1 /* $Id: d1_absyn.c,v 1.4 2002-12-16 20:27:18 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
32 #define D1_MAX_NESTING 128
37 struct data1_systag *next;
40 struct data1_absyn_cache_info
44 data1_absyn_cache next;
47 struct data1_attset_cache_info
51 data1_attset_cache next;
54 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
56 data1_absyn_cache p = *data1_absyn_cache_get (dh);
60 if (!strcmp (name, p->name))
67 We need to destroy DFAs, in xp_element (xelm) definitions
71 void data1_absyn_destroy (data1_handle dh)
73 data1_absyn_cache p = *data1_absyn_cache_get (dh);
77 data1_absyn *abs = p->absyn;
78 data1_xpelement *xpe = abs->xp_elements;
80 logf (LOG_DEBUG,"Destroy xp element %s",xpe->xpath_expr);
81 if (xpe->dfa) { dfa_delete (&xpe->dfa); }
89 void data1_absyn_trav (data1_handle dh, void *handle,
90 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
92 data1_absyn_cache p = *data1_absyn_cache_get (dh);
96 (*fh)(dh, handle, p->absyn);
101 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
104 NMEM mem = data1_nmem_get (dh);
106 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
107 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
109 sprintf(fname, "%s.abs", name);
110 p->absyn = data1_read_absyn (dh, fname, 0);
111 p->name = nmem_strdup (mem, name);
117 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
121 if (!(absyn = data1_absyn_search (dh, name)))
122 absyn = data1_absyn_add (dh, name);
126 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
128 data1_attset_cache p = *data1_attset_cache_get (dh);
132 if (!strcmp (name, p->name))
139 data1_attset *data1_attset_search_id (data1_handle dh, int id)
141 data1_attset_cache p = *data1_attset_cache_get (dh);
145 if (id == p->attset->reference)
152 data1_attset *data1_attset_add (data1_handle dh, const char *name)
154 char fname[512], aname[512];
155 NMEM mem = data1_nmem_get (dh);
156 data1_attset *attset;
158 strcpy (aname, name);
159 sprintf(fname, "%s.att", name);
160 attset = data1_read_attset (dh, fname);
164 attset = data1_read_attset (dh, name);
165 if (attset && (cp = strrchr (aname, '.')))
169 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
172 data1_attset_cache p = (data1_attset_cache)
173 nmem_malloc (mem, sizeof(*p));
174 data1_attset_cache *pp = data1_attset_cache_get (dh);
176 attset->name = p->name = nmem_strdup (mem, aname);
184 data1_attset *data1_get_attset (data1_handle dh, const char *name)
186 data1_attset *attset;
188 if (!(attset = data1_attset_search_name (dh, name)))
189 attset = data1_attset_add (dh, name);
193 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
198 for (r = a->esetnames; r; r = r->next)
199 if (!data1_matchstr(r->name, name))
204 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
205 data1_element *parent,
210 /* It's now possible to have a data1 tree with no abstract syntax */
215 r = abs->main_elements;
217 r = parent->children;
219 for (; r; r = r->next)
223 for (n = r->tag->names; n; n = n->next)
224 if (!data1_matchstr(tagname, n->name))
230 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
235 /* It's now possible to have a data1 tree with no abstract syntax */
238 for (r = absyn->main_elements; r; r = r->next)
239 if (!data1_matchstr(r->name, name))
245 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
247 /* It's now possible to have a data1 tree with no abstract syntax */
251 for (; e; e = e->next)
256 fix_element_ref (dh, absyn, e->children);
260 data1_sub_elements *sub_e = absyn->sub_elements;
261 while (sub_e && strcmp (e->sub_name, sub_e->name))
264 e->children = sub_e->elements;
266 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
273 New function, a bit dummy now... I've seen it in zrpn.c... We should build
274 more clever regexps...
287 const char * mk_xpath_regexp (data1_handle dh, char *expr)
294 static char *stack[32];
295 static char res[1024];
298 if (*p != '/') { return (""); }
300 if (*p == '/') { abs=0; p++; }
304 while (*p && !strchr("/",*p)) { i++; p++; }
305 stack[e] = (char *) nmem_malloc (data1_nmem_get (dh), i+1);
306 memcpy (stack[e], p - i, i);
311 e--; p = &res[0]; i=0;
312 sprintf (p, "^"); p++;
314 /* !!! res size is not checked !!! */
315 sprintf (p, "%s/",stack[e]);
316 p += strlen(stack[e]) + 1;
319 if (!abs) { sprintf (p, ".*"); p+=2; }
320 sprintf (p, "$"); p++;
321 r = nmem_strdup (data1_nmem_get (dh), res);
327 added arg xpelement... when called from xelm context, it's 1, saying
328 that ! means xpath, not element name as attribute name...
332 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
333 char *p, const char *file, int lineno,
334 const char *element_name, data1_absyn *res,
337 data1_termlist **tp = *tpp;
340 char attname[512], structure[512];
344 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
348 "%s:%d: Syntax error in termlistspec '%s'",
353 *tp = (data1_termlist *)
354 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
359 strcpy(attname, element_name);
361 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
363 if ((!xpelement) || (*attname != '!')) {
365 "%s:%d: Couldn't find att '%s' in attset",
366 file, lineno, attname);
373 if (r == 2 && (source = strchr(structure, ':')))
374 *source++ = '\0'; /* cut off structure .. */
376 source = "data"; /* ok: default is leaf data */
377 (*tp)->source = (char *)
378 nmem_strdup (data1_nmem_get (dh), source);
380 if (r < 2) /* is the structure qualified? */
381 (*tp)->structure = "w";
383 (*tp)->structure = (char *)
384 nmem_strdup (data1_nmem_get (dh), structure);
387 while ((p = strchr(p, ',')) && *(++p));
392 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
393 const char *default_value)
395 struct data1_systag *p = absyn->systags;
396 for (; p; p = p->next)
397 if (!strcmp(p->name, tag))
399 return default_value;
402 data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
405 data1_sub_elements *cur_elements = NULL;
406 data1_xpelement *cur_xpelement = NULL;
408 data1_absyn *res = 0;
410 data1_element **ppl[D1_MAX_NESTING];
411 data1_esetname **esetpp;
412 data1_maptab **maptabp;
413 data1_marctab **marcp;
414 data1_termlist *all = 0;
415 data1_attset_child **attset_childp;
416 data1_tagset **tagset_childp;
417 struct data1_systag **systagsp;
421 char *argv[50], line[512];
423 if (!(f = data1_path_fopen(dh, file, "r")))
425 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
430 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
432 res->reference = VAL_NONE;
435 res->enable_xpath_indexing = (f ? 0 : 1);
437 systagsp = &res->systags;
438 tagset_childp = &res->tagset;
440 res->attset = data1_empty_attset (dh);
441 attset_childp = &res->attset->children;
445 esetpp = &res->esetnames;
447 maptabp = &res->maptabs;
450 res->sub_elements = NULL;
451 res->main_elements = NULL;
452 res->xp_elements = NULL;
454 while (f && (argc = readconf_line(f, &lineno, line, 512, argv, 50)))
457 if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
459 data1_element *new_element;
461 char *p, *sub_p, *path, *name, *termlists;
467 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
476 cur_elements = (data1_sub_elements *)
477 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
478 cur_elements->next = res->sub_elements;
479 cur_elements->elements = NULL;
480 cur_elements->name = "main";
481 res->sub_elements = cur_elements;
484 ppl[level] = &cur_elements->elements;
491 if ((e = strchr(p, '/')))
498 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
503 new_element = *ppl[level-1] = (data1_element *)
504 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
505 new_element->next = new_element->children = 0;
506 new_element->tag = 0;
507 new_element->termlists = 0;
508 new_element->sub_name = 0;
510 tp = &new_element->termlists;
511 ppl[level-1] = &new_element->next;
512 ppl[level] = &new_element->children;
514 /* consider subtree (if any) ... */
515 if ((sub_p = strchr (p, ':')) && sub_p[1])
518 new_element->sub_name =
519 nmem_strdup (data1_nmem_get(dh), sub_p);
521 /* well-defined tag */
522 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
526 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
530 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
533 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
543 new_element->tag = (data1_tag *)
544 nmem_malloc(data1_nmem_get (dh),
545 sizeof(*new_element->tag));
546 nt->which = DATA1T_string;
547 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
548 nt->names = (data1_name *)
549 nmem_malloc(data1_nmem_get(dh),
550 sizeof(*new_element->tag->names));
551 nt->names->name = nt->value.string;
553 nt->kind = DATA1K_string;
559 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
563 /* parse termList definitions */
567 assert (res->attset);
569 if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
574 *tp = all; /* append any ALL entries to the list */
576 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
579 New code to support xelm directive
580 for each xelm a dfa is built. xelms are stored in res->xp_elements
582 maybe we should use a simple sscanf instead of dfa?
587 else if (!strcmp(cmd, "xelm")) {
590 char *p, *xpath_expr, *termlists;
593 struct DFA *dfa = dfa = dfa_init();
598 yaz_log(LOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
601 xpath_expr = argv[1];
603 regexp = mk_xpath_regexp(dh, xpath_expr);
604 i = dfa_parse (dfa, ®exp);
606 yaz_log(LOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
613 cur_xpelement = (data1_xpelement *)
614 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
615 res->xp_elements = cur_xpelement;
617 cur_xpelement->next = (data1_xpelement *)
618 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
619 cur_xpelement = cur_xpelement->next;
621 cur_xpelement->next = NULL;
622 cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh),
626 cur_xpelement->dfa = dfa;
628 cur_xpelement->termlists = 0;
629 tp = &cur_xpelement->termlists;
631 /* parse termList definitions */
635 assert (res->attset);
637 if (parse_termlists (dh, &tp, p, file, lineno,
643 *tp = all; /* append any ALL entries to the list */
647 else if (!strcmp(cmd, "section"))
653 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
659 cur_elements = (data1_sub_elements *)
660 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
661 cur_elements->next = res->sub_elements;
662 cur_elements->elements = NULL;
663 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
664 res->sub_elements = cur_elements;
667 ppl[level] = &cur_elements->elements;
669 else if (!strcmp(cmd, "xpath"))
673 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
677 if (!strcmp(argv[1], "enable"))
678 res->enable_xpath_indexing = 1;
679 else if (!strcmp (argv[1], "disable"))
680 res->enable_xpath_indexing = 0;
683 yaz_log(LOG_WARN, "%s:%d: Expecting disable/enable "
684 "after 'xpath' directive", file, lineno);
687 else if (!strcmp(cmd, "all"))
689 data1_termlist **tp = &all;
692 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
698 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
702 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
708 else if (!strcmp(cmd, "name"))
712 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
716 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
718 else if (!strcmp(cmd, "reference"))
724 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
729 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
731 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
736 else if (!strcmp(cmd, "attset"))
739 data1_attset *attset;
743 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
748 if (!(attset = data1_get_attset (dh, name)))
750 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
754 *attset_childp = (data1_attset_child *)
755 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
756 (*attset_childp)->child = attset;
757 (*attset_childp)->next = 0;
758 attset_childp = &(*attset_childp)->next;
760 else if (!strcmp(cmd, "tagset"))
766 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
772 type = atoi(argv[2]);
773 *tagset_childp = data1_read_tagset (dh, name, type);
774 if (!(*tagset_childp))
776 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
780 tagset_childp = &(*tagset_childp)->next;
782 else if (!strcmp(cmd, "varset"))
788 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
793 if (!(res->varset = data1_read_varset (dh, name)))
795 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
800 else if (!strcmp(cmd, "esetname"))
806 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
813 *esetpp = (data1_esetname *)
814 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
815 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
819 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
821 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
822 file, lineno, fname);
825 esetpp = &(*esetpp)->next;
827 else if (!strcmp(cmd, "maptab"))
833 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
838 if (!(*maptabp = data1_read_maptab (dh, name)))
840 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
844 maptabp = &(*maptabp)->next;
846 else if (!strcmp(cmd, "marc"))
852 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
857 if (!(*marcp = data1_read_marctab (dh, name)))
859 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
863 marcp = &(*marcp)->next;
865 else if (!strcmp(cmd, "encoding"))
869 yaz_log(LOG_WARN, "%s:%d: Bad # or args for encoding",
873 res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
875 else if (!strcmp(cmd, "systag"))
877 struct data1_systag *st;
880 yaz_log(LOG_WARN, "%s:%d: Bad # or args for systag",
884 *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
886 (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
887 (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
888 systagsp = &(*systagsp)->next;
892 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file,
900 for (cur_elements = res->sub_elements; cur_elements;
901 cur_elements = cur_elements->next)
903 if (!strcmp (cur_elements->name, "main"))
904 res->main_elements = cur_elements->elements;
905 fix_element_ref (dh, res, cur_elements->elements);
908 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);