2 * Copyright (c) 1995-1999, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.25 1999-10-21 12:06:29 adam
8 * Retrieval module no longer uses ctype.h - functions.
10 * Revision 1.24 1999/08/27 09:40:32 adam
11 * Renamed logf function to yaz_log. Removed VC++ project files.
13 * Revision 1.23 1998/10/15 08:29:16 adam
14 * Tag set type may be specified in reference to it using "tagset"
15 * directive in .abs-files and "include" directive in .tag-files.
17 * Revision 1.22 1998/10/13 16:09:47 adam
18 * Added support for arbitrary OID's for tagsets, schemas and attribute sets.
19 * Added support for multiple attribute set references and tagset references
20 * from an abstract syntax file.
21 * Fixed many bad logs-calls in routines that read the various
22 * specifications regarding data1 (*.abs,*.att,...) and made the messages
23 * consistent whenever possible.
24 * Added extra 'lineno' argument to function readconf_line.
26 * Revision 1.21 1998/06/09 13:55:07 adam
29 * Revision 1.20 1998/05/18 13:07:02 adam
30 * Changed the way attribute sets are handled by the retriaval module.
31 * Extended Explain conversion / schema.
32 * Modified server and client to work with ASN.1 compiled protocol handlers.
34 * Revision 1.19 1998/03/05 08:15:32 adam
35 * Implemented data1_add_insert_taggeddata utility which is more flexible
36 * than data1_insert_taggeddata.
38 * Revision 1.18 1998/02/27 14:08:04 adam
39 * Added const to some char pointer arguments.
40 * Reworked data1_read_node so that it doesn't create a tree with
41 * pointers to original "SGML"-buffer.
43 * Revision 1.17 1998/02/11 11:53:34 adam
44 * Changed code so that it compiles as C++.
46 * Revision 1.16 1997/12/18 10:51:30 adam
47 * Implemented sub-trees feature for schemas - including forward
50 * Revision 1.15 1997/12/09 16:18:16 adam
51 * Work on EXPLAIN schema. First implementation of sub-schema facility
54 * Revision 1.14 1997/10/31 12:20:09 adam
55 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
56 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
57 * Bug fix: missing fclose in data1_read_espec1.
59 * Revision 1.13 1997/10/27 13:54:18 adam
60 * Changed structure field in data1 node to be simple string which
61 * is "unknown" to the retrieval system itself.
63 * Revision 1.12 1997/09/17 12:10:34 adam
66 * Revision 1.11 1997/09/05 09:50:55 adam
67 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
69 * Revision 1.10 1997/05/14 06:54:01 adam
72 * Revision 1.9 1997/02/19 14:46:15 adam
73 * The "all" specifier only affects elements that are indexed (and not
76 * Revision 1.8 1997/01/02 10:47:59 quinn
77 * Added optional, physical ANY
79 * Revision 1.7 1996/06/10 08:56:01 quinn
82 * Revision 1.6 1996/05/31 13:52:21 quinn
83 * Fixed uninitialized variable for local tags in abstract syntax.
85 * Revision 1.5 1996/05/09 07:27:43 quinn
86 * Multiple local attributes values supported.
88 * Revision 1.4 1996/05/01 12:45:28 quinn
89 * Support use of local tag names in abs file.
91 * Revision 1.3 1995/11/01 16:34:55 quinn
92 * Making data1 look for tables in data1_tabpath
94 * Revision 1.2 1995/11/01 13:54:44 quinn
97 * Revision 1.1 1995/11/01 11:56:06 quinn
98 * Added Retrieval (data management) functions en masse.
111 #define D1_MAX_NESTING 128
113 struct data1_absyn_cache_info
117 data1_absyn_cache next;
120 struct data1_attset_cache_info
123 data1_attset *attset;
124 data1_attset_cache next;
127 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
129 data1_absyn_cache p = *data1_absyn_cache_get (dh);
133 if (!strcmp (name, p->name))
140 void data1_absyn_trav (data1_handle dh, void *handle,
141 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
143 data1_absyn_cache p = *data1_absyn_cache_get (dh);
147 (*fh)(dh, handle, p->absyn);
152 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
155 NMEM mem = data1_nmem_get (dh);
157 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
158 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
160 sprintf(fname, "%s.abs", name);
161 p->absyn = data1_read_absyn (dh, fname);
162 p->name = nmem_strdup (mem, name);
168 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
172 if (!(absyn = data1_absyn_search (dh, name)))
173 absyn = data1_absyn_add (dh, name);
177 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
179 data1_attset_cache p = *data1_attset_cache_get (dh);
183 if (!strcmp (name, p->name))
190 data1_attset *data1_attset_search_id (data1_handle dh, int id)
192 data1_attset_cache p = *data1_attset_cache_get (dh);
196 if (id == p->attset->reference)
203 data1_attset *data1_attset_add (data1_handle dh, const char *name)
205 char fname[512], aname[512];
206 NMEM mem = data1_nmem_get (dh);
207 data1_attset *attset;
209 strcpy (aname, name);
210 sprintf(fname, "%s.att", name);
211 attset = data1_read_attset (dh, fname);
215 attset = data1_read_attset (dh, name);
216 if (attset && (cp = strrchr (aname, '.')))
220 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
223 data1_attset_cache p = (data1_attset_cache)
224 nmem_malloc (mem, sizeof(*p));
225 data1_attset_cache *pp = data1_attset_cache_get (dh);
227 attset->name = p->name = nmem_strdup (mem, aname);
235 data1_attset *data1_get_attset (data1_handle dh, const char *name)
237 data1_attset *attset;
239 if (!(attset = data1_attset_search_name (dh, name)))
240 attset = data1_attset_add (dh, name);
244 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
249 for (r = a->esetnames; r; r = r->next)
250 if (!data1_matchstr(r->name, name))
255 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
256 data1_element *parent,
262 r = abs->main_elements;
264 r = parent->children;
265 assert (abs->main_elements);
266 for (; r; r = r->next)
270 for (n = r->tag->names; n; n = n->next)
271 if (!data1_matchstr(tagname, n->name))
277 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
281 assert (absyn->main_elements);
282 for (r = absyn->main_elements; r; r = r->next)
283 if (!data1_matchstr(r->name, name))
289 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
291 for (; e; e = e->next)
296 fix_element_ref (dh, absyn, e->children);
300 data1_sub_elements *sub_e = absyn->sub_elements;
301 while (sub_e && strcmp (e->sub_name, sub_e->name))
304 e->children = sub_e->elements;
306 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
312 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
314 data1_sub_elements *cur_elements = NULL;
315 data1_absyn *res = 0;
317 data1_element **ppl[D1_MAX_NESTING];
318 data1_esetname **esetpp;
319 data1_maptab **maptabp;
320 data1_marctab **marcp;
321 data1_termlist *all = 0;
322 data1_attset_child **attset_childp;
323 data1_tagset **tagset_childp;
327 char *argv[50], line[512];
329 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
331 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
335 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
337 res->reference = VAL_NONE;
339 tagset_childp = &res->tagset;
341 res->attset = data1_empty_attset (dh);
342 attset_childp = &res->attset->children;
346 esetpp = &res->esetnames;
348 maptabp = &res->maptabs;
352 res->sub_elements = NULL;
353 res->main_elements = NULL;
355 while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
358 if (!strcmp(cmd, "elm"))
360 data1_element *new_element;
362 char *p, *sub_p, *path, *name, *termlists;
368 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
377 cur_elements = (data1_sub_elements *)
378 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
379 cur_elements->next = res->sub_elements;
380 cur_elements->elements = NULL;
381 cur_elements->name = "main";
382 res->sub_elements = cur_elements;
385 ppl[level] = &cur_elements->elements;
392 if ((e = strchr(p, '/')))
399 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
404 new_element = *ppl[level] = (data1_element *)
405 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
406 new_element->next = new_element->children = 0;
407 new_element->tag = 0;
408 new_element->termlists = 0;
409 new_element->sub_name = 0;
411 tp = &new_element->termlists;
412 ppl[level] = &new_element->next;
413 ppl[level+1] = &new_element->children;
415 /* consider subtree (if any) ... */
416 if ((sub_p = strchr (p, ':')) && sub_p[1])
419 new_element->sub_name =
420 nmem_strdup (data1_nmem_get(dh), sub_p);
422 /* well-defined tag */
423 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
427 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
431 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
434 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
444 new_element->tag = (data1_tag *)
445 nmem_malloc(data1_nmem_get (dh),
446 sizeof(*new_element->tag));
447 nt->which = DATA1T_string;
448 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
449 nt->names = (data1_name *)
450 nmem_malloc(data1_nmem_get(dh),
451 sizeof(*new_element->tag->names));
452 nt->names->name = nt->value.string;
454 nt->kind = DATA1K_string;
460 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
464 /* parse termList definitions */
467 new_element->termlists = 0;
470 assert (res->attset);
473 char attname[512], structure[512];
476 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
480 "%s:%d: Syntax error in termlistspec '%s'",
486 strcpy(attname, name);
487 *tp = (data1_termlist *)
488 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
490 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
494 "%s:%d: Couldn't find att '%s' in attset",
495 file, lineno, attname);
499 if (r < 2) /* is the structure qualified? */
500 (*tp)->structure = "w";
503 (*tp)->structure = (char *)
504 nmem_malloc (data1_nmem_get (dh),
505 strlen(structure)+1);
506 strcpy ((*tp)->structure, structure);
510 while ((p = strchr(p, ',')) && *(++p));
511 *tp = all; /* append any ALL entries to the list */
513 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
515 else if (!strcmp(cmd, "section"))
521 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
527 cur_elements = (data1_sub_elements *)
528 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
529 cur_elements->next = res->sub_elements;
530 cur_elements->elements = NULL;
531 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
532 res->sub_elements = cur_elements;
535 ppl[level] = &cur_elements->elements;
537 else if (!strcmp(cmd, "all"))
540 data1_termlist **tp = &all;
544 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
551 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
556 assert (res->attset);
559 char attname[512], structure[512];
562 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
565 yaz_log(LOG_WARN, "%s:%d: Syntax error in termlistspec",
570 *tp = (data1_termlist *)
571 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
573 data1_getattbyname (dh, res->attset, attname)))
575 yaz_log(LOG_WARN, "%s:%d: Couldn't find att '%s' in attset",
576 file, lineno, attname);
580 if (r < 2) /* is the structure qualified? */
581 (*tp)->structure = "w";
585 (char *)nmem_malloc (data1_nmem_get (dh),
586 strlen(structure)+1);
587 strcpy ((*tp)->structure, structure);
592 while ((p = strchr(p, ',')) && *(++p));
594 else if (!strcmp(cmd, "name"))
598 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
602 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
604 else if (!strcmp(cmd, "reference"))
610 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
615 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
617 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
622 else if (!strcmp(cmd, "attset"))
625 data1_attset *attset;
629 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
634 if (!(attset = data1_get_attset (dh, name)))
636 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
640 *attset_childp = (data1_attset_child *)
641 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
642 (*attset_childp)->child = attset;
643 (*attset_childp)->next = 0;
644 attset_childp = &(*attset_childp)->next;
646 else if (!strcmp(cmd, "tagset"))
652 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
658 type = atoi(argv[2]);
659 *tagset_childp = data1_read_tagset (dh, name, type);
660 if (!(*tagset_childp))
662 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
666 tagset_childp = &(*tagset_childp)->next;
668 else if (!strcmp(cmd, "varset"))
674 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
679 if (!(res->varset = data1_read_varset (dh, name)))
681 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
686 else if (!strcmp(cmd, "esetname"))
692 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
699 *esetpp = (data1_esetname *)
700 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
701 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
705 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
707 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
708 file, lineno, fname);
711 esetpp = &(*esetpp)->next;
713 else if (!strcmp(cmd, "maptab"))
719 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
724 if (!(*maptabp = data1_read_maptab (dh, name)))
726 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
730 maptabp = &(*maptabp)->next;
732 else if (!strcmp(cmd, "marc"))
738 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
743 if (!(*marcp = data1_read_marctab (dh, name)))
745 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
749 marcp = &(*marcp)->next;
753 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file, lineno, cmd);
759 for (cur_elements = res->sub_elements; cur_elements;
760 cur_elements = cur_elements->next)
762 if (!strcmp (cur_elements->name, "main"))
763 res->main_elements = cur_elements->elements;
764 fix_element_ref (dh, res, cur_elements->elements);
766 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);