2 * Copyright (c) 1995-1998, Index Data.
3 * See the file LICENSE for details.
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.21 1998-06-09 13:55:07 adam
10 * Revision 1.20 1998/05/18 13:07:02 adam
11 * Changed the way attribute sets are handled by the retriaval module.
12 * Extended Explain conversion / schema.
13 * Modified server and client to work with ASN.1 compiled protocol handlers.
15 * Revision 1.19 1998/03/05 08:15:32 adam
16 * Implemented data1_add_insert_taggeddata utility which is more flexible
17 * than data1_insert_taggeddata.
19 * Revision 1.18 1998/02/27 14:08:04 adam
20 * Added const to some char pointer arguments.
21 * Reworked data1_read_node so that it doesn't create a tree with
22 * pointers to original "SGML"-buffer.
24 * Revision 1.17 1998/02/11 11:53:34 adam
25 * Changed code so that it compiles as C++.
27 * Revision 1.16 1997/12/18 10:51:30 adam
28 * Implemented sub-trees feature for schemas - including forward
31 * Revision 1.15 1997/12/09 16:18:16 adam
32 * Work on EXPLAIN schema. First implementation of sub-schema facility
35 * Revision 1.14 1997/10/31 12:20:09 adam
36 * Improved memory debugging for xmalloc/nmem.c. References to NMEM
37 * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
38 * Bug fix: missing fclose in data1_read_espec1.
40 * Revision 1.13 1997/10/27 13:54:18 adam
41 * Changed structure field in data1 node to be simple string which
42 * is "unknown" to the retrieval system itself.
44 * Revision 1.12 1997/09/17 12:10:34 adam
47 * Revision 1.11 1997/09/05 09:50:55 adam
48 * Removed global data1_tabpath - uses data1_get_tabpath() instead.
50 * Revision 1.10 1997/05/14 06:54:01 adam
53 * Revision 1.9 1997/02/19 14:46:15 adam
54 * The "all" specifier only affects elements that are indexed (and not
57 * Revision 1.8 1997/01/02 10:47:59 quinn
58 * Added optional, physical ANY
60 * Revision 1.7 1996/06/10 08:56:01 quinn
63 * Revision 1.6 1996/05/31 13:52:21 quinn
64 * Fixed uninitialized variable for local tags in abstract syntax.
66 * Revision 1.5 1996/05/09 07:27:43 quinn
67 * Multiple local attributes values supported.
69 * Revision 1.4 1996/05/01 12:45:28 quinn
70 * Support use of local tag names in abs file.
72 * Revision 1.3 1995/11/01 16:34:55 quinn
73 * Making data1 look for tables in data1_tabpath
75 * Revision 1.2 1995/11/01 13:54:44 quinn
78 * Revision 1.1 1995/11/01 11:56:06 quinn
79 * Added Retrieval (data management) functions en masse.
96 #define D1_MAX_NESTING 128
98 struct data1_absyn_cache_info
102 data1_absyn_cache next;
105 struct data1_attset_cache_info
108 data1_attset *attset;
109 data1_attset_cache next;
112 data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
114 data1_absyn_cache p = *data1_absyn_cache_get (dh);
118 if (!strcmp (name, p->name))
125 void data1_absyn_trav (data1_handle dh, void *handle,
126 void (*fh)(data1_handle dh, void *h, data1_absyn *a))
128 data1_absyn_cache p = *data1_absyn_cache_get (dh);
132 (*fh)(dh, handle, p->absyn);
137 data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
140 NMEM mem = data1_nmem_get (dh);
142 data1_absyn_cache p = (data1_absyn_cache)nmem_malloc (mem, sizeof(*p));
143 data1_absyn_cache *pp = data1_absyn_cache_get (dh);
145 sprintf(fname, "%s.abs", name);
146 p->absyn = data1_read_absyn (dh, fname);
147 p->name = nmem_strdup (mem, name);
153 data1_absyn *data1_get_absyn (data1_handle dh, const char *name)
157 if (!(absyn = data1_absyn_search (dh, name)))
158 absyn = data1_absyn_add (dh, name);
162 data1_attset *data1_attset_search_name (data1_handle dh, const char *name)
164 data1_attset_cache p = *data1_attset_cache_get (dh);
168 if (!strcmp (name, p->name))
175 data1_attset *data1_attset_search_id (data1_handle dh, int id)
177 data1_attset_cache p = *data1_attset_cache_get (dh);
181 if (id == p->attset->reference)
188 data1_attset *data1_attset_add (data1_handle dh, const char *name)
190 char fname[512], aname[512];
191 NMEM mem = data1_nmem_get (dh);
192 data1_attset *attset;
194 strcpy (aname, name);
195 sprintf(fname, "%s.att", name);
196 attset = data1_read_attset (dh, fname);
200 attset = data1_read_attset (dh, name);
201 if (attset && (cp = strrchr (aname, '.')))
205 logf (LOG_WARN|LOG_ERRNO, "couldn't load attribute set %s", name);
208 data1_attset_cache p = (data1_attset_cache)
209 nmem_malloc (mem, sizeof(*p));
210 data1_attset_cache *pp = data1_attset_cache_get (dh);
212 attset->name = p->name = nmem_strdup (mem, aname);
220 data1_attset *data1_get_attset (data1_handle dh, const char *name)
222 data1_attset *attset;
224 if (!(attset = data1_attset_search_name (dh, name)))
225 attset = data1_attset_add (dh, name);
229 data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
234 for (r = a->esetnames; r; r = r->next)
235 if (!data1_matchstr(r->name, name))
240 data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
241 data1_element *parent,
247 r = abs->main_elements;
249 r = parent->children;
250 assert (abs->main_elements);
251 for (; r; r = r->next)
255 for (n = r->tag->names; n; n = n->next)
256 if (!data1_matchstr(tagname, n->name))
262 data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
266 assert (absyn->main_elements);
267 for (r = absyn->main_elements; r; r = r->next)
268 if (!data1_matchstr(r->name, name))
274 void fix_element_ref (data1_handle dh, data1_absyn *absyn, data1_element *e)
276 for (; e; e = e->next)
281 fix_element_ref (dh, absyn, e->children);
285 data1_sub_elements *sub_e = absyn->sub_elements;
286 while (sub_e && strcmp (e->sub_name, sub_e->name))
289 e->children = sub_e->elements;
291 logf (LOG_WARN, "Unresolved reference to sub-elements %s",
297 data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
299 char line[512], *r, cmd[512], args[512];
300 data1_sub_elements *cur_elements = NULL;
301 data1_absyn *res = 0;
303 data1_element **ppl[D1_MAX_NESTING];
304 data1_esetname **esetpp;
305 data1_maptab **maptabp;
306 data1_marctab **marcp;
307 data1_termlist *all = 0;
310 if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
312 logf(LOG_WARN|LOG_ERRNO, "Couldn't open %s", file);
316 res = (data1_absyn *)nmem_malloc(data1_nmem_get(dh), sizeof(*res));
318 res->reference = VAL_NONE;
323 esetpp = &res->esetnames;
325 maptabp = &res->maptabs;
329 res->sub_elements = NULL;
330 res->main_elements = NULL;
334 while ((r = fgets(line, 512, f)))
336 while (*r && isspace(*r))
343 if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
345 if (!strcmp(cmd, "elm"))
347 data1_element *new_element;
349 char path[512], name[512], termlists[512], *p, *sub_p;
355 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
356 sizeof(*cur_elements));
357 cur_elements->next = res->sub_elements;
358 cur_elements->elements = NULL;
359 cur_elements->name = "main";
360 res->sub_elements = cur_elements;
363 ppl[level] = &cur_elements->elements;
365 if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
367 logf(LOG_WARN, "Bad # of args to elm in %s: '%s'",
377 if ((e = strchr(p, '/')))
384 logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
389 new_element = *ppl[level] =
390 (data1_element *)nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
391 new_element->next = new_element->children = 0;
392 new_element->tag = 0;
393 new_element->termlists = 0;
394 new_element->sub_name = 0;
396 tp = &new_element->termlists;
397 ppl[level] = &new_element->next;
398 ppl[level+1] = &new_element->children;
400 /* consider subtree (if any) ... */
401 if ((sub_p = strchr (p, ':')) && sub_p[1])
404 new_element->sub_name =
405 nmem_strdup (data1_nmem_get(dh), sub_p);
407 /* well-defined tag */
408 if (sscanf(p, "(%d,%d)", &type, &value) == 2)
412 logf(LOG_WARN, "No tagset loaded in %s", file);
416 if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
419 logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
429 new_element->tag = (data1_tag *)nmem_malloc(data1_nmem_get (dh),
430 sizeof(*new_element->tag));
431 nt->which = DATA1T_string;
432 nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
433 nt->names = (data1_name *)nmem_malloc(data1_nmem_get(dh),
434 sizeof(*new_element->tag->names));
435 nt->names->name = nt->value.string;
437 nt->kind = DATA1K_string;
443 logf(LOG_WARN, "Bad element is %s", file);
447 /* parse termList definitions */
450 new_element->termlists = 0;
455 logf(LOG_WARN, "No attset loaded in %s", file);
461 char attname[512], structure[512];
464 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
467 logf(LOG_WARN, "Syntax error in termlistspec in %s",
473 strcpy(attname, name);
474 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
476 if (!((*tp)->att = data1_getattbyname(dh, res->attset,
479 logf(LOG_WARN, "Couldn't find att '%s' in attset",
484 if (r < 2) /* is the structure qualified? */
485 (*tp)->structure = "w";
488 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
489 strlen(structure)+1);
490 strcpy ((*tp)->structure, structure);
494 while ((p = strchr(p, ',')) && *(++p));
495 *tp = all; /* append any ALL entries to the list */
497 new_element->name = nmem_strdup(data1_nmem_get (dh), name);
499 else if (!strcmp(cmd, "section"))
502 if (sscanf(args, "%511s", name) < 1)
504 logf(LOG_WARN, "Bad # of args to sub in %s: '%s'",
508 cur_elements = (data1_sub_elements *)nmem_malloc(data1_nmem_get(dh),
509 sizeof(*cur_elements));
510 cur_elements->next = res->sub_elements;
511 cur_elements->elements = NULL;
512 cur_elements->name = nmem_strdup (data1_nmem_get(dh), name);
513 res->sub_elements = cur_elements;
516 ppl[level] = &cur_elements->elements;
518 else if (!strcmp(cmd, "all"))
521 data1_termlist **tp = &all;
525 logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
533 logf(LOG_WARN, "No attset loaded in %s", file);
539 char attname[512], structure[512];
542 if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
545 logf(LOG_WARN, "Syntax error in termlistspec in %s",
550 *tp = (data1_termlist *)nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
551 if (!((*tp)->att = data1_getattbyname (dh, res->attset,
554 logf(LOG_WARN, "Couldn't find att '%s' in attset",
559 if (r < 2) /* is the structure qualified? */
560 (*tp)->structure = "w";
563 (*tp)->structure = (char *)nmem_malloc (data1_nmem_get (dh),
564 strlen(structure)+1);
565 strcpy ((*tp)->structure, structure);
570 while ((p = strchr(p, ',')) && *(++p));
572 else if (!strcmp(cmd, "name"))
576 if (!sscanf(args, "%511s", name))
578 logf(LOG_WARN, "Malformed name directive in %s", file);
582 res->name = nmem_strdup(data1_nmem_get(dh), name);
584 else if (!strcmp(cmd, "reference"))
588 if (!sscanf(args, "%s", name))
590 logf(LOG_WARN, "Malformed reference in %s", file);
594 if ((res->reference = oid_getvalbyname(name)) == VAL_NONE)
596 logf(LOG_WARN, "Unknown tagset ref '%s' in %s", name, file);
601 else if (!strcmp(cmd, "attset"))
605 if (!sscanf(args, "%s", name))
607 logf(LOG_WARN, "Malformed attset directive in %s", file);
611 if (!(res->attset = data1_get_attset (dh, name)))
613 logf(LOG_WARN, "Attset failed in %s", file);
618 else if (!strcmp(cmd, "tagset"))
622 if (!sscanf(args, "%s", name))
624 logf(LOG_WARN, "Malformed tagset directive in %s", file);
628 if (!(res->tagset = data1_read_tagset (dh, name)))
630 logf(LOG_WARN, "Tagset failed in %s", file);
635 else if (!strcmp(cmd, "varset"))
639 if (!sscanf(args, "%s", name))
641 logf(LOG_WARN, "Malformed varset directive in %s", file);
645 if (!(res->varset = data1_read_varset (dh, name)))
647 logf(LOG_WARN, "Varset failed in %s", file);
652 else if (!strcmp(cmd, "esetname"))
654 char name[512], fname[512];
656 if (sscanf(args, "%s %s", name, fname) != 2)
658 logf(LOG_WARN, "Two arg's required for esetname in %s",
663 *esetpp = (data1_esetname *)nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
664 (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
668 else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
670 logf(LOG_WARN, "%s: Espec-1 read failed", file);
674 esetpp = &(*esetpp)->next;
676 else if (!strcmp(cmd, "maptab"))
680 if (sscanf(args, "%s", name) != 1)
682 logf(LOG_WARN, "One argument for maptab directive in %s",
686 if (!(*maptabp = data1_read_maptab (dh, name)))
688 logf(LOG_WARN, "Failed to read maptab %s in %s",
692 maptabp = &(*maptabp)->next;
694 else if (!strcmp(cmd, "marc"))
698 if (sscanf(args, "%s", name) != 1)
700 logf(LOG_WARN, "One argument for marc directive in %s",
704 if (!(*marcp = data1_read_marctab (dh, name)))
706 logf(LOG_WARN, "%Failed to read marctab %s in %s",
710 marcp = &(*marcp)->next;
714 logf(LOG_WARN, "Unknown directive '%s' in %s", cmd, file);
721 for (cur_elements = res->sub_elements; cur_elements;
722 cur_elements = cur_elements->next)
724 if (!strcmp (cur_elements->name, "main"))
725 res->main_elements = cur_elements->elements;
726 fix_element_ref (dh, res, cur_elements->elements);
728 logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);