2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.35 2000-11-29 15:21:31 adam
8 * Fixed problem with passwd db.
10 * Revision 1.34 2000/02/25 13:24:49 adam
11 * Fixed bug regarding pointer conversion that showed up on OSF V5.
13 * Revision 1.33 1999/11/30 13:48:04 adam
14 * Improved installation. Updated for inclusion of YAZ header files.
16 * Revision 1.32 1999/09/07 07:19:21 adam
17 * Work on character mapping. Implemented replace rules.
19 * Revision 1.31 1999/07/14 10:56:43 adam
20 * Fixed potential memory leak.
22 * Revision 1.30 1999/07/06 12:26:41 adam
23 * Retrieval handler obeys schema and handles XML transfer syntax.
25 * Revision 1.29 1999/05/26 07:49:14 adam
28 * Revision 1.28 1999/05/21 12:00:17 adam
29 * Better diagnostics for extraction process.
31 * Revision 1.27 1999/05/20 12:57:18 adam
32 * Implemented TCL filter. Updated recctrl system.
34 * Revision 1.26 1999/03/02 16:15:44 quinn
35 * Added "tagsysno" and "tagrank" directives to zebra.cfg.
37 * Revision 1.25 1999/02/18 15:01:26 adam
40 * Revision 1.24 1999/02/02 14:51:28 adam
41 * Updated WIN32 code specific sections. Changed header.
43 * Revision 1.23 1998/10/18 07:51:10 adam
44 * Changed one logf call.
46 * Revision 1.22 1998/10/16 08:14:37 adam
47 * Updated record control system.
49 * Revision 1.21 1998/07/01 09:16:10 adam
50 * Element localno only added when it's greater than 0.
52 * Revision 1.20 1998/05/20 10:12:26 adam
53 * Implemented automatic EXPLAIN database maintenance.
54 * Modified Zebra to work with ASN.1 compiled version of YAZ.
56 * Revision 1.19 1998/03/11 11:19:05 adam
57 * Changed the way sequence numbers are generated.
59 * Revision 1.18 1998/03/05 08:41:31 adam
62 * Revision 1.17 1998/02/10 12:03:06 adam
65 * Revision 1.16 1998/01/29 13:38:17 adam
66 * Fixed problem with mapping to record with unknown schema.
68 * Revision 1.15 1998/01/26 10:37:57 adam
71 * Revision 1.14 1997/11/06 11:41:01 adam
72 * Implemented "begin variant" for the sgml.regx filter.
74 * Revision 1.13 1997/10/31 12:35:44 adam
75 * Added a few log statements.
77 * Revision 1.12 1997/10/29 12:02:22 adam
78 * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent.
80 * Revision 1.11 1997/10/27 14:34:00 adam
81 * Work on generic character mapping depending on "structure" field
82 * in abstract syntax file.
84 * Revision 1.10 1997/09/18 08:59:21 adam
85 * Extra generic handle for the character mapping routines.
87 * Revision 1.9 1997/09/17 12:19:21 adam
88 * Zebra version corresponds to YAZ version 1.4.
89 * Changed Zebra server so that it doesn't depend on global common_resource.
91 * Revision 1.8 1997/09/09 13:38:14 adam
92 * Partial port to WIN95/NT.
94 * Revision 1.7 1997/09/05 15:30:10 adam
95 * Changed prototype for chr_map_input - added const.
96 * Added support for C++, headers uses extern "C" for public definitions.
98 * Revision 1.6 1997/09/04 13:54:40 adam
99 * Added MARC filter - type grs.marc.<syntax> where syntax refers
100 * to abstract syntax. New method tellf in retrieve/extract method.
102 * Revision 1.5 1997/07/15 16:29:03 adam
103 * Initialized dummy variable to keep checker gcc happy.
105 * Revision 1.4 1997/04/30 08:56:08 quinn
108 * Revision 1.2 1996/10/11 16:06:43 quinn
109 * Revision 1.3 1997/02/24 10:41:50 adam
110 * Cleanup of code and commented out the "end element-end-record" code.
112 * Revision 1.2 1996/10/11 16:06:43 quinn
113 * Fixed arguments to nodetogr
115 * Revision 1.1 1996/10/11 10:57:25 adam
116 * New module recctrl. Used to manage records (extract/retrieval).
118 * Revision 1.29 1996/10/08 10:30:21 quinn
119 * Fixed type mismatch
121 * Revision 1.28 1996/10/07 16:06:40 quinn
124 * Revision 1.27 1996/06/11 10:54:12 quinn
127 * Revision 1.26 1996/06/06 12:08:45 quinn
128 * Added showRecord function
130 * Revision 1.25 1996/06/04 14:18:53 quinn
133 * Revision 1.24 1996/06/04 13:27:54 quinn
134 * More work on charmapping
136 * Revision 1.23 1996/06/04 10:19:01 adam
137 * Minor changes - removed include of ctype.h.
139 * Revision 1.22 1996/06/03 10:15:27 quinn
140 * Various character-mapping.
142 * Revision 1.21 1996/05/31 13:27:24 quinn
143 * Character-conversion in phrases, too.
145 * Revision 1.19 1996/05/16 15:31:14 quinn
148 * Revision 1.18 1996/05/09 07:28:56 quinn
149 * Work towards phrases and multiple registers
151 * Revision 1.17 1996/05/01 13:46:37 adam
152 * First work on multiple records in one file.
153 * New option, -offset, to the "unread" command in the filter module.
155 * Revision 1.16 1996/01/17 14:57:54 adam
156 * Prototype changed for reader functions in extract/retrieve. File
157 * is identified by 'void *' instead of 'int.
159 * Revision 1.15 1996/01/08 19:15:47 adam
160 * New input filter that works!
162 * Revision 1.14 1995/12/15 12:36:11 adam
163 * Retrieval calls data1_read_regx when subType is specified.
165 * Revision 1.13 1995/12/15 12:24:43 quinn
166 * *** empty log message ***
168 * Revision 1.12 1995/12/15 12:20:28 quinn
169 * *** empty log message ***
171 * Revision 1.11 1995/12/15 12:07:57 quinn
172 * Changed extraction strategy.
174 * Revision 1.10 1995/12/14 11:10:48 quinn
177 * Revision 1.9 1995/12/13 17:14:05 quinn
178 * *** empty log message ***
180 * Revision 1.8 1995/12/13 15:33:18 quinn
181 * *** empty log message ***
183 * Revision 1.7 1995/12/13 13:45:39 quinn
184 * Changed data1 to use nmem.
186 * Revision 1.6 1995/12/04 14:22:30 adam
187 * Extra arg to recType_byName.
188 * Started work on new regular expression parsed input to
189 * structured records.
191 * Revision 1.5 1995/11/28 14:18:37 quinn
194 * Revision 1.4 1995/11/21 13:14:49 quinn
195 * Fixed end-of-data-field problem (maybe).
197 * Revision 1.3 1995/11/15 19:13:09 adam
198 * Work on record management.
204 #include <sys/types.h>
215 #define GRS_MAX_WORD 512
221 struct grs_handler *next;
224 struct grs_handlers {
225 struct grs_handler *handlers;
228 static int read_grs_type (struct grs_handlers *h,
229 struct grs_read_info *p, const char *type,
232 struct grs_handler *gh = h->handlers;
233 const char *cp = strchr (type, '.');
235 if (cp == NULL || cp == type)
237 cp = strlen(type) + type;
241 strcpy (p->type, cp+1);
242 for (gh = h->handlers; gh; gh = gh->next)
244 if (!memcmp (type, gh->type->type, cp-type))
249 gh->clientData = (*gh->type->init)();
251 p->clientData = gh->clientData;
252 *root = (gh->type->read)(p);
253 gh->clientData = p->clientData;
260 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
262 struct grs_handler *gh = (struct grs_handler *) malloc (sizeof(*gh));
263 gh->next = h->handlers;
270 static void *grs_init(RecType recType)
272 struct grs_handlers *h = (struct grs_handlers *) malloc (sizeof(*h));
275 grs_add_handler (h, recTypeGrs_sgml);
276 grs_add_handler (h, recTypeGrs_regx);
278 grs_add_handler (h, recTypeGrs_tcl);
280 grs_add_handler (h, recTypeGrs_marc);
284 static void grs_destroy(void *clientData)
286 struct grs_handlers *h = (struct grs_handlers *) clientData;
287 struct grs_handler *gh = h->handlers, *gh_next;
292 (*gh->type->destroy)(gh->clientData);
299 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
302 (*p->init)(p, &wrd); /* set defaults */
303 for (; n; n = n->next)
305 if (p->flagShowRecords) /* display element description to user */
307 if (n->which == DATA1N_root)
309 printf("%*s", level * 4, "");
310 printf("Record type: '%s'\n", n->u.root.absyn->name);
312 else if (n->which == DATA1N_tag)
316 printf("%*s", level * 4, "");
317 if (!(e = n->u.tag.element))
318 printf("Local tag: '%s'\n", n->u.tag.tag);
321 printf("Elm: '%s' ", e->name);
324 data1_tag *t = e->tag;
326 printf("TagNam: '%s' ", t->names->name);
329 printf("%s[%d],", t->tagset->name, t->tagset->type);
332 if (t->which == DATA1T_numeric)
333 printf("%d)", t->value.numeric);
335 printf("'%s')", t->value.string);
343 if (dumpkeys(n->child, p, level + 1) < 0)
346 if (n->which == DATA1N_data)
348 data1_node *par = get_parent_tag(p->dh, n);
349 data1_termlist *tlist = 0;
350 data1_datatype dtype = DATA1K_string;
352 if (p->flagShowRecords)
354 printf("%*s", level * 4, "");
356 if (n->u.data.len > 32)
357 printf("'%.24s ... %.6s'\n", n->u.data.data,
358 n->u.data.data + n->u.data.len-6);
359 else if (n->u.data.len > 0)
360 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
368 * cycle up towards the root until we find a tag with an att..
369 * this has the effect of indexing locally defined tags with
370 * the attribute of their ancestor in the record.
373 while (!par->u.tag.element)
374 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
376 if (!par || !(tlist = par->u.tag.element->termlists))
378 if (par->u.tag.element->tag)
379 dtype = par->u.tag.element->tag->kind;
380 for (; tlist; tlist = tlist->next)
382 if (p->flagShowRecords)
384 printf("%*sIdx: [%s]", (level + 1) * 4, "",
386 printf("%s:%s [%d]\n",
387 tlist->att->parent->name,
388 tlist->att->name, tlist->att->value);
392 wrd.reg_type = *tlist->structure;
393 wrd.string = n->u.data.data;
394 wrd.length = n->u.data.len;
395 wrd.attrSet = (int) (tlist->att->parent->reference);
396 wrd.attrUse = tlist->att->locals->local;
397 (*p->tokenAdd)(&wrd);
401 if (p->flagShowRecords && n->which == DATA1N_root)
403 printf("%*s-------------\n\n", level * 4, "");
409 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
412 int oidtmp[OID_SIZE];
414 oe.proto = PROTO_Z3950;
415 oe.oclass = CLASS_SCHEMA;
416 oe.value = n->u.root.absyn->reference;
418 if ((oid_ent_to_oid (&oe, oidtmp)))
419 (*p->schemaAdd)(p, oidtmp);
421 return dumpkeys(n, p, 0);
424 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
428 struct grs_read_info gri;
430 int oidtmp[OID_SIZE];
432 gri.readf = p->readf;
433 gri.seekf = p->seekf;
434 gri.tellf = p->tellf;
437 gri.offset = p->offset;
441 if (read_grs_type (h, &gri, p->subType, &n))
442 return RECCTRL_EXTRACT_ERROR;
444 return RECCTRL_EXTRACT_EOF;
445 oe.proto = PROTO_Z3950;
446 oe.oclass = CLASS_SCHEMA;
447 oe.value = n->u.root.absyn->reference;
448 if ((oid_ent_to_oid (&oe, oidtmp)))
449 (*p->schemaAdd)(p, oidtmp);
451 if (dumpkeys(n, p, 0) < 0)
453 data1_free_tree(p->dh, n);
454 return RECCTRL_EXTRACT_ERROR;
456 data1_free_tree(p->dh, n);
457 return RECCTRL_EXTRACT_OK;
460 static int grs_extract(void *clientData, struct recExtractCtrl *p)
463 NMEM mem = nmem_create ();
464 struct grs_handlers *h = (struct grs_handlers *) clientData;
466 ret = grs_extract_sub(h, p, mem);
472 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
474 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
476 data1_esetname *eset;
482 case Z_RecordComp_simple:
483 if (c->u.simple->which != Z_ElementSetNames_generic)
484 return 26; /* only generic form supported. Fix this later */
485 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
486 c->u.simple->u.generic)))
488 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
489 return 25; /* invalid esetname */
491 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
492 c->u.simple->u.generic);
495 case Z_RecordComp_complex:
496 if (c->u.complex->generic)
498 /* insert check for schema */
499 if ((p = c->u.complex->generic->elementSpec))
503 case Z_ElementSpec_elementSetName:
505 data1_getesetbyname(dh, n->u.root.absyn,
506 p->u.elementSetName)))
508 logf(LOG_LOG, "Unknown esetname '%s'",
509 p->u.elementSetName);
510 return 25; /* invalid esetname */
512 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
513 p->u.elementSetName);
516 case Z_ElementSpec_externalSpec:
517 if (p->u.externalSpec->which == Z_External_espec1)
519 logf(LOG_DEBUG, "Got Espec-1");
520 espec = p->u.externalSpec-> u.espec1;
524 logf(LOG_LOG, "Unknown external espec.");
525 return 25; /* bad. what is proper diagnostic? */
536 logf (LOG_DEBUG, "Element: Espec-1 match");
537 return data1_doespec1(dh, n, espec);
541 logf (LOG_DEBUG, "Element: all match");
546 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
548 data1_node *node = 0, *onode = 0;
551 int res, selected = 0;
553 struct grs_read_info gri;
555 struct grs_handlers *h = (struct grs_handlers *) clientData;
556 int requested_schema = VAL_NONE;
559 gri.readf = p->readf;
560 gri.seekf = p->seekf;
561 gri.tellf = p->tellf;
568 logf (LOG_DEBUG, "grs_retrieve");
569 if (read_grs_type (h, &gri, p->subType, &node))
581 data1_pr_tree (p->dh, node, stdout);
582 logf (LOG_DEBUG, "grs_retrieve: size");
583 if ((dnew = data1_insert_taggeddata(p->dh, node, node,
586 dnew->u.data.what = DATA1I_text;
587 dnew->u.data.data = dnew->lbuf;
588 sprintf(dnew->u.data.data, "%d", p->recordSize);
589 dnew->u.data.len = strlen(dnew->u.data.data);
592 tagname = res_get_def(p->res, "tagrank", "rank");
593 if (strcmp(tagname, "0") && p->score >= 0 &&
594 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
596 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
597 dnew->u.data.what = DATA1I_num;
598 dnew->u.data.data = dnew->lbuf;
599 sprintf(dnew->u.data.data, "%d", p->score);
600 dnew->u.data.len = strlen(dnew->u.data.data);
603 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
604 if (strcmp(tagname, "0") && p->localno > 0 &&
605 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
607 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
608 dnew->u.data.what = DATA1I_text;
609 dnew->u.data.data = dnew->lbuf;
610 sprintf(dnew->u.data.data, "%d", p->localno);
611 dnew->u.data.len = strlen(dnew->u.data.data);
614 if (p->comp && p->comp->which == Z_RecordComp_complex &&
615 p->comp->u.complex->generic &&
616 p->comp->u.complex->generic->schema)
618 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
620 requested_schema = oe->value;
623 /* If schema has been specified, map if possible, then check that
624 * we got the right one
626 if (requested_schema != VAL_NONE)
628 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
629 for (map = node->u.root.absyn->maptabs; map; map = map->next)
631 if (map->target_absyn_ref == requested_schema)
634 if (!(node = data1_map_record(p->dh, onode, map, mem)))
643 if (node->u.root.absyn &&
644 requested_schema != node->u.root.absyn->reference)
652 * Does the requested format match a known syntax-mapping? (this reflects
653 * the overlap of schema and formatting which is inherent in the MARC
656 logf (LOG_DEBUG, "grs_retrieve: syntax mapping");
657 for (map = node->u.root.absyn->maptabs; map; map = map->next)
659 if (map->target_absyn_ref == p->input_format)
662 if (!(node = data1_map_record(p->dh, onode, map, mem)))
671 logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
672 if (node->u.root.absyn &&
673 node->u.root.absyn->reference != VAL_NONE &&
674 p->input_format == VAL_GRS1)
678 int oidtmp[OID_SIZE];
680 oe.proto = PROTO_Z3950;
681 oe.oclass = CLASS_SCHEMA;
682 oe.value = node->u.root.absyn->reference;
684 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
687 data1_handle dh = p->dh;
691 for (ii = oid; *ii >= 0; ii++)
695 sprintf(p, "%d", *ii);
700 if ((dnew = data1_insert_taggeddata(dh, node, node,
701 "schemaIdentifier", mem)))
703 dnew->u.data.what = DATA1I_oid;
704 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
705 memcpy(dnew->u.data.data, tmp, p - tmp);
706 dnew->u.data.len = p - tmp;
711 logf (LOG_DEBUG, "grs_retrieve: element spec");
712 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
716 data1_free_tree(p->dh, onode);
717 data1_free_tree(p->dh, node);
721 else if (p->comp && !res)
724 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
725 switch (p->output_format = (p->input_format != VAL_NONE ?
726 p->input_format : VAL_SUTRS))
728 data1_marctab *marctab;
732 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
737 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
738 memcpy (new_buf, p->rec_buf, p->rec_len);
739 p->rec_buf = new_buf;
744 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
746 p->diagnostic = 238; /* not available in requested syntax */
748 p->rec_len = (size_t) (-1);
751 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
755 p->rec_len = (size_t) (-1);
758 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
762 p->rec_len = (size_t) (-1);
765 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
770 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
771 memcpy (new_buf, p->rec_buf, p->rec_len);
772 p->rec_buf = new_buf;
776 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
781 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
782 memcpy (new_buf, p->rec_buf, p->rec_len);
783 p->rec_buf = new_buf;
787 if (!node->u.root.absyn)
792 for (marctab = node->u.root.absyn->marc; marctab;
793 marctab = marctab->next)
794 if (marctab->reference == p->input_format)
801 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
802 selected, &p->rec_len)))
806 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
807 memcpy (new_buf, p->rec_buf, p->rec_len);
808 p->rec_buf = new_buf;
812 data1_free_tree(p->dh, node);
814 data1_free_tree(p->dh, onode);
819 static struct recType grs_type =
828 RecType recTypeGrs = &grs_type;