2 * Copyright (c) 1995-1999, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.26 1999-11-30 13:47:12 adam
8 * Improved installation. Moved header files to include/yaz.
10 * Revision 1.25 1999/10/21 12:06:29 adam
11 * Retrieval module no longer uses ctype.h - functions.
13 * Revision 1.24 1999/08/27 09:40:32 adam
14 * Renamed logf function to yaz_log. Removed VC++ project files.
16 * Revision 1.23 1998/10/15 08:29:16 adam
17 * Tag set type may be specified in reference to it using "tagset"
18 * directive in .abs-files and "include" directive in .tag-files.
20 * Revision 1.22 1998/10/13 16:09:47 adam
21 * Added support for arbitrary OID's for tagsets, schemas and attribute sets.
22 * Added support for multiple attribute set references and tagset references
23 * from an abstract syntax file.
24 * Fixed many bad logs-calls in routines that read the various
25 * specifications regarding data1 (*.abs,*.att,...) and made the messages
26 * consistent whenever possible.
27 * Added extra 'lineno' argument to function readconf_line.
29 * Revision 1.21 1998/06/09 13:55:07 adam
32 * Revision 1.20 1998/05/18 13:07:02 adam
33 * Changed the way attribute sets are handled by the retriaval module.
34 * Extended Explain conversion / schema.
35 * Modified server and client to work with ASN.1 compiled protocol handlers.
37 * Revision 1.19 1998/03/05 08:15:32 adam
38 * Implemented data1_add_insert_taggeddata utility which is more flexible
39 * than data1_insert_taggeddata.
41 * Revision 1.18 1998/02/27 14:08:04 adam
42 * Added const to some char pointer arguments.
43 * Reworked data1_read_node so that it doesn't create a tree with
44 * pointers to original "SGML"-buffer.
46 * Revision 1.17 1998/02/11 11:53:34 adam
47 * Changed code so that it compiles as C++.
49 * Revision 1.16 1997/12/18 10:51:30 adam
50 * Implemented sub-trees feature for schemas - including forward
53 * Revision 1.15 1997/12/09 16:18:16 adam
54 * Work on EXPLAIN schema. First implementation of sub-schema facility
57 * Revision 1.14 1997/10/31 12:20:09 adam
58 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
59 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
60 * Bug fix: missing fclose in data1_read_espec1.
62 * Revision 1.13 1997/10/27 13:54:18 adam
63 * Changed structure field in data1 node to be simple string which
64 * is "unknown" to the retrieval system itself.
66 * Revision 1.12 1997/09/17 12:10:34 adam
69 * Revision 1.11 1997/09/05 09:50:55 adam
70 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
72 * Revision 1.10 1997/05/14 06:54:01 adam
75 * Revision 1.9 1997/02/19 14:46:15 adam
76 * The "all" specifier only affects elements that are indexed (and not
79 * Revision 1.8 1997/01/02 10:47:59 quinn
80 * Added optional, physical ANY
82 * Revision 1.7 1996/06/10 08:56:01 quinn
85 * Revision 1.6 1996/05/31 13:52:21 quinn
86 * Fixed uninitialized variable for local tags in abstract syntax.
88 * Revision 1.5 1996/05/09 07:27:43 quinn
89 * Multiple local attributes values supported.
91 * Revision 1.4 1996/05/01 12:45:28 quinn
92 * Support use of local tag names in abs file.
94 * Revision 1.3 1995/11/01 16:34:55 quinn
95 * Making data1 look for tables in data1_tabpath
97 * Revision 1.2 1995/11/01 13:54:44 quinn
100 * Revision 1.1 1995/11/01 11:56:06 quinn
101 * Added Retrieval (data management) functions en masse.
112 #include <yaz/data1.h>
114 #define D1_MAX_NESTING 128
116 struct data1_absyn_cache_info
120 data1_absyn_cache next;
123 struct data1_attset_cache_info
126 data1_attset *attset;
127 data1_attset_cache next;
130 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
132 data1_absyn_cache p = *data1_absyn_cache_get (dh);
136 if (!strcmp (name, p->name))
143 void data1_absyn_trav (data1_handle dh, void *handle,
144 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
146 data1_absyn_cache p = *data1_absyn_cache_get (dh);
150 (*fh)(dh, handle, p->absyn);
155 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
158 NMEM mem = data1_nmem_get (dh);
160 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
161 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
163 sprintf(fname, "%s.abs", name);
164 p->absyn = data1_read_absyn (dh, fname);
165 p->name = nmem_strdup (mem, name);
171 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
175 if (!(absyn = data1_absyn_search (dh, name)))
176 absyn = data1_absyn_add (dh, name);
180 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
182 data1_attset_cache p = *data1_attset_cache_get (dh);
186 if (!strcmp (name, p->name))
193 data1_attset *data1_attset_search_id (data1_handle dh, int id)
195 data1_attset_cache p = *data1_attset_cache_get (dh);
199 if (id == p->attset->reference)
206 data1_attset *data1_attset_add (data1_handle dh, const char *name)
208 char fname[512], aname[512];
209 NMEM mem = data1_nmem_get (dh);
210 data1_attset *attset;
212 strcpy (aname, name);
213 sprintf(fname, "%s.att", name);
214 attset = data1_read_attset (dh, fname);
218 attset = data1_read_attset (dh, name);
219 if (attset && (cp = strrchr (aname, '.')))
223 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
226 data1_attset_cache p = (data1_attset_cache)
227 nmem_malloc (mem, sizeof(*p));
228 data1_attset_cache *pp = data1_attset_cache_get (dh);
230 attset->name = p->name = nmem_strdup (mem, aname);
238 data1_attset *data1_get_attset (data1_handle dh, const char *name)
240 data1_attset *attset;
242 if (!(attset = data1_attset_search_name (dh, name)))
243 attset = data1_attset_add (dh, name);
247 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
252 for (r = a->esetnames; r; r = r->next)
253 if (!data1_matchstr(r->name, name))
258 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
259 data1_element *parent,
265 r = abs->main_elements;
267 r = parent->children;
268 assert (abs->main_elements);
269 for (; r; r = r->next)
273 for (n = r->tag->names; n; n = n->next)
274 if (!data1_matchstr(tagname, n->name))
280 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
284 assert (absyn->main_elements);
285 for (r = absyn->main_elements; r; r = r->next)
286 if (!data1_matchstr(r->name, name))
292 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
294 for (; e; e = e->next)
299 fix_element_ref (dh, absyn, e->children);
303 data1_sub_elements *sub_e = absyn->sub_elements;
304 while (sub_e && strcmp (e->sub_name, sub_e->name))
307 e->children = sub_e->elements;
309 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
315 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
317 data1_sub_elements *cur_elements = NULL;
318 data1_absyn *res = 0;
320 data1_element **ppl[D1_MAX_NESTING];
321 data1_esetname **esetpp;
322 data1_maptab **maptabp;
323 data1_marctab **marcp;
324 data1_termlist *all = 0;
325 data1_attset_child **attset_childp;
326 data1_tagset **tagset_childp;
330 char *argv[50], line[512];
332 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
334 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
338 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
340 res->reference = VAL_NONE;
342 tagset_childp = &res->tagset;
344 res->attset = data1_empty_attset (dh);
345 attset_childp = &res->attset->children;
349 esetpp = &res->esetnames;
351 maptabp = &res->maptabs;
355 res->sub_elements = NULL;
356 res->main_elements = NULL;
358 while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
361 if (!strcmp(cmd, "elm"))
363 data1_element *new_element;
365 char *p, *sub_p, *path, *name, *termlists;
371 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
380 cur_elements = (data1_sub_elements *)
381 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
382 cur_elements->next = res->sub_elements;
383 cur_elements->elements = NULL;
384 cur_elements->name = "main";
385 res->sub_elements = cur_elements;
388 ppl[level] = &cur_elements->elements;
395 if ((e = strchr(p, '/')))
402 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
407 new_element = *ppl[level] = (data1_element *)
408 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
409 new_element->next = new_element->children = 0;
410 new_element->tag = 0;
411 new_element->termlists = 0;
412 new_element->sub_name = 0;
414 tp = &new_element->termlists;
415 ppl[level] = &new_element->next;
416 ppl[level+1] = &new_element->children;
418 /* consider subtree (if any) ... */
419 if ((sub_p = strchr (p, ':')) && sub_p[1])
422 new_element->sub_name =
423 nmem_strdup (data1_nmem_get(dh), sub_p);
425 /* well-defined tag */
426 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
430 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
434 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
437 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
447 new_element->tag = (data1_tag *)
448 nmem_malloc(data1_nmem_get (dh),
449 sizeof(*new_element->tag));
450 nt->which = DATA1T_string;
451 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
452 nt->names = (data1_name *)
453 nmem_malloc(data1_nmem_get(dh),
454 sizeof(*new_element->tag->names));
455 nt->names->name = nt->value.string;
457 nt->kind = DATA1K_string;
463 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
467 /* parse termList definitions */
470 new_element->termlists = 0;
473 assert (res->attset);
476 char attname[512], structure[512];
479 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
483 "%s:%d: Syntax error in termlistspec '%s'",
489 strcpy(attname, name);
490 *tp = (data1_termlist *)
491 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
493 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
497 "%s:%d: Couldn't find att '%s' in attset",
498 file, lineno, attname);
502 if (r < 2) /* is the structure qualified? */
503 (*tp)->structure = "w";
506 (*tp)->structure = (char *)
507 nmem_malloc (data1_nmem_get (dh),
508 strlen(structure)+1);
509 strcpy ((*tp)->structure, structure);
513 while ((p = strchr(p, ',')) && *(++p));
514 *tp = all; /* append any ALL entries to the list */
516 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
518 else if (!strcmp(cmd, "section"))
524 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
530 cur_elements = (data1_sub_elements *)
531 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
532 cur_elements->next = res->sub_elements;
533 cur_elements->elements = NULL;
534 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
535 res->sub_elements = cur_elements;
538 ppl[level] = &cur_elements->elements;
540 else if (!strcmp(cmd, "all"))
543 data1_termlist **tp = &all;
547 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
554 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
559 assert (res->attset);
562 char attname[512], structure[512];
565 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
568 yaz_log(LOG_WARN, "%s:%d: Syntax error in termlistspec",
573 *tp = (data1_termlist *)
574 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
576 data1_getattbyname (dh, res->attset, attname)))
578 yaz_log(LOG_WARN, "%s:%d: Couldn't find att '%s' in attset",
579 file, lineno, attname);
583 if (r < 2) /* is the structure qualified? */
584 (*tp)->structure = "w";
588 (char *)nmem_malloc (data1_nmem_get (dh),
589 strlen(structure)+1);
590 strcpy ((*tp)->structure, structure);
595 while ((p = strchr(p, ',')) && *(++p));
597 else if (!strcmp(cmd, "name"))
601 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
605 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
607 else if (!strcmp(cmd, "reference"))
613 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
618 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
620 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
625 else if (!strcmp(cmd, "attset"))
628 data1_attset *attset;
632 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
637 if (!(attset = data1_get_attset (dh, name)))
639 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
643 *attset_childp = (data1_attset_child *)
644 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
645 (*attset_childp)->child = attset;
646 (*attset_childp)->next = 0;
647 attset_childp = &(*attset_childp)->next;
649 else if (!strcmp(cmd, "tagset"))
655 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
661 type = atoi(argv[2]);
662 *tagset_childp = data1_read_tagset (dh, name, type);
663 if (!(*tagset_childp))
665 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
669 tagset_childp = &(*tagset_childp)->next;
671 else if (!strcmp(cmd, "varset"))
677 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
682 if (!(res->varset = data1_read_varset (dh, name)))
684 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
689 else if (!strcmp(cmd, "esetname"))
695 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
702 *esetpp = (data1_esetname *)
703 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
704 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
708 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
710 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
711 file, lineno, fname);
714 esetpp = &(*esetpp)->next;
716 else if (!strcmp(cmd, "maptab"))
722 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
727 if (!(*maptabp = data1_read_maptab (dh, name)))
729 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
733 maptabp = &(*maptabp)->next;
735 else if (!strcmp(cmd, "marc"))
741 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
746 if (!(*marcp = data1_read_marctab (dh, name)))
748 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
752 marcp = &(*marcp)->next;
756 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file, lineno, cmd);
762 for (cur_elements = res->sub_elements; cur_elements;
763 cur_elements = cur_elements->next)
765 if (!strcmp (cur_elements->name, "main"))
766 res->main_elements = cur_elements->elements;
767 fix_element_ref (dh, res, cur_elements->elements);
769 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);