2 * Copyright (C) 1994-1999, Index Data
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.34 2000-02-25 13:24:49 adam
8 * Fixed bug regarding pointer conversion that showed up on OSF V5.
10 * Revision 1.33 1999/11/30 13:48:04 adam
11 * Improved installation. Updated for inclusion of YAZ header files.
13 * Revision 1.32 1999/09/07 07:19:21 adam
14 * Work on character mapping. Implemented replace rules.
16 * Revision 1.31 1999/07/14 10:56:43 adam
17 * Fixed potential memory leak.
19 * Revision 1.30 1999/07/06 12:26:41 adam
20 * Retrieval handler obeys schema and handles XML transfer syntax.
22 * Revision 1.29 1999/05/26 07:49:14 adam
25 * Revision 1.28 1999/05/21 12:00:17 adam
26 * Better diagnostics for extraction process.
28 * Revision 1.27 1999/05/20 12:57:18 adam
29 * Implemented TCL filter. Updated recctrl system.
31 * Revision 1.26 1999/03/02 16:15:44 quinn
32 * Added "tagsysno" and "tagrank" directives to zebra.cfg.
34 * Revision 1.25 1999/02/18 15:01:26 adam
37 * Revision 1.24 1999/02/02 14:51:28 adam
38 * Updated WIN32 code specific sections. Changed header.
40 * Revision 1.23 1998/10/18 07:51:10 adam
41 * Changed one logf call.
43 * Revision 1.22 1998/10/16 08:14:37 adam
44 * Updated record control system.
46 * Revision 1.21 1998/07/01 09:16:10 adam
47 * Element localno only added when it's greater than 0.
49 * Revision 1.20 1998/05/20 10:12:26 adam
50 * Implemented automatic EXPLAIN database maintenance.
51 * Modified Zebra to work with ASN.1 compiled version of YAZ.
53 * Revision 1.19 1998/03/11 11:19:05 adam
54 * Changed the way sequence numbers are generated.
56 * Revision 1.18 1998/03/05 08:41:31 adam
59 * Revision 1.17 1998/02/10 12:03:06 adam
62 * Revision 1.16 1998/01/29 13:38:17 adam
63 * Fixed problem with mapping to record with unknown schema.
65 * Revision 1.15 1998/01/26 10:37:57 adam
68 * Revision 1.14 1997/11/06 11:41:01 adam
69 * Implemented "begin variant" for the sgml.regx filter.
71 * Revision 1.13 1997/10/31 12:35:44 adam
72 * Added a few log statements.
74 * Revision 1.12 1997/10/29 12:02:22 adam
75 * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent.
77 * Revision 1.11 1997/10/27 14:34:00 adam
78 * Work on generic character mapping depending on "structure" field
79 * in abstract syntax file.
81 * Revision 1.10 1997/09/18 08:59:21 adam
82 * Extra generic handle for the character mapping routines.
84 * Revision 1.9 1997/09/17 12:19:21 adam
85 * Zebra version corresponds to YAZ version 1.4.
86 * Changed Zebra server so that it doesn't depend on global common_resource.
88 * Revision 1.8 1997/09/09 13:38:14 adam
89 * Partial port to WIN95/NT.
91 * Revision 1.7 1997/09/05 15:30:10 adam
92 * Changed prototype for chr_map_input - added const.
93 * Added support for C++, headers uses extern "C" for public definitions.
95 * Revision 1.6 1997/09/04 13:54:40 adam
96 * Added MARC filter - type grs.marc.<syntax> where syntax refers
97 * to abstract syntax. New method tellf in retrieve/extract method.
99 * Revision 1.5 1997/07/15 16:29:03 adam
100 * Initialized dummy variable to keep checker gcc happy.
102 * Revision 1.4 1997/04/30 08:56:08 quinn
105 * Revision 1.2 1996/10/11 16:06:43 quinn
106 * Revision 1.3 1997/02/24 10:41:50 adam
107 * Cleanup of code and commented out the "end element-end-record" code.
109 * Revision 1.2 1996/10/11 16:06:43 quinn
110 * Fixed arguments to nodetogr
112 * Revision 1.1 1996/10/11 10:57:25 adam
113 * New module recctrl. Used to manage records (extract/retrieval).
115 * Revision 1.29 1996/10/08 10:30:21 quinn
116 * Fixed type mismatch
118 * Revision 1.28 1996/10/07 16:06:40 quinn
121 * Revision 1.27 1996/06/11 10:54:12 quinn
124 * Revision 1.26 1996/06/06 12:08:45 quinn
125 * Added showRecord function
127 * Revision 1.25 1996/06/04 14:18:53 quinn
130 * Revision 1.24 1996/06/04 13:27:54 quinn
131 * More work on charmapping
133 * Revision 1.23 1996/06/04 10:19:01 adam
134 * Minor changes - removed include of ctype.h.
136 * Revision 1.22 1996/06/03 10:15:27 quinn
137 * Various character-mapping.
139 * Revision 1.21 1996/05/31 13:27:24 quinn
140 * Character-conversion in phrases, too.
142 * Revision 1.19 1996/05/16 15:31:14 quinn
145 * Revision 1.18 1996/05/09 07:28:56 quinn
146 * Work towards phrases and multiple registers
148 * Revision 1.17 1996/05/01 13:46:37 adam
149 * First work on multiple records in one file.
150 * New option, -offset, to the "unread" command in the filter module.
152 * Revision 1.16 1996/01/17 14:57:54 adam
153 * Prototype changed for reader functions in extract/retrieve. File
154 * is identified by 'void *' instead of 'int.
156 * Revision 1.15 1996/01/08 19:15:47 adam
157 * New input filter that works!
159 * Revision 1.14 1995/12/15 12:36:11 adam
160 * Retrieval calls data1_read_regx when subType is specified.
162 * Revision 1.13 1995/12/15 12:24:43 quinn
163 * *** empty log message ***
165 * Revision 1.12 1995/12/15 12:20:28 quinn
166 * *** empty log message ***
168 * Revision 1.11 1995/12/15 12:07:57 quinn
169 * Changed extraction strategy.
171 * Revision 1.10 1995/12/14 11:10:48 quinn
174 * Revision 1.9 1995/12/13 17:14:05 quinn
175 * *** empty log message ***
177 * Revision 1.8 1995/12/13 15:33:18 quinn
178 * *** empty log message ***
180 * Revision 1.7 1995/12/13 13:45:39 quinn
181 * Changed data1 to use nmem.
183 * Revision 1.6 1995/12/04 14:22:30 adam
184 * Extra arg to recType_byName.
185 * Started work on new regular expression parsed input to
186 * structured records.
188 * Revision 1.5 1995/11/28 14:18:37 quinn
191 * Revision 1.4 1995/11/21 13:14:49 quinn
192 * Fixed end-of-data-field problem (maybe).
194 * Revision 1.3 1995/11/15 19:13:09 adam
195 * Work on record management.
201 #include <sys/types.h>
212 #define GRS_MAX_WORD 512
218 struct grs_handler *next;
221 struct grs_handlers {
222 struct grs_handler *handlers;
225 static int read_grs_type (struct grs_handlers *h,
226 struct grs_read_info *p, const char *type,
229 struct grs_handler *gh = h->handlers;
230 const char *cp = strchr (type, '.');
232 if (cp == NULL || cp == type)
234 cp = strlen(type) + type;
238 strcpy (p->type, cp+1);
239 for (gh = h->handlers; gh; gh = gh->next)
241 if (!memcmp (type, gh->type->type, cp-type))
246 gh->clientData = (*gh->type->init)();
248 p->clientData = gh->clientData;
249 *root = (gh->type->read)(p);
250 gh->clientData = p->clientData;
257 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
259 struct grs_handler *gh = (struct grs_handler *) malloc (sizeof(*gh));
260 gh->next = h->handlers;
267 static void *grs_init(RecType recType)
269 struct grs_handlers *h = (struct grs_handlers *) malloc (sizeof(*h));
272 grs_add_handler (h, recTypeGrs_sgml);
273 grs_add_handler (h, recTypeGrs_regx);
275 grs_add_handler (h, recTypeGrs_tcl);
277 grs_add_handler (h, recTypeGrs_marc);
281 static void grs_destroy(void *clientData)
283 struct grs_handlers *h = (struct grs_handlers *) clientData;
284 struct grs_handler *gh = h->handlers, *gh_next;
289 (*gh->type->destroy)(gh->clientData);
296 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
299 (*p->init)(p, &wrd); /* set defaults */
300 for (; n; n = n->next)
302 if (p->flagShowRecords) /* display element description to user */
304 if (n->which == DATA1N_root)
306 printf("%*s", level * 4, "");
307 printf("Record type: '%s'\n", n->u.root.absyn->name);
309 else if (n->which == DATA1N_tag)
313 printf("%*s", level * 4, "");
314 if (!(e = n->u.tag.element))
315 printf("Local tag: '%s'\n", n->u.tag.tag);
318 printf("Elm: '%s' ", e->name);
321 data1_tag *t = e->tag;
323 printf("TagNam: '%s' ", t->names->name);
326 printf("%s[%d],", t->tagset->name, t->tagset->type);
329 if (t->which == DATA1T_numeric)
330 printf("%d)", t->value.numeric);
332 printf("'%s')", t->value.string);
340 if (dumpkeys(n->child, p, level + 1) < 0)
343 if (n->which == DATA1N_data)
345 data1_node *par = get_parent_tag(p->dh, n);
346 data1_termlist *tlist = 0;
347 data1_datatype dtype = DATA1K_string;
349 if (p->flagShowRecords)
351 printf("%*s", level * 4, "");
353 if (n->u.data.len > 32)
354 printf("'%.24s ... %.6s'\n", n->u.data.data,
355 n->u.data.data + n->u.data.len-6);
356 else if (n->u.data.len > 0)
357 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
365 * cycle up towards the root until we find a tag with an att..
366 * this has the effect of indexing locally defined tags with
367 * the attribute of their ancestor in the record.
370 while (!par->u.tag.element)
371 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
373 if (!par || !(tlist = par->u.tag.element->termlists))
375 if (par->u.tag.element->tag)
376 dtype = par->u.tag.element->tag->kind;
377 for (; tlist; tlist = tlist->next)
379 if (p->flagShowRecords)
381 printf("%*sIdx: [%s]", (level + 1) * 4, "",
383 printf("%s:%s [%d]\n",
384 tlist->att->parent->name,
385 tlist->att->name, tlist->att->value);
389 wrd.reg_type = *tlist->structure;
390 wrd.string = n->u.data.data;
391 wrd.length = n->u.data.len;
392 wrd.attrSet = (int) (tlist->att->parent->reference);
393 wrd.attrUse = tlist->att->locals->local;
394 (*p->tokenAdd)(&wrd);
398 if (p->flagShowRecords && n->which == DATA1N_root)
400 printf("%*s-------------\n\n", level * 4, "");
406 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
409 int oidtmp[OID_SIZE];
411 oe.proto = PROTO_Z3950;
412 oe.oclass = CLASS_SCHEMA;
413 oe.value = n->u.root.absyn->reference;
415 if ((oid_ent_to_oid (&oe, oidtmp)))
416 (*p->schemaAdd)(p, oidtmp);
418 return dumpkeys(n, p, 0);
421 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
425 struct grs_read_info gri;
427 int oidtmp[OID_SIZE];
429 gri.readf = p->readf;
430 gri.seekf = p->seekf;
431 gri.tellf = p->tellf;
434 gri.offset = p->offset;
438 if (read_grs_type (h, &gri, p->subType, &n))
439 return RECCTRL_EXTRACT_ERROR;
441 return RECCTRL_EXTRACT_EOF;
442 oe.proto = PROTO_Z3950;
443 oe.oclass = CLASS_SCHEMA;
444 oe.value = n->u.root.absyn->reference;
445 if ((oid_ent_to_oid (&oe, oidtmp)))
446 (*p->schemaAdd)(p, oidtmp);
448 if (dumpkeys(n, p, 0) < 0)
450 data1_free_tree(p->dh, n);
451 return RECCTRL_EXTRACT_ERROR;
453 data1_free_tree(p->dh, n);
454 return RECCTRL_EXTRACT_OK;
457 static int grs_extract(void *clientData, struct recExtractCtrl *p)
460 NMEM mem = nmem_create ();
461 struct grs_handlers *h = (struct grs_handlers *) clientData;
463 ret = grs_extract_sub(h, p, mem);
469 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
471 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
473 data1_esetname *eset;
479 case Z_RecordComp_simple:
480 if (c->u.simple->which != Z_ElementSetNames_generic)
481 return 26; /* only generic form supported. Fix this later */
482 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
483 c->u.simple->u.generic)))
485 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
486 return 25; /* invalid esetname */
488 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
489 c->u.simple->u.generic);
492 case Z_RecordComp_complex:
493 if (c->u.complex->generic)
495 /* insert check for schema */
496 if ((p = c->u.complex->generic->elementSpec))
500 case Z_ElementSpec_elementSetName:
502 data1_getesetbyname(dh, n->u.root.absyn,
503 p->u.elementSetName)))
505 logf(LOG_LOG, "Unknown esetname '%s'",
506 p->u.elementSetName);
507 return 25; /* invalid esetname */
509 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
510 p->u.elementSetName);
513 case Z_ElementSpec_externalSpec:
514 if (p->u.externalSpec->which == Z_External_espec1)
516 logf(LOG_DEBUG, "Got Espec-1");
517 espec = p->u.externalSpec-> u.espec1;
521 logf(LOG_LOG, "Unknown external espec.");
522 return 25; /* bad. what is proper diagnostic? */
533 logf (LOG_DEBUG, "Element: Espec-1 match");
534 return data1_doespec1(dh, n, espec);
538 logf (LOG_DEBUG, "Element: all match");
543 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
545 data1_node *node = 0, *onode = 0;
548 int res, selected = 0;
550 struct grs_read_info gri;
552 struct grs_handlers *h = (struct grs_handlers *) clientData;
553 int requested_schema = VAL_NONE;
556 gri.readf = p->readf;
557 gri.seekf = p->seekf;
558 gri.tellf = p->tellf;
565 logf (LOG_DEBUG, "grs_retrieve");
566 if (read_grs_type (h, &gri, p->subType, &node))
578 logf (LOG_DEBUG, "grs_retrieve: size");
579 if ((dnew = data1_insert_taggeddata(p->dh, node, node,
582 dnew->u.data.what = DATA1I_text;
583 dnew->u.data.data = dnew->lbuf;
584 sprintf(dnew->u.data.data, "%d", p->recordSize);
585 dnew->u.data.len = strlen(dnew->u.data.data);
588 tagname = res_get_def(p->res, "tagrank", "rank");
589 if (strcmp(tagname, "0") && p->score >= 0 &&
590 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
592 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
593 dnew->u.data.what = DATA1I_num;
594 dnew->u.data.data = dnew->lbuf;
595 sprintf(dnew->u.data.data, "%d", p->score);
596 dnew->u.data.len = strlen(dnew->u.data.data);
599 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
600 if (strcmp(tagname, "0") && p->localno > 0 &&
601 (dnew = data1_insert_taggeddata(p->dh, node, node, tagname, mem)))
603 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
604 dnew->u.data.what = DATA1I_text;
605 dnew->u.data.data = dnew->lbuf;
606 sprintf(dnew->u.data.data, "%d", p->localno);
607 dnew->u.data.len = strlen(dnew->u.data.data);
610 if (p->comp && p->comp->which == Z_RecordComp_complex &&
611 p->comp->u.complex->generic &&
612 p->comp->u.complex->generic->schema)
614 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
616 requested_schema = oe->value;
619 /* If schema has been specified, map if possible, then check that
620 * we got the right one
622 if (requested_schema != VAL_NONE)
624 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
625 for (map = node->u.root.absyn->maptabs; map; map = map->next)
627 if (map->target_absyn_ref == requested_schema)
630 if (!(node = data1_map_record(p->dh, onode, map, mem)))
639 if (node->u.root.absyn &&
640 requested_schema != node->u.root.absyn->reference)
648 * Does the requested format match a known syntax-mapping? (this reflects
649 * the overlap of schema and formatting which is inherent in the MARC
652 logf (LOG_DEBUG, "grs_retrieve: syntax mapping");
653 for (map = node->u.root.absyn->maptabs; map; map = map->next)
655 if (map->target_absyn_ref == p->input_format)
658 if (!(node = data1_map_record(p->dh, onode, map, mem)))
667 logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
668 if (node->u.root.absyn &&
669 node->u.root.absyn->reference != VAL_NONE &&
670 p->input_format == VAL_GRS1)
674 int oidtmp[OID_SIZE];
676 oe.proto = PROTO_Z3950;
677 oe.oclass = CLASS_SCHEMA;
678 oe.value = node->u.root.absyn->reference;
680 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
683 data1_handle dh = p->dh;
687 for (ii = oid; *ii >= 0; ii++)
691 sprintf(p, "%d", *ii);
696 if ((dnew = data1_insert_taggeddata(dh, node, node,
697 "schemaIdentifier", mem)))
699 dnew->u.data.what = DATA1I_oid;
700 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
701 memcpy(dnew->u.data.data, tmp, p - tmp);
702 dnew->u.data.len = p - tmp;
707 logf (LOG_DEBUG, "grs_retrieve: element spec");
708 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
712 data1_free_tree(p->dh, onode);
713 data1_free_tree(p->dh, node);
717 else if (p->comp && !res)
720 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
721 switch (p->output_format = (p->input_format != VAL_NONE ?
722 p->input_format : VAL_SUTRS))
724 data1_marctab *marctab;
728 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
733 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
734 memcpy (new_buf, p->rec_buf, p->rec_len);
735 p->rec_buf = new_buf;
740 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
742 p->diagnostic = 238; /* not available in requested syntax */
744 p->rec_len = (size_t) (-1);
747 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
751 p->rec_len = (size_t) (-1);
754 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
758 p->rec_len = (size_t) (-1);
761 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
766 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
767 memcpy (new_buf, p->rec_buf, p->rec_len);
768 p->rec_buf = new_buf;
772 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
777 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
778 memcpy (new_buf, p->rec_buf, p->rec_len);
779 p->rec_buf = new_buf;
783 if (!node->u.root.absyn)
788 for (marctab = node->u.root.absyn->marc; marctab;
789 marctab = marctab->next)
790 if (marctab->reference == p->input_format)
797 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
798 selected, &p->rec_len)))
802 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
803 memcpy (new_buf, p->rec_buf, p->rec_len);
804 p->rec_buf = new_buf;
808 data1_free_tree(p->dh, node);
810 data1_free_tree(p->dh, onode);
815 static struct recType grs_type =
824 RecType recTypeGrs = &grs_type;