2 * Copyright (C) 1994-2000, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.36 2000-12-05 10:01:44 adam
8 * Fixed bug regarding user-defined attribute sets.
10 * Revision 1.35 2000/11/29 15:21:31 adam
11 * Fixed problem with passwd db.
13 * Revision 1.34 2000/02/25 13:24:49 adam
14 * Fixed bug regarding pointer conversion that showed up on OSF V5.
16 * Revision 1.33 1999/11/30 13:48:04 adam
17 * Improved installation. Updated for inclusion of YAZ header files.
19 * Revision 1.32 1999/09/07 07:19:21 adam
20 * Work on character mapping. Implemented replace rules.
22 * Revision 1.31 1999/07/14 10:56:43 adam
23 * Fixed potential memory leak.
25 * Revision 1.30 1999/07/06 12:26:41 adam
26 * Retrieval handler obeys schema and handles XML transfer syntax.
28 * Revision 1.29 1999/05/26 07:49:14 adam
31 * Revision 1.28 1999/05/21 12:00:17 adam
32 * Better diagnostics for extraction process.
34 * Revision 1.27 1999/05/20 12:57:18 adam
35 * Implemented TCL filter. Updated recctrl system.
37 * Revision 1.26 1999/03/02 16:15:44 quinn
38 * Added "tagsysno" and "tagrank" directives to zebra.cfg.
40 * Revision 1.25 1999/02/18 15:01:26 adam
43 * Revision 1.24 1999/02/02 14:51:28 adam
44 * Updated WIN32 code specific sections. Changed header.
46 * Revision 1.23 1998/10/18 07:51:10 adam
47 * Changed one logf call.
49 * Revision 1.22 1998/10/16 08:14:37 adam
50 * Updated record control system.
52 * Revision 1.21 1998/07/01 09:16:10 adam
53 * Element localno only added when it's greater than 0.
55 * Revision 1.20 1998/05/20 10:12:26 adam
56 * Implemented automatic EXPLAIN database maintenance.
57 * Modified Zebra to work with ASN.1 compiled version of YAZ.
59 * Revision 1.19 1998/03/11 11:19:05 adam
60 * Changed the way sequence numbers are generated.
62 * Revision 1.18 1998/03/05 08:41:31 adam
65 * Revision 1.17 1998/02/10 12:03:06 adam
68 * Revision 1.16 1998/01/29 13:38:17 adam
69 * Fixed problem with mapping to record with unknown schema.
71 * Revision 1.15 1998/01/26 10:37:57 adam
74 * Revision 1.14 1997/11/06 11:41:01 adam
75 * Implemented "begin variant" for the sgml.regx filter.
77 * Revision 1.13 1997/10/31 12:35:44 adam
78 * Added a few log statements.
80 * Revision 1.12 1997/10/29 12:02:22 adam
81 * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent.
83 * Revision 1.11 1997/10/27 14:34:00 adam
84 * Work on generic character mapping depending on "structure" field
85 * in abstract syntax file.
87 * Revision 1.10 1997/09/18 08:59:21 adam
88 * Extra generic handle for the character mapping routines.
90 * Revision 1.9 1997/09/17 12:19:21 adam
91 * Zebra version corresponds to YAZ version 1.4.
92 * Changed Zebra server so that it doesn't depend on global common_resource.
94 * Revision 1.8 1997/09/09 13:38:14 adam
95 * Partial port to WIN95/NT.
97 * Revision 1.7 1997/09/05 15:30:10 adam
98 * Changed prototype for chr_map_input - added const.
99 * Added support for C++, headers uses extern "C" for public definitions.
101 * Revision 1.6 1997/09/04 13:54:40 adam
102 * Added MARC filter - type grs.marc.<syntax> where syntax refers
103 * to abstract syntax. New method tellf in retrieve/extract method.
105 * Revision 1.5 1997/07/15 16:29:03 adam
106 * Initialized dummy variable to keep checker gcc happy.
108 * Revision 1.4 1997/04/30 08:56:08 quinn
111 * Revision 1.2 1996/10/11 16:06:43 quinn
112 * Revision 1.3 1997/02/24 10:41:50 adam
113 * Cleanup of code and commented out the "end element-end-record" code.
115 * Revision 1.2 1996/10/11 16:06:43 quinn
116 * Fixed arguments to nodetogr
118 * Revision 1.1 1996/10/11 10:57:25 adam
119 * New module recctrl. Used to manage records (extract/retrieval).
121 * Revision 1.29 1996/10/08 10:30:21 quinn
122 * Fixed type mismatch
124 * Revision 1.28 1996/10/07 16:06:40 quinn
127 * Revision 1.27 1996/06/11 10:54:12 quinn
130 * Revision 1.26 1996/06/06 12:08:45 quinn
131 * Added showRecord function
133 * Revision 1.25 1996/06/04 14:18:53 quinn
136 * Revision 1.24 1996/06/04 13:27:54 quinn
137 * More work on charmapping
139 * Revision 1.23 1996/06/04 10:19:01 adam
140 * Minor changes - removed include of ctype.h.
142 * Revision 1.22 1996/06/03 10:15:27 quinn
143 * Various character-mapping.
145 * Revision 1.21 1996/05/31 13:27:24 quinn
146 * Character-conversion in phrases, too.
148 * Revision 1.19 1996/05/16 15:31:14 quinn
151 * Revision 1.18 1996/05/09 07:28:56 quinn
152 * Work towards phrases and multiple registers
154 * Revision 1.17 1996/05/01 13:46:37 adam
155 * First work on multiple records in one file.
156 * New option, -offset, to the "unread" command in the filter module.
158 * Revision 1.16 1996/01/17 14:57:54 adam
159 * Prototype changed for reader functions in extract/retrieve. File
160 * is identified by 'void *' instead of 'int.
162 * Revision 1.15 1996/01/08 19:15:47 adam
163 * New input filter that works!
165 * Revision 1.14 1995/12/15 12:36:11 adam
166 * Retrieval calls data1_read_regx when subType is specified.
168 * Revision 1.13 1995/12/15 12:24:43 quinn
169 * *** empty log message ***
171 * Revision 1.12 1995/12/15 12:20:28 quinn
172 * *** empty log message ***
174 * Revision 1.11 1995/12/15 12:07:57 quinn
175 * Changed extraction strategy.
177 * Revision 1.10 1995/12/14 11:10:48 quinn
180 * Revision 1.9 1995/12/13 17:14:05 quinn
181 * *** empty log message ***
183 * Revision 1.8 1995/12/13 15:33:18 quinn
184 * *** empty log message ***
186 * Revision 1.7 1995/12/13 13:45:39 quinn
187 * Changed data1 to use nmem.
189 * Revision 1.6 1995/12/04 14:22:30 adam
190 * Extra arg to recType_byName.
191 * Started work on new regular expression parsed input to
192 * structured records.
194 * Revision 1.5 1995/11/28 14:18:37 quinn
197 * Revision 1.4 1995/11/21 13:14:49 quinn
198 * Fixed end-of-data-field problem (maybe).
200 * Revision 1.3 1995/11/15 19:13:09 adam
201 * Work on record management.
207 #include <sys/types.h>
218 #define GRS_MAX_WORD 512
224 struct grs_handler *next;
227 struct grs_handlers {
228 struct grs_handler *handlers;
231 static int read_grs_type (struct grs_handlers *h,
232 struct grs_read_info *p, const char *type,
235 struct grs_handler *gh = h->handlers;
236 const char *cp = strchr (type, '.');
238 if (cp == NULL || cp == type)
240 cp = strlen(type) + type;
244 strcpy (p->type, cp+1);
245 for (gh = h->handlers; gh; gh = gh->next)
247 if (!memcmp (type, gh->type->type, cp-type))
252 gh->clientData = (*gh->type->init)();
254 p->clientData = gh->clientData;
255 *root = (gh->type->read)(p);
256 gh->clientData = p->clientData;
263 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
265 struct grs_handler *gh = (struct grs_handler *) malloc (sizeof(*gh));
266 gh->next = h->handlers;
273 static void *grs_init(RecType recType)
275 struct grs_handlers *h = (struct grs_handlers *) malloc (sizeof(*h));
278 grs_add_handler (h, recTypeGrs_sgml);
279 grs_add_handler (h, recTypeGrs_regx);
281 grs_add_handler (h, recTypeGrs_tcl);
283 grs_add_handler (h, recTypeGrs_marc);
287 static void grs_destroy(void *clientData)
289 struct grs_handlers *h = (struct grs_handlers *) clientData;
290 struct grs_handler *gh = h->handlers, *gh_next;
295 (*gh->type->destroy)(gh->clientData);
302 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
305 (*p->init)(p, &wrd); /* set defaults */
306 for (; n; n = n->next)
308 if (p->flagShowRecords) /* display element description to user */
310 if (n->which == DATA1N_root)
312 printf("%*s", level * 4, "");
313 printf("Record type: '%s'\n", n->u.root.absyn->name);
315 else if (n->which == DATA1N_tag)
319 printf("%*s", level * 4, "");
320 if (!(e = n->u.tag.element))
321 printf("Local tag: '%s'\n", n->u.tag.tag);
324 printf("Elm: '%s' ", e->name);
327 data1_tag *t = e->tag;
329 printf("TagNam: '%s' ", t->names->name);
332 printf("%s[%d],", t->tagset->name, t->tagset->type);
335 if (t->which == DATA1T_numeric)
336 printf("%d)", t->value.numeric);
338 printf("'%s')", t->value.string);
346 if (dumpkeys(n->child, p, level + 1) < 0)
349 if (n->which == DATA1N_data)
351 data1_node *par = get_parent_tag(p->dh, n);
352 data1_termlist *tlist = 0;
353 data1_datatype dtype = DATA1K_string;
355 if (p->flagShowRecords)
357 printf("%*s", level * 4, "");
359 if (n->u.data.len > 32)
360 printf("'%.24s ... %.6s'\n", n->u.data.data,
361 n->u.data.data + n->u.data.len-6);
362 else if (n->u.data.len > 0)
363 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
371 * cycle up towards the root until we find a tag with an att..
372 * this has the effect of indexing locally defined tags with
373 * the attribute of their ancestor in the record.
376 while (!par->u.tag.element)
377 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
379 if (!par || !(tlist = par->u.tag.element->termlists))
381 if (par->u.tag.element->tag)
382 dtype = par->u.tag.element->tag->kind;
383 for (; tlist; tlist = tlist->next)
385 if (p->flagShowRecords)
387 printf("%*sIdx: [%s]", (level + 1) * 4, "",
389 printf("%s:%s [%d]\n",
390 tlist->att->parent->name,
391 tlist->att->name, tlist->att->value);
395 wrd.reg_type = *tlist->structure;
396 wrd.string = n->u.data.data;
397 wrd.length = n->u.data.len;
398 wrd.attrSet = (int) (tlist->att->parent->reference);
399 wrd.attrUse = tlist->att->locals->local;
400 (*p->tokenAdd)(&wrd);
404 if (p->flagShowRecords && n->which == DATA1N_root)
406 printf("%*s-------------\n\n", level * 4, "");
412 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
415 int oidtmp[OID_SIZE];
417 oe.proto = PROTO_Z3950;
418 oe.oclass = CLASS_SCHEMA;
419 oe.value = n->u.root.absyn->reference;
421 if ((oid_ent_to_oid (&oe, oidtmp)))
422 (*p->schemaAdd)(p, oidtmp);
424 return dumpkeys(n, p, 0);
427 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
431 struct grs_read_info gri;
433 int oidtmp[OID_SIZE];
435 gri.readf = p->readf;
436 gri.seekf = p->seekf;
437 gri.tellf = p->tellf;
440 gri.offset = p->offset;
444 if (read_grs_type (h, &gri, p->subType, &n))
445 return RECCTRL_EXTRACT_ERROR;
447 return RECCTRL_EXTRACT_EOF;
448 oe.proto = PROTO_Z3950;
449 oe.oclass = CLASS_SCHEMA;
450 oe.value = n->u.root.absyn->reference;
451 if ((oid_ent_to_oid (&oe, oidtmp)))
452 (*p->schemaAdd)(p, oidtmp);
454 if (dumpkeys(n, p, 0) < 0)
456 data1_free_tree(p->dh, n);
457 return RECCTRL_EXTRACT_ERROR;
459 data1_free_tree(p->dh, n);
460 return RECCTRL_EXTRACT_OK;
463 static int grs_extract(void *clientData, struct recExtractCtrl *p)
466 NMEM mem = nmem_create ();
467 struct grs_handlers *h = (struct grs_handlers *) clientData;
469 ret = grs_extract_sub(h, p, mem);
475 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
477 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
479 data1_esetname *eset;
485 case Z_RecordComp_simple:
486 if (c->u.simple->which != Z_ElementSetNames_generic)
487 return 26; /* only generic form supported. Fix this later */
488 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
489 c->u.simple->u.generic)))
491 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
492 return 25; /* invalid esetname */
494 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
495 c->u.simple->u.generic);
498 case Z_RecordComp_complex:
499 if (c->u.complex->generic)
501 /* insert check for schema */
502 if ((p = c->u.complex->generic->elementSpec))
506 case Z_ElementSpec_elementSetName:
508 data1_getesetbyname(dh, n->u.root.absyn,
509 p->u.elementSetName)))
511 logf(LOG_LOG, "Unknown esetname '%s'",
512 p->u.elementSetName);
513 return 25; /* invalid esetname */
515 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
516 p->u.elementSetName);
519 case Z_ElementSpec_externalSpec:
520 if (p->u.externalSpec->which == Z_External_espec1)
522 logf(LOG_DEBUG, "Got Espec-1");
523 espec = p->u.externalSpec-> u.espec1;
527 logf(LOG_LOG, "Unknown external espec.");
528 return 25; /* bad. what is proper diagnostic? */
539 logf (LOG_DEBUG, "Element: Espec-1 match");
540 return data1_doespec1(dh, n, espec);
544 logf (LOG_DEBUG, "Element: all match");
549 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
551 data1_node *node = 0, *onode = 0;
554 int res, selected = 0;
556 struct grs_read_info gri;
558 struct grs_handlers *h = (struct grs_handlers *) clientData;
559 int requested_schema = VAL_NONE;
562 gri.readf = p->readf;
563 gri.seekf = p->seekf;
564 gri.tellf = p->tellf;
571 logf (LOG_DEBUG, "grs_retrieve");
572 if (read_grs_type (h, &gri, p->subType, &node))
585 data1_pr_tree (p->dh, node, stdout);
587 logf (LOG_DEBUG, "grs_retrieve: size");
588 if ((dnew = data1_insert_taggeddata(p->dh, node, node,
591 dnew->u.data.what = DATA1I_text;
592 dnew->u.data.data = dnew->lbuf;
593 sprintf(dnew->u.data.data, "%d", p->recordSize);
594 dnew->u.data.len = strlen(dnew->u.data.data);
597 tagname = res_get_def(p->res, "tagrank", "rank");
598 if (strcmp(tagname, "0") && p->score >= 0 &&
599 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
601 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
602 dnew->u.data.what = DATA1I_num;
603 dnew->u.data.data = dnew->lbuf;
604 sprintf(dnew->u.data.data, "%d", p->score);
605 dnew->u.data.len = strlen(dnew->u.data.data);
608 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
609 if (strcmp(tagname, "0") && p->localno > 0 &&
610 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
612 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
613 dnew->u.data.what = DATA1I_text;
614 dnew->u.data.data = dnew->lbuf;
615 sprintf(dnew->u.data.data, "%d", p->localno);
616 dnew->u.data.len = strlen(dnew->u.data.data);
619 if (p->comp && p->comp->which == Z_RecordComp_complex &&
620 p->comp->u.complex->generic &&
621 p->comp->u.complex->generic->schema)
623 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
625 requested_schema = oe->value;
628 /* If schema has been specified, map if possible, then check that
629 * we got the right one
631 if (requested_schema != VAL_NONE)
633 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
634 for (map = node->u.root.absyn->maptabs; map; map = map->next)
636 if (map->target_absyn_ref == requested_schema)
639 if (!(node = data1_map_record(p->dh, onode, map, mem)))
648 if (node->u.root.absyn &&
649 requested_schema != node->u.root.absyn->reference)
657 * Does the requested format match a known syntax-mapping? (this reflects
658 * the overlap of schema and formatting which is inherent in the MARC
661 logf (LOG_DEBUG, "grs_retrieve: syntax mapping");
662 for (map = node->u.root.absyn->maptabs; map; map = map->next)
664 if (map->target_absyn_ref == p->input_format)
667 if (!(node = data1_map_record(p->dh, onode, map, mem)))
676 logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
677 if (node->u.root.absyn &&
678 node->u.root.absyn->reference != VAL_NONE &&
679 p->input_format == VAL_GRS1)
683 int oidtmp[OID_SIZE];
685 oe.proto = PROTO_Z3950;
686 oe.oclass = CLASS_SCHEMA;
687 oe.value = node->u.root.absyn->reference;
689 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
692 data1_handle dh = p->dh;
696 for (ii = oid; *ii >= 0; ii++)
700 sprintf(p, "%d", *ii);
705 if ((dnew = data1_insert_taggeddata(dh, node, node,
706 "schemaIdentifier", mem)))
708 dnew->u.data.what = DATA1I_oid;
709 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
710 memcpy(dnew->u.data.data, tmp, p - tmp);
711 dnew->u.data.len = p - tmp;
716 logf (LOG_DEBUG, "grs_retrieve: element spec");
717 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
721 data1_free_tree(p->dh, onode);
722 data1_free_tree(p->dh, node);
726 else if (p->comp && !res)
729 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
730 switch (p->output_format = (p->input_format != VAL_NONE ?
731 p->input_format : VAL_SUTRS))
733 data1_marctab *marctab;
737 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
742 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
743 memcpy (new_buf, p->rec_buf, p->rec_len);
744 p->rec_buf = new_buf;
749 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
751 p->diagnostic = 238; /* not available in requested syntax */
753 p->rec_len = (size_t) (-1);
756 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
760 p->rec_len = (size_t) (-1);
763 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
767 p->rec_len = (size_t) (-1);
770 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
775 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
776 memcpy (new_buf, p->rec_buf, p->rec_len);
777 p->rec_buf = new_buf;
781 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
786 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
787 memcpy (new_buf, p->rec_buf, p->rec_len);
788 p->rec_buf = new_buf;
792 if (!node->u.root.absyn)
797 for (marctab = node->u.root.absyn->marc; marctab;
798 marctab = marctab->next)
799 if (marctab->reference == p->input_format)
806 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
807 selected, &p->rec_len)))
811 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
812 memcpy (new_buf, p->rec_buf, p->rec_len);
813 p->rec_buf = new_buf;
817 data1_free_tree(p->dh, node);
819 data1_free_tree(p->dh, onode);
824 static struct recType grs_type =
833 RecType recTypeGrs = &grs_type;