1 /* $Id: d1_absyn.c,v 1.9 2003-06-12 18:20:24 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
5 This file is part of the Zebra server.
7 Zebra is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 2, or (at your option) any later
12 Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with Zebra; see the file LICENSE.zebra. If not, write to the
19 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
31 #include <zebra_xpath.h>
33 #define D1_MAX_NESTING 128
38 struct data1_systag *next;
41 struct data1_absyn_cache_info
45 data1_absyn_cache next;
48 struct data1_attset_cache_info
52 data1_attset_cache next;
55 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
57 data1_absyn_cache p = *data1_absyn_cache_get (dh);
61 if (!strcmp (name, p->name))
68 We need to destroy DFAs, in xp_element (xelm) definitions
72 void data1_absyn_destroy (data1_handle dh)
74 data1_absyn_cache p = *data1_absyn_cache_get (dh);
78 data1_absyn *abs = p->absyn;
81 data1_xpelement *xpe = abs->xp_elements;
83 logf (LOG_DEBUG,"Destroy xp element %s",xpe->xpath_expr);
84 if (xpe->dfa) { dfa_delete (&xpe->dfa); }
93 void data1_absyn_trav (data1_handle dh, void *handle,
94 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
96 data1_absyn_cache p = *data1_absyn_cache_get (dh);
100 (*fh)(dh, handle, p->absyn);
105 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
108 NMEM mem = data1_nmem_get (dh);
110 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
111 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
113 sprintf(fname, "%s.abs", name);
114 p->absyn = data1_read_absyn (dh, fname, 0);
115 p->name = nmem_strdup (mem, name);
121 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
125 if (!(absyn = data1_absyn_search (dh, name)))
126 absyn = data1_absyn_add (dh, name);
130 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
132 data1_attset_cache p = *data1_attset_cache_get (dh);
136 if (!strcmp (name, p->name))
143 data1_attset *data1_attset_search_id (data1_handle dh, int id)
145 data1_attset_cache p = *data1_attset_cache_get (dh);
149 if (id == p->attset->reference)
156 data1_attset *data1_attset_add (data1_handle dh, const char *name)
158 char fname[512], aname[512];
159 NMEM mem = data1_nmem_get (dh);
160 data1_attset *attset;
162 strcpy (aname, name);
163 sprintf(fname, "%s.att", name);
164 attset = data1_read_attset (dh, fname);
168 attset = data1_read_attset (dh, name);
169 if (attset && (cp = strrchr (aname, '.')))
173 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
176 data1_attset_cache p = (data1_attset_cache)
177 nmem_malloc (mem, sizeof(*p));
178 data1_attset_cache *pp = data1_attset_cache_get (dh);
180 attset->name = p->name = nmem_strdup (mem, aname);
188 data1_attset *data1_get_attset (data1_handle dh, const char *name)
190 data1_attset *attset;
192 if (!(attset = data1_attset_search_name (dh, name)))
193 attset = data1_attset_add (dh, name);
197 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
202 for (r = a->esetnames; r; r = r->next)
203 if (!data1_matchstr(r->name, name))
208 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
209 data1_element *parent,
214 /* It's now possible to have a data1 tree with no abstract syntax */
219 r = abs->main_elements;
221 r = parent->children;
223 for (; r; r = r->next)
227 for (n = r->tag->names; n; n = n->next)
228 if (!data1_matchstr(tagname, n->name))
234 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
239 /* It's now possible to have a data1 tree with no abstract syntax */
242 for (r = absyn->main_elements; r; r = r->next)
243 if (!data1_matchstr(r->name, name))
249 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
251 /* It's now possible to have a data1 tree with no abstract syntax */
255 for (; e; e = e->next)
260 fix_element_ref (dh, absyn, e->children);
264 data1_sub_elements *sub_e = absyn->sub_elements;
265 while (sub_e && strcmp (e->sub_name, sub_e->name))
268 e->children = sub_e->elements;
270 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
277 New function, a bit dummy now... I've seen it in zrpn.c... We should build
278 more clever regexps...
290 Now [] predicates are supported
296 const char * mk_xpath_regexp (data1_handle dh, char *expr)
305 int is_predicate = 0;
307 static char *stack[32];
308 static char res[1024];
311 if (*p != '/') { return (""); }
313 if (*p == '/') { abs=0; p++; }
317 while (*p && !strchr("/",*p)) {
320 stack[e] = (char *) nmem_malloc (data1_nmem_get (dh), i+1);
322 for (j=0; j< i; j++) {
327 else if (*pp == ']') {
342 e--; p = &res[0]; i=0;
343 sprintf (p, "^"); p++;
345 /* !!! res size is not checked !!! */
346 sprintf (p, "%s/",stack[e]);
347 p += strlen(stack[e]) + 1;
350 if (!abs) { sprintf (p, ".*"); p+=2; }
351 sprintf (p, "$"); p++;
352 r = nmem_strdup (data1_nmem_get (dh), res);
353 yaz_log(LOG_DEBUG,"Got regexp: %s",r);
359 added arg xpelement... when called from xelm context, it's 1, saying
360 that ! means xpath, not element name as attribute name...
364 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
365 char *p, const char *file, int lineno,
366 const char *element_name, data1_absyn *res,
369 data1_termlist **tp = *tpp;
372 char attname[512], structure[512];
376 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
380 "%s:%d: Syntax error in termlistspec '%s'",
385 *tp = (data1_termlist *)
386 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
391 strcpy(attname, element_name);
393 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
395 if ((!xpelement) || (*attname != '!')) {
397 "%s:%d: Couldn't find att '%s' in attset",
398 file, lineno, attname);
405 if (r == 2 && (source = strchr(structure, ':')))
406 *source++ = '\0'; /* cut off structure .. */
408 source = "data"; /* ok: default is leaf data */
409 (*tp)->source = (char *)
410 nmem_strdup (data1_nmem_get (dh), source);
412 if (r < 2) /* is the structure qualified? */
413 (*tp)->structure = "w";
415 (*tp)->structure = (char *)
416 nmem_strdup (data1_nmem_get (dh), structure);
419 while ((p = strchr(p, ',')) && *(++p));
424 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
425 const char *default_value)
427 struct data1_systag *p = absyn->systags;
428 for (; p; p = p->next)
429 if (!strcmp(p->name, tag))
431 return default_value;
434 #define l_isspace(c) ((c) == '\t' || (c) == ' ' || (c) == '\n' || (c) == '\r')
436 int read_absyn_line(FILE *f, int *lineno, char *line, int len,
437 char *argv[], int num)
443 while ((p = fgets(line, len, f)))
446 while (*p && l_isspace(*p))
454 for (argc = 0; *p ; argc++)
456 if (*p == '#') /* trailing comment */
459 while (*p && !(l_isspace(*p) && !quoted)) {
460 if (*p =='"') quoted = 1 - quoted;
461 if (*p =='[') quoted = 1;
462 if (*p ==']') quoted = 0;
468 while (*p && l_isspace(*p))
476 data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
479 data1_sub_elements *cur_elements = NULL;
480 data1_xpelement *cur_xpelement = NULL;
482 data1_absyn *res = 0;
484 data1_element **ppl[D1_MAX_NESTING];
485 data1_esetname **esetpp;
486 data1_maptab **maptabp;
487 data1_marctab **marcp;
488 data1_termlist *all = 0;
489 data1_attset_child **attset_childp;
490 data1_tagset **tagset_childp;
491 struct data1_systag **systagsp;
495 char *argv[50], line[512];
497 if (!(f = data1_path_fopen(dh, file, "r")))
499 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
504 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
506 res->reference = VAL_NONE;
509 res->enable_xpath_indexing = (f ? 0 : 1);
511 systagsp = &res->systags;
512 tagset_childp = &res->tagset;
514 res->attset = data1_empty_attset (dh);
515 attset_childp = &res->attset->children;
519 esetpp = &res->esetnames;
521 maptabp = &res->maptabs;
524 res->sub_elements = NULL;
525 res->main_elements = NULL;
526 res->xp_elements = NULL;
528 while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
531 if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
533 data1_element *new_element;
535 char *p, *sub_p, *path, *name, *termlists;
541 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
550 cur_elements = (data1_sub_elements *)
551 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
552 cur_elements->next = res->sub_elements;
553 cur_elements->elements = NULL;
554 cur_elements->name = "main";
555 res->sub_elements = cur_elements;
558 ppl[level] = &cur_elements->elements;
565 if ((e = strchr(p, '/')))
572 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
577 new_element = *ppl[level-1] = (data1_element *)
578 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
579 new_element->next = new_element->children = 0;
580 new_element->tag = 0;
581 new_element->termlists = 0;
582 new_element->sub_name = 0;
584 tp = &new_element->termlists;
585 ppl[level-1] = &new_element->next;
586 ppl[level] = &new_element->children;
588 /* consider subtree (if any) ... */
589 if ((sub_p = strchr (p, ':')) && sub_p[1])
592 new_element->sub_name =
593 nmem_strdup (data1_nmem_get(dh), sub_p);
595 /* well-defined tag */
596 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
600 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
604 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
607 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
617 new_element->tag = (data1_tag *)
618 nmem_malloc(data1_nmem_get (dh),
619 sizeof(*new_element->tag));
620 nt->which = DATA1T_string;
621 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
622 nt->names = (data1_name *)
623 nmem_malloc(data1_nmem_get(dh),
624 sizeof(*new_element->tag->names));
625 nt->names->name = nt->value.string;
627 nt->kind = DATA1K_string;
633 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
637 /* parse termList definitions */
641 assert (res->attset);
643 if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
648 *tp = all; /* append any ALL entries to the list */
650 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
653 New code to support xelm directive
654 for each xelm a dfa is built. xelms are stored in res->xp_elements
656 maybe we should use a simple sscanf instead of dfa?
660 Now [] predicates are supported. regexps and xpath structure is
661 a bit redundant, however it's comfortable later...
666 else if (!strcmp(cmd, "xelm")) {
669 char *p, *xpath_expr, *termlists;
671 struct DFA *dfa = dfa = dfa_init();
676 yaz_log(LOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
679 xpath_expr = argv[1];
681 regexp = mk_xpath_regexp(dh, xpath_expr);
682 i = dfa_parse (dfa, ®exp);
684 yaz_log(LOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
691 cur_xpelement = (data1_xpelement *)
692 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
693 res->xp_elements = cur_xpelement;
695 cur_xpelement->next = (data1_xpelement *)
696 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
697 cur_xpelement = cur_xpelement->next;
699 cur_xpelement->next = NULL;
700 cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh),
704 cur_xpelement->dfa = dfa;
707 cur_xpelement->xpath_len =
708 zebra_parse_xpath_str(xpath_expr,
709 cur_xpelement->xpath, XPATH_STEP_COUNT,
713 dump_xp_steps(cur_xpelement->xpath,cur_xpelement->xpath_len);
716 cur_xpelement->termlists = 0;
717 tp = &cur_xpelement->termlists;
719 /* parse termList definitions */
723 assert (res->attset);
725 if (parse_termlists (dh, &tp, p, file, lineno,
731 *tp = all; /* append any ALL entries to the list */
734 else if (!strcmp(cmd, "section"))
740 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
746 cur_elements = (data1_sub_elements *)
747 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
748 cur_elements->next = res->sub_elements;
749 cur_elements->elements = NULL;
750 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
751 res->sub_elements = cur_elements;
754 ppl[level] = &cur_elements->elements;
756 else if (!strcmp(cmd, "xpath"))
760 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
764 if (!strcmp(argv[1], "enable"))
765 res->enable_xpath_indexing = 1;
766 else if (!strcmp (argv[1], "disable"))
767 res->enable_xpath_indexing = 0;
770 yaz_log(LOG_WARN, "%s:%d: Expecting disable/enable "
771 "after 'xpath' directive", file, lineno);
774 else if (!strcmp(cmd, "all"))
776 data1_termlist **tp = &all;
779 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
785 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
789 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
795 else if (!strcmp(cmd, "name"))
799 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
803 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
805 else if (!strcmp(cmd, "reference"))
811 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
816 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
818 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
823 else if (!strcmp(cmd, "attset"))
826 data1_attset *attset;
830 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
835 if (!(attset = data1_get_attset (dh, name)))
837 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
841 *attset_childp = (data1_attset_child *)
842 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
843 (*attset_childp)->child = attset;
844 (*attset_childp)->next = 0;
845 attset_childp = &(*attset_childp)->next;
847 else if (!strcmp(cmd, "tagset"))
853 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
859 type = atoi(argv[2]);
860 *tagset_childp = data1_read_tagset (dh, name, type);
861 if (!(*tagset_childp))
863 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
867 tagset_childp = &(*tagset_childp)->next;
869 else if (!strcmp(cmd, "varset"))
875 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
880 if (!(res->varset = data1_read_varset (dh, name)))
882 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
887 else if (!strcmp(cmd, "esetname"))
893 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
900 *esetpp = (data1_esetname *)
901 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
902 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
906 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
908 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
909 file, lineno, fname);
912 esetpp = &(*esetpp)->next;
914 else if (!strcmp(cmd, "maptab"))
920 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
925 if (!(*maptabp = data1_read_maptab (dh, name)))
927 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
931 maptabp = &(*maptabp)->next;
933 else if (!strcmp(cmd, "marc"))
939 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
944 if (!(*marcp = data1_read_marctab (dh, name)))
946 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
950 marcp = &(*marcp)->next;
952 else if (!strcmp(cmd, "encoding"))
956 yaz_log(LOG_WARN, "%s:%d: Bad # or args for encoding",
960 res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
962 else if (!strcmp(cmd, "systag"))
966 yaz_log(LOG_WARN, "%s:%d: Bad # or args for systag",
970 *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
972 (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
973 (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
974 systagsp = &(*systagsp)->next;
978 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file,
986 for (cur_elements = res->sub_elements; cur_elements;
987 cur_elements = cur_elements->next)
989 if (!strcmp (cur_elements->name, "main"))
990 res->main_elements = cur_elements->elements;
991 fix_element_ref (dh, res, cur_elements->elements);
994 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);