2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.33 1999-11-30 13:48:04 adam
8 * Improved installation. Updated for inclusion of YAZ header files.
10 * Revision 1.32 1999/09/07 07:19:21 adam
11 * Work on character mapping. Implemented replace rules.
13 * Revision 1.31 1999/07/14 10:56:43 adam
14 * Fixed potential memory leak.
16 * Revision 1.30 1999/07/06 12:26:41 adam
17 * Retrieval handler obeys schema and handles XML transfer syntax.
19 * Revision 1.29 1999/05/26 07:49:14 adam
22 * Revision 1.28 1999/05/21 12:00:17 adam
23 * Better diagnostics for extraction process.
25 * Revision 1.27 1999/05/20 12:57:18 adam
26 * Implemented TCL filter. Updated recctrl system.
28 * Revision 1.26 1999/03/02 16:15:44 quinn
29 * Added "tagsysno" and "tagrank" directives to zebra.cfg.
31 * Revision 1.25 1999/02/18 15:01:26 adam
34 * Revision 1.24 1999/02/02 14:51:28 adam
35 * Updated WIN32 code specific sections. Changed header.
37 * Revision 1.23 1998/10/18 07:51:10 adam
38 * Changed one logf call.
40 * Revision 1.22 1998/10/16 08:14:37 adam
41 * Updated record control system.
43 * Revision 1.21 1998/07/01 09:16:10 adam
44 * Element localno only added when it's greater than 0.
46 * Revision 1.20 1998/05/20 10:12:26 adam
47 * Implemented automatic EXPLAIN database maintenance.
48 * Modified Zebra to work with ASN.1 compiled version of YAZ.
50 * Revision 1.19 1998/03/11 11:19:05 adam
51 * Changed the way sequence numbers are generated.
53 * Revision 1.18 1998/03/05 08:41:31 adam
56 * Revision 1.17 1998/02/10 12:03:06 adam
59 * Revision 1.16 1998/01/29 13:38:17 adam
60 * Fixed problem with mapping to record with unknown schema.
62 * Revision 1.15 1998/01/26 10:37:57 adam
65 * Revision 1.14 1997/11/06 11:41:01 adam
66 * Implemented "begin variant" for the sgml.regx filter.
68 * Revision 1.13 1997/10/31 12:35:44 adam
69 * Added a few log statements.
71 * Revision 1.12 1997/10/29 12:02:22 adam
72 * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent.
74 * Revision 1.11 1997/10/27 14:34:00 adam
75 * Work on generic character mapping depending on "structure" field
76 * in abstract syntax file.
78 * Revision 1.10 1997/09/18 08:59:21 adam
79 * Extra generic handle for the character mapping routines.
81 * Revision 1.9 1997/09/17 12:19:21 adam
82 * Zebra version corresponds to YAZ version 1.4.
83 * Changed Zebra server so that it doesn't depend on global common_resource.
85 * Revision 1.8 1997/09/09 13:38:14 adam
86 * Partial port to WIN95/NT.
88 * Revision 1.7 1997/09/05 15:30:10 adam
89 * Changed prototype for chr_map_input - added const.
90 * Added support for C++, headers uses extern "C" for public definitions.
92 * Revision 1.6 1997/09/04 13:54:40 adam
93 * Added MARC filter - type grs.marc.<syntax> where syntax refers
94 * to abstract syntax. New method tellf in retrieve/extract method.
96 * Revision 1.5 1997/07/15 16:29:03 adam
97 * Initialized dummy variable to keep checker gcc happy.
99 * Revision 1.4 1997/04/30 08:56:08 quinn
102 * Revision 1.2 1996/10/11 16:06:43 quinn
103 * Revision 1.3 1997/02/24 10:41:50 adam
104 * Cleanup of code and commented out the "end element-end-record" code.
106 * Revision 1.2 1996/10/11 16:06:43 quinn
107 * Fixed arguments to nodetogr
109 * Revision 1.1 1996/10/11 10:57:25 adam
110 * New module recctrl. Used to manage records (extract/retrieval).
112 * Revision 1.29 1996/10/08 10:30:21 quinn
113 * Fixed type mismatch
115 * Revision 1.28 1996/10/07 16:06:40 quinn
118 * Revision 1.27 1996/06/11 10:54:12 quinn
121 * Revision 1.26 1996/06/06 12:08:45 quinn
122 * Added showRecord function
124 * Revision 1.25 1996/06/04 14:18:53 quinn
127 * Revision 1.24 1996/06/04 13:27:54 quinn
128 * More work on charmapping
130 * Revision 1.23 1996/06/04 10:19:01 adam
131 * Minor changes - removed include of ctype.h.
133 * Revision 1.22 1996/06/03 10:15:27 quinn
134 * Various character-mapping.
136 * Revision 1.21 1996/05/31 13:27:24 quinn
137 * Character-conversion in phrases, too.
139 * Revision 1.19 1996/05/16 15:31:14 quinn
142 * Revision 1.18 1996/05/09 07:28:56 quinn
143 * Work towards phrases and multiple registers
145 * Revision 1.17 1996/05/01 13:46:37 adam
146 * First work on multiple records in one file.
147 * New option, -offset, to the "unread" command in the filter module.
149 * Revision 1.16 1996/01/17 14:57:54 adam
150 * Prototype changed for reader functions in extract/retrieve. File
151 * is identified by 'void *' instead of 'int.
153 * Revision 1.15 1996/01/08 19:15:47 adam
154 * New input filter that works!
156 * Revision 1.14 1995/12/15 12:36:11 adam
157 * Retrieval calls data1_read_regx when subType is specified.
159 * Revision 1.13 1995/12/15 12:24:43 quinn
160 * *** empty log message ***
162 * Revision 1.12 1995/12/15 12:20:28 quinn
163 * *** empty log message ***
165 * Revision 1.11 1995/12/15 12:07:57 quinn
166 * Changed extraction strategy.
168 * Revision 1.10 1995/12/14 11:10:48 quinn
171 * Revision 1.9 1995/12/13 17:14:05 quinn
172 * *** empty log message ***
174 * Revision 1.8 1995/12/13 15:33:18 quinn
175 * *** empty log message ***
177 * Revision 1.7 1995/12/13 13:45:39 quinn
178 * Changed data1 to use nmem.
180 * Revision 1.6 1995/12/04 14:22:30 adam
181 * Extra arg to recType_byName.
182 * Started work on new regular expression parsed input to
183 * structured records.
185 * Revision 1.5 1995/11/28 14:18:37 quinn
188 * Revision 1.4 1995/11/21 13:14:49 quinn
189 * Fixed end-of-data-field problem (maybe).
191 * Revision 1.3 1995/11/15 19:13:09 adam
192 * Work on record management.
198 #include <sys/types.h>
209 #define GRS_MAX_WORD 512
215 struct grs_handler *next;
218 struct grs_handlers {
219 struct grs_handler *handlers;
222 static int read_grs_type (struct grs_handlers *h,
223 struct grs_read_info *p, const char *type,
226 struct grs_handler *gh = h->handlers;
227 const char *cp = strchr (type, '.');
229 if (cp == NULL || cp == type)
231 cp = strlen(type) + type;
235 strcpy (p->type, cp+1);
236 for (gh = h->handlers; gh; gh = gh->next)
238 if (!memcmp (type, gh->type->type, cp-type))
243 gh->clientData = (*gh->type->init)();
245 p->clientData = gh->clientData;
246 *root = (gh->type->read)(p);
247 gh->clientData = p->clientData;
254 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
256 struct grs_handler *gh = (struct grs_handler *) malloc (sizeof(*gh));
257 gh->next = h->handlers;
264 static void *grs_init(RecType recType)
266 struct grs_handlers *h = (struct grs_handlers *) malloc (sizeof(*h));
269 grs_add_handler (h, recTypeGrs_sgml);
270 grs_add_handler (h, recTypeGrs_regx);
272 grs_add_handler (h, recTypeGrs_tcl);
274 grs_add_handler (h, recTypeGrs_marc);
278 static void grs_destroy(void *clientData)
280 struct grs_handlers *h = (struct grs_handlers *) clientData;
281 struct grs_handler *gh = h->handlers, *gh_next;
286 (*gh->type->destroy)(gh->clientData);
293 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
296 (*p->init)(p, &wrd); /* set defaults */
297 for (; n; n = n->next)
299 if (p->flagShowRecords) /* display element description to user */
301 if (n->which == DATA1N_root)
303 printf("%*s", level * 4, "");
304 printf("Record type: '%s'\n", n->u.root.absyn->name);
306 else if (n->which == DATA1N_tag)
310 printf("%*s", level * 4, "");
311 if (!(e = n->u.tag.element))
312 printf("Local tag: '%s'\n", n->u.tag.tag);
315 printf("Elm: '%s' ", e->name);
318 data1_tag *t = e->tag;
320 printf("TagNam: '%s' ", t->names->name);
323 printf("%s[%d],", t->tagset->name, t->tagset->type);
326 if (t->which == DATA1T_numeric)
327 printf("%d)", t->value.numeric);
329 printf("'%s')", t->value.string);
337 if (dumpkeys(n->child, p, level + 1) < 0)
340 if (n->which == DATA1N_data)
342 data1_node *par = get_parent_tag(p->dh, n);
343 data1_termlist *tlist = 0;
344 data1_datatype dtype = DATA1K_string;
346 if (p->flagShowRecords)
348 printf("%*s", level * 4, "");
350 if (n->u.data.len > 32)
351 printf("'%.24s ... %.6s'\n", n->u.data.data,
352 n->u.data.data + n->u.data.len-6);
353 else if (n->u.data.len > 0)
354 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
362 * cycle up towards the root until we find a tag with an att..
363 * this has the effect of indexing locally defined tags with
364 * the attribute of their ancestor in the record.
367 while (!par->u.tag.element)
368 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
370 if (!par || !(tlist = par->u.tag.element->termlists))
372 if (par->u.tag.element->tag)
373 dtype = par->u.tag.element->tag->kind;
374 for (; tlist; tlist = tlist->next)
376 if (p->flagShowRecords)
378 printf("%*sIdx: [%s]", (level + 1) * 4, "",
380 printf("%s:%s [%d]\n",
381 tlist->att->parent->name,
382 tlist->att->name, tlist->att->value);
386 wrd.reg_type = *tlist->structure;
387 wrd.string = n->u.data.data;
388 wrd.length = n->u.data.len;
389 wrd.attrSet = (int) (tlist->att->parent->reference);
390 wrd.attrUse = tlist->att->locals->local;
391 (*p->tokenAdd)(&wrd);
395 if (p->flagShowRecords && n->which == DATA1N_root)
397 printf("%*s-------------\n\n", level * 4, "");
403 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
406 int oidtmp[OID_SIZE];
408 oe.proto = PROTO_Z3950;
409 oe.oclass = CLASS_SCHEMA;
410 oe.value = n->u.root.absyn->reference;
412 if ((oid_ent_to_oid (&oe, oidtmp)))
413 (*p->schemaAdd)(p, oidtmp);
415 return dumpkeys(n, p, 0);
418 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
422 struct grs_read_info gri;
424 int oidtmp[OID_SIZE];
426 gri.readf = p->readf;
427 gri.seekf = p->seekf;
428 gri.tellf = p->tellf;
431 gri.offset = p->offset;
435 if (read_grs_type (h, &gri, p->subType, &n))
436 return RECCTRL_EXTRACT_ERROR;
438 return RECCTRL_EXTRACT_EOF;
439 oe.proto = PROTO_Z3950;
440 oe.oclass = CLASS_SCHEMA;
441 oe.value = n->u.root.absyn->reference;
442 if ((oid_ent_to_oid (&oe, oidtmp)))
443 (*p->schemaAdd)(p, oidtmp);
445 if (dumpkeys(n, p, 0) < 0)
447 data1_free_tree(p->dh, n);
448 return RECCTRL_EXTRACT_ERROR;
450 data1_free_tree(p->dh, n);
451 return RECCTRL_EXTRACT_OK;
454 static int grs_extract(void *clientData, struct recExtractCtrl *p)
457 NMEM mem = nmem_create ();
458 struct grs_handlers *h = (struct grs_handlers *) clientData;
460 ret = grs_extract_sub(h, p, mem);
466 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
468 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
470 data1_esetname *eset;
476 case Z_RecordComp_simple:
477 if (c->u.simple->which != Z_ElementSetNames_generic)
478 return 26; /* only generic form supported. Fix this later */
479 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
480 c->u.simple->u.generic)))
482 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
483 return 25; /* invalid esetname */
485 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
486 c->u.simple->u.generic);
489 case Z_RecordComp_complex:
490 if (c->u.complex->generic)
492 /* insert check for schema */
493 if ((p = c->u.complex->generic->elementSpec))
497 case Z_ElementSpec_elementSetName:
499 data1_getesetbyname(dh, n->u.root.absyn,
500 p->u.elementSetName)))
502 logf(LOG_LOG, "Unknown esetname '%s'",
503 p->u.elementSetName);
504 return 25; /* invalid esetname */
506 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
507 p->u.elementSetName);
510 case Z_ElementSpec_externalSpec:
511 if (p->u.externalSpec->which == Z_External_espec1)
513 logf(LOG_DEBUG, "Got Espec-1");
514 espec = p->u.externalSpec-> u.espec1;
518 logf(LOG_LOG, "Unknown external espec.");
519 return 25; /* bad. what is proper diagnostic? */
530 logf (LOG_DEBUG, "Element: Espec-1 match");
531 return data1_doespec1(dh, n, espec);
535 logf (LOG_DEBUG, "Element: all match");
540 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
542 data1_node *node = 0, *onode = 0;
545 int res, selected = 0;
547 struct grs_read_info gri;
549 struct grs_handlers *h = (struct grs_handlers *) clientData;
550 int requested_schema = VAL_NONE;
553 gri.readf = p->readf;
554 gri.seekf = p->seekf;
555 gri.tellf = p->tellf;
562 logf (LOG_DEBUG, "grs_retrieve");
563 if (read_grs_type (h, &gri, p->subType, &node))
575 logf (LOG_DEBUG, "grs_retrieve: size");
576 if ((dnew = data1_insert_taggeddata(p->dh, node, node,
579 dnew->u.data.what = DATA1I_text;
580 dnew->u.data.data = dnew->lbuf;
581 sprintf(dnew->u.data.data, "%d", p->recordSize);
582 dnew->u.data.len = strlen(dnew->u.data.data);
585 tagname = res_get_def(p->res, "tagrank", "rank");
586 if (strcmp(tagname, "0") && p->score >= 0 &&
587 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
589 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
590 dnew->u.data.what = DATA1I_num;
591 dnew->u.data.data = dnew->lbuf;
592 sprintf(dnew->u.data.data, "%d", p->score);
593 dnew->u.data.len = strlen(dnew->u.data.data);
596 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
597 if (strcmp(tagname, "0") && p->localno > 0 &&
598 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
600 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
601 dnew->u.data.what = DATA1I_text;
602 dnew->u.data.data = dnew->lbuf;
603 sprintf(dnew->u.data.data, "%d", p->localno);
604 dnew->u.data.len = strlen(dnew->u.data.data);
607 if (p->comp && p->comp->which == Z_RecordComp_complex &&
608 p->comp->u.complex->generic &&
609 p->comp->u.complex->generic->schema)
611 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
613 requested_schema = oe->value;
616 /* If schema has been specified, map if possible, then check that
617 * we got the right one
619 if (requested_schema != VAL_NONE)
621 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
622 for (map = node->u.root.absyn->maptabs; map; map = map->next)
624 if (map->target_absyn_ref == requested_schema)
627 if (!(node = data1_map_record(p->dh, onode, map, mem)))
636 if (node->u.root.absyn &&
637 requested_schema != node->u.root.absyn->reference)
645 * Does the requested format match a known syntax-mapping? (this reflects
646 * the overlap of schema and formatting which is inherent in the MARC
649 logf (LOG_DEBUG, "grs_retrieve: syntax mapping");
650 for (map = node->u.root.absyn->maptabs; map; map = map->next)
652 if (map->target_absyn_ref == p->input_format)
655 if (!(node = data1_map_record(p->dh, onode, map, mem)))
664 logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
665 if (node->u.root.absyn &&
666 node->u.root.absyn->reference != VAL_NONE &&
667 p->input_format == VAL_GRS1)
671 int oidtmp[OID_SIZE];
673 oe.proto = PROTO_Z3950;
674 oe.oclass = CLASS_SCHEMA;
675 oe.value = node->u.root.absyn->reference;
677 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
680 data1_handle dh = p->dh;
684 for (ii = oid; *ii >= 0; ii++)
688 sprintf(p, "%d", *ii);
693 if ((dnew = data1_insert_taggeddata(dh, node, node,
694 "schemaIdentifier", mem)))
696 dnew->u.data.what = DATA1I_oid;
697 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
698 memcpy(dnew->u.data.data, tmp, p - tmp);
699 dnew->u.data.len = p - tmp;
704 logf (LOG_DEBUG, "grs_retrieve: element spec");
705 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
709 data1_free_tree(p->dh, onode);
710 data1_free_tree(p->dh, node);
714 else if (p->comp && !res)
717 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
718 switch (p->output_format = (p->input_format != VAL_NONE ?
719 p->input_format : VAL_SUTRS))
721 data1_marctab *marctab;
725 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
730 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
731 memcpy (new_buf, p->rec_buf, p->rec_len);
732 p->rec_buf = new_buf;
737 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
739 p->diagnostic = 238; /* not available in requested syntax */
741 p->rec_len = (size_t) (-1);
744 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
748 p->rec_len = (size_t) (-1);
751 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
755 p->rec_len = (size_t) (-1);
758 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
763 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
764 memcpy (new_buf, p->rec_buf, p->rec_len);
765 p->rec_buf = new_buf;
769 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
774 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
775 memcpy (new_buf, p->rec_buf, p->rec_len);
776 p->rec_buf = new_buf;
780 if (!node->u.root.absyn)
785 for (marctab = node->u.root.absyn->marc; marctab;
786 marctab = marctab->next)
787 if (marctab->reference == p->input_format)
794 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
800 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
801 memcpy (new_buf, p->rec_buf, p->rec_len);
802 p->rec_buf = new_buf;
806 data1_free_tree(p->dh, node);
808 data1_free_tree(p->dh, onode);
813 static struct recType grs_type =
822 RecType recTypeGrs = &grs_type;