2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.31 1999-07-14 10:56:43 adam
8 * Fixed potential memory leak.
10 * Revision 1.30 1999/07/06 12:26:41 adam
11 * Retrieval handler obeys schema and handles XML transfer syntax.
13 * Revision 1.29 1999/05/26 07:49:14 adam
16 * Revision 1.28 1999/05/21 12:00:17 adam
17 * Better diagnostics for extraction process.
19 * Revision 1.27 1999/05/20 12:57:18 adam
20 * Implemented TCL filter. Updated recctrl system.
22 * Revision 1.26 1999/03/02 16:15:44 quinn
23 * Added "tagsysno" and "tagrank" directives to zebra.cfg.
25 * Revision 1.25 1999/02/18 15:01:26 adam
28 * Revision 1.24 1999/02/02 14:51:28 adam
29 * Updated WIN32 code specific sections. Changed header.
31 * Revision 1.23 1998/10/18 07:51:10 adam
32 * Changed one logf call.
34 * Revision 1.22 1998/10/16 08:14:37 adam
35 * Updated record control system.
37 * Revision 1.21 1998/07/01 09:16:10 adam
38 * Element localno only added when it's greater than 0.
40 * Revision 1.20 1998/05/20 10:12:26 adam
41 * Implemented automatic EXPLAIN database maintenance.
42 * Modified Zebra to work with ASN.1 compiled version of YAZ.
44 * Revision 1.19 1998/03/11 11:19:05 adam
45 * Changed the way sequence numbers are generated.
47 * Revision 1.18 1998/03/05 08:41:31 adam
50 * Revision 1.17 1998/02/10 12:03:06 adam
53 * Revision 1.16 1998/01/29 13:38:17 adam
54 * Fixed problem with mapping to record with unknown schema.
56 * Revision 1.15 1998/01/26 10:37:57 adam
59 * Revision 1.14 1997/11/06 11:41:01 adam
60 * Implemented "begin variant" for the sgml.regx filter.
62 * Revision 1.13 1997/10/31 12:35:44 adam
63 * Added a few log statements.
65 * Revision 1.12 1997/10/29 12:02:22 adam
66 * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent.
68 * Revision 1.11 1997/10/27 14:34:00 adam
69 * Work on generic character mapping depending on "structure" field
70 * in abstract syntax file.
72 * Revision 1.10 1997/09/18 08:59:21 adam
73 * Extra generic handle for the character mapping routines.
75 * Revision 1.9 1997/09/17 12:19:21 adam
76 * Zebra version corresponds to YAZ version 1.4.
77 * Changed Zebra server so that it doesn't depend on global common_resource.
79 * Revision 1.8 1997/09/09 13:38:14 adam
80 * Partial port to WIN95/NT.
82 * Revision 1.7 1997/09/05 15:30:10 adam
83 * Changed prototype for chr_map_input - added const.
84 * Added support for C++, headers uses extern "C" for public definitions.
86 * Revision 1.6 1997/09/04 13:54:40 adam
87 * Added MARC filter - type grs.marc.<syntax> where syntax refers
88 * to abstract syntax. New method tellf in retrieve/extract method.
90 * Revision 1.5 1997/07/15 16:29:03 adam
91 * Initialized dummy variable to keep checker gcc happy.
93 * Revision 1.4 1997/04/30 08:56:08 quinn
96 * Revision 1.2 1996/10/11 16:06:43 quinn
97 * Revision 1.3 1997/02/24 10:41:50 adam
98 * Cleanup of code and commented out the "end element-end-record" code.
100 * Revision 1.2 1996/10/11 16:06:43 quinn
101 * Fixed arguments to nodetogr
103 * Revision 1.1 1996/10/11 10:57:25 adam
104 * New module recctrl. Used to manage records (extract/retrieval).
106 * Revision 1.29 1996/10/08 10:30:21 quinn
107 * Fixed type mismatch
109 * Revision 1.28 1996/10/07 16:06:40 quinn
112 * Revision 1.27 1996/06/11 10:54:12 quinn
115 * Revision 1.26 1996/06/06 12:08:45 quinn
116 * Added showRecord function
118 * Revision 1.25 1996/06/04 14:18:53 quinn
121 * Revision 1.24 1996/06/04 13:27:54 quinn
122 * More work on charmapping
124 * Revision 1.23 1996/06/04 10:19:01 adam
125 * Minor changes - removed include of ctype.h.
127 * Revision 1.22 1996/06/03 10:15:27 quinn
128 * Various character-mapping.
130 * Revision 1.21 1996/05/31 13:27:24 quinn
131 * Character-conversion in phrases, too.
133 * Revision 1.19 1996/05/16 15:31:14 quinn
136 * Revision 1.18 1996/05/09 07:28:56 quinn
137 * Work towards phrases and multiple registers
139 * Revision 1.17 1996/05/01 13:46:37 adam
140 * First work on multiple records in one file.
141 * New option, -offset, to the "unread" command in the filter module.
143 * Revision 1.16 1996/01/17 14:57:54 adam
144 * Prototype changed for reader functions in extract/retrieve. File
145 * is identified by 'void *' instead of 'int.
147 * Revision 1.15 1996/01/08 19:15:47 adam
148 * New input filter that works!
150 * Revision 1.14 1995/12/15 12:36:11 adam
151 * Retrieval calls data1_read_regx when subType is specified.
153 * Revision 1.13 1995/12/15 12:24:43 quinn
154 * *** empty log message ***
156 * Revision 1.12 1995/12/15 12:20:28 quinn
157 * *** empty log message ***
159 * Revision 1.11 1995/12/15 12:07:57 quinn
160 * Changed extraction strategy.
162 * Revision 1.10 1995/12/14 11:10:48 quinn
165 * Revision 1.9 1995/12/13 17:14:05 quinn
166 * *** empty log message ***
168 * Revision 1.8 1995/12/13 15:33:18 quinn
169 * *** empty log message ***
171 * Revision 1.7 1995/12/13 13:45:39 quinn
172 * Changed data1 to use nmem.
174 * Revision 1.6 1995/12/04 14:22:30 adam
175 * Extra arg to recType_byName.
176 * Started work on new regular expression parsed input to
177 * structured records.
179 * Revision 1.5 1995/11/28 14:18:37 quinn
182 * Revision 1.4 1995/11/21 13:14:49 quinn
183 * Fixed end-of-data-field problem (maybe).
185 * Revision 1.3 1995/11/15 19:13:09 adam
186 * Work on record management.
192 #include <sys/types.h>
203 #define GRS_MAX_WORD 512
209 struct grs_handler *next;
212 struct grs_handlers {
213 struct grs_handler *handlers;
216 static int read_grs_type (struct grs_handlers *h,
217 struct grs_read_info *p, const char *type,
220 struct grs_handler *gh = h->handlers;
221 const char *cp = strchr (type, '.');
223 if (cp == NULL || cp == type)
225 cp = strlen(type) + type;
229 strcpy (p->type, cp+1);
230 for (gh = h->handlers; gh; gh = gh->next)
232 if (!memcmp (type, gh->type->type, cp-type))
237 gh->clientData = (*gh->type->init)();
239 p->clientData = gh->clientData;
240 *root = (gh->type->read)(p);
241 gh->clientData = p->clientData;
248 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
250 struct grs_handler *gh = (struct grs_handler *) malloc (sizeof(*gh));
251 gh->next = h->handlers;
258 static void *grs_init(RecType recType)
260 struct grs_handlers *h = (struct grs_handlers *) malloc (sizeof(*h));
263 grs_add_handler (h, recTypeGrs_sgml);
264 grs_add_handler (h, recTypeGrs_regx);
266 grs_add_handler (h, recTypeGrs_tcl);
268 grs_add_handler (h, recTypeGrs_marc);
272 static void grs_destroy(void *clientData)
274 struct grs_handlers *h = (struct grs_handlers *) clientData;
275 struct grs_handler *gh = h->handlers, *gh_next;
280 (*gh->type->destroy)(gh->clientData);
287 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
290 (*p->init)(p, &wrd); /* set defaults */
291 for (; n; n = n->next)
293 if (p->flagShowRecords) /* display element description to user */
295 if (n->which == DATA1N_root)
297 printf("%*s", level * 4, "");
298 printf("Record type: '%s'\n", n->u.root.absyn->name);
300 else if (n->which == DATA1N_tag)
304 printf("%*s", level * 4, "");
305 if (!(e = n->u.tag.element))
306 printf("Local tag: '%s'\n", n->u.tag.tag);
309 printf("Elm: '%s' ", e->name);
312 data1_tag *t = e->tag;
314 printf("TagNam: '%s' ", t->names->name);
317 printf("%s[%d],", t->tagset->name, t->tagset->type);
320 if (t->which == DATA1T_numeric)
321 printf("%d)", t->value.numeric);
323 printf("'%s')", t->value.string);
331 if (dumpkeys(n->child, p, level + 1) < 0)
334 if (n->which == DATA1N_data)
336 data1_node *par = get_parent_tag(p->dh, n);
337 data1_termlist *tlist = 0;
338 data1_datatype dtype = DATA1K_string;
340 if (p->flagShowRecords)
342 printf("%*s", level * 4, "");
344 if (n->u.data.len > 32)
345 printf("'%.24s ... %.6s'\n", n->u.data.data,
346 n->u.data.data + n->u.data.len-6);
347 else if (n->u.data.len > 0)
348 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
356 * cycle up towards the root until we find a tag with an att..
357 * this has the effect of indexing locally defined tags with
358 * the attribute of their ancestor in the record.
361 while (!par->u.tag.element)
362 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
364 if (!par || !(tlist = par->u.tag.element->termlists))
366 if (par->u.tag.element->tag)
367 dtype = par->u.tag.element->tag->kind;
368 for (; tlist; tlist = tlist->next)
370 if (p->flagShowRecords)
372 printf("%*sIdx: [%s]", (level + 1) * 4, "",
374 printf("%s:%s [%d]\n",
375 tlist->att->parent->name,
376 tlist->att->name, tlist->att->value);
380 wrd.reg_type = *tlist->structure;
381 wrd.string = n->u.data.data;
382 wrd.length = n->u.data.len;
383 wrd.attrSet = (int) (tlist->att->parent->reference);
384 wrd.attrUse = tlist->att->locals->local;
389 if (p->flagShowRecords && n->which == DATA1N_root)
391 printf("%*s-------------\n\n", level * 4, "");
397 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
400 int oidtmp[OID_SIZE];
402 oe.proto = PROTO_Z3950;
403 oe.oclass = CLASS_SCHEMA;
404 oe.value = n->u.root.absyn->reference;
406 if ((oid_ent_to_oid (&oe, oidtmp)))
407 (*p->addSchema)(p, oidtmp);
409 return dumpkeys(n, p, 0);
412 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
416 struct grs_read_info gri;
418 int oidtmp[OID_SIZE];
420 gri.readf = p->readf;
421 gri.seekf = p->seekf;
422 gri.tellf = p->tellf;
425 gri.offset = p->offset;
429 if (read_grs_type (h, &gri, p->subType, &n))
430 return RECCTRL_EXTRACT_ERROR;
432 return RECCTRL_EXTRACT_EOF;
433 oe.proto = PROTO_Z3950;
434 oe.oclass = CLASS_SCHEMA;
435 oe.value = n->u.root.absyn->reference;
436 if ((oid_ent_to_oid (&oe, oidtmp)))
437 (*p->addSchema)(p, oidtmp);
439 if (dumpkeys(n, p, 0) < 0)
441 data1_free_tree(p->dh, n);
442 return RECCTRL_EXTRACT_ERROR;
444 data1_free_tree(p->dh, n);
445 return RECCTRL_EXTRACT_OK;
448 static int grs_extract(void *clientData, struct recExtractCtrl *p)
451 NMEM mem = nmem_create ();
452 struct grs_handlers *h = (struct grs_handlers *) clientData;
454 ret = grs_extract_sub(h, p, mem);
460 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
462 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
464 data1_esetname *eset;
470 case Z_RecordComp_simple:
471 if (c->u.simple->which != Z_ElementSetNames_generic)
472 return 26; /* only generic form supported. Fix this later */
473 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
474 c->u.simple->u.generic)))
476 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
477 return 25; /* invalid esetname */
479 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
480 c->u.simple->u.generic);
483 case Z_RecordComp_complex:
484 if (c->u.complex->generic)
486 /* insert check for schema */
487 if ((p = c->u.complex->generic->elementSpec))
491 case Z_ElementSpec_elementSetName:
493 data1_getesetbyname(dh, n->u.root.absyn,
494 p->u.elementSetName)))
496 logf(LOG_LOG, "Unknown esetname '%s'",
497 p->u.elementSetName);
498 return 25; /* invalid esetname */
500 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
501 p->u.elementSetName);
504 case Z_ElementSpec_externalSpec:
505 if (p->u.externalSpec->which == Z_External_espec1)
507 logf(LOG_DEBUG, "Got Espec-1");
508 espec = p->u.externalSpec-> u.espec1;
512 logf(LOG_LOG, "Unknown external espec.");
513 return 25; /* bad. what is proper diagnostic? */
524 logf (LOG_DEBUG, "Element: Espec-1 match");
525 return data1_doespec1(dh, n, espec);
529 logf (LOG_DEBUG, "Element: all match");
534 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
536 data1_node *node = 0, *onode = 0;
539 int res, selected = 0;
541 struct grs_read_info gri;
543 struct grs_handlers *h = (struct grs_handlers *) clientData;
544 int requested_schema = VAL_NONE;
547 gri.readf = p->readf;
548 gri.seekf = p->seekf;
549 gri.tellf = p->tellf;
556 logf (LOG_DEBUG, "grs_retrieve");
557 if (read_grs_type (h, &gri, p->subType, &node))
569 logf (LOG_DEBUG, "grs_retrieve: size");
570 if ((dnew = data1_insert_taggeddata(p->dh, node, node,
573 dnew->u.data.what = DATA1I_text;
574 dnew->u.data.data = dnew->lbuf;
575 sprintf(dnew->u.data.data, "%d", p->recordSize);
576 dnew->u.data.len = strlen(dnew->u.data.data);
579 tagname = res_get_def(p->res, "tagrank", "rank");
580 if (strcmp(tagname, "0") && p->score >= 0 &&
581 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
583 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
584 dnew->u.data.what = DATA1I_num;
585 dnew->u.data.data = dnew->lbuf;
586 sprintf(dnew->u.data.data, "%d", p->score);
587 dnew->u.data.len = strlen(dnew->u.data.data);
590 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
591 if (strcmp(tagname, "0") && p->localno > 0 &&
592 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
594 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
595 dnew->u.data.what = DATA1I_text;
596 dnew->u.data.data = dnew->lbuf;
597 sprintf(dnew->u.data.data, "%d", p->localno);
598 dnew->u.data.len = strlen(dnew->u.data.data);
601 if (p->comp && p->comp->which == Z_RecordComp_complex &&
602 p->comp->u.complex->generic &&
603 p->comp->u.complex->generic->schema)
605 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
607 requested_schema = oe->value;
610 /* If schema has been specified, map if possible, then check that
611 * we got the right one
613 if (requested_schema != VAL_NONE)
615 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
616 for (map = node->u.root.absyn->maptabs; map; map = map->next)
618 if (map->target_absyn_ref == requested_schema)
621 if (!(node = data1_map_record(p->dh, onode, map, mem)))
630 if (node->u.root.absyn &&
631 requested_schema != node->u.root.absyn->reference)
639 * Does the requested format match a known syntax-mapping? (this reflects
640 * the overlap of schema and formatting which is inherent in the MARC
643 logf (LOG_DEBUG, "grs_retrieve: syntax mapping");
644 for (map = node->u.root.absyn->maptabs; map; map = map->next)
646 if (map->target_absyn_ref == p->input_format)
649 if (!(node = data1_map_record(p->dh, onode, map, mem)))
658 logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
659 if (node->u.root.absyn &&
660 node->u.root.absyn->reference != VAL_NONE &&
661 p->input_format == VAL_GRS1)
665 int oidtmp[OID_SIZE];
667 oe.proto = PROTO_Z3950;
668 oe.oclass = CLASS_SCHEMA;
669 oe.value = node->u.root.absyn->reference;
671 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
674 data1_handle dh = p->dh;
678 for (ii = oid; *ii >= 0; ii++)
682 sprintf(p, "%d", *ii);
687 if ((dnew = data1_insert_taggeddata(dh, node, node,
688 "schemaIdentifier", mem)))
690 dnew->u.data.what = DATA1I_oid;
691 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
692 memcpy(dnew->u.data.data, tmp, p - tmp);
693 dnew->u.data.len = p - tmp;
698 logf (LOG_DEBUG, "grs_retrieve: element spec");
699 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
703 data1_free_tree(p->dh, onode);
704 data1_free_tree(p->dh, node);
708 else if (p->comp && !res)
711 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
712 switch (p->output_format = (p->input_format != VAL_NONE ?
713 p->input_format : VAL_SUTRS))
715 data1_marctab *marctab;
719 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
724 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
725 memcpy (new_buf, p->rec_buf, p->rec_len);
726 p->rec_buf = new_buf;
731 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
733 p->diagnostic = 238; /* not available in requested syntax */
735 p->rec_len = (size_t) (-1);
738 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
742 p->rec_len = (size_t) (-1);
745 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
749 p->rec_len = (size_t) (-1);
752 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
757 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
758 memcpy (new_buf, p->rec_buf, p->rec_len);
759 p->rec_buf = new_buf;
763 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
768 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
769 memcpy (new_buf, p->rec_buf, p->rec_len);
770 p->rec_buf = new_buf;
774 if (!node->u.root.absyn)
779 for (marctab = node->u.root.absyn->marc; marctab;
780 marctab = marctab->next)
781 if (marctab->reference == p->input_format)
788 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
794 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
795 memcpy (new_buf, p->rec_buf, p->rec_len);
796 p->rec_buf = new_buf;
800 data1_free_tree(p->dh, node);
802 data1_free_tree(p->dh, onode);
807 static struct recType grs_type =
816 RecType recTypeGrs = &grs_type;