2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.18 1998-02-27 14:08:04 adam
8 * Added const to some char pointer arguments.
9 * Reworked data1_read_node so that it doesn't create a tree with
10 * pointers to original "SGML"-buffer.
12 * Revision 1.17 1998/02/11 11:53:34 adam
13 * Changed code so that it compiles as C++.
15 * Revision 1.16 1997/12/18 10:51:30 adam
16 * Implemented sub-trees feature for schemas - including forward
19 * Revision 1.15 1997/12/09 16:18:16 adam
20 * Work on EXPLAIN schema. First implementation of sub-schema facility
23 * Revision 1.14 1997/10/31 12:20:09 adam
24 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
25 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
26 * Bug fix: missing fclose in data1_read_espec1.
28 * Revision 1.13 1997/10/27 13:54:18 adam
29 * Changed structure field in data1 node to be simple string which
30 * is "unknown" to the retrieval system itself.
32 * Revision 1.12 1997/09/17 12:10:34 adam
35 * Revision 1.11 1997/09/05 09:50:55 adam
36 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
38 * Revision 1.10 1997/05/14 06:54:01 adam
41 * Revision 1.9 1997/02/19 14:46:15 adam
42 * The "all" specifier only affects elements that are indexed (and not
45 * Revision 1.8 1997/01/02 10:47:59 quinn
46 * Added optional, physical ANY
48 * Revision 1.7 1996/06/10 08:56:01 quinn
51 * Revision 1.6 1996/05/31 13:52:21 quinn
52 * Fixed uninitialized variable for local tags in abstract syntax.
54 * Revision 1.5 1996/05/09 07:27:43 quinn
55 * Multiple local attributes values supported.
57 * Revision 1.4 1996/05/01 12:45:28 quinn
58 * Support use of local tag names in abs file.
60 * Revision 1.3 1995/11/01 16:34:55 quinn
61 * Making data1 look for tables in data1_tabpath
63 * Revision 1.2 1995/11/01 13:54:44 quinn
66 * Revision 1.1 1995/11/01 11:56:06 quinn
67 * Added Retrieval (data management) functions en masse.
84 #define D1_MAX_NESTING 128
86 struct data1_absyn_cache_info
90 data1_absyn_cache next;
93 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
95 data1_absyn_cache p = *data1_absyn_cache_get (dh);
99 if (!strcmp (name, p->name))
106 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
109 NMEM mem = data1_nmem_get (dh);
111 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
112 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
114 sprintf(fname, "%s.abs", name);
115 p->absyn = data1_read_absyn (dh, fname);
116 p->name = nmem_strdup (mem, name);
122 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
126 if (!(absyn = data1_absyn_search (dh, name)))
127 absyn = data1_absyn_add (dh, name);
131 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
136 for (r = a->esetnames; r; r = r->next)
137 if (!data1_matchstr(r->name, name))
142 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
143 data1_element *parent,
149 r = abs->main_elements;
151 r = parent->children;
152 assert (abs->main_elements);
153 for (; r; r = r->next)
157 for (n = r->tag->names; n; n = n->next)
158 if (!data1_matchstr(tagname, n->name))
164 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
168 assert (absyn->main_elements);
169 for (r = absyn->main_elements; r; r = r->next)
170 if (!data1_matchstr(r->name, name))
176 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
178 for (; e; e = e->next)
183 fix_element_ref (dh, absyn, e->children);
187 data1_sub_elements *sub_e = absyn->sub_elements;
188 while (sub_e && strcmp (e->sub_name, sub_e->name))
191 e->children = sub_e->elements;
193 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
199 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
201 char line[512], *r, cmd[512], args[512];
202 data1_sub_elements *cur_elements = NULL;
203 data1_absyn *res = 0;
205 data1_element **ppl[D1_MAX_NESTING];
206 data1_esetname **esetpp;
207 data1_maptab **maptabp;
208 data1_marctab **marcp;
209 data1_termlist *all = 0;
212 logf (LOG_DEBUG, "begin data1_read_absyn file=%s", file);
213 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
215 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
219 res = (data1_absyn *)nmem_malloc(data1_nmem_get(dh), sizeof(*res));
221 res->reference = VAL_NONE;
226 esetpp = &res->esetnames;
228 maptabp = &res->maptabs;
232 res->sub_elements = NULL;
233 res->main_elements = NULL;
237 while ((r = fgets(line, 512, f)))
239 while (*r && isspace(*r))
246 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
248 if (!strcmp(cmd, "elm"))
250 data1_element *new_element;
252 char path[512], name[512], termlists[512], *p, *sub_p;
258 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
259 sizeof(*cur_elements));
260 cur_elements->next = res->sub_elements;
261 cur_elements->elements = NULL;
262 cur_elements->name = "main";
263 res->sub_elements = cur_elements;
266 ppl[level] = &cur_elements->elements;
268 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
270 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
280 if ((e = strchr(p, '/')))
287 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
292 new_element = *ppl[level] =
293 (data1_element *)nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
294 new_element->next = new_element->children = 0;
295 new_element->tag = 0;
296 new_element->termlists = 0;
297 new_element->sub_name = 0;
299 tp = &new_element->termlists;
300 ppl[level] = &new_element->next;
301 ppl[level+1] = &new_element->children;
303 /* consider subtree (if any) ... */
304 if ((sub_p = strchr (p, ':')) && sub_p[1])
307 new_element->sub_name =
308 nmem_strdup (data1_nmem_get(dh), sub_p);
310 /* well-defined tag */
311 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
315 logf(LOG_WARN, "No tagset loaded in %s", file);
319 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
322 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
332 new_element->tag = (data1_tag *)nmem_malloc(data1_nmem_get (dh),
333 sizeof(*new_element->tag));
334 nt->which = DATA1T_string;
335 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
336 nt->names = (data1_name *)nmem_malloc(data1_nmem_get(dh),
337 sizeof(*new_element->tag->names));
338 nt->names->name = nt->value.string;
340 nt->kind = DATA1K_string;
346 logf(LOG_WARN, "Bad element is %s", file);
350 /* parse termList definitions */
353 new_element->termlists = 0;
358 logf(LOG_WARN, "No attset loaded in %s", file);
364 char attname[512], structure[512];
367 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
370 logf(LOG_WARN, "Syntax error in termlistspec in %s",
376 strcpy(attname, name);
377 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
379 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
382 logf(LOG_WARN, "Couldn't find att '%s' in attset",
387 if (r < 2) /* is the structure qualified? */
388 (*tp)->structure = "w";
391 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
392 strlen(structure)+1);
393 strcpy ((*tp)->structure, structure);
397 while ((p = strchr(p, ',')) && *(++p));
398 *tp = all; /* append any ALL entries to the list */
400 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
402 else if (!strcmp(cmd, "section"))
405 if (sscanf(args, "%511s", name) < 1)
407 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
411 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
412 sizeof(*cur_elements));
413 cur_elements->next = res->sub_elements;
414 cur_elements->elements = NULL;
415 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
416 res->sub_elements = cur_elements;
419 ppl[level] = &cur_elements->elements;
421 else if (!strcmp(cmd, "all"))
424 data1_termlist **tp = &all;
428 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
436 logf(LOG_WARN, "No attset loaded in %s", file);
442 char attname[512], structure[512];
445 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
448 logf(LOG_WARN, "Syntax error in termlistspec in %s",
453 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
454 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
457 logf(LOG_WARN, "Couldn't find att '%s' in attset",
462 if (r < 2) /* is the structure qualified? */
463 (*tp)->structure = "w";
466 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
467 strlen(structure)+1);
468 strcpy ((*tp)->structure, structure);
473 while ((p = strchr(p, ',')) && *(++p));
475 else if (!strcmp(cmd, "name"))
479 if (!sscanf(args, "%511s", name))
481 logf(LOG_WARN, "Malformed name directive in %s", file);
485 res->name = nmem_strdup(data1_nmem_get(dh), name);
487 else if (!strcmp(cmd, "reference"))
491 if (!sscanf(args, "%s", name))
493 logf(LOG_WARN, "Malformed reference in %s", file);
497 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
499 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
504 else if (!strcmp(cmd, "attset"))
508 if (!sscanf(args, "%s", name))
510 logf(LOG_WARN, "Malformed attset directive in %s", file);
514 if (!(res->attset = data1_read_attset (dh, name)))
516 logf(LOG_WARN, "Attset failed in %s", file);
521 else if (!strcmp(cmd, "tagset"))
525 if (!sscanf(args, "%s", name))
527 logf(LOG_WARN, "Malformed tagset directive in %s", file);
531 if (!(res->tagset = data1_read_tagset (dh, name)))
533 logf(LOG_WARN, "Tagset failed in %s", file);
538 else if (!strcmp(cmd, "varset"))
542 if (!sscanf(args, "%s", name))
544 logf(LOG_WARN, "Malformed varset directive in %s", file);
548 if (!(res->varset = data1_read_varset (dh, name)))
550 logf(LOG_WARN, "Varset failed in %s", file);
555 else if (!strcmp(cmd, "esetname"))
557 char name[512], fname[512];
559 if (sscanf(args, "%s %s", name, fname) != 2)
561 logf(LOG_WARN, "Two arg's required for esetname in %s",
566 *esetpp = (data1_esetname *)nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
567 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
571 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
573 logf(LOG_WARN, "%s: Espec-1 read failed", file);
577 esetpp = &(*esetpp)->next;
579 else if (!strcmp(cmd, "maptab"))
583 if (sscanf(args, "%s", name) != 1)
585 logf(LOG_WARN, "One argument for maptab directive in %s",
589 if (!(*maptabp = data1_read_maptab (dh, name)))
591 logf(LOG_WARN, "Failed to read maptab %s in %s",
595 maptabp = &(*maptabp)->next;
597 else if (!strcmp(cmd, "marc"))
601 if (sscanf(args, "%s", name) != 1)
603 logf(LOG_WARN, "One argument for marc directive in %s",
607 if (!(*marcp = data1_read_marctab (dh, name)))
609 logf(LOG_WARN, "%Failed to read marctab %s in %s",
613 marcp = &(*marcp)->next;
617 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
624 for (cur_elements = res->sub_elements; cur_elements;
625 cur_elements = cur_elements->next)
627 if (!strcmp (cur_elements->name, "main"))
628 res->main_elements = cur_elements->elements;
629 fix_element_ref (dh, res, cur_elements->elements);
631 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);