2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.19 1998-03-05 08:15:32 adam
8 * Implemented data1_add_insert_taggeddata utility which is more flexible
9 * than data1_insert_taggeddata.
11 * Revision 1.18 1998/02/27 14:08:04 adam
12 * Added const to some char pointer arguments.
13 * Reworked data1_read_node so that it doesn't create a tree with
14 * pointers to original "SGML"-buffer.
16 * Revision 1.17 1998/02/11 11:53:34 adam
17 * Changed code so that it compiles as C++.
19 * Revision 1.16 1997/12/18 10:51:30 adam
20 * Implemented sub-trees feature for schemas - including forward
23 * Revision 1.15 1997/12/09 16:18:16 adam
24 * Work on EXPLAIN schema. First implementation of sub-schema facility
27 * Revision 1.14 1997/10/31 12:20:09 adam
28 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
29 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
30 * Bug fix: missing fclose in data1_read_espec1.
32 * Revision 1.13 1997/10/27 13:54:18 adam
33 * Changed structure field in data1 node to be simple string which
34 * is "unknown" to the retrieval system itself.
36 * Revision 1.12 1997/09/17 12:10:34 adam
39 * Revision 1.11 1997/09/05 09:50:55 adam
40 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
42 * Revision 1.10 1997/05/14 06:54:01 adam
45 * Revision 1.9 1997/02/19 14:46:15 adam
46 * The "all" specifier only affects elements that are indexed (and not
49 * Revision 1.8 1997/01/02 10:47:59 quinn
50 * Added optional, physical ANY
52 * Revision 1.7 1996/06/10 08:56:01 quinn
55 * Revision 1.6 1996/05/31 13:52:21 quinn
56 * Fixed uninitialized variable for local tags in abstract syntax.
58 * Revision 1.5 1996/05/09 07:27:43 quinn
59 * Multiple local attributes values supported.
61 * Revision 1.4 1996/05/01 12:45:28 quinn
62 * Support use of local tag names in abs file.
64 * Revision 1.3 1995/11/01 16:34:55 quinn
65 * Making data1 look for tables in data1_tabpath
67 * Revision 1.2 1995/11/01 13:54:44 quinn
70 * Revision 1.1 1995/11/01 11:56:06 quinn
71 * Added Retrieval (data management) functions en masse.
88 #define D1_MAX_NESTING 128
90 struct data1_absyn_cache_info
94 data1_absyn_cache next;
97 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
99 data1_absyn_cache p = *data1_absyn_cache_get (dh);
103 if (!strcmp (name, p->name))
110 void data1_absyn_trav (data1_handle dh, void *handle,
111 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
113 data1_absyn_cache p = *data1_absyn_cache_get (dh);
117 (*fh)(dh, handle, p->absyn);
122 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
125 NMEM mem = data1_nmem_get (dh);
127 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
128 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
130 sprintf(fname, "%s.abs", name);
131 p->absyn = data1_read_absyn (dh, fname);
132 p->name = nmem_strdup (mem, name);
138 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
142 if (!(absyn = data1_absyn_search (dh, name)))
143 absyn = data1_absyn_add (dh, name);
147 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
152 for (r = a->esetnames; r; r = r->next)
153 if (!data1_matchstr(r->name, name))
158 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
159 data1_element *parent,
165 r = abs->main_elements;
167 r = parent->children;
168 assert (abs->main_elements);
169 for (; r; r = r->next)
173 for (n = r->tag->names; n; n = n->next)
174 if (!data1_matchstr(tagname, n->name))
180 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
184 assert (absyn->main_elements);
185 for (r = absyn->main_elements; r; r = r->next)
186 if (!data1_matchstr(r->name, name))
192 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
194 for (; e; e = e->next)
199 fix_element_ref (dh, absyn, e->children);
203 data1_sub_elements *sub_e = absyn->sub_elements;
204 while (sub_e && strcmp (e->sub_name, sub_e->name))
207 e->children = sub_e->elements;
209 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
215 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
217 char line[512], *r, cmd[512], args[512];
218 data1_sub_elements *cur_elements = NULL;
219 data1_absyn *res = 0;
221 data1_element **ppl[D1_MAX_NESTING];
222 data1_esetname **esetpp;
223 data1_maptab **maptabp;
224 data1_marctab **marcp;
225 data1_termlist *all = 0;
228 logf (LOG_DEBUG, "begin data1_read_absyn file=%s", file);
229 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
231 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
235 res = (data1_absyn *)nmem_malloc(data1_nmem_get(dh), sizeof(*res));
237 res->reference = VAL_NONE;
242 esetpp = &res->esetnames;
244 maptabp = &res->maptabs;
248 res->sub_elements = NULL;
249 res->main_elements = NULL;
253 while ((r = fgets(line, 512, f)))
255 while (*r && isspace(*r))
262 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
264 if (!strcmp(cmd, "elm"))
266 data1_element *new_element;
268 char path[512], name[512], termlists[512], *p, *sub_p;
274 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
275 sizeof(*cur_elements));
276 cur_elements->next = res->sub_elements;
277 cur_elements->elements = NULL;
278 cur_elements->name = "main";
279 res->sub_elements = cur_elements;
282 ppl[level] = &cur_elements->elements;
284 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
286 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
296 if ((e = strchr(p, '/')))
303 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
308 new_element = *ppl[level] =
309 (data1_element *)nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
310 new_element->next = new_element->children = 0;
311 new_element->tag = 0;
312 new_element->termlists = 0;
313 new_element->sub_name = 0;
315 tp = &new_element->termlists;
316 ppl[level] = &new_element->next;
317 ppl[level+1] = &new_element->children;
319 /* consider subtree (if any) ... */
320 if ((sub_p = strchr (p, ':')) && sub_p[1])
323 new_element->sub_name =
324 nmem_strdup (data1_nmem_get(dh), sub_p);
326 /* well-defined tag */
327 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
331 logf(LOG_WARN, "No tagset loaded in %s", file);
335 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
338 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
348 new_element->tag = (data1_tag *)nmem_malloc(data1_nmem_get (dh),
349 sizeof(*new_element->tag));
350 nt->which = DATA1T_string;
351 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
352 nt->names = (data1_name *)nmem_malloc(data1_nmem_get(dh),
353 sizeof(*new_element->tag->names));
354 nt->names->name = nt->value.string;
356 nt->kind = DATA1K_string;
362 logf(LOG_WARN, "Bad element is %s", file);
366 /* parse termList definitions */
369 new_element->termlists = 0;
374 logf(LOG_WARN, "No attset loaded in %s", file);
380 char attname[512], structure[512];
383 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
386 logf(LOG_WARN, "Syntax error in termlistspec in %s",
392 strcpy(attname, name);
393 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
395 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
398 logf(LOG_WARN, "Couldn't find att '%s' in attset",
403 if (r < 2) /* is the structure qualified? */
404 (*tp)->structure = "w";
407 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
408 strlen(structure)+1);
409 strcpy ((*tp)->structure, structure);
413 while ((p = strchr(p, ',')) && *(++p));
414 *tp = all; /* append any ALL entries to the list */
416 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
418 else if (!strcmp(cmd, "section"))
421 if (sscanf(args, "%511s", name) < 1)
423 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
427 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
428 sizeof(*cur_elements));
429 cur_elements->next = res->sub_elements;
430 cur_elements->elements = NULL;
431 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
432 res->sub_elements = cur_elements;
435 ppl[level] = &cur_elements->elements;
437 else if (!strcmp(cmd, "all"))
440 data1_termlist **tp = &all;
444 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
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",
469 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
470 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
473 logf(LOG_WARN, "Couldn't find att '%s' in attset",
478 if (r < 2) /* is the structure qualified? */
479 (*tp)->structure = "w";
482 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
483 strlen(structure)+1);
484 strcpy ((*tp)->structure, structure);
489 while ((p = strchr(p, ',')) && *(++p));
491 else if (!strcmp(cmd, "name"))
495 if (!sscanf(args, "%511s", name))
497 logf(LOG_WARN, "Malformed name directive in %s", file);
501 res->name = nmem_strdup(data1_nmem_get(dh), name);
503 else if (!strcmp(cmd, "reference"))
507 if (!sscanf(args, "%s", name))
509 logf(LOG_WARN, "Malformed reference in %s", file);
513 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
515 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
520 else if (!strcmp(cmd, "attset"))
524 if (!sscanf(args, "%s", name))
526 logf(LOG_WARN, "Malformed attset directive in %s", file);
530 if (!(res->attset = data1_read_attset (dh, name)))
532 logf(LOG_WARN, "Attset failed in %s", file);
537 else if (!strcmp(cmd, "tagset"))
541 if (!sscanf(args, "%s", name))
543 logf(LOG_WARN, "Malformed tagset directive in %s", file);
547 if (!(res->tagset = data1_read_tagset (dh, name)))
549 logf(LOG_WARN, "Tagset failed in %s", file);
554 else if (!strcmp(cmd, "varset"))
558 if (!sscanf(args, "%s", name))
560 logf(LOG_WARN, "Malformed varset directive in %s", file);
564 if (!(res->varset = data1_read_varset (dh, name)))
566 logf(LOG_WARN, "Varset failed in %s", file);
571 else if (!strcmp(cmd, "esetname"))
573 char name[512], fname[512];
575 if (sscanf(args, "%s %s", name, fname) != 2)
577 logf(LOG_WARN, "Two arg's required for esetname in %s",
582 *esetpp = (data1_esetname *)nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
583 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
587 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
589 logf(LOG_WARN, "%s: Espec-1 read failed", file);
593 esetpp = &(*esetpp)->next;
595 else if (!strcmp(cmd, "maptab"))
599 if (sscanf(args, "%s", name) != 1)
601 logf(LOG_WARN, "One argument for maptab directive in %s",
605 if (!(*maptabp = data1_read_maptab (dh, name)))
607 logf(LOG_WARN, "Failed to read maptab %s in %s",
611 maptabp = &(*maptabp)->next;
613 else if (!strcmp(cmd, "marc"))
617 if (sscanf(args, "%s", name) != 1)
619 logf(LOG_WARN, "One argument for marc directive in %s",
623 if (!(*marcp = data1_read_marctab (dh, name)))
625 logf(LOG_WARN, "%Failed to read marctab %s in %s",
629 marcp = &(*marcp)->next;
633 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
640 for (cur_elements = res->sub_elements; cur_elements;
641 cur_elements = cur_elements->next)
643 if (!strcmp (cur_elements->name, "main"))
644 res->main_elements = cur_elements->elements;
645 fix_element_ref (dh, res, cur_elements->elements);
647 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);