *ap = (struct lexRuleAction *) xmalloc (sizeof(**ap));
(*ap)->which = tok;
break;
}
actionListDel (&spec->context->endActionList);
actionListMk (spec, s, &spec->context->endActionList);
break;
+ case REGX_INIT:
+ actionListDel (&spec->context->initActionList);
+ actionListMk (spec, s, &spec->context->initActionList);
+ break;
case REGX_PATTERN:
#if REGX_DEBUG
logf (LOG_DEBUG, "rule %d %s", spec->context->ruleNo, s);
return -1;
}
s++;
- rp = xmalloc (sizeof(*rp));
+ rp = (struct lexRule *) xmalloc (sizeof(*rp));
rp->info.no = spec->context->ruleNo++;
rp->next = spec->context->rules;
spec->context->rules = rp;
char *lineBuf;
int lineSize = 512;
int c, i, errors = 0;
- FILE *spec_inf;
+ FILE *spec_inf = 0;
- lineBuf = xmalloc (1+lineSize);
- logf (LOG_LOG, "reading regx filter %s.flt", spec->name);
- sprintf (lineBuf, "%s.flt", spec->name);
- if (!(spec_inf = yaz_path_fopen (data1_get_tabpath(spec->dh),
- lineBuf, "r")))
+ lineBuf = (char *) xmalloc (1+lineSize);
+#if HAVE_TCL_H
+ if (spec->tcl_interp)
+ {
+ sprintf (lineBuf, "%s.tflt", spec->name);
+ spec_inf = yaz_path_fopen (data1_get_tabpath(spec->dh), lineBuf, "r");
+ }
+#endif
+ if (!spec_inf)
+ {
+ sprintf (lineBuf, "%s.flt", spec->name);
+ spec_inf = yaz_path_fopen (data1_get_tabpath(spec->dh), lineBuf, "r");
+ }
+ if (!spec_inf)
{
logf (LOG_ERRNO|LOG_WARN, "cannot read spec file %s", spec->name);
xfree (lineBuf);
return -1;
}
+ logf (LOG_LOG, "reading regx filter %s", lineBuf);
+#if HAVE_TCL_H
+ if (spec->tcl_interp)
+ logf (LOG_LOG, "Tcl enabled");
+#endif
spec->lineNo = 0;
c = getc (spec_inf);
while (c != EOF)
for (lc = spec->context; lc; lc = lc->next)
{
struct lexRule *rp;
- lc->fastRule = xmalloc (sizeof(*lc->fastRule) * lc->ruleNo);
+ lc->fastRule = (struct lexRuleInfo **)
+ xmalloc (sizeof(*lc->fastRule) * lc->ruleNo);
for (i = 0; i < lc->ruleNo; i++)
lc->fastRule[i] = NULL;
for (rp = lc->rules; rp; rp = rp->next)
}
if (errors)
return -1;
+
return 0;
}
+#if 0
static struct lexSpec *curLexSpec = NULL;
-
-static void destroy_data (struct data1_node *n)
-{
- assert (n->which == DATA1N_data);
- xfree (n->u.data.data);
-}
+#endif
static void execData (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level,
const char *ebuf, int elen, int formatted_text)
{
struct data1_node *res, *parent;
+ int org_len;
if (elen == 0) /* shouldn't happen, but it does! */
return ;
logf (LOG_DEBUG, "data (%d bytes)", elen);
#endif
- if (*d1_level <= 1)
+ if (spec->d1_level <= 1)
return;
- parent = d1_stack[*d1_level -1];
+ parent = spec->d1_stack[spec->d1_level -1];
assert (parent);
- if ((res=d1_stack[*d1_level]) && res->which == DATA1N_data)
+
+ if ((res = spec->d1_stack[spec->d1_level]) && res->which == DATA1N_data)
+ org_len = res->u.data.len;
+ else
{
- if (elen + res->u.data.len <= DATA1_LOCALDATA)
- memcpy (res->u.data.data + res->u.data.len, ebuf, elen);
- else
- {
- char *nb = xmalloc (elen + res->u.data.len);
- memcpy (nb, res->u.data.data, res->u.data.len);
- memcpy (nb + res->u.data.len, ebuf, elen);
- res->u.data.data = nb;
- res->destroy = destroy_data;
- }
- res->u.data.len += elen;
+ org_len = 0;
+
+ res = data1_mk_node (spec->dh, spec->m);
+ res->parent = parent;
+ res->which = DATA1N_data;
+ res->u.data.what = DATA1I_text;
+ res->u.data.len = 0;
+ res->u.data.formatted_text = formatted_text;
+#if 0
+ if (elen > DATA1_LOCALDATA)
+ res->u.data.data = nmem_malloc (spec->m, elen);
+ else
+ res->u.data.data = res->lbuf;
+ memcpy (res->u.data.data, ebuf, elen);
+#else
+ res->u.data.data = 0;
+#endif
+ res->root = parent->root;
+
+ parent->last_child = res;
+ if (spec->d1_stack[spec->d1_level])
+ spec->d1_stack[spec->d1_level]->next = res;
+ else
+ parent->child = res;
+ spec->d1_stack[spec->d1_level] = res;
}
- else
+ if (org_len + elen >= spec->concatBuf[spec->d1_level]->max)
{
- res = data1_mk_node (spec->dh, spec->m);
- res->parent = parent;
- res->which = DATA1N_data;
- res->u.data.what = DATA1I_text;
- res->u.data.len = elen;
- res->u.data.formatted_text = formatted_text;
- if (elen > DATA1_LOCALDATA)
- res->u.data.data = nmem_malloc (spec->m, elen);
- else
- res->u.data.data = res->lbuf;
- memcpy (res->u.data.data, ebuf, elen);
- res->root = parent->root;
-
- parent->last_child = res;
- if (d1_stack[*d1_level])
- d1_stack[*d1_level]->next = res;
- else
- parent->child = res;
- d1_stack[*d1_level] = res;
+ char *old_buf, *new_buf;
+
+ spec->concatBuf[spec->d1_level]->max = org_len + elen + 256;
+ new_buf = (char *) xmalloc (spec->concatBuf[spec->d1_level]->max);
+ if ((old_buf = spec->concatBuf[spec->d1_level]->buf))
+ {
+ memcpy (new_buf, old_buf, org_len);
+ xfree (old_buf);
+ }
+ spec->concatBuf[spec->d1_level]->buf = new_buf;
}
+ assert (spec->concatBuf[spec->d1_level]);
+ memcpy (spec->concatBuf[spec->d1_level]->buf + org_len, ebuf, elen);
+ res->u.data.len += elen;
}
static void execDataP (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level,
const char *ebuf, int elen, int formatted_text)
{
- execData (spec, d1_stack, d1_level, ebuf, elen, formatted_text);
+ execData (spec, ebuf, elen, formatted_text);
+}
+
+static void tagDataRelease (struct lexSpec *spec)
+{
+ data1_node *res;
+
+ if ((res = spec->d1_stack[spec->d1_level]) &&
+ res->which == DATA1N_data &&
+ res->u.data.what == DATA1I_text)
+ {
+ assert (!res->u.data.data);
+ assert (res->u.data.len > 0);
+ if (res->u.data.len > DATA1_LOCALDATA)
+ res->u.data.data = (char *) nmem_malloc (spec->m, res->u.data.len);
+ else
+ res->u.data.data = res->lbuf;
+ memcpy (res->u.data.data, spec->concatBuf[spec->d1_level]->buf,
+ res->u.data.len);
+ }
}
static void variantBegin (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level,
const char *class_str, int class_len,
const char *type_str, int type_len,
const char *value_str, int value_len)
{
- struct data1_node *parent = d1_stack[*d1_level -1];
+ struct data1_node *parent = spec->d1_stack[spec->d1_level -1];
char tclass[DATA1_MAX_SYMBOL], ttype[DATA1_MAX_SYMBOL];
data1_vartype *tp;
int i;
data1_node *res;
- if (*d1_level == 0)
+ if (spec->d1_level == 0)
{
logf (LOG_WARN, "in variant begin. No record type defined");
return ;
ttype[type_len] = '\0';
#if REGX_DEBUG
- logf (LOG_DEBUG, "variant begin %s %s (%d)", tclass, ttype, *d1_level);
+ logf (LOG_DEBUG, "variant begin %s %s (%d)", tclass, ttype,
+ spec->d1_level);
#endif
if (!(tp =
res->root = parent->root;
parent->last_child = res;
- if (d1_stack[*d1_level])
- d1_stack[*d1_level]->next = res;
+ if (spec->d1_stack[spec->d1_level])
+ {
+ tagDataRelease (spec);
+ spec->d1_stack[spec->d1_level]->next = res;
+ }
else
parent->child = res;
- d1_stack[*d1_level] = res;
- d1_stack[++(*d1_level)] = NULL;
+ spec->d1_stack[spec->d1_level] = res;
+ spec->d1_stack[++(spec->d1_level)] = NULL;
}
- for (i = *d1_level-1; d1_stack[i]->which == DATA1N_variant; i--)
- if (d1_stack[i]->u.variant.type == tp)
+ for (i = spec->d1_level-1; spec->d1_stack[i]->which == DATA1N_variant; i--)
+ if (spec->d1_stack[i]->u.variant.type == tp)
{
- *d1_level = i;
+ spec->d1_level = i;
break;
}
#if REGX_DEBUG
- logf (LOG_DEBUG, "variant node (%d)", *d1_level);
+ logf (LOG_DEBUG, "variant node (%d)", spec->d1_level);
#endif
- parent = d1_stack[*d1_level-1];
+ parent = spec->d1_stack[spec->d1_level-1];
res = data1_mk_node (spec->dh, spec->m);
res->parent = parent;
res->which = DATA1N_variant;
res->u.variant.value = res->lbuf;
parent->last_child = res;
- if (d1_stack[*d1_level])
- d1_stack[*d1_level]->next = res;
+ if (spec->d1_stack[spec->d1_level])
+ {
+ tagDataRelease (spec);
+ spec->d1_stack[spec->d1_level]->next = res;
+ }
else
parent->child = res;
- d1_stack[*d1_level] = res;
- d1_stack[++(*d1_level)] = NULL;
+ spec->d1_stack[spec->d1_level] = res;
+ spec->d1_stack[++(spec->d1_level)] = NULL;
}
static void tagStrip (const char **tag, int *len)
}
static void tagBegin (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level,
const char *tag, int len)
{
- struct data1_node *parent = d1_stack[*d1_level -1];
+ struct data1_node *parent = spec->d1_stack[spec->d1_level -1];
data1_element *elem = NULL;
data1_node *partag = get_parent_tag(spec->dh, parent);
data1_node *res;
data1_element *e = NULL;
int localtag = 0;
- if (*d1_level == 0)
+ if (spec->d1_level == 0)
{
logf (LOG_WARN, "in element begin. No record type defined");
return ;
res->u.tag.get_bytes = -1;
if (len >= DATA1_LOCALDATA)
- len = DATA1_LOCALDATA-1;
- memcpy (res->lbuf, tag, len);
- res->lbuf[len] = '\0';
- res->u.tag.tag = res->lbuf;
+ res->u.tag.tag = (char *) nmem_malloc (spec->m, len+1);
+ else
+ res->u.tag.tag = res->lbuf;
+
+ memcpy (res->u.tag.tag, tag, len);
+ res->u.tag.tag[len] = '\0';
#if REGX_DEBUG
- logf (LOG_DEBUG, "begin tag %s (%d)", res->u.tag.tag, *d1_level);
+ logf (LOG_DEBUG, "begin tag %s (%d)", res->u.tag.tag, spec->d1_level);
#endif
if (parent->which == DATA1N_variant)
return ;
if (!(e = partag->u.tag.element))
localtag = 1;
- elem = data1_getelementbytagname (spec->dh, d1_stack[0]->u.root.absyn,
+ elem = data1_getelementbytagname (spec->dh,
+ spec->d1_stack[0]->u.root.absyn,
e, res->u.tag.tag);
res->u.tag.element = elem;
res->u.tag.node_selected = 0;
res->root = parent->root;
parent->last_child = res;
- if (d1_stack[*d1_level])
- d1_stack[*d1_level]->next = res;
+ if (spec->d1_stack[spec->d1_level])
+ {
+ tagDataRelease (spec);
+ spec->d1_stack[spec->d1_level]->next = res;
+ }
else
parent->child = res;
- d1_stack[*d1_level] = res;
- d1_stack[++(*d1_level)] = NULL;
+ spec->d1_stack[spec->d1_level] = res;
+ spec->d1_stack[++(spec->d1_level)] = NULL;
}
-static void tagEnd (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level,
+static void tagEnd (struct lexSpec *spec, int min_level,
const char *tag, int len)
{
tagStrip (&tag, &len);
- while (*d1_level > 1)
+ while (spec->d1_level > min_level)
{
- (*d1_level)--;
- if ((d1_stack[*d1_level]->which == DATA1N_tag) &&
+ tagDataRelease (spec);
+ (spec->d1_level)--;
+ if (spec->d1_level == 0)
+ break;
+ if ((spec->d1_stack[spec->d1_level]->which == DATA1N_tag) &&
(!tag ||
- (strlen(d1_stack[*d1_level]->u.tag.tag) == (size_t) len &&
- !memcmp (d1_stack[*d1_level]->u.tag.tag, tag, len))))
+ (strlen(spec->d1_stack[spec->d1_level]->u.tag.tag) ==
+ (size_t) len &&
+ !memcmp (spec->d1_stack[spec->d1_level]->u.tag.tag, tag, len))))
break;
}
#if REGX_DEBUG
- logf (LOG_DEBUG, "end tag (%d)", *d1_level);
+ logf (LOG_DEBUG, "end tag (%d)", spec->d1_level);
#endif
}
}
static int execTok (struct lexSpec *spec, const char **src,
- int arg_no, int *arg_start, int *arg_end,
const char **tokBuf, int *tokLen)
{
const char *s = *src;
s++;
while (*s >= '0' && *s <= '9')
n = n*10 + (*s++ -'0');
- if (arg_no == 0)
+ if (spec->arg_no == 0)
{
*tokBuf = "";
*tokLen = 0;
}
else
{
- if (n >= arg_no)
- n = arg_no-1;
- *tokBuf = f_win_get (spec, arg_start[n], arg_end[n], tokLen);
+ if (n >= spec->arg_no)
+ n = spec->arg_no-1;
+ *tokBuf = f_win_get (spec, spec->arg_start[n], spec->arg_end[n],
+ tokLen);
}
}
else if (*s == '\"')
return str;
}
-static int execCode (struct lexSpec *spec,
- int arg_no, int *arg_start, int *arg_end, int *pptr,
- struct regxCode *code,
- data1_node **d1_stack, int *d1_level)
+#if HAVE_TCL_H
+static int cmd_tcl_begin (ClientData clientData, Tcl_Interp *interp,
+ int argc, char **argv)
+{
+ struct lexSpec *spec = (struct lexSpec *) clientData;
+ if (argc < 2)
+ return TCL_ERROR;
+ if (!strcmp(argv[1], "record") && argc == 3)
+ {
+ char *absynName = argv[2];
+ data1_absyn *absyn;
+
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "begin record %s", absynName);
+#endif
+ if (!(absyn = data1_get_absyn (spec->dh, absynName)))
+ logf (LOG_WARN, "Unknown tagset: %s", absynName);
+ else
+ {
+ data1_node *res;
+
+ res = data1_mk_node (spec->dh, spec->m);
+ res->which = DATA1N_root;
+ res->u.root.type = absynName;
+ res->u.root.absyn = absyn;
+ res->root = res;
+
+ spec->d1_stack[spec->d1_level] = res;
+ spec->d1_stack[++(spec->d1_level)] = NULL;
+ }
+ }
+ else if (!strcmp(argv[1], "element") && argc == 3)
+ {
+ tagBegin (spec, argv[2], strlen(argv[2]));
+ }
+ else if (!strcmp (argv[1], "variant") && argc == 5)
+ {
+ variantBegin (spec, argv[2], strlen(argv[2]),
+ argv[3], strlen(argv[3]),
+ argv[4], strlen(argv[4]));
+ }
+ else if (!strcmp (argv[1], "context") && argc == 3)
+ {
+ struct lexContext *lc = spec->context;
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "begin context %s",argv[2]);
+#endif
+ while (lc && strcmp (argv[2], lc->name))
+ lc = lc->next;
+ if (lc)
+ {
+ spec->context_stack[++(spec->context_stack_top)] = lc;
+ }
+ else
+ logf (LOG_WARN, "unknown context %s", argv[2]);
+ }
+ else
+ return TCL_ERROR;
+ return TCL_OK;
+}
+
+static int cmd_tcl_end (ClientData clientData, Tcl_Interp *interp,
+ int argc, char **argv)
+{
+ struct lexSpec *spec = (struct lexSpec *) clientData;
+ if (argc < 2)
+ return TCL_ERROR;
+
+ if (!strcmp (argv[1], "record"))
+ {
+ while (spec->d1_level)
+ {
+ tagDataRelease (spec);
+ (spec->d1_level)--;
+ }
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "end record");
+#endif
+ spec->stop_flag = 1;
+ }
+ else if (!strcmp (argv[1], "element"))
+ {
+ int min_level = 1;
+ char *element = 0;
+ if (argc >= 3 && !strcmp(argv[2], "-record"))
+ {
+ min_level = 0;
+ if (argc == 4)
+ element = argv[3];
+ }
+ else
+ if (argc == 3)
+ element = argv[2];
+ tagEnd (spec, min_level, element, (element ? strlen(element) : 0));
+ if (spec->d1_level == 0)
+ {
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "end element end records");
+#endif
+ spec->stop_flag = 1;
+ }
+ }
+ else if (!strcmp (argv[1], "context"))
+ {
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "end context");
+#endif
+ if (spec->context_stack_top)
+ (spec->context_stack_top)--;
+ }
+ else
+ return TCL_ERROR;
+ return TCL_OK;
+}
+
+static int cmd_tcl_data (ClientData clientData, Tcl_Interp *interp,
+ int argc, char **argv)
+{
+ int argi = 1;
+ int textFlag = 0;
+ const char *element = 0;
+ struct lexSpec *spec = (struct lexSpec *) clientData;
+
+ while (argi < argc)
+ {
+ if (!strcmp("-text", argv[argi]))
+ {
+ textFlag = 1;
+ argi++;
+ }
+ else if (!strcmp("-element", argv[argi]))
+ {
+ argi++;
+ if (argi < argc)
+ element = argv[argi++];
+ }
+ else
+ break;
+ }
+ if (element)
+ tagBegin (spec, element, strlen(element));
+
+ while (argi < argc)
+ {
+ execData (spec, argv[argi], strlen(argv[argi]), textFlag);
+ argi++;
+ }
+ if (element)
+ tagEnd (spec, 1, NULL, 0);
+ return TCL_OK;
+}
+
+static int cmd_tcl_unread (ClientData clientData, Tcl_Interp *interp,
+ int argc, char **argv)
+{
+ struct lexSpec *spec = (struct lexSpec *) clientData;
+ int argi = 1;
+ int offset = 0;
+ int no;
+
+ while (argi < argc)
+ {
+ if (!strcmp("-offset", argv[argi]))
+ {
+ argi++;
+ if (argi < argc)
+ {
+ offset = atoi(argv[argi]);
+ argi++;
+ }
+ }
+ else
+ break;
+ }
+ if (argi != argc-1)
+ return TCL_ERROR;
+ no = atoi(argv[argi]);
+ if (no >= spec->arg_no)
+ no = spec->arg_no - 1;
+ spec->ptr = spec->arg_start[no] + offset;
+ return TCL_OK;
+}
+
+static void execTcl (struct lexSpec *spec, struct regxCode *code)
+{
+ int i;
+ for (i = 0; i < spec->arg_no; i++)
+ {
+ char var_name[10], *var_buf;
+ int var_len, ch;
+
+ sprintf (var_name, "%d", i);
+ var_buf = f_win_get (spec, spec->arg_start[i], spec->arg_end[i],
+ &var_len);
+ if (var_buf)
+ {
+ ch = var_buf[var_len];
+ var_buf[var_len] = '\0';
+ Tcl_SetVar (spec->tcl_interp, var_name, var_buf, 0);
+ var_buf[var_len] = ch;
+ }
+ }
+ Tcl_Eval (spec->tcl_interp, code->str);
+}
+/* HAVE_TCL_H */
+#endif
+
+static void execCode (struct lexSpec *spec, struct regxCode *code)
{
const char *s = code->str;
int cmd_len, r;
- int returnCode = 1;
const char *cmd_str;
- r = execTok (spec, &s, arg_no, arg_start, arg_end, &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
while (r)
{
char *p, ptmp[64];
if (r == 1)
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
continue;
}
p = regxStrz (cmd_str, cmd_len, ptmp);
if (!strcmp (p, "begin"))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
{
logf (LOG_WARN, "missing keyword after 'begin'");
p = regxStrz (cmd_str, cmd_len, ptmp);
if (!strcmp (p, "record"))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
- if (*d1_level == 0)
+ if (spec->d1_level == 0)
{
static char absynName[64];
data1_absyn *absyn;
res->u.root.absyn = absyn;
res->root = res;
- d1_stack[*d1_level] = res;
- d1_stack[++(*d1_level)] = NULL;
+ spec->d1_stack[spec->d1_level] = res;
+ spec->d1_stack[++(spec->d1_level)] = NULL;
}
}
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else if (!strcmp (p, "element"))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
- tagBegin (spec, d1_stack, d1_level, cmd_str, cmd_len);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ tagBegin (spec, cmd_str, cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else if (!strcmp (p, "variant"))
{
const char *type_str = NULL;
int value_len;
const char *value_str = NULL;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
class_str = cmd_str;
class_len = cmd_len;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
type_str = cmd_str;
type_len = cmd_len;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
continue;
value_str = cmd_str;
value_len = cmd_len;
- variantBegin (spec, d1_stack, d1_level, class_str, class_len,
+ variantBegin (spec, class_str, class_len,
type_str, type_len, value_str, value_len);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else if (!strcmp (p, "context"))
{
if (r > 1)
{
struct lexContext *lc = spec->context;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
p = regxStrz (cmd_str, cmd_len, ptmp);
#if REGX_DEBUG
logf (LOG_DEBUG, "begin context %s", p);
logf (LOG_WARN, "unknown context %s", p);
}
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else
{
}
else if (!strcmp (p, "end"))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
{
logf (LOG_WARN, "missing keyword after 'end'");
p = regxStrz (cmd_str, cmd_len, ptmp);
if (!strcmp (p, "record"))
{
- *d1_level = 0;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ while (spec->d1_level)
+ {
+ tagDataRelease (spec);
+ (spec->d1_level)--;
+ }
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
#if REGX_DEBUG
logf (LOG_DEBUG, "end record");
#endif
- returnCode = 0;
+ spec->stop_flag = 1;
}
else if (!strcmp (p, "element"))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
-#if 0
- if (*d1_level == 1)
- {
- *d1_level = 0;
- returnCode = 0;
- }
-#endif
+ int min_level = 1;
+ while ((r = execTok (spec, &s, &cmd_str, &cmd_len)) == 3)
+ {
+ if (cmd_len==7 && !memcmp ("-record", cmd_str, cmd_len))
+ min_level = 0;
+ }
if (r > 2)
{
- tagEnd (spec, d1_stack, d1_level, cmd_str, cmd_len);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ tagEnd (spec, min_level, cmd_str, cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else
- tagEnd (spec, d1_stack, d1_level, NULL, 0);
+ tagEnd (spec, min_level, NULL, 0);
+ if (spec->d1_level == 0)
+ {
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "end element end records");
+#endif
+ spec->stop_flag = 1;
+ }
+
}
else if (!strcmp (p, "context"))
{
#endif
if (spec->context_stack_top)
(spec->context_stack_top)--;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else
logf (LOG_WARN, "bad keyword '%s' after end", p);
int element_len;
const char *element_str = NULL;
- while ((r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len)) == 3)
+ while ((r = execTok (spec, &s, &cmd_str, &cmd_len)) == 3)
{
if (cmd_len==5 && !memcmp ("-text", cmd_str, cmd_len))
textFlag = 1;
else if (cmd_len==8 && !memcmp ("-element", cmd_str, cmd_len))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &element_str, &element_len);
+ r = execTok (spec, &s, &element_str, &element_len);
if (r < 2)
break;
}
continue;
}
if (element_str)
- tagBegin (spec, d1_stack, d1_level, element_str, element_len);
+ tagBegin (spec, element_str, element_len);
do
{
- execData (spec, d1_stack, d1_level, cmd_str, cmd_len,
- textFlag);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ execData (spec, cmd_str, cmd_len,textFlag);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
} while (r > 1);
if (element_str)
- tagEnd (spec, d1_stack, d1_level, NULL, 0);
+ tagEnd (spec, 1, NULL, 0);
}
else if (!strcmp (p, "unread"))
{
int no, offset;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r==3 && cmd_len == 7 && !memcmp ("-offset", cmd_str, cmd_len))
{
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
if (r < 2)
{
logf (LOG_WARN, "missing number after -offset");
}
p = regxStrz (cmd_str, cmd_len, ptmp);
offset = atoi (p);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else
offset = 0;
else
{
no = *cmd_str - '0';
- if (no >= arg_no)
- no = arg_no - 1;
- *pptr = arg_start[no] + offset;
+ if (no >= spec->arg_no)
+ no = spec->arg_no - 1;
+ spec->ptr = spec->arg_start[no] + offset;
}
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else if (!strcmp (p, "context"))
{
if (r > 1)
{
struct lexContext *lc = spec->context;
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
p = regxStrz (cmd_str, cmd_len, ptmp);
while (lc && strcmp (p, lc->name))
logf (LOG_WARN, "unknown context %s", p);
}
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
}
else
{
logf (LOG_WARN, "unknown code command '%.*s'", cmd_len, cmd_str);
- r = execTok (spec, &s, arg_no, arg_start, arg_end,
- &cmd_str, &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
continue;
}
if (r > 1)
{
logf (LOG_WARN, "ignoring token %.*s", cmd_len, cmd_str);
do {
- r = execTok (spec, &s, arg_no, arg_start, arg_end, &cmd_str,
- &cmd_len);
+ r = execTok (spec, &s, &cmd_str, &cmd_len);
} while (r > 1);
}
}
- return returnCode;
}
-/*
- * execAction: Execute action specified by 'ap'. Returns 0 if
- * the pattern(s) associated by rule and code could be executed
- * ok; returns 1 if code couldn't be executed.
- */
static int execAction (struct lexSpec *spec, struct lexRuleAction *ap,
- data1_node **d1_stack, int *d1_level,
int start_ptr, int *pptr)
{
int sptr;
int arg_end[20];
int arg_no = 1;
+ if (!ap)
+ return 1;
arg_start[0] = start_ptr;
arg_end[0] = *pptr;
+ spec->arg_start = arg_start;
+ spec->arg_end = arg_end;
while (ap)
{
arg_no++;
break;
case REGX_CODE:
- if (!execCode (spec, arg_no, arg_start, arg_end, pptr,
- ap->u.code, d1_stack, d1_level))
- return 0;
+ spec->arg_no = arg_no;
+ spec->ptr = *pptr;
+#if HAVE_TCL_H
+ if (spec->tcl_interp)
+ execTcl(spec, ap->u.code);
+ else
+ execCode (spec, ap->u.code);
+#else
+ execCode (spec, ap->u.code);
+#endif
+ *pptr = spec->ptr;
+ if (spec->stop_flag)
+ return 0;
break;
case REGX_END:
arg_start[arg_no] = *pptr;
}
static int execRule (struct lexSpec *spec, struct lexContext *context,
- data1_node **d1_stack, int *d1_level,
int ruleNo, int start_ptr, int *pptr)
{
#if REGX_DEBUG
logf (LOG_DEBUG, "exec rule %d", ruleNo);
#endif
return execAction (spec, context->fastRule[ruleNo]->actionList,
- d1_stack, d1_level, start_ptr, pptr);
+ start_ptr, pptr);
}
-data1_node *lexNode (struct lexSpec *spec,
- data1_node **d1_stack, int *d1_level, int *ptr)
+data1_node *lexNode (struct lexSpec *spec, int *ptr)
{
struct lexContext *context = spec->context_stack[spec->context_stack_top];
struct DFA_state *state = context->dfa->states[0];
int size;
char *buf;
buf = f_win_get (spec, skip_ptr, start_ptr, &size);
- execDataP (spec, d1_stack, d1_level, buf, size, 0);
+ execDataP (spec, buf, size, 0);
}
/* restore pointer */
*ptr = last_ptr;
/* execute rule */
- if (!execRule (spec, context, d1_stack, d1_level,
- last_rule, start_ptr, ptr))
- break;
+ if (!execRule (spec, context, last_rule, start_ptr, ptr))
+ break;
/* restore skip pointer */
skip_ptr = *ptr;
last_rule = 0;
int size;
char *buf;
buf = f_win_get (spec, skip_ptr, *ptr, &size);
- execDataP (spec, d1_stack, d1_level, buf, size, 0);
+ execDataP (spec, buf, size, 0);
}
if (*ptr == F_WIN_EOF)
break;
int size;
char *buf;
buf = f_win_get (spec, skip_ptr, start_ptr, &size);
- execDataP (spec, d1_stack, d1_level, buf, size, 0);
+ execDataP (spec, buf, size, 0);
}
/* restore pointer */
*ptr = last_ptr;
- if (!execRule (spec, context, d1_stack, d1_level,
- last_rule, start_ptr, ptr))
+ if (!execRule (spec, context, last_rule, start_ptr, ptr))
{
if (spec->f_win_ef && *ptr != F_WIN_EOF)
{
const char *context_name)
{
struct lexContext *lt = spec->context;
- data1_node *d1_stack[512];
- int d1_level = 0;
int ptr = offset;
+ spec->stop_flag = 0;
+ spec->d1_level = 0;
spec->context_stack_top = 0;
while (lt)
{
return NULL;
}
spec->context_stack[spec->context_stack_top] = lt;
- d1_stack[d1_level] = NULL;
- if (lt->beginActionList)
- execAction (spec, lt->beginActionList, d1_stack, &d1_level, 0, &ptr);
- lexNode (spec, d1_stack, &d1_level, &ptr);
- if (lt->endActionList)
- execAction (spec, lt->endActionList, d1_stack, &d1_level, ptr, &ptr);
- return *d1_stack;
+ spec->d1_stack[spec->d1_level] = NULL;
+#if 1
+ if (!lt->initFlag)
+ {
+ lt->initFlag = 1;
+ execAction (spec, lt->initActionList, ptr, &ptr);
+ }
+#endif
+ execAction (spec, lt->beginActionList, ptr, &ptr);
+ lexNode (spec, &ptr);
+ while (spec->d1_level)
+ {
+ tagDataRelease (spec);
+ (spec->d1_level)--;
+ }
+ execAction (spec, lt->endActionList, ptr, &ptr);
+ return spec->d1_stack[0];
+}
+
+void grs_destroy(void *clientData)
+{
+ struct lexSpecs *specs = (struct lexSpecs *) clientData;
+ if (specs->spec)
+ {
+ lexSpecDestroy(&specs->spec);
+ }
+ xfree (specs);
+}
+
+void *grs_init(void)
+{
+ struct lexSpecs *specs = (struct lexSpecs *) xmalloc (sizeof(*specs));
+ specs->spec = 0;
+ return specs;
}
data1_node *grs_read_regx (struct grs_read_info *p)
{
int res;
+ struct lexSpecs *specs = (struct lexSpecs *) p->clientData;
+ struct lexSpec **curLexSpec = &specs->spec;
#if REGX_DEBUG
logf (LOG_DEBUG, "grs_read_regx");
#endif
- if (!curLexSpec || strcmp (curLexSpec->name, p->type))
+ if (!*curLexSpec || strcmp ((*curLexSpec)->name, p->type))
{
- if (curLexSpec)
- lexSpecDestroy (&curLexSpec);
- curLexSpec = lexSpecCreate (p->type);
- curLexSpec->dh = p->dh;
- res = readFileSpec (curLexSpec);
+ if (*curLexSpec)
+ lexSpecDestroy (curLexSpec);
+ *curLexSpec = lexSpecCreate (p->type, p->dh);
+ res = readFileSpec (*curLexSpec);
if (res)
{
- lexSpecDestroy (&curLexSpec);
+ lexSpecDestroy (curLexSpec);
return NULL;
}
}
+ (*curLexSpec)->dh = p->dh;
if (!p->offset)
{
- curLexSpec->f_win_start = 0;
- curLexSpec->f_win_end = 0;
- curLexSpec->f_win_rf = p->readf;
- curLexSpec->f_win_sf = p->seekf;
- curLexSpec->f_win_fh = p->fh;
- curLexSpec->f_win_ef = p->endf;
- curLexSpec->f_win_size = 500000;
+ (*curLexSpec)->f_win_start = 0;
+ (*curLexSpec)->f_win_end = 0;
+ (*curLexSpec)->f_win_rf = p->readf;
+ (*curLexSpec)->f_win_sf = p->seekf;
+ (*curLexSpec)->f_win_fh = p->fh;
+ (*curLexSpec)->f_win_ef = p->endf;
+ (*curLexSpec)->f_win_size = 500000;
}
- curLexSpec->m = p->mem;
- return lexRoot (curLexSpec, p->offset, "main");
+ (*curLexSpec)->m = p->mem;
+ return lexRoot (*curLexSpec, p->offset, "main");
}
+
+static struct recTypeGrs regx_type = {
+ "regx",
+ grs_init,
+ grs_destroy,
+ grs_read_regx
+};
+
+RecTypeGrs recTypeGrs_regx = ®x_type;
+
+#if HAVE_TCL_H
+data1_node *grs_read_tcl (struct grs_read_info *p)
+{
+ int res;
+ struct lexSpecs *specs = (struct lexSpecs *) p->clientData;
+ struct lexSpec **curLexSpec = &specs->spec;
+
+#if REGX_DEBUG
+ logf (LOG_DEBUG, "grs_read_tcl");
+#endif
+ if (!*curLexSpec || strcmp ((*curLexSpec)->name, p->type))
+ {
+ Tcl_Interp *tcl_interp;
+ if (*curLexSpec)
+ lexSpecDestroy (curLexSpec);
+ *curLexSpec = lexSpecCreate (p->type, p->dh);
+ tcl_interp = (*curLexSpec)->tcl_interp = Tcl_CreateInterp();
+ Tcl_CreateCommand (tcl_interp, "begin", cmd_tcl_begin, *curLexSpec, 0);
+ Tcl_CreateCommand (tcl_interp, "end", cmd_tcl_end, *curLexSpec, 0);
+ Tcl_CreateCommand (tcl_interp, "data", cmd_tcl_data, *curLexSpec, 0);
+ Tcl_CreateCommand (tcl_interp, "unread", cmd_tcl_unread,
+ *curLexSpec, 0);
+ res = readFileSpec (*curLexSpec);
+ if (res)
+ {
+ lexSpecDestroy (curLexSpec);
+ return NULL;
+ }
+ }
+ (*curLexSpec)->dh = p->dh;
+ if (!p->offset)
+ {
+ (*curLexSpec)->f_win_start = 0;
+ (*curLexSpec)->f_win_end = 0;
+ (*curLexSpec)->f_win_rf = p->readf;
+ (*curLexSpec)->f_win_sf = p->seekf;
+ (*curLexSpec)->f_win_fh = p->fh;
+ (*curLexSpec)->f_win_ef = p->endf;
+ (*curLexSpec)->f_win_size = 500000;
+ }
+ (*curLexSpec)->m = p->mem;
+ return lexRoot (*curLexSpec, p->offset, "main");
+}
+
+static struct recTypeGrs tcl_type = {
+ "tcl",
+ grs_init,
+ grs_destroy,
+ grs_read_tcl
+};
+
+RecTypeGrs recTypeGrs_tcl = &tcl_type;
+#endif
projects / idzebra-moved-to-github.git / blobdiff