2 * Copyright (c) 1995-2000, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.31 2002-04-04 20:49:46 adam
8 * New functions yaz_is_abspath, yaz_path_fopen_base
10 * Revision 1.30 2000/12/05 19:07:24 adam
11 * Fixed problem with element level in reading of abstract syntax.
13 * Revision 1.29 2000/12/05 14:34:49 adam
14 * Fixed bug with termlists (introduced by previous commit).
16 * Revision 1.28 2000/12/05 12:21:45 adam
17 * Added termlist source for data1 system.
19 * Revision 1.27 1999/12/21 14:16:19 ian
20 * Changed retrieval module to allow data1 trees with no associated absyn.
21 * Also added a simple interface for extracting values from data1 trees using
22 * a string based tagpath.
24 * Revision 1.26 1999/11/30 13:47:12 adam
25 * Improved installation. Moved header files to include/yaz.
27 * Revision 1.25 1999/10/21 12:06:29 adam
28 * Retrieval module no longer uses ctype.h - functions.
30 * Revision 1.24 1999/08/27 09:40:32 adam
31 * Renamed logf function to yaz_log. Removed VC++ project files.
33 * Revision 1.23 1998/10/15 08:29:16 adam
34 * Tag set type may be specified in reference to it using "tagset"
35 * directive in .abs-files and "include" directive in .tag-files.
37 * Revision 1.22 1998/10/13 16:09:47 adam
38 * Added support for arbitrary OID's for tagsets, schemas and attribute sets.
39 * Added support for multiple attribute set references and tagset references
40 * from an abstract syntax file.
41 * Fixed many bad logs-calls in routines that read the various
42 * specifications regarding data1 (*.abs,*.att,...) and made the messages
43 * consistent whenever possible.
44 * Added extra 'lineno' argument to function readconf_line.
46 * Revision 1.21 1998/06/09 13:55:07 adam
49 * Revision 1.20 1998/05/18 13:07:02 adam
50 * Changed the way attribute sets are handled by the retriaval module.
51 * Extended Explain conversion / schema.
52 * Modified server and client to work with ASN.1 compiled protocol handlers.
54 * Revision 1.19 1998/03/05 08:15:32 adam
55 * Implemented data1_add_insert_taggeddata utility which is more flexible
56 * than data1_insert_taggeddata.
58 * Revision 1.18 1998/02/27 14:08:04 adam
59 * Added const to some char pointer arguments.
60 * Reworked data1_read_node so that it doesn't create a tree with
61 * pointers to original "SGML"-buffer.
63 * Revision 1.17 1998/02/11 11:53:34 adam
64 * Changed code so that it compiles as C++.
66 * Revision 1.16 1997/12/18 10:51:30 adam
67 * Implemented sub-trees feature for schemas - including forward
70 * Revision 1.15 1997/12/09 16:18:16 adam
71 * Work on EXPLAIN schema. First implementation of sub-schema facility
74 * Revision 1.14 1997/10/31 12:20:09 adam
75 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
76 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
77 * Bug fix: missing fclose in data1_read_espec1.
79 * Revision 1.13 1997/10/27 13:54:18 adam
80 * Changed structure field in data1 node to be simple string which
81 * is "unknown" to the retrieval system itself.
83 * Revision 1.12 1997/09/17 12:10:34 adam
86 * Revision 1.11 1997/09/05 09:50:55 adam
87 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
89 * Revision 1.10 1997/05/14 06:54:01 adam
92 * Revision 1.9 1997/02/19 14:46:15 adam
93 * The "all" specifier only affects elements that are indexed (and not
96 * Revision 1.8 1997/01/02 10:47:59 quinn
97 * Added optional, physical ANY
99 * Revision 1.7 1996/06/10 08:56:01 quinn
102 * Revision 1.6 1996/05/31 13:52:21 quinn
103 * Fixed uninitialized variable for local tags in abstract syntax.
105 * Revision 1.5 1996/05/09 07:27:43 quinn
106 * Multiple local attributes values supported.
108 * Revision 1.4 1996/05/01 12:45:28 quinn
109 * Support use of local tag names in abs file.
111 * Revision 1.3 1995/11/01 16:34:55 quinn
112 * Making data1 look for tables in data1_tabpath
114 * Revision 1.2 1995/11/01 13:54:44 quinn
117 * Revision 1.1 1995/11/01 11:56:06 quinn
118 * Added Retrieval (data management) functions en masse.
129 #include <yaz/data1.h>
131 #define D1_MAX_NESTING 128
133 struct data1_absyn_cache_info
137 data1_absyn_cache next;
140 struct data1_attset_cache_info
143 data1_attset *attset;
144 data1_attset_cache next;
147 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
149 data1_absyn_cache p = *data1_absyn_cache_get (dh);
153 if (!strcmp (name, p->name))
160 void data1_absyn_trav (data1_handle dh, void *handle,
161 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
163 data1_absyn_cache p = *data1_absyn_cache_get (dh);
167 (*fh)(dh, handle, p->absyn);
172 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
175 NMEM mem = data1_nmem_get (dh);
177 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
178 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
180 sprintf(fname, "%s.abs", name);
181 p->absyn = data1_read_absyn (dh, fname);
182 p->name = nmem_strdup (mem, name);
188 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
192 if (!(absyn = data1_absyn_search (dh, name)))
193 absyn = data1_absyn_add (dh, name);
197 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
199 data1_attset_cache p = *data1_attset_cache_get (dh);
203 if (!strcmp (name, p->name))
210 data1_attset *data1_attset_search_id (data1_handle dh, int id)
212 data1_attset_cache p = *data1_attset_cache_get (dh);
216 if (id == p->attset->reference)
223 data1_attset *data1_attset_add (data1_handle dh, const char *name)
225 char fname[512], aname[512];
226 NMEM mem = data1_nmem_get (dh);
227 data1_attset *attset;
229 strcpy (aname, name);
230 sprintf(fname, "%s.att", name);
231 attset = data1_read_attset (dh, fname);
235 attset = data1_read_attset (dh, name);
236 if (attset && (cp = strrchr (aname, '.')))
240 yaz_log (LOG_WARN|LOG_ERRNO, "Couldn't load attribute set %s", name);
243 data1_attset_cache p = (data1_attset_cache)
244 nmem_malloc (mem, sizeof(*p));
245 data1_attset_cache *pp = data1_attset_cache_get (dh);
247 attset->name = p->name = nmem_strdup (mem, aname);
255 data1_attset *data1_get_attset (data1_handle dh, const char *name)
257 data1_attset *attset;
259 if (!(attset = data1_attset_search_name (dh, name)))
260 attset = data1_attset_add (dh, name);
264 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
269 for (r = a->esetnames; r; r = r->next)
270 if (!data1_matchstr(r->name, name))
275 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
276 data1_element *parent,
281 /* It's now possible to have a data1 tree with no abstract syntax */
286 r = abs->main_elements;
288 r = parent->children;
289 assert (abs->main_elements);
290 for (; r; r = r->next)
294 for (n = r->tag->names; n; n = n->next)
295 if (!data1_matchstr(tagname, n->name))
301 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
306 /* It's now possible to have a data1 tree with no abstract syntax */
310 assert (absyn->main_elements);
311 for (r = absyn->main_elements; r; r = r->next)
312 if (!data1_matchstr(r->name, name))
318 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
320 /* It's now possible to have a data1 tree with no abstract syntax */
324 for (; e; e = e->next)
329 fix_element_ref (dh, absyn, e->children);
333 data1_sub_elements *sub_e = absyn->sub_elements;
334 while (sub_e && strcmp (e->sub_name, sub_e->name))
337 e->children = sub_e->elements;
339 yaz_log (LOG_WARN, "Unresolved reference to sub-elements %s",
346 static int parse_termlists (data1_handle dh, data1_termlist ***tpp,
347 char *p, const char *file, int lineno,
348 const char *element_name, data1_absyn *res)
350 data1_termlist **tp = *tpp;
353 char attname[512], structure[512];
357 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
361 "%s:%d: Syntax error in termlistspec '%s'",
370 strcpy(attname, element_name);
371 *tp = (data1_termlist *)
372 nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
374 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
378 "%s:%d: Couldn't find att '%s' in attset",
379 file, lineno, attname);
386 if (r == 2 && (source = strchr(structure, ':')))
387 *source++ = '\0'; /* cut off structure .. */
389 source = "data"; /* ok: default is leaf data */
390 (*tp)->source = (char *)
391 nmem_strdup (data1_nmem_get (dh), source);
393 if (r < 2) /* is the structure qualified? */
394 (*tp)->structure = "w";
396 (*tp)->structure = (char *)
397 nmem_strdup (data1_nmem_get (dh), structure);
400 while ((p = strchr(p, ',')) && *(++p));
405 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
407 data1_sub_elements *cur_elements = NULL;
408 data1_absyn *res = 0;
410 data1_element **ppl[D1_MAX_NESTING];
411 data1_esetname **esetpp;
412 data1_maptab **maptabp;
413 data1_marctab **marcp;
414 data1_termlist *all = 0;
415 data1_attset_child **attset_childp;
416 data1_tagset **tagset_childp;
420 char *argv[50], line[512];
422 if (!(f = data1_path_fopen(dh, file, "r")))
424 yaz_log(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
428 res = (data1_absyn *) nmem_malloc(data1_nmem_get(dh), sizeof(*res));
430 res->reference = VAL_NONE;
432 tagset_childp = &res->tagset;
434 res->attset = data1_empty_attset (dh);
435 attset_childp = &res->attset->children;
439 esetpp = &res->esetnames;
441 maptabp = &res->maptabs;
445 res->sub_elements = NULL;
446 res->main_elements = NULL;
448 while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
451 if (!strcmp(cmd, "elm"))
453 data1_element *new_element;
455 char *p, *sub_p, *path, *name, *termlists;
461 yaz_log(LOG_WARN, "%s:%d: Bad # of args to elm", file, lineno);
470 cur_elements = (data1_sub_elements *)
471 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
472 cur_elements->next = res->sub_elements;
473 cur_elements->elements = NULL;
474 cur_elements->name = "main";
475 res->sub_elements = cur_elements;
478 ppl[level] = &cur_elements->elements;
485 if ((e = strchr(p, '/')))
492 yaz_log(LOG_WARN, "%s:%d: Bad level increase", file, lineno);
497 new_element = *ppl[level-1] = (data1_element *)
498 nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
499 new_element->next = new_element->children = 0;
500 new_element->tag = 0;
501 new_element->termlists = 0;
502 new_element->sub_name = 0;
504 tp = &new_element->termlists;
505 ppl[level-1] = &new_element->next;
506 ppl[level] = &new_element->children;
508 /* consider subtree (if any) ... */
509 if ((sub_p = strchr (p, ':')) && sub_p[1])
512 new_element->sub_name =
513 nmem_strdup (data1_nmem_get(dh), sub_p);
515 /* well-defined tag */
516 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
520 yaz_log(LOG_WARN, "%s:%d: No tagset loaded", file, lineno);
524 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
527 yaz_log(LOG_WARN, "%s:%d: Couldn't find tag %s in tagset",
537 new_element->tag = (data1_tag *)
538 nmem_malloc(data1_nmem_get (dh),
539 sizeof(*new_element->tag));
540 nt->which = DATA1T_string;
541 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
542 nt->names = (data1_name *)
543 nmem_malloc(data1_nmem_get(dh),
544 sizeof(*new_element->tag->names));
545 nt->names->name = nt->value.string;
547 nt->kind = DATA1K_string;
553 yaz_log(LOG_WARN, "%s:%d: Bad element", file, lineno);
557 /* parse termList definitions */
561 assert (res->attset);
563 if (parse_termlists (dh, &tp, p, file, lineno, name, res))
568 *tp = all; /* append any ALL entries to the list */
570 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
572 else if (!strcmp(cmd, "section"))
578 yaz_log(LOG_WARN, "%s:%d: Bad # of args to section",
584 cur_elements = (data1_sub_elements *)
585 nmem_malloc(data1_nmem_get(dh), sizeof(*cur_elements));
586 cur_elements->next = res->sub_elements;
587 cur_elements->elements = NULL;
588 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
589 res->sub_elements = cur_elements;
592 ppl[level] = &cur_elements->elements;
594 else if (!strcmp(cmd, "all"))
596 data1_termlist **tp = &all;
599 yaz_log(LOG_WARN, "%s:%d: Too many 'all' directives - ignored",
605 yaz_log(LOG_WARN, "%s:%d: Bad # of args to 'all' directive",
609 if (parse_termlists (dh, &tp, argv[1], file, lineno, 0, res))
615 else if (!strcmp(cmd, "name"))
619 yaz_log(LOG_WARN, "%s:%d: Bad # of args to name directive",
623 res->name = nmem_strdup(data1_nmem_get(dh), argv[1]);
625 else if (!strcmp(cmd, "reference"))
631 yaz_log(LOG_WARN, "%s:%d: Bad # of args to reference",
636 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
638 yaz_log(LOG_WARN, "%s:%d: Unknown tagset ref '%s'",
643 else if (!strcmp(cmd, "attset"))
646 data1_attset *attset;
650 yaz_log(LOG_WARN, "%s:%d: Bad # of args to attset",
655 if (!(attset = data1_get_attset (dh, name)))
657 yaz_log(LOG_WARN, "%s:%d: Couldn't find attset %s",
661 *attset_childp = (data1_attset_child *)
662 nmem_malloc (data1_nmem_get(dh), sizeof(**attset_childp));
663 (*attset_childp)->child = attset;
664 (*attset_childp)->next = 0;
665 attset_childp = &(*attset_childp)->next;
667 else if (!strcmp(cmd, "tagset"))
673 yaz_log(LOG_WARN, "%s:%d: Bad # of args to tagset",
679 type = atoi(argv[2]);
680 *tagset_childp = data1_read_tagset (dh, name, type);
681 if (!(*tagset_childp))
683 yaz_log(LOG_WARN, "%s:%d: Couldn't load tagset %s",
687 tagset_childp = &(*tagset_childp)->next;
689 else if (!strcmp(cmd, "varset"))
695 yaz_log(LOG_WARN, "%s:%d: Bad # of args in varset",
700 if (!(res->varset = data1_read_varset (dh, name)))
702 yaz_log(LOG_WARN, "%s:%d: Couldn't load Varset %s",
707 else if (!strcmp(cmd, "esetname"))
713 yaz_log(LOG_WARN, "%s:%d: Bad # of args in esetname",
720 *esetpp = (data1_esetname *)
721 nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
722 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
726 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
728 yaz_log(LOG_WARN, "%s:%d: Espec-1 read failed for %s",
729 file, lineno, fname);
732 esetpp = &(*esetpp)->next;
734 else if (!strcmp(cmd, "maptab"))
740 yaz_log(LOG_WARN, "%s:%d: Bad # of args for maptab",
745 if (!(*maptabp = data1_read_maptab (dh, name)))
747 yaz_log(LOG_WARN, "%s:%d: Couldn't load maptab %s",
751 maptabp = &(*maptabp)->next;
753 else if (!strcmp(cmd, "marc"))
759 yaz_log(LOG_WARN, "%s:%d: Bad # or args for marc",
764 if (!(*marcp = data1_read_marctab (dh, name)))
766 yaz_log(LOG_WARN, "%s:%d: Couldn't read marctab %s",
770 marcp = &(*marcp)->next;
774 yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'", file, lineno, cmd);
780 for (cur_elements = res->sub_elements; cur_elements;
781 cur_elements = cur_elements->next)
783 if (!strcmp (cur_elements->name, "main"))
784 res->main_elements = cur_elements->elements;
785 fix_element_ref (dh, res, cur_elements->elements);
787 yaz_log (LOG_DEBUG, "%s: data1_read_absyn end", file);