2 * Copyright (c) 1995-1997, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.16 1997-12-18 10:51:30 adam
8 * Implemented sub-trees feature for schemas - including forward
11 * Revision 1.15 1997/12/09 16:18:16 adam
12 * Work on EXPLAIN schema. First implementation of sub-schema facility
15 * Revision 1.14 1997/10/31 12:20:09 adam
16 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
17 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
18 * Bug fix: missing fclose in data1_read_espec1.
20 * Revision 1.13 1997/10/27 13:54:18 adam
21 * Changed structure field in data1 node to be simple string which
22 * is "unknown" to the retrieval system itself.
24 * Revision 1.12 1997/09/17 12:10:34 adam
27 * Revision 1.11 1997/09/05 09:50:55 adam
28 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
30 * Revision 1.10 1997/05/14 06:54:01 adam
33 * Revision 1.9 1997/02/19 14:46:15 adam
34 * The "all" specifier only affects elements that are indexed (and not
37 * Revision 1.8 1997/01/02 10:47:59 quinn
38 * Added optional, physical ANY
40 * Revision 1.7 1996/06/10 08:56:01 quinn
43 * Revision 1.6 1996/05/31 13:52:21 quinn
44 * Fixed uninitialized variable for local tags in abstract syntax.
46 * Revision 1.5 1996/05/09 07:27:43 quinn
47 * Multiple local attributes values supported.
49 * Revision 1.4 1996/05/01 12:45:28 quinn
50 * Support use of local tag names in abs file.
52 * Revision 1.3 1995/11/01 16:34:55 quinn
53 * Making data1 look for tables in data1_tabpath
55 * Revision 1.2 1995/11/01 13:54:44 quinn
58 * Revision 1.1 1995/11/01 11:56:06 quinn
59 * Added Retrieval (data management) functions en masse.
76 #define D1_MAX_NESTING 128
78 struct data1_absyn_cache_info
82 data1_absyn_cache next;
85 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
87 data1_absyn_cache p = *data1_absyn_cache_get (dh);
91 if (!strcmp (name, p->name))
98 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
101 NMEM mem = data1_nmem_get (dh);
103 data1_absyn_cache p = nmem_malloc (mem, sizeof(*p));
104 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
106 sprintf(fname, "%s.abs", name);
107 p->absyn = data1_read_absyn (dh, fname);
108 p->name = nmem_strdup (mem, name);
114 data1_absyn *data1_get_absyn (data1_handle dh, char *name)
118 if (!(absyn = data1_absyn_search (dh, name)))
119 absyn = data1_absyn_add (dh, name);
123 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
128 for (r = a->esetnames; r; r = r->next)
129 if (!data1_matchstr(r->name, name))
134 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
135 data1_element *parent,
141 r = abs->main_elements;
143 r = parent->children;
144 assert (abs->main_elements);
145 for (; r; r = r->next)
149 for (n = r->tag->names; n; n = n->next)
150 if (!data1_matchstr(tagname, n->name))
156 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
160 assert (absyn->main_elements);
161 for (r = absyn->main_elements; r; r = r->next)
162 if (!data1_matchstr(r->name, name))
168 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
170 for (; e; e = e->next)
175 fix_element_ref (dh, absyn, e->children);
179 data1_sub_elements *sub_e = absyn->sub_elements;
180 while (sub_e && strcmp (e->sub_name, sub_e->name))
183 e->children = sub_e->elements;
185 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
191 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
193 char line[512], *r, cmd[512], args[512];
194 data1_sub_elements *cur_elements = NULL;
195 data1_absyn *res = 0;
197 data1_element **ppl[D1_MAX_NESTING];
198 data1_esetname **esetpp;
199 data1_maptab **maptabp;
200 data1_marctab **marcp;
201 data1_termlist *all = 0;
204 logf (LOG_DEBUG, "begin data1_read_absyn file=%s", file);
205 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
207 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
211 res = nmem_malloc(data1_nmem_get(dh), sizeof(*res));
213 res->reference = VAL_NONE;
218 esetpp = &res->esetnames;
220 maptabp = &res->maptabs;
224 res->sub_elements = NULL;
225 res->main_elements = NULL;
229 while ((r = fgets(line, 512, f)))
231 while (*r && isspace(*r))
238 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
240 if (!strcmp(cmd, "elm"))
242 data1_element *new_element;
244 char path[512], name[512], termlists[512], *p, *sub_p;
250 cur_elements = nmem_malloc(data1_nmem_get(dh),
251 sizeof(*cur_elements));
252 cur_elements->next = res->sub_elements;
253 cur_elements->elements = NULL;
254 cur_elements->name = "main";
255 res->sub_elements = cur_elements;
258 ppl[level] = &cur_elements->elements;
260 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
262 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
272 if ((e = strchr(p, '/')))
279 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
284 new_element = *ppl[level] =
285 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
286 new_element->next = new_element->children = 0;
287 new_element->tag = 0;
288 new_element->termlists = 0;
289 new_element->sub_name = 0;
291 tp = &new_element->termlists;
292 ppl[level] = &new_element->next;
293 ppl[level+1] = &new_element->children;
295 /* consider subtree (if any) ... */
296 if ((sub_p = strchr (p, ':')) && sub_p[1])
299 new_element->sub_name =
300 nmem_strdup (data1_nmem_get(dh), sub_p);
302 /* well-defined tag */
303 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
307 logf(LOG_WARN, "No tagset loaded in %s", file);
311 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
314 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
324 new_element->tag = nmem_malloc(data1_nmem_get (dh),
325 sizeof(*new_element->tag));
326 nt->which = DATA1T_string;
327 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
328 nt->names = nmem_malloc(data1_nmem_get(dh),
329 sizeof(*new_element->tag->names));
330 nt->names->name = nt->value.string;
332 nt->kind = DATA1K_string;
338 logf(LOG_WARN, "Bad element is %s", file);
342 /* parse termList definitions */
345 new_element->termlists = 0;
350 logf(LOG_WARN, "No attset loaded in %s", file);
356 char attname[512], structure[512];
359 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
362 logf(LOG_WARN, "Syntax error in termlistspec in %s",
368 strcpy(attname, name);
369 *tp = nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
371 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
374 logf(LOG_WARN, "Couldn't find att '%s' in attset",
379 if (r < 2) /* is the structure qualified? */
380 (*tp)->structure = "w";
383 (*tp)->structure = nmem_malloc (data1_nmem_get (dh),
384 strlen(structure)+1);
385 strcpy ((*tp)->structure, structure);
389 while ((p = strchr(p, ',')) && *(++p));
390 *tp = all; /* append any ALL entries to the list */
392 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
394 else if (!strcmp(cmd, "section"))
397 if (sscanf(args, "%511s", name) < 1)
399 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
403 cur_elements = nmem_malloc(data1_nmem_get(dh),
404 sizeof(*cur_elements));
405 cur_elements->next = res->sub_elements;
406 cur_elements->elements = NULL;
407 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
408 res->sub_elements = cur_elements;
411 ppl[level] = &cur_elements->elements;
413 else if (!strcmp(cmd, "all"))
416 data1_termlist **tp = &all;
420 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
428 logf(LOG_WARN, "No attset loaded in %s", file);
434 char attname[512], structure[512];
437 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
440 logf(LOG_WARN, "Syntax error in termlistspec in %s",
445 *tp = nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
446 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
449 logf(LOG_WARN, "Couldn't find att '%s' in attset",
454 if (r < 2) /* is the structure qualified? */
455 (*tp)->structure = "w";
458 (*tp)->structure = nmem_malloc (data1_nmem_get (dh),
459 strlen(structure)+1);
460 strcpy ((*tp)->structure, structure);
465 while ((p = strchr(p, ',')) && *(++p));
467 else if (!strcmp(cmd, "name"))
471 if (!sscanf(args, "%511s", name))
473 logf(LOG_WARN, "Malformed name directive in %s", file);
477 res->name = nmem_strdup(data1_nmem_get(dh), name);
479 else if (!strcmp(cmd, "reference"))
483 if (!sscanf(args, "%s", name))
485 logf(LOG_WARN, "Malformed reference in %s", file);
489 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
491 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
496 else if (!strcmp(cmd, "attset"))
500 if (!sscanf(args, "%s", name))
502 logf(LOG_WARN, "Malformed attset directive in %s", file);
506 if (!(res->attset = data1_read_attset (dh, name)))
508 logf(LOG_WARN, "Attset failed in %s", file);
513 else if (!strcmp(cmd, "tagset"))
517 if (!sscanf(args, "%s", name))
519 logf(LOG_WARN, "Malformed tagset directive in %s", file);
523 if (!(res->tagset = data1_read_tagset (dh, name)))
525 logf(LOG_WARN, "Tagset failed in %s", file);
530 else if (!strcmp(cmd, "varset"))
534 if (!sscanf(args, "%s", name))
536 logf(LOG_WARN, "Malformed varset directive in %s", file);
540 if (!(res->varset = data1_read_varset (dh, name)))
542 logf(LOG_WARN, "Varset failed in %s", file);
547 else if (!strcmp(cmd, "esetname"))
549 char name[512], fname[512];
551 if (sscanf(args, "%s %s", name, fname) != 2)
553 logf(LOG_WARN, "Two arg's required for esetname in %s",
558 *esetpp = nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
559 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
563 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
565 logf(LOG_WARN, "%s: Espec-1 read failed", file);
569 esetpp = &(*esetpp)->next;
571 else if (!strcmp(cmd, "maptab"))
575 if (sscanf(args, "%s", name) != 1)
577 logf(LOG_WARN, "One argument for maptab directive in %s",
581 if (!(*maptabp = data1_read_maptab (dh, name)))
583 logf(LOG_WARN, "Failed to read maptab %s in %s",
587 maptabp = &(*maptabp)->next;
589 else if (!strcmp(cmd, "marc"))
593 if (sscanf(args, "%s", name) != 1)
595 logf(LOG_WARN, "One argument for marc directive in %s",
599 if (!(*marcp = data1_read_marctab (dh, name)))
601 logf(LOG_WARN, "%Failed to read marctab %s in %s",
605 marcp = &(*marcp)->next;
609 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
616 for (cur_elements = res->sub_elements; cur_elements;
617 cur_elements = cur_elements->next)
619 if (!strcmp (cur_elements->name, "main"))
620 res->main_elements = cur_elements->elements;
621 fix_element_ref (dh, res, cur_elements->elements);
623 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);