1 /* $Id: d1_absyn.c,v 1.6 2003-02-04 12:06:46 pop 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;
79 data1_xpelement *xpe = abs->xp_elements;
81 logf (LOG_DEBUG,"Destroy xp element %s",xpe->xpath_expr);
82 if (xpe->dfa) { dfa_delete (&xpe->dfa); }
90 void data1_absyn_trav (data1_handle dh, void *handle,
91 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
93 data1_absyn_cache p = *data1_absyn_cache_get (dh);
97 (*fh)(dh, handle, p->absyn);
102 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
105 NMEM mem = data1_nmem_get (dh);
107 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
108 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
110 sprintf(fname, "%s.abs", name);
111 p->absyn = data1_read_absyn (dh, fname, 0);
112 p->name = nmem_strdup (mem, name);
118 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
122 if (!(absyn = data1_absyn_search (dh, name)))
123 absyn = data1_absyn_add (dh, name);
127 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
129 data1_attset_cache p = *data1_attset_cache_get (dh);
133 if (!strcmp (name, p->name))
140 data1_attset *data1_attset_search_id (data1_handle dh, int id)
142 data1_attset_cache p = *data1_attset_cache_get (dh);
146 if (id == p->attset->reference)
153 data1_attset *data1_attset_add (data1_handle dh, const char *name)
155 char fname[512], aname[512];
156 NMEM mem = data1_nmem_get (dh);
157 data1_attset *attset;
159 strcpy (aname, name);
160 sprintf(fname, "%s.att", name);
161 attset = data1_read_attset (dh, fname);
165 attset = data1_read_attset (dh, name);
166 if (attset && (cp = strrchr (aname, '.')))
170 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
173 data1_attset_cache p = (data1_attset_cache)
174 nmem_malloc (mem, sizeof(*p));
175 data1_attset_cache *pp = data1_attset_cache_get (dh);
177 attset->name = p->name = nmem_strdup (mem, aname);
185 data1_attset *data1_get_attset (data1_handle dh, const char *name)
187 data1_attset *attset;
189 if (!(attset = data1_attset_search_name (dh, name)))
190 attset = data1_attset_add (dh, name);
194 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
199 for (r = a->esetnames; r; r = r->next)
200 if (!data1_matchstr(r->name, name))
205 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
206 data1_element *parent,
211 /* It's now possible to have a data1 tree with no abstract syntax */
216 r = abs->main_elements;
218 r = parent->children;
220 for (; r; r = r->next)
224 for (n = r->tag->names; n; n = n->next)
225 if (!data1_matchstr(tagname, n->name))
231 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
236 /* It's now possible to have a data1 tree with no abstract syntax */
239 for (r = absyn->main_elements; r; r = r->next)
240 if (!data1_matchstr(r->name, name))
246 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
248 /* It's now possible to have a data1 tree with no abstract syntax */
252 for (; e; e = e->next)
257 fix_element_ref (dh, absyn, e->children);
261 data1_sub_elements *sub_e = absyn->sub_elements;
262 while (sub_e && strcmp (e->sub_name, sub_e->name))
265 e->children = sub_e->elements;
267 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
274 New function, a bit dummy now... I've seen it in zrpn.c... We should build
275 more clever regexps...
287 Now [] predicates are supported
293 const char * mk_xpath_regexp (data1_handle dh, char *expr)
302 int is_predicate = 0;
304 static char *stack[32];
305 static char res[1024];
308 if (*p != '/') { return (""); }
310 if (*p == '/') { abs=0; p++; }
314 while (*p && !strchr("/",*p)) {
317 stack[e] = (char *) nmem_malloc (data1_nmem_get (dh), i+1);
319 for (j=0; j< i; j++) {
324 else if (*pp == ']') {
339 e--; p = &res[0]; i=0;
340 sprintf (p, "^"); p++;
342 /* !!! res size is not checked !!! */
343 sprintf (p, "%s/",stack[e]);
344 p += strlen(stack[e]) + 1;
347 if (!abs) { sprintf (p, ".*"); p+=2; }
348 sprintf (p, "$"); p++;
349 r = nmem_strdup (data1_nmem_get (dh), res);
350 yaz_log(LOG_DEBUG,"Got regexp: %s",r);
356 added arg xpelement... when called from xelm context, it's 1, saying
357 that ! means xpath, not element name as attribute name...
361 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
362 char *p, const char *file, int lineno,
363 const char *element_name, data1_absyn *res,
366 data1_termlist **tp = *tpp;
369 char attname[512], structure[512];
373 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
377 "%s:%d: Syntax error in termlistspec '%s'",
382 *tp = (data1_termlist *)
383 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
388 strcpy(attname, element_name);
390 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
392 if ((!xpelement) || (*attname != '!')) {
394 "%s:%d: Couldn't find att '%s' in attset",
395 file, lineno, attname);
402 if (r == 2 && (source = strchr(structure, ':')))
403 *source++ = '\0'; /* cut off structure .. */
405 source = "data"; /* ok: default is leaf data */
406 (*tp)->source = (char *)
407 nmem_strdup (data1_nmem_get (dh), source);
409 if (r < 2) /* is the structure qualified? */
410 (*tp)->structure = "w";
412 (*tp)->structure = (char *)
413 nmem_strdup (data1_nmem_get (dh), structure);
416 while ((p = strchr(p, ',')) && *(++p));
421 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
422 const char *default_value)
424 struct data1_systag *p = absyn->systags;
425 for (; p; p = p->next)
426 if (!strcmp(p->name, tag))
428 return default_value;
431 #define l_isspace(c) ((c) == '\t' || (c) == ' ' || (c) == '\n' || (c) == '\r')
433 int read_absyn_line(FILE *f, int *lineno, char *line, int len,
434 char *argv[], int num)
440 while ((p = fgets(line, len, f)))
443 while (*p && l_isspace(*p))
451 for (argc = 0; *p ; argc++)
453 if (*p == '#') /* trailing comment */
456 while (*p && !(l_isspace(*p) && !quoted)) {
457 if (*p =='"') quoted = 1 - quoted;
458 if (*p =='[') quoted = 1;
459 if (*p ==']') quoted = 0;
465 while (*p && l_isspace(*p))
473 data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
476 data1_sub_elements *cur_elements = NULL;
477 data1_xpelement *cur_xpelement = NULL;
479 data1_absyn *res = 0;
481 data1_element **ppl[D1_MAX_NESTING];
482 data1_esetname **esetpp;
483 data1_maptab **maptabp;
484 data1_marctab **marcp;
485 data1_termlist *all = 0;
486 data1_attset_child **attset_childp;
487 data1_tagset **tagset_childp;
488 struct data1_systag **systagsp;
492 char *argv[50], line[512];
494 if (!(f = data1_path_fopen(dh, file, "r")))
496 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
501 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
503 res->reference = VAL_NONE;
506 res->enable_xpath_indexing = (f ? 0 : 1);
508 systagsp = &res->systags;
509 tagset_childp = &res->tagset;
511 res->attset = data1_empty_attset (dh);
512 attset_childp = &res->attset->children;
516 esetpp = &res->esetnames;
518 maptabp = &res->maptabs;
521 res->sub_elements = NULL;
522 res->main_elements = NULL;
523 res->xp_elements = NULL;
525 while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
528 if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
530 data1_element *new_element;
532 char *p, *sub_p, *path, *name, *termlists;
538 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
547 cur_elements = (data1_sub_elements *)
548 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
549 cur_elements->next = res->sub_elements;
550 cur_elements->elements = NULL;
551 cur_elements->name = "main";
552 res->sub_elements = cur_elements;
555 ppl[level] = &cur_elements->elements;
562 if ((e = strchr(p, '/')))
569 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
574 new_element = *ppl[level-1] = (data1_element *)
575 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
576 new_element->next = new_element->children = 0;
577 new_element->tag = 0;
578 new_element->termlists = 0;
579 new_element->sub_name = 0;
581 tp = &new_element->termlists;
582 ppl[level-1] = &new_element->next;
583 ppl[level] = &new_element->children;
585 /* consider subtree (if any) ... */
586 if ((sub_p = strchr (p, ':')) && sub_p[1])
589 new_element->sub_name =
590 nmem_strdup (data1_nmem_get(dh), sub_p);
592 /* well-defined tag */
593 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
597 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
601 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
604 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
614 new_element->tag = (data1_tag *)
615 nmem_malloc(data1_nmem_get (dh),
616 sizeof(*new_element->tag));
617 nt->which = DATA1T_string;
618 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
619 nt->names = (data1_name *)
620 nmem_malloc(data1_nmem_get(dh),
621 sizeof(*new_element->tag->names));
622 nt->names->name = nt->value.string;
624 nt->kind = DATA1K_string;
630 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
634 /* parse termList definitions */
638 assert (res->attset);
640 if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
645 *tp = all; /* append any ALL entries to the list */
647 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
650 New code to support xelm directive
651 for each xelm a dfa is built. xelms are stored in res->xp_elements
653 maybe we should use a simple sscanf instead of dfa?
657 Now [] predicates are supported. regexps and xpath structure is
658 a bit redundant, however it's comfortable later...
663 else if (!strcmp(cmd, "xelm")) {
666 char *p, *xpath_expr, *termlists;
669 struct DFA *dfa = dfa = dfa_init();
674 yaz_log(LOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
677 xpath_expr = argv[1];
679 regexp = mk_xpath_regexp(dh, xpath_expr);
680 i = dfa_parse (dfa, ®exp);
682 yaz_log(LOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
689 cur_xpelement = (data1_xpelement *)
690 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
691 res->xp_elements = cur_xpelement;
693 cur_xpelement->next = (data1_xpelement *)
694 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
695 cur_xpelement = cur_xpelement->next;
697 cur_xpelement->next = NULL;
698 cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh),
702 cur_xpelement->dfa = dfa;
705 cur_xpelement->xpath_len = parse_xpath_str(xpath_expr,
706 cur_xpelement->xpath,
710 dump_xp_steps(cur_xpelement->xpath,cur_xpelement->xpath_len);
713 cur_xpelement->termlists = 0;
714 tp = &cur_xpelement->termlists;
716 /* parse termList definitions */
720 assert (res->attset);
722 if (parse_termlists (dh, &tp, p, file, lineno,
728 *tp = all; /* append any ALL entries to the list */
731 else if (!strcmp(cmd, "section"))
737 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
743 cur_elements = (data1_sub_elements *)
744 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
745 cur_elements->next = res->sub_elements;
746 cur_elements->elements = NULL;
747 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
748 res->sub_elements = cur_elements;
751 ppl[level] = &cur_elements->elements;
753 else if (!strcmp(cmd, "xpath"))
757 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
761 if (!strcmp(argv[1], "enable"))
762 res->enable_xpath_indexing = 1;
763 else if (!strcmp (argv[1], "disable"))
764 res->enable_xpath_indexing = 0;
767 yaz_log(LOG_WARN, "%s:%d: Expecting disable/enable "
768 "after 'xpath' directive", file, lineno);
771 else if (!strcmp(cmd, "all"))
773 data1_termlist **tp = &all;
776 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
782 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
786 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
792 else if (!strcmp(cmd, "name"))
796 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
800 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
802 else if (!strcmp(cmd, "reference"))
808 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
813 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
815 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
820 else if (!strcmp(cmd, "attset"))
823 data1_attset *attset;
827 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
832 if (!(attset = data1_get_attset (dh, name)))
834 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
838 *attset_childp = (data1_attset_child *)
839 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
840 (*attset_childp)->child = attset;
841 (*attset_childp)->next = 0;
842 attset_childp = &(*attset_childp)->next;
844 else if (!strcmp(cmd, "tagset"))
850 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
856 type = atoi(argv[2]);
857 *tagset_childp = data1_read_tagset (dh, name, type);
858 if (!(*tagset_childp))
860 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
864 tagset_childp = &(*tagset_childp)->next;
866 else if (!strcmp(cmd, "varset"))
872 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
877 if (!(res->varset = data1_read_varset (dh, name)))
879 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
884 else if (!strcmp(cmd, "esetname"))
890 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
897 *esetpp = (data1_esetname *)
898 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
899 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
903 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
905 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
906 file, lineno, fname);
909 esetpp = &(*esetpp)->next;
911 else if (!strcmp(cmd, "maptab"))
917 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
922 if (!(*maptabp = data1_read_maptab (dh, name)))
924 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
928 maptabp = &(*maptabp)->next;
930 else if (!strcmp(cmd, "marc"))
936 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
941 if (!(*marcp = data1_read_marctab (dh, name)))
943 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
947 marcp = &(*marcp)->next;
949 else if (!strcmp(cmd, "encoding"))
953 yaz_log(LOG_WARN, "%s:%d: Bad # or args for encoding",
957 res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
959 else if (!strcmp(cmd, "systag"))
961 struct data1_systag *st;
964 yaz_log(LOG_WARN, "%s:%d: Bad # or args for systag",
968 *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
970 (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
971 (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
972 systagsp = &(*systagsp)->next;
976 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file,
984 for (cur_elements = res->sub_elements; cur_elements;
985 cur_elements = cur_elements->next)
987 if (!strcmp (cur_elements->name, "main"))
988 res->main_elements = cur_elements->elements;
989 fix_element_ref (dh, res, cur_elements->elements);
992 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);