2 * Copyright (c) 1995-1997, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.17 1998-02-11 11:53:34 adam
8 * Changed code so that it compiles as C++.
10 * Revision 1.16 1997/12/18 10:51:30 adam
11 * Implemented sub-trees feature for schemas - including forward
14 * Revision 1.15 1997/12/09 16:18:16 adam
15 * Work on EXPLAIN schema. First implementation of sub-schema facility
18 * Revision 1.14 1997/10/31 12:20:09 adam
19 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
20 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
21 * Bug fix: missing fclose in data1_read_espec1.
23 * Revision 1.13 1997/10/27 13:54:18 adam
24 * Changed structure field in data1 node to be simple string which
25 * is "unknown" to the retrieval system itself.
27 * Revision 1.12 1997/09/17 12:10:34 adam
30 * Revision 1.11 1997/09/05 09:50:55 adam
31 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
33 * Revision 1.10 1997/05/14 06:54:01 adam
36 * Revision 1.9 1997/02/19 14:46:15 adam
37 * The "all" specifier only affects elements that are indexed (and not
40 * Revision 1.8 1997/01/02 10:47:59 quinn
41 * Added optional, physical ANY
43 * Revision 1.7 1996/06/10 08:56:01 quinn
46 * Revision 1.6 1996/05/31 13:52:21 quinn
47 * Fixed uninitialized variable for local tags in abstract syntax.
49 * Revision 1.5 1996/05/09 07:27:43 quinn
50 * Multiple local attributes values supported.
52 * Revision 1.4 1996/05/01 12:45:28 quinn
53 * Support use of local tag names in abs file.
55 * Revision 1.3 1995/11/01 16:34:55 quinn
56 * Making data1 look for tables in data1_tabpath
58 * Revision 1.2 1995/11/01 13:54:44 quinn
61 * Revision 1.1 1995/11/01 11:56:06 quinn
62 * Added Retrieval (data management) functions en masse.
79 #define D1_MAX_NESTING 128
81 struct data1_absyn_cache_info
85 data1_absyn_cache next;
88 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
90 data1_absyn_cache p = *data1_absyn_cache_get (dh);
94 if (!strcmp (name, p->name))
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);
111 p->name = nmem_strdup (mem, name);
117 data1_absyn *data1_get_absyn (data1_handle dh, char *name)
121 if (!(absyn = data1_absyn_search (dh, name)))
122 absyn = data1_absyn_add (dh, name);
126 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
131 for (r = a->esetnames; r; r = r->next)
132 if (!data1_matchstr(r->name, name))
137 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
138 data1_element *parent,
144 r = abs->main_elements;
146 r = parent->children;
147 assert (abs->main_elements);
148 for (; r; r = r->next)
152 for (n = r->tag->names; n; n = n->next)
153 if (!data1_matchstr(tagname, n->name))
159 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
163 assert (absyn->main_elements);
164 for (r = absyn->main_elements; r; r = r->next)
165 if (!data1_matchstr(r->name, name))
171 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
173 for (; e; e = e->next)
178 fix_element_ref (dh, absyn, e->children);
182 data1_sub_elements *sub_e = absyn->sub_elements;
183 while (sub_e && strcmp (e->sub_name, sub_e->name))
186 e->children = sub_e->elements;
188 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
194 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
196 char line[512], *r, cmd[512], args[512];
197 data1_sub_elements *cur_elements = NULL;
198 data1_absyn *res = 0;
200 data1_element **ppl[D1_MAX_NESTING];
201 data1_esetname **esetpp;
202 data1_maptab **maptabp;
203 data1_marctab **marcp;
204 data1_termlist *all = 0;
207 logf (LOG_DEBUG, "begin data1_read_absyn file=%s", file);
208 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
210 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
214 res = (data1_absyn *)nmem_malloc(data1_nmem_get(dh), sizeof(*res));
216 res->reference = VAL_NONE;
221 esetpp = &res->esetnames;
223 maptabp = &res->maptabs;
227 res->sub_elements = NULL;
228 res->main_elements = NULL;
232 while ((r = fgets(line, 512, f)))
234 while (*r && isspace(*r))
241 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
243 if (!strcmp(cmd, "elm"))
245 data1_element *new_element;
247 char path[512], name[512], termlists[512], *p, *sub_p;
253 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
254 sizeof(*cur_elements));
255 cur_elements->next = res->sub_elements;
256 cur_elements->elements = NULL;
257 cur_elements->name = "main";
258 res->sub_elements = cur_elements;
261 ppl[level] = &cur_elements->elements;
263 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
265 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
275 if ((e = strchr(p, '/')))
282 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
287 new_element = *ppl[level] =
288 (data1_element *)nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
289 new_element->next = new_element->children = 0;
290 new_element->tag = 0;
291 new_element->termlists = 0;
292 new_element->sub_name = 0;
294 tp = &new_element->termlists;
295 ppl[level] = &new_element->next;
296 ppl[level+1] = &new_element->children;
298 /* consider subtree (if any) ... */
299 if ((sub_p = strchr (p, ':')) && sub_p[1])
302 new_element->sub_name =
303 nmem_strdup (data1_nmem_get(dh), sub_p);
305 /* well-defined tag */
306 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
310 logf(LOG_WARN, "No tagset loaded in %s", file);
314 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
317 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
327 new_element->tag = (data1_tag *)nmem_malloc(data1_nmem_get (dh),
328 sizeof(*new_element->tag));
329 nt->which = DATA1T_string;
330 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
331 nt->names = (data1_name *)nmem_malloc(data1_nmem_get(dh),
332 sizeof(*new_element->tag->names));
333 nt->names->name = nt->value.string;
335 nt->kind = DATA1K_string;
341 logf(LOG_WARN, "Bad element is %s", file);
345 /* parse termList definitions */
348 new_element->termlists = 0;
353 logf(LOG_WARN, "No attset loaded in %s", file);
359 char attname[512], structure[512];
362 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
365 logf(LOG_WARN, "Syntax error in termlistspec in %s",
371 strcpy(attname, name);
372 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
374 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
377 logf(LOG_WARN, "Couldn't find att '%s' in attset",
382 if (r < 2) /* is the structure qualified? */
383 (*tp)->structure = "w";
386 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
387 strlen(structure)+1);
388 strcpy ((*tp)->structure, structure);
392 while ((p = strchr(p, ',')) && *(++p));
393 *tp = all; /* append any ALL entries to the list */
395 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
397 else if (!strcmp(cmd, "section"))
400 if (sscanf(args, "%511s", name) < 1)
402 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
406 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
407 sizeof(*cur_elements));
408 cur_elements->next = res->sub_elements;
409 cur_elements->elements = NULL;
410 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
411 res->sub_elements = cur_elements;
414 ppl[level] = &cur_elements->elements;
416 else if (!strcmp(cmd, "all"))
419 data1_termlist **tp = &all;
423 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
431 logf(LOG_WARN, "No attset loaded in %s", file);
437 char attname[512], structure[512];
440 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
443 logf(LOG_WARN, "Syntax error in termlistspec in %s",
448 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
449 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
452 logf(LOG_WARN, "Couldn't find att '%s' in attset",
457 if (r < 2) /* is the structure qualified? */
458 (*tp)->structure = "w";
461 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
462 strlen(structure)+1);
463 strcpy ((*tp)->structure, structure);
468 while ((p = strchr(p, ',')) && *(++p));
470 else if (!strcmp(cmd, "name"))
474 if (!sscanf(args, "%511s", name))
476 logf(LOG_WARN, "Malformed name directive in %s", file);
480 res->name = nmem_strdup(data1_nmem_get(dh), name);
482 else if (!strcmp(cmd, "reference"))
486 if (!sscanf(args, "%s", name))
488 logf(LOG_WARN, "Malformed reference in %s", file);
492 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
494 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
499 else if (!strcmp(cmd, "attset"))
503 if (!sscanf(args, "%s", name))
505 logf(LOG_WARN, "Malformed attset directive in %s", file);
509 if (!(res->attset = data1_read_attset (dh, name)))
511 logf(LOG_WARN, "Attset failed in %s", file);
516 else if (!strcmp(cmd, "tagset"))
520 if (!sscanf(args, "%s", name))
522 logf(LOG_WARN, "Malformed tagset directive in %s", file);
526 if (!(res->tagset = data1_read_tagset (dh, name)))
528 logf(LOG_WARN, "Tagset failed in %s", file);
533 else if (!strcmp(cmd, "varset"))
537 if (!sscanf(args, "%s", name))
539 logf(LOG_WARN, "Malformed varset directive in %s", file);
543 if (!(res->varset = data1_read_varset (dh, name)))
545 logf(LOG_WARN, "Varset failed in %s", file);
550 else if (!strcmp(cmd, "esetname"))
552 char name[512], fname[512];
554 if (sscanf(args, "%s %s", name, fname) != 2)
556 logf(LOG_WARN, "Two arg's required for esetname in %s",
561 *esetpp = (data1_esetname *)nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
562 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
566 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
568 logf(LOG_WARN, "%s: Espec-1 read failed", file);
572 esetpp = &(*esetpp)->next;
574 else if (!strcmp(cmd, "maptab"))
578 if (sscanf(args, "%s", name) != 1)
580 logf(LOG_WARN, "One argument for maptab directive in %s",
584 if (!(*maptabp = data1_read_maptab (dh, name)))
586 logf(LOG_WARN, "Failed to read maptab %s in %s",
590 maptabp = &(*maptabp)->next;
592 else if (!strcmp(cmd, "marc"))
596 if (sscanf(args, "%s", name) != 1)
598 logf(LOG_WARN, "One argument for marc directive in %s",
602 if (!(*marcp = data1_read_marctab (dh, name)))
604 logf(LOG_WARN, "%Failed to read marctab %s in %s",
608 marcp = &(*marcp)->next;
612 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
619 for (cur_elements = res->sub_elements; cur_elements;
620 cur_elements = cur_elements->next)
622 if (!strcmp (cur_elements->name, "main"))
623 res->main_elements = cur_elements->elements;
624 fix_element_ref (dh, res, cur_elements->elements);
626 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);