2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.20 1998-05-18 13:07:02 adam
8 * Changed the way attribute sets are handled by the retriaval module.
9 * Extended Explain conversion / schema.
10 * Modified server and client to work with ASN.1 compiled protocol handlers.
12 * Revision 1.19 1998/03/05 08:15:32 adam
13 * Implemented data1_add_insert_taggeddata utility which is more flexible
14 * than data1_insert_taggeddata.
16 * Revision 1.18 1998/02/27 14:08:04 adam
17 * Added const to some char pointer arguments.
18 * Reworked data1_read_node so that it doesn't create a tree with
19 * pointers to original "SGML"-buffer.
21 * Revision 1.17 1998/02/11 11:53:34 adam
22 * Changed code so that it compiles as C++.
24 * Revision 1.16 1997/12/18 10:51:30 adam
25 * Implemented sub-trees feature for schemas - including forward
28 * Revision 1.15 1997/12/09 16:18:16 adam
29 * Work on EXPLAIN schema. First implementation of sub-schema facility
32 * Revision 1.14 1997/10/31 12:20:09 adam
33 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
34 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
35 * Bug fix: missing fclose in data1_read_espec1.
37 * Revision 1.13 1997/10/27 13:54:18 adam
38 * Changed structure field in data1 node to be simple string which
39 * is "unknown" to the retrieval system itself.
41 * Revision 1.12 1997/09/17 12:10:34 adam
44 * Revision 1.11 1997/09/05 09:50:55 adam
45 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
47 * Revision 1.10 1997/05/14 06:54:01 adam
50 * Revision 1.9 1997/02/19 14:46:15 adam
51 * The "all" specifier only affects elements that are indexed (and not
54 * Revision 1.8 1997/01/02 10:47:59 quinn
55 * Added optional, physical ANY
57 * Revision 1.7 1996/06/10 08:56:01 quinn
60 * Revision 1.6 1996/05/31 13:52:21 quinn
61 * Fixed uninitialized variable for local tags in abstract syntax.
63 * Revision 1.5 1996/05/09 07:27:43 quinn
64 * Multiple local attributes values supported.
66 * Revision 1.4 1996/05/01 12:45:28 quinn
67 * Support use of local tag names in abs file.
69 * Revision 1.3 1995/11/01 16:34:55 quinn
70 * Making data1 look for tables in data1_tabpath
72 * Revision 1.2 1995/11/01 13:54:44 quinn
75 * Revision 1.1 1995/11/01 11:56:06 quinn
76 * Added Retrieval (data management) functions en masse.
93 #define D1_MAX_NESTING 128
95 struct data1_absyn_cache_info
99 data1_absyn_cache next;
102 struct data1_attset_cache_info
105 data1_attset *attset;
106 data1_attset_cache next;
109 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
111 data1_absyn_cache p = *data1_absyn_cache_get (dh);
115 if (!strcmp (name, p->name))
122 void data1_absyn_trav (data1_handle dh, void *handle,
123 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
125 data1_absyn_cache p = *data1_absyn_cache_get (dh);
129 (*fh)(dh, handle, p->absyn);
134 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
137 NMEM mem = data1_nmem_get (dh);
139 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
140 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
142 sprintf(fname, "%s.abs", name);
143 p->absyn = data1_read_absyn (dh, fname);
144 p->name = nmem_strdup (mem, name);
150 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
154 if (!(absyn = data1_absyn_search (dh, name)))
155 absyn = data1_absyn_add (dh, name);
159 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
161 data1_attset_cache p = *data1_attset_cache_get (dh);
165 if (!strcmp (name, p->name))
172 data1_attset *data1_attset_search_id (data1_handle dh, int id)
174 data1_attset_cache p = *data1_attset_cache_get (dh);
178 if (id == p->attset->reference)
185 data1_attset *data1_attset_add (data1_handle dh, const char *name)
187 char fname[512], aname[512];
188 NMEM mem = data1_nmem_get (dh);
189 data1_attset *attset;
191 strcpy (aname, name);
192 sprintf(fname, "%s.att", name);
193 attset = data1_read_attset (dh, fname);
197 attset = data1_read_attset (dh, name);
198 if (attset && (cp = strrchr (aname, '.')))
202 logf (LOG_WARN|LOG_ERRNO, "couldn't load attribute set %s", name);
205 data1_attset_cache p = (data1_attset_cache)
206 nmem_malloc (mem, sizeof(*p));
207 data1_attset_cache *pp = data1_attset_cache_get (dh);
209 attset->name = p->name = nmem_strdup (mem, aname);
217 data1_attset *data1_get_attset (data1_handle dh, const char *name)
219 data1_attset *attset;
221 if (!(attset = data1_attset_search_name (dh, name)))
222 attset = data1_attset_add (dh, name);
226 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
231 for (r = a->esetnames; r; r = r->next)
232 if (!data1_matchstr(r->name, name))
237 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
238 data1_element *parent,
244 r = abs->main_elements;
246 r = parent->children;
247 assert (abs->main_elements);
248 for (; r; r = r->next)
252 for (n = r->tag->names; n; n = n->next)
253 if (!data1_matchstr(tagname, n->name))
259 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
263 assert (absyn->main_elements);
264 for (r = absyn->main_elements; r; r = r->next)
265 if (!data1_matchstr(r->name, name))
271 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
273 for (; e; e = e->next)
278 fix_element_ref (dh, absyn, e->children);
282 data1_sub_elements *sub_e = absyn->sub_elements;
283 while (sub_e && strcmp (e->sub_name, sub_e->name))
286 e->children = sub_e->elements;
288 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
294 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
296 char line[512], *r, cmd[512], args[512];
297 data1_sub_elements *cur_elements = NULL;
298 data1_absyn *res = 0;
300 data1_element **ppl[D1_MAX_NESTING];
301 data1_esetname **esetpp;
302 data1_maptab **maptabp;
303 data1_marctab **marcp;
304 data1_termlist *all = 0;
307 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
309 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
313 res = (data1_absyn *)nmem_malloc(data1_nmem_get(dh), sizeof(*res));
315 res->reference = VAL_NONE;
320 esetpp = &res->esetnames;
322 maptabp = &res->maptabs;
326 res->sub_elements = NULL;
327 res->main_elements = NULL;
331 while ((r = fgets(line, 512, f)))
333 while (*r && isspace(*r))
340 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
342 if (!strcmp(cmd, "elm"))
344 data1_element *new_element;
346 char path[512], name[512], termlists[512], *p, *sub_p;
352 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
353 sizeof(*cur_elements));
354 cur_elements->next = res->sub_elements;
355 cur_elements->elements = NULL;
356 cur_elements->name = "main";
357 res->sub_elements = cur_elements;
360 ppl[level] = &cur_elements->elements;
362 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
364 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
374 if ((e = strchr(p, '/')))
381 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
386 new_element = *ppl[level] =
387 (data1_element *)nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
388 new_element->next = new_element->children = 0;
389 new_element->tag = 0;
390 new_element->termlists = 0;
391 new_element->sub_name = 0;
393 tp = &new_element->termlists;
394 ppl[level] = &new_element->next;
395 ppl[level+1] = &new_element->children;
397 /* consider subtree (if any) ... */
398 if ((sub_p = strchr (p, ':')) && sub_p[1])
401 new_element->sub_name =
402 nmem_strdup (data1_nmem_get(dh), sub_p);
404 /* well-defined tag */
405 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
409 logf(LOG_WARN, "No tagset loaded in %s", file);
413 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
416 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
426 new_element->tag = (data1_tag *)nmem_malloc(data1_nmem_get (dh),
427 sizeof(*new_element->tag));
428 nt->which = DATA1T_string;
429 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
430 nt->names = (data1_name *)nmem_malloc(data1_nmem_get(dh),
431 sizeof(*new_element->tag->names));
432 nt->names->name = nt->value.string;
434 nt->kind = DATA1K_string;
440 logf(LOG_WARN, "Bad element is %s", file);
444 /* parse termList definitions */
447 new_element->termlists = 0;
452 logf(LOG_WARN, "No attset loaded in %s", file);
458 char attname[512], structure[512];
461 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
464 logf(LOG_WARN, "Syntax error in termlistspec in %s",
470 strcpy(attname, name);
471 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
473 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
476 logf(LOG_WARN, "Couldn't find att '%s' in attset",
481 if (r < 2) /* is the structure qualified? */
482 (*tp)->structure = "w";
485 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
486 strlen(structure)+1);
487 strcpy ((*tp)->structure, structure);
491 while ((p = strchr(p, ',')) && *(++p));
492 *tp = all; /* append any ALL entries to the list */
494 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
496 else if (!strcmp(cmd, "section"))
499 if (sscanf(args, "%511s", name) < 1)
501 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
505 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
506 sizeof(*cur_elements));
507 cur_elements->next = res->sub_elements;
508 cur_elements->elements = NULL;
509 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
510 res->sub_elements = cur_elements;
513 ppl[level] = &cur_elements->elements;
515 else if (!strcmp(cmd, "all"))
518 data1_termlist **tp = &all;
522 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
530 logf(LOG_WARN, "No attset loaded in %s", file);
536 char attname[512], structure[512];
539 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
542 logf(LOG_WARN, "Syntax error in termlistspec in %s",
547 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
548 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
551 logf(LOG_WARN, "Couldn't find att '%s' in attset",
556 if (r < 2) /* is the structure qualified? */
557 (*tp)->structure = "w";
560 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
561 strlen(structure)+1);
562 strcpy ((*tp)->structure, structure);
567 while ((p = strchr(p, ',')) && *(++p));
569 else if (!strcmp(cmd, "name"))
573 if (!sscanf(args, "%511s", name))
575 logf(LOG_WARN, "Malformed name directive in %s", file);
579 res->name = nmem_strdup(data1_nmem_get(dh), name);
581 else if (!strcmp(cmd, "reference"))
585 if (!sscanf(args, "%s", name))
587 logf(LOG_WARN, "Malformed reference in %s", file);
591 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
593 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
598 else if (!strcmp(cmd, "attset"))
602 if (!sscanf(args, "%s", name))
604 logf(LOG_WARN, "Malformed attset directive in %s", file);
608 if (!(res->attset = data1_get_attset (dh, name)))
610 logf(LOG_WARN, "Attset failed in %s", file);
615 else if (!strcmp(cmd, "tagset"))
619 if (!sscanf(args, "%s", name))
621 logf(LOG_WARN, "Malformed tagset directive in %s", file);
625 if (!(res->tagset = data1_read_tagset (dh, name)))
627 logf(LOG_WARN, "Tagset failed in %s", file);
632 else if (!strcmp(cmd, "varset"))
636 if (!sscanf(args, "%s", name))
638 logf(LOG_WARN, "Malformed varset directive in %s", file);
642 if (!(res->varset = data1_read_varset (dh, name)))
644 logf(LOG_WARN, "Varset failed in %s", file);
649 else if (!strcmp(cmd, "esetname"))
651 char name[512], fname[512];
653 if (sscanf(args, "%s %s", name, fname) != 2)
655 logf(LOG_WARN, "Two arg's required for esetname in %s",
660 *esetpp = (data1_esetname *)nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
661 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
665 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
667 logf(LOG_WARN, "%s: Espec-1 read failed", file);
671 esetpp = &(*esetpp)->next;
673 else if (!strcmp(cmd, "maptab"))
677 if (sscanf(args, "%s", name) != 1)
679 logf(LOG_WARN, "One argument for maptab directive in %s",
683 if (!(*maptabp = data1_read_maptab (dh, name)))
685 logf(LOG_WARN, "Failed to read maptab %s in %s",
689 maptabp = &(*maptabp)->next;
691 else if (!strcmp(cmd, "marc"))
695 if (sscanf(args, "%s", name) != 1)
697 logf(LOG_WARN, "One argument for marc directive in %s",
701 if (!(*marcp = data1_read_marctab (dh, name)))
703 logf(LOG_WARN, "%Failed to read marctab %s in %s",
707 marcp = &(*marcp)->next;
711 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
718 for (cur_elements = res->sub_elements; cur_elements;
719 cur_elements = cur_elements->next)
721 if (!strcmp (cur_elements->name, "main"))
722 res->main_elements = cur_elements->elements;
723 fix_element_ref (dh, res, cur_elements->elements);
725 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);