2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.23 1998-10-15 08:29:16 adam
8 * Tag set type may be specified in reference to it using "tagset"
9 * directive in .abs-files and "include" directive in .tag-files.
11 * Revision 1.22 1998/10/13 16:09:47 adam
12 * Added support for arbitrary OID's for tagsets, schemas and attribute sets.
13 * Added support for multiple attribute set references and tagset references
14 * from an abstract syntax file.
15 * Fixed many bad logs-calls in routines that read the various
16 * specifications regarding data1 (*.abs,*.att,...) and made the messages
17 * consistent whenever possible.
18 * Added extra 'lineno' argument to function readconf_line.
20 * Revision 1.21 1998/06/09 13:55:07 adam
23 * Revision 1.20 1998/05/18 13:07:02 adam
24 * Changed the way attribute sets are handled by the retriaval module.
25 * Extended Explain conversion / schema.
26 * Modified server and client to work with ASN.1 compiled protocol handlers.
28 * Revision 1.19 1998/03/05 08:15:32 adam
29 * Implemented data1_add_insert_taggeddata utility which is more flexible
30 * than data1_insert_taggeddata.
32 * Revision 1.18 1998/02/27 14:08:04 adam
33 * Added const to some char pointer arguments.
34 * Reworked data1_read_node so that it doesn't create a tree with
35 * pointers to original "SGML"-buffer.
37 * Revision 1.17 1998/02/11 11:53:34 adam
38 * Changed code so that it compiles as C++.
40 * Revision 1.16 1997/12/18 10:51:30 adam
41 * Implemented sub-trees feature for schemas - including forward
44 * Revision 1.15 1997/12/09 16:18:16 adam
45 * Work on EXPLAIN schema. First implementation of sub-schema facility
48 * Revision 1.14 1997/10/31 12:20:09 adam
49 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
50 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
51 * Bug fix: missing fclose in data1_read_espec1.
53 * Revision 1.13 1997/10/27 13:54:18 adam
54 * Changed structure field in data1 node to be simple string which
55 * is "unknown" to the retrieval system itself.
57 * Revision 1.12 1997/09/17 12:10:34 adam
60 * Revision 1.11 1997/09/05 09:50:55 adam
61 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
63 * Revision 1.10 1997/05/14 06:54:01 adam
66 * Revision 1.9 1997/02/19 14:46:15 adam
67 * The "all" specifier only affects elements that are indexed (and not
70 * Revision 1.8 1997/01/02 10:47:59 quinn
71 * Added optional, physical ANY
73 * Revision 1.7 1996/06/10 08:56:01 quinn
76 * Revision 1.6 1996/05/31 13:52:21 quinn
77 * Fixed uninitialized variable for local tags in abstract syntax.
79 * Revision 1.5 1996/05/09 07:27:43 quinn
80 * Multiple local attributes values supported.
82 * Revision 1.4 1996/05/01 12:45:28 quinn
83 * Support use of local tag names in abs file.
85 * Revision 1.3 1995/11/01 16:34:55 quinn
86 * Making data1 look for tables in data1_tabpath
88 * Revision 1.2 1995/11/01 13:54:44 quinn
91 * Revision 1.1 1995/11/01 11:56:06 quinn
92 * Added Retrieval (data management) functions en masse.
106 #define D1_MAX_NESTING 128
108 struct data1_absyn_cache_info
112 data1_absyn_cache next;
115 struct data1_attset_cache_info
118 data1_attset *attset;
119 data1_attset_cache next;
122 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
124 data1_absyn_cache p = *data1_absyn_cache_get (dh);
128 if (!strcmp (name, p->name))
135 void data1_absyn_trav (data1_handle dh, void *handle,
136 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
138 data1_absyn_cache p = *data1_absyn_cache_get (dh);
142 (*fh)(dh, handle, p->absyn);
147 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
150 NMEM mem = data1_nmem_get (dh);
152 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
153 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
155 sprintf(fname, "%s.abs", name);
156 p->absyn = data1_read_absyn (dh, fname);
157 p->name = nmem_strdup (mem, name);
163 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
167 if (!(absyn = data1_absyn_search (dh, name)))
168 absyn = data1_absyn_add (dh, name);
172 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
174 data1_attset_cache p = *data1_attset_cache_get (dh);
178 if (!strcmp (name, p->name))
185 data1_attset *data1_attset_search_id (data1_handle dh, int id)
187 data1_attset_cache p = *data1_attset_cache_get (dh);
191 if (id == p->attset->reference)
198 data1_attset *data1_attset_add (data1_handle dh, const char *name)
200 char fname[512], aname[512];
201 NMEM mem = data1_nmem_get (dh);
202 data1_attset *attset;
204 strcpy (aname, name);
205 sprintf(fname, "%s.att", name);
206 attset = data1_read_attset (dh, fname);
210 attset = data1_read_attset (dh, name);
211 if (attset && (cp = strrchr (aname, '.')))
215 logf (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
218 data1_attset_cache p = (data1_attset_cache)
219 nmem_malloc (mem, sizeof(*p));
220 data1_attset_cache *pp = data1_attset_cache_get (dh);
222 attset->name = p->name = nmem_strdup (mem, aname);
230 data1_attset *data1_get_attset (data1_handle dh, const char *name)
232 data1_attset *attset;
234 if (!(attset = data1_attset_search_name (dh, name)))
235 attset = data1_attset_add (dh, name);
239 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
244 for (r = a->esetnames; r; r = r->next)
245 if (!data1_matchstr(r->name, name))
250 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
251 data1_element *parent,
257 r = abs->main_elements;
259 r = parent->children;
260 assert (abs->main_elements);
261 for (; r; r = r->next)
265 for (n = r->tag->names; n; n = n->next)
266 if (!data1_matchstr(tagname, n->name))
272 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
276 assert (absyn->main_elements);
277 for (r = absyn->main_elements; r; r = r->next)
278 if (!data1_matchstr(r->name, name))
284 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
286 for (; e; e = e->next)
291 fix_element_ref (dh, absyn, e->children);
295 data1_sub_elements *sub_e = absyn->sub_elements;
296 while (sub_e && strcmp (e->sub_name, sub_e->name))
299 e->children = sub_e->elements;
301 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
307 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
309 data1_sub_elements *cur_elements = NULL;
310 data1_absyn *res = 0;
312 data1_element **ppl[D1_MAX_NESTING];
313 data1_esetname **esetpp;
314 data1_maptab **maptabp;
315 data1_marctab **marcp;
316 data1_termlist *all = 0;
317 data1_attset_child **attset_childp;
318 data1_tagset **tagset_childp;
322 char *argv[50], line[512];
324 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
326 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
330 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
332 res->reference = VAL_NONE;
334 tagset_childp = &res->tagset;
336 res->attset = data1_empty_attset (dh);
337 attset_childp = &res->attset->children;
341 esetpp = &res->esetnames;
343 maptabp = &res->maptabs;
347 res->sub_elements = NULL;
348 res->main_elements = NULL;
350 while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
353 if (!strcmp(cmd, "elm"))
355 data1_element *new_element;
357 char *p, *sub_p, *path, *name, *termlists;
363 logf(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
372 cur_elements = (data1_sub_elements *)
373 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
374 cur_elements->next = res->sub_elements;
375 cur_elements->elements = NULL;
376 cur_elements->name = "main";
377 res->sub_elements = cur_elements;
380 ppl[level] = &cur_elements->elements;
387 if ((e = strchr(p, '/')))
394 logf(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
399 new_element = *ppl[level] = (data1_element *)
400 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
401 new_element->next = new_element->children = 0;
402 new_element->tag = 0;
403 new_element->termlists = 0;
404 new_element->sub_name = 0;
406 tp = &new_element->termlists;
407 ppl[level] = &new_element->next;
408 ppl[level+1] = &new_element->children;
410 /* consider subtree (if any) ... */
411 if ((sub_p = strchr (p, ':')) && sub_p[1])
414 new_element->sub_name =
415 nmem_strdup (data1_nmem_get(dh), sub_p);
417 /* well-defined tag */
418 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
422 logf(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
426 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
429 logf(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
439 new_element->tag = (data1_tag *)
440 nmem_malloc(data1_nmem_get (dh),
441 sizeof(*new_element->tag));
442 nt->which = DATA1T_string;
443 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
444 nt->names = (data1_name *)
445 nmem_malloc(data1_nmem_get(dh),
446 sizeof(*new_element->tag->names));
447 nt->names->name = nt->value.string;
449 nt->kind = DATA1K_string;
455 logf(LOG_WARN, "%s:%d: Bad element", file, lineno);
459 /* parse termList definitions */
462 new_element->termlists = 0;
465 assert (res->attset);
468 char attname[512], structure[512];
471 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
475 "%s:%d: Syntax error in termlistspec '%s'",
481 strcpy(attname, name);
482 *tp = (data1_termlist *)
483 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
485 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
489 "%s:%d: Couldn't find att '%s' in attset",
490 file, lineno, attname);
494 if (r < 2) /* is the structure qualified? */
495 (*tp)->structure = "w";
498 (*tp)->structure = (char *)
499 nmem_malloc (data1_nmem_get (dh),
500 strlen(structure)+1);
501 strcpy ((*tp)->structure, structure);
505 while ((p = strchr(p, ',')) && *(++p));
506 *tp = all; /* append any ALL entries to the list */
508 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
510 else if (!strcmp(cmd, "section"))
516 logf(LOG_WARN, "%s:%d: Bad # of args to section",
522 cur_elements = (data1_sub_elements *)
523 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
524 cur_elements->next = res->sub_elements;
525 cur_elements->elements = NULL;
526 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
527 res->sub_elements = cur_elements;
530 ppl[level] = &cur_elements->elements;
532 else if (!strcmp(cmd, "all"))
535 data1_termlist **tp = &all;
539 logf(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
546 logf(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
551 assert (res->attset);
554 char attname[512], structure[512];
557 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
560 logf(LOG_WARN, "%s:%d: Syntax error in termlistspec",
565 *tp = (data1_termlist *)
566 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
568 data1_getattbyname (dh, res->attset, attname)))
570 logf(LOG_WARN, "%s:%d: Couldn't find att '%s' in attset",
571 file, lineno, attname);
575 if (r < 2) /* is the structure qualified? */
576 (*tp)->structure = "w";
580 (char *)nmem_malloc (data1_nmem_get (dh),
581 strlen(structure)+1);
582 strcpy ((*tp)->structure, structure);
587 while ((p = strchr(p, ',')) && *(++p));
589 else if (!strcmp(cmd, "name"))
593 logf(LOG_WARN, "%s:%d: Bad # of args to name directive",
597 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
599 else if (!strcmp(cmd, "reference"))
605 logf(LOG_WARN, "%s:%d: Bad # of args to reference",
610 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
612 logf(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
617 else if (!strcmp(cmd, "attset"))
620 data1_attset *attset;
624 logf(LOG_WARN, "%s:%d: Bad # of args to attset",
629 if (!(attset = data1_get_attset (dh, name)))
631 logf(LOG_WARN, "%s:%d: Couldn't find attset %s",
635 *attset_childp = (data1_attset_child *)
636 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
637 (*attset_childp)->child = attset;
638 (*attset_childp)->next = 0;
639 attset_childp = &(*attset_childp)->next;
641 else if (!strcmp(cmd, "tagset"))
647 logf(LOG_WARN, "%s:%d: Bad # of args to tagset",
653 type = atoi(argv[2]);
654 *tagset_childp = data1_read_tagset (dh, name, type);
655 if (!(*tagset_childp))
657 logf(LOG_WARN, "%s:%d: Couldn't load tagset %s",
661 tagset_childp = &(*tagset_childp)->next;
663 else if (!strcmp(cmd, "varset"))
669 logf(LOG_WARN, "%s:%d: Bad # of args in varset",
674 if (!(res->varset = data1_read_varset (dh, name)))
676 logf(LOG_WARN, "%s:%d: Couldn't load Varset %s",
681 else if (!strcmp(cmd, "esetname"))
687 logf(LOG_WARN, "%s:%d: Bad # of args in esetname",
694 *esetpp = (data1_esetname *)
695 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
696 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
700 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
702 logf(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
703 file, lineno, fname);
706 esetpp = &(*esetpp)->next;
708 else if (!strcmp(cmd, "maptab"))
714 logf(LOG_WARN, "%s:%d: Bad # of args for maptab",
719 if (!(*maptabp = data1_read_maptab (dh, name)))
721 logf(LOG_WARN, "%s:%d: Couldn't load maptab %s",
725 maptabp = &(*maptabp)->next;
727 else if (!strcmp(cmd, "marc"))
733 logf(LOG_WARN, "%s:%d: Bad # or args for marc",
738 if (!(*marcp = data1_read_marctab (dh, name)))
740 logf(LOG_WARN, "%s:%d: Couldn't read marctab %s",
744 marcp = &(*marcp)->next;
748 logf(LOG_WARN, "%s:%d: Unknown directive '%s'", file, lineno, cmd);
754 for (cur_elements = res->sub_elements; cur_elements;
755 cur_elements = cur_elements->next)
757 if (!strcmp (cur_elements->name, "main"))
758 res->main_elements = cur_elements->elements;
759 fix_element_ref (dh, res, cur_elements->elements);
761 logf (LOG_DEBUG, "%s: data1_read_absyn end", file);