1 /* $Id: d1_absyn.c,v 1.9.2.4 2005-07-19 18:34:16 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 static const char * mk_xpath_regexp (data1_handle dh, const char *expr)
298 const char *p = expr;
303 char *res_p, *res = 0;
316 int is_predicate = 0;
319 for (i = 0; *p && !strchr("/",*p); i++, p++)
321 res_size += (i+3); /* we'll add / between later .. */
322 stack[e] = (char *) nmem_malloc(data1_nmem_get(dh), i+1);
324 for (j = 0; j < i; j++)
326 const char *pp = p-i+j;
345 res_p = res = nmem_malloc(data1_nmem_get(dh), res_size + 10);
348 sprintf(res_p, ".*/");
349 res_p = res_p + strlen(res_p);
351 sprintf(res_p, "%s/", stack[e]);
352 res_p += strlen(stack[e]) + 1;
356 sprintf(res_p, ".*");
359 sprintf (res_p, "$");
361 yaz_log(YLOG_DEBUG, "Got regexp: %s", res);
367 added arg xpelement... when called from xelm context, it's 1, saying
368 that ! means xpath, not element name as attribute name...
372 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
373 char *cp, const char *file, int lineno,
374 const char *element_name, data1_absyn *res,
377 data1_termlist **tp = *tpp;
380 char attname[512], structure[512];
385 for (i = 0; cp[i] && i<sizeof(attname)-1; i++)
386 if (strchr(":,", cp[i]))
394 "%s:%d: Syntax error in termlistspec '%s'",
404 for (i = 0; cp[i] && i<sizeof(structure)-1; i++)
405 if (level == 0 && strchr(",", cp[i]))
409 structure[i] = cp[i];
412 else if (cp[i] == ')')
422 *tp = (data1_termlist *)
423 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
428 strcpy(attname, element_name);
430 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
432 if ((!xpelement) || (*attname != '!')) {
434 "%s:%d: Couldn't find att '%s' in attset",
435 file, lineno, attname);
442 if (r == 2 && (source = strchr(structure, ':')))
443 *source++ = '\0'; /* cut off structure .. */
445 source = "data"; /* ok: default is leaf data */
446 (*tp)->source = (char *)
447 nmem_strdup (data1_nmem_get (dh), source);
449 if (r < 2) /* is the structure qualified? */
450 (*tp)->structure = "w";
452 (*tp)->structure = (char *)
453 nmem_strdup (data1_nmem_get (dh), structure);
462 * Converts a 'melm' field[$subfield] pattern to a simple xpath
464 static int melm2xpath(char *melm, char *buf)
470 if ((dollar = strchr(melm, '$'))) {
475 if (field[0] == '0' && field[1] == '0')
476 fieldtype = "controlfield";
478 fieldtype = "datafield";
479 sprintf(buf, "/*/%s[@tag=\"%s\"]", fieldtype, field);
481 sprintf(buf + strlen(buf), "/subfield[@code=\"%s\"]", subfield);
482 else if (field[0] != '0' || field[1] != '0')
483 strcat(buf, "/subfield");
484 yaz_log(YLOG_DEBUG, "Created xpath: '%s'", buf);
488 const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
489 const char *default_value)
491 struct data1_systag *p = absyn->systags;
492 for (; p; p = p->next)
493 if (!strcmp(p->name, tag))
495 return default_value;
498 #define l_isspace(c) ((c) == '\t' || (c) == ' ' || (c) == '\n' || (c) == '\r')
500 int read_absyn_line(FILE *f, int *lineno, char *line, int len,
501 char *argv[], int num)
507 while ((p = fgets(line, len, f)))
510 while (*p && l_isspace(*p))
518 for (argc = 0; *p ; argc++)
520 if (*p == '#') /* trailing comment */
523 while (*p && !(l_isspace(*p) && !quoted)) {
524 if (*p =='"') quoted = 1 - quoted;
525 if (*p =='[') quoted = 1;
526 if (*p ==']') quoted = 0;
532 while (*p && l_isspace(*p))
540 data1_absyn *data1_read_absyn (data1_handle dh, const char *file,
543 data1_sub_elements *cur_elements = NULL;
544 data1_xpelement *cur_xpelement = NULL;
546 data1_absyn *res = 0;
548 data1_element **ppl[D1_MAX_NESTING];
549 data1_esetname **esetpp;
550 data1_maptab **maptabp;
551 data1_marctab **marcp;
552 data1_termlist *all = 0;
553 data1_attset_child **attset_childp;
554 data1_tagset **tagset_childp;
555 struct data1_systag **systagsp;
559 char *argv[50], line[512];
561 if (!(f = data1_path_fopen(dh, file, "r")))
563 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
568 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
570 res->reference = VAL_NONE;
573 res->enable_xpath_indexing = (f ? 0 : 1);
575 systagsp = &res->systags;
576 tagset_childp = &res->tagset;
578 res->attset = data1_empty_attset (dh);
579 attset_childp = &res->attset->children;
583 esetpp = &res->esetnames;
585 maptabp = &res->maptabs;
588 res->sub_elements = NULL;
589 res->main_elements = NULL;
590 res->xp_elements = NULL;
592 while (f && (argc = read_absyn_line(f, &lineno, line, 512, argv, 50)))
595 if (!strcmp(cmd, "elm") || !strcmp(cmd, "element"))
597 data1_element *new_element;
599 char *p, *sub_p, *path, *name, *termlists;
605 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
614 cur_elements = (data1_sub_elements *)
615 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
616 cur_elements->next = res->sub_elements;
617 cur_elements->elements = NULL;
618 cur_elements->name = "main";
619 res->sub_elements = cur_elements;
622 ppl[level] = &cur_elements->elements;
629 if ((e = strchr(p, '/')))
636 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
641 new_element = *ppl[level-1] = (data1_element *)
642 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
643 new_element->next = new_element->children = 0;
644 new_element->tag = 0;
645 new_element->termlists = 0;
646 new_element->sub_name = 0;
648 tp = &new_element->termlists;
649 ppl[level-1] = &new_element->next;
650 ppl[level] = &new_element->children;
652 /* consider subtree (if any) ... */
653 if ((sub_p = strchr (p, ':')) && sub_p[1])
656 new_element->sub_name =
657 nmem_strdup (data1_nmem_get(dh), sub_p);
659 /* well-defined tag */
660 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
664 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
668 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
671 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
681 new_element->tag = (data1_tag *)
682 nmem_malloc(data1_nmem_get (dh),
683 sizeof(*new_element->tag));
684 nt->which = DATA1T_string;
685 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
686 nt->names = (data1_name *)
687 nmem_malloc(data1_nmem_get(dh),
688 sizeof(*new_element->tag->names));
689 nt->names->name = nt->value.string;
691 nt->kind = DATA1K_string;
697 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
701 /* parse termList definitions */
705 assert (res->attset);
707 if (parse_termlists (dh, &tp, p, file, lineno, name, res, 0))
712 *tp = all; /* append any ALL entries to the list */
714 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
717 New code to support xelm directive
718 for each xelm a dfa is built. xelms are stored in res->xp_elements
720 maybe we should use a simple sscanf instead of dfa?
724 Now [] predicates are supported. regexps and xpath structure is
725 a bit redundant, however it's comfortable later...
730 else if (!strcmp(cmd, "xelm") || !strcmp(cmd, "melm")) {
733 char *p, *xpath_expr, *termlists;
735 struct DFA *dfa = dfa = dfa_init();
737 char melm_xpath[128];
741 yaz_log(LOG_WARN, "%s:%d: Bad # of args to xelm", file, lineno);
745 if (!strcmp(cmd, "melm")) {
746 if (melm2xpath(argv[1], melm_xpath) < 0)
748 xpath_expr = melm_xpath;
750 xpath_expr = argv[1];
753 regexp = mk_xpath_regexp(dh, xpath_expr);
754 i = dfa_parse (dfa, ®exp);
756 yaz_log(LOG_WARN, "%s:%d: Bad xpath to xelm", file, lineno);
763 cur_xpelement = (data1_xpelement *)
764 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
765 res->xp_elements = cur_xpelement;
767 cur_xpelement->next = (data1_xpelement *)
768 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_xpelement));
769 cur_xpelement = cur_xpelement->next;
771 cur_xpelement->next = NULL;
772 cur_xpelement->xpath_expr = nmem_strdup(data1_nmem_get (dh),
776 cur_xpelement->dfa = dfa;
779 cur_xpelement->xpath_len =
780 zebra_parse_xpath_str(xpath_expr,
781 cur_xpelement->xpath, XPATH_STEP_COUNT,
785 dump_xp_steps(cur_xpelement->xpath,cur_xpelement->xpath_len);
788 cur_xpelement->termlists = 0;
789 tp = &cur_xpelement->termlists;
791 /* parse termList definitions */
795 assert (res->attset);
797 if (parse_termlists (dh, &tp, p, file, lineno,
803 *tp = all; /* append any ALL entries to the list */
806 else if (!strcmp(cmd, "section"))
812 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
818 cur_elements = (data1_sub_elements *)
819 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
820 cur_elements->next = res->sub_elements;
821 cur_elements->elements = NULL;
822 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
823 res->sub_elements = cur_elements;
826 ppl[level] = &cur_elements->elements;
828 else if (!strcmp(cmd, "xpath"))
832 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'xpath' directive",
836 if (!strcmp(argv[1], "enable"))
837 res->enable_xpath_indexing = 1;
838 else if (!strcmp (argv[1], "disable"))
839 res->enable_xpath_indexing = 0;
842 yaz_log(LOG_WARN, "%s:%d: Expecting disable/enable "
843 "after 'xpath' directive", file, lineno);
846 else if (!strcmp(cmd, "all"))
848 data1_termlist **tp = &all;
851 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
857 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
861 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res, 0))
867 else if (!strcmp(cmd, "name"))
871 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
875 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
877 else if (!strcmp(cmd, "reference"))
883 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
888 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
890 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
895 else if (!strcmp(cmd, "attset"))
898 data1_attset *attset;
902 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
907 if (!(attset = data1_get_attset (dh, name)))
909 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
913 *attset_childp = (data1_attset_child *)
914 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
915 (*attset_childp)->child = attset;
916 (*attset_childp)->next = 0;
917 attset_childp = &(*attset_childp)->next;
919 else if (!strcmp(cmd, "tagset"))
925 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
931 type = atoi(argv[2]);
932 *tagset_childp = data1_read_tagset (dh, name, type);
933 if (!(*tagset_childp))
935 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
939 tagset_childp = &(*tagset_childp)->next;
941 else if (!strcmp(cmd, "varset"))
947 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
952 if (!(res->varset = data1_read_varset (dh, name)))
954 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
959 else if (!strcmp(cmd, "esetname"))
965 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
972 *esetpp = (data1_esetname *)
973 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
974 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
978 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
980 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
981 file, lineno, fname);
984 esetpp = &(*esetpp)->next;
986 else if (!strcmp(cmd, "maptab"))
992 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
997 if (!(*maptabp = data1_read_maptab (dh, name)))
999 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
1000 file, lineno, name);
1003 maptabp = &(*maptabp)->next;
1005 else if (!strcmp(cmd, "marc"))
1011 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
1016 if (!(*marcp = data1_read_marctab (dh, name)))
1018 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
1019 file, lineno, name);
1022 marcp = &(*marcp)->next;
1024 else if (!strcmp(cmd, "encoding"))
1028 yaz_log(LOG_WARN, "%s:%d: Bad # or args for encoding",
1032 res->encoding = nmem_strdup (data1_nmem_get(dh), argv[1]);
1034 else if (!strcmp(cmd, "systag"))
1038 yaz_log(LOG_WARN, "%s:%d: Bad # or args for systag",
1042 *systagsp = nmem_malloc (data1_nmem_get(dh), sizeof(**systagsp));
1044 (*systagsp)->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
1045 (*systagsp)->value = nmem_strdup(data1_nmem_get(dh), argv[2]);
1046 systagsp = &(*systagsp)->next;
1050 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file,
1058 for (cur_elements = res->sub_elements; cur_elements;
1059 cur_elements = cur_elements->next)
1061 if (!strcmp (cur_elements->name, "main"))
1062 res->main_elements = cur_elements->elements;
1063 fix_element_ref (dh, res, cur_elements->elements);
1066 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);