1 /* $Id: d1_absyn.c,v 1.12 2004-09-28 10:15:02 adam Exp $
2 Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
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
30 #include <idzebra/data1.h>
31 #include <zebra_xpath.h>
34 #define D1_MAX_NESTING 128
36 #if PRIVATE_DATA1_ABSYN
42 struct data1_systag *next;
45 struct data1_absyn_cache_info
49 data1_absyn_cache next;
52 struct data1_attset_cache_info
56 data1_attset_cache next;
59 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
61 data1_absyn_cache p = *data1_absyn_cache_get (dh);
65 if (!yaz_matchstr (name, p->name))
72 We need to destroy DFAs, in xp_element (xelm) definitions
76 void data1_absyn_destroy (data1_handle dh)
78 data1_absyn_cache p = *data1_absyn_cache_get (dh);
82 data1_absyn *abs = p->absyn;
85 data1_xpelement *xpe = abs->xp_elements;
87 logf (LOG_DEBUG,"Destroy xp element %s",xpe->xpath_expr);
88 if (xpe->dfa) { dfa_delete (&xpe->dfa); }
97 void data1_absyn_trav (data1_handle dh, void *handle,
98 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
100 data1_absyn_cache p = *data1_absyn_cache_get (dh);
104 (*fh)(dh, handle, p->absyn);
109 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
112 NMEM mem = data1_nmem_get (dh);
114 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
115 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
117 sprintf(fname, "%s.abs", name);
118 p->absyn = data1_read_absyn (dh, fname, 0);
119 p->name = nmem_strdup (mem, name);
125 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
129 if (!(absyn = data1_absyn_search (dh, name)))
130 absyn = data1_absyn_add (dh, name);
134 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
136 data1_attset_cache p = *data1_attset_cache_get (dh);
140 if (!yaz_matchstr (name, p->name))
147 data1_attset *data1_attset_search_id (data1_handle dh, int id)
149 data1_attset_cache p = *data1_attset_cache_get (dh);
153 if (id == p->attset->reference)
160 data1_attset *data1_attset_add (data1_handle dh, const char *name)
162 char fname[512], aname[512];
163 NMEM mem = data1_nmem_get (dh);
164 data1_attset *attset;
166 strcpy (aname, name);
167 sprintf(fname, "%s.att", name);
168 attset = data1_read_attset (dh, fname);
172 attset = data1_read_attset (dh, name);
173 if (attset && (cp = strrchr (aname, '.')))
177 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
180 data1_attset_cache p = (data1_attset_cache)
181 nmem_malloc (mem, sizeof(*p));
182 data1_attset_cache *pp = data1_attset_cache_get (dh);
184 attset->name = p->name = nmem_strdup (mem, aname);
192 data1_attset *data1_get_attset (data1_handle dh, const char *name)
194 data1_attset *attset;
196 if (!(attset = data1_attset_search_name (dh, name)))
197 attset = data1_attset_add (dh, name);
201 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
206 for (r = a->esetnames; r; r = r->next)
207 if (!data1_matchstr(r->name, name))
212 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
213 data1_element *parent,
218 /* It's now possible to have a data1 tree with no abstract syntax */
223 r = abs->main_elements;
225 r = parent->children;
227 for (; r; r = r->next)
231 for (n = r->tag->names; n; n = n->next)
232 if (!data1_matchstr(tagname, n->name))
238 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
243 /* It's now possible to have a data1 tree with no abstract syntax */
246 for (r = absyn->main_elements; r; r = r->next)
247 if (!data1_matchstr(r->name, name))
253 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
255 /* It's now possible to have a data1 tree with no abstract syntax */
259 for (; e; e = e->next)
264 fix_element_ref (dh, absyn, e->children);
268 data1_sub_elements *sub_e = absyn->sub_elements;
269 while (sub_e && strcmp (e->sub_name, sub_e->name))
272 e->children = sub_e->elements;
274 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
281 New function, a bit dummy now... I've seen it in zrpn.c... We should build
282 more clever regexps...
294 Now [] predicates are supported
300 const char * mk_xpath_regexp (data1_handle dh, char *expr)
309 int is_predicate = 0;
311 static char *stack[32];
312 static char res[1024];
315 if (*p != '/') { return (""); }
317 if (*p == '/') { abs=0; p++; }
321 while (*p && !strchr("/",*p)) {
324 stack[e] = (char *) nmem_malloc (data1_nmem_get (dh), i+1);
326 for (j=0; j< i; j++) {
331 else if (*pp == ']') {
346 e--; p = &res[0]; i=0;
347 sprintf (p, "^"); p++;
349 /* !!! res size is not checked !!! */
350 sprintf (p, "%s/",stack[e]);
351 p += strlen(stack[e]) + 1;
354 if (!abs) { sprintf (p, ".*"); p+=2; }
355 sprintf (p, "$"); p++;
356 r = nmem_strdup (data1_nmem_get (dh), res);
357 yaz_log(LOG_DEBUG,"Got regexp: %s",r);
363 added arg xpelement... when called from xelm context, it's 1, saying
364 that ! means xpath, not element name as attribute name...
368 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
369 char *cp, const char *file, int lineno,
370 const char *element_name, data1_absyn *res,
373 data1_termlist **tp = *tpp;
376 char attname[512], structure[512];
381 for (i = 0; cp[i] && i<sizeof(attname)-1; i++)
382 if (strchr(":,", cp[i]))
390 "%s:%d: Syntax error in termlistspec '%s'",
400 for (i = 0; cp[i] && i<sizeof(structure)-1; i++)
401 if (level == 0 && strchr(",", cp[i]))
405 structure[i] = cp[i];
408 else if (cp[i] == ')')
418 *tp = (data1_termlist *)
419 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
424 strcpy(attname, element_name);
426 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
428 if ((!xpelement) || (*attname != '!')) {
430 "%s:%d: Couldn't find att '%s' in attset",
431 file, lineno, attname);
438 if (r == 2 && (source = strchr(structure, ':')))
439 *source++ = '\0'; /* cut off structure .. */
441 source = "data"; /* ok: default is leaf data */
442 (*tp)->source = (char *)
443 nmem_strdup (data1_nmem_get (dh), source);
445 if (r < 2) /* is the structure qualified? */
446 (*tp)->structure = "w";
448 (*tp)->structure = (char *)
449 nmem_strdup (data1_nmem_get (dh), structure);
457 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
458 const char *default_value)
460 struct data1_systag *p = absyn->systags;
461 for (; p; p = p->next)
462 if (!strcmp(p->name, tag))
464 return default_value;
467 #define l_isspace(c) ((c) == '\t' || (c) == ' ' || (c) == '\n' || (c) == '\r')
469 int read_absyn_line(FILE *f, int *lineno, char *line, int len,
470 char *argv[], int num)
476 while ((p = fgets(line, len, f)))
479 while (*p && l_isspace(*p))
487 for (argc = 0; *p ; argc++)
489 if (*p == '#') /* trailing comment */
492 while (*p && !(l_isspace(*p) && !quoted)) {
493 if (*p =='"') quoted = 1 - quoted;
494 if (*p =='[') quoted = 1;
495 if (*p ==']') quoted = 0;
501 while (*p && l_isspace(*p))
508 data1_marctab *data1_absyn_getmarctab(data1_handle dh, data1_absyn *absyn)
513 YAZ_EXPORT data1_element *data1_absyn_getelements(data1_handle dh,
516 return absyn->main_elements;
519 data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
522 data1_sub_elements *cur_elements = NULL;
523 data1_xpelement *cur_xpelement = NULL;
525 data1_absyn *res = 0;
527 data1_element **ppl[D1_MAX_NESTING];
528 data1_esetname **esetpp;
529 data1_maptab **maptabp;
530 data1_marctab **marcp;
531 data1_termlist *all = 0;
532 data1_attset_child **attset_childp;
533 data1_tagset **tagset_childp;
534 struct data1_systag **systagsp;
538 char *argv[50], line[512];
540 if (!(f = data1_path_fopen(dh, file, "r")))
542 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
547 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
549 res->reference = VAL_NONE;
552 res->enable_xpath_indexing = (f ? 0 : 1);
554 systagsp = &res->systags;
555 tagset_childp = &res->tagset;
557 res->attset = data1_empty_attset (dh);
558 attset_childp = &res->attset->children;
562 esetpp = &res->esetnames;
564 maptabp = &res->maptabs;
567 res->sub_elements = NULL;
568 res->main_elements = NULL;
569 res->xp_elements = NULL;
571 while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
574 if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
576 data1_element *new_element;
578 char *p, *sub_p, *path, *name, *termlists;
584 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
593 cur_elements = (data1_sub_elements *)
594 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
595 cur_elements->next = res->sub_elements;
596 cur_elements->elements = NULL;
597 cur_elements->name = "main";
598 res->sub_elements = cur_elements;
601 ppl[level] = &cur_elements->elements;
608 if ((e = strchr(p, '/')))
615 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
620 new_element = *ppl[level-1] = (data1_element *)
621 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
622 new_element->next = new_element->children = 0;
623 new_element->tag = 0;
624 new_element->termlists = 0;
625 new_element->sub_name = 0;
627 tp = &new_element->termlists;
628 ppl[level-1] = &new_element->next;
629 ppl[level] = &new_element->children;
631 /* consider subtree (if any) ... */
632 if ((sub_p = strchr (p, ':')) && sub_p[1])
635 new_element->sub_name =
636 nmem_strdup (data1_nmem_get(dh), sub_p);
638 /* well-defined tag */
639 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
643 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
647 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
650 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
660 new_element->tag = (data1_tag *)
661 nmem_malloc(data1_nmem_get (dh),
662 sizeof(*new_element->tag));
663 nt->which = DATA1T_string;
664 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
665 nt->names = (data1_name *)
666 nmem_malloc(data1_nmem_get(dh),
667 sizeof(*new_element->tag->names));
668 nt->names->name = nt->value.string;
670 nt->kind = DATA1K_string;
676 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
680 /* parse termList definitions */
684 assert (res->attset);
686 if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
691 *tp = all; /* append any ALL entries to the list */
693 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
696 New code to support xelm directive
697 for each xelm a dfa is built. xelms are stored in res->xp_elements
699 maybe we should use a simple sscanf instead of dfa?
703 Now [] predicates are supported. regexps and xpath structure is
704 a bit redundant, however it's comfortable later...
709 else if (!strcmp(cmd, "xelm")) {
712 char *p, *xpath_expr, *termlists;
714 struct DFA *dfa = dfa = dfa_init();
719 yaz_log(LOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
722 xpath_expr = argv[1];
724 regexp = mk_xpath_regexp(dh, xpath_expr);
725 i = dfa_parse (dfa, ®exp);
727 yaz_log(LOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
734 cur_xpelement = (data1_xpelement *)
735 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
736 res->xp_elements = cur_xpelement;
738 cur_xpelement->next = (data1_xpelement *)
739 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
740 cur_xpelement = cur_xpelement->next;
742 cur_xpelement->next = NULL;
743 cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh),
747 cur_xpelement->dfa = dfa;
750 cur_xpelement->xpath_len =
751 zebra_parse_xpath_str(xpath_expr,
752 cur_xpelement->xpath, XPATH_STEP_COUNT,
756 dump_xp_steps(cur_xpelement->xpath,cur_xpelement->xpath_len);
759 cur_xpelement->termlists = 0;
760 tp = &cur_xpelement->termlists;
762 /* parse termList definitions */
766 assert (res->attset);
768 if (parse_termlists (dh, &tp, p, file, lineno,
774 *tp = all; /* append any ALL entries to the list */
777 else if (!strcmp(cmd, "section"))
783 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
789 cur_elements = (data1_sub_elements *)
790 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
791 cur_elements->next = res->sub_elements;
792 cur_elements->elements = NULL;
793 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
794 res->sub_elements = cur_elements;
797 ppl[level] = &cur_elements->elements;
799 else if (!strcmp(cmd, "xpath"))
803 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
807 if (!strcmp(argv[1], "enable"))
808 res->enable_xpath_indexing = 1;
809 else if (!strcmp (argv[1], "disable"))
810 res->enable_xpath_indexing = 0;
813 yaz_log(LOG_WARN, "%s:%d: Expecting disable/enable "
814 "after 'xpath' directive", file, lineno);
817 else if (!strcmp(cmd, "all"))
819 data1_termlist **tp = &all;
822 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
828 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
832 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
838 else if (!strcmp(cmd, "name"))
842 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
846 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
848 else if (!strcmp(cmd, "reference"))
854 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
859 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
861 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
866 else if (!strcmp(cmd, "attset"))
869 data1_attset *attset;
873 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
878 if (!(attset = data1_get_attset (dh, name)))
880 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
884 *attset_childp = (data1_attset_child *)
885 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
886 (*attset_childp)->child = attset;
887 (*attset_childp)->next = 0;
888 attset_childp = &(*attset_childp)->next;
890 else if (!strcmp(cmd, "tagset"))
896 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
902 type = atoi(argv[2]);
903 *tagset_childp = data1_read_tagset (dh, name, type);
904 if (!(*tagset_childp))
906 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
910 tagset_childp = &(*tagset_childp)->next;
912 else if (!strcmp(cmd, "varset"))
918 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
923 if (!(res->varset = data1_read_varset (dh, name)))
925 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
930 else if (!strcmp(cmd, "esetname"))
936 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
943 *esetpp = (data1_esetname *)
944 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
945 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
949 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
951 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
952 file, lineno, fname);
955 esetpp = &(*esetpp)->next;
957 else if (!strcmp(cmd, "maptab"))
963 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
968 if (!(*maptabp = data1_read_maptab (dh, name)))
970 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
974 maptabp = &(*maptabp)->next;
976 else if (!strcmp(cmd, "marc"))
982 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
987 if (!(*marcp = data1_read_marctab (dh, name)))
989 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
993 marcp = &(*marcp)->next;
995 else if (!strcmp(cmd, "encoding"))
999 yaz_log(LOG_WARN, "%s:%d: Bad # or args for encoding",
1003 res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
1005 else if (!strcmp(cmd, "systag"))
1009 yaz_log(LOG_WARN, "%s:%d: Bad # or args for systag",
1013 *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
1015 (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
1016 (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
1017 systagsp = &(*systagsp)->next;
1021 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file,
1029 for (cur_elements = res->sub_elements; cur_elements;
1030 cur_elements = cur_elements->next)
1032 if (!strcmp (cur_elements->name, "main"))
1033 res->main_elements = cur_elements->elements;
1034 fix_element_ref (dh, res, cur_elements->elements);
1037 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);