2 * Copyright (C) 1994-1996, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.5 1997-07-15 16:29:03 adam
8 * Initialized dummy variable to keep checker gcc happy.
10 * Revision 1.4 1997/04/30 08:56:08 quinn
13 * Revision 1.2 1996/10/11 16:06:43 quinn
14 * Revision 1.3 1997/02/24 10:41:50 adam
15 * Cleanup of code and commented out the "end element-end-record" code.
17 * Revision 1.2 1996/10/11 16:06:43 quinn
18 * Fixed arguments to nodetogr
20 * Revision 1.1 1996/10/11 10:57:25 adam
21 * New module recctrl. Used to manage records (extract/retrieval).
23 * Revision 1.29 1996/10/08 10:30:21 quinn
26 * Revision 1.28 1996/10/07 16:06:40 quinn
29 * Revision 1.27 1996/06/11 10:54:12 quinn
32 * Revision 1.26 1996/06/06 12:08:45 quinn
33 * Added showRecord function
35 * Revision 1.25 1996/06/04 14:18:53 quinn
38 * Revision 1.24 1996/06/04 13:27:54 quinn
39 * More work on charmapping
41 * Revision 1.23 1996/06/04 10:19:01 adam
42 * Minor changes - removed include of ctype.h.
44 * Revision 1.22 1996/06/03 10:15:27 quinn
45 * Various character-mapping.
47 * Revision 1.21 1996/05/31 13:27:24 quinn
48 * Character-conversion in phrases, too.
50 * Revision 1.19 1996/05/16 15:31:14 quinn
53 * Revision 1.18 1996/05/09 07:28:56 quinn
54 * Work towards phrases and multiple registers
56 * Revision 1.17 1996/05/01 13:46:37 adam
57 * First work on multiple records in one file.
58 * New option, -offset, to the "unread" command in the filter module.
60 * Revision 1.16 1996/01/17 14:57:54 adam
61 * Prototype changed for reader functions in extract/retrieve. File
62 * is identified by 'void *' instead of 'int.
64 * Revision 1.15 1996/01/08 19:15:47 adam
65 * New input filter that works!
67 * Revision 1.14 1995/12/15 12:36:11 adam
68 * Retrieval calls data1_read_regx when subType is specified.
70 * Revision 1.13 1995/12/15 12:24:43 quinn
71 * *** empty log message ***
73 * Revision 1.12 1995/12/15 12:20:28 quinn
74 * *** empty log message ***
76 * Revision 1.11 1995/12/15 12:07:57 quinn
77 * Changed extraction strategy.
79 * Revision 1.10 1995/12/14 11:10:48 quinn
82 * Revision 1.9 1995/12/13 17:14:05 quinn
83 * *** empty log message ***
85 * Revision 1.8 1995/12/13 15:33:18 quinn
86 * *** empty log message ***
88 * Revision 1.7 1995/12/13 13:45:39 quinn
89 * Changed data1 to use nmem.
91 * Revision 1.6 1995/12/04 14:22:30 adam
92 * Extra arg to recType_byName.
93 * Started work on new regular expression parsed input to
96 * Revision 1.5 1995/11/28 14:18:37 quinn
99 * Revision 1.4 1995/11/21 13:14:49 quinn
100 * Fixed end-of-data-field problem (maybe).
102 * Revision 1.3 1995/11/15 19:13:09 adam
103 * Work on record management.
109 #include <sys/types.h>
119 #define GRS_MAX_WORD 512
121 static int seqno = 0;
123 static data1_node *read_grs_type (struct grs_read_info *p, const char *type)
127 data1_node *(*func)(struct grs_read_info *p);
129 { "sgml", grs_read_sgml },
130 { "regx", grs_read_regx },
133 const char *cp = strchr (type, '.');
136 if (cp == NULL || cp == type)
138 cp = strlen(type) + type;
142 strcpy (p->type, cp+1);
143 for (i=0; tab[i].type; i++)
145 if (!memcmp (type, tab[i].type, cp-type))
146 return (tab[i].func)(p);
151 static void grs_init(void)
155 static void dumpkeys_word(data1_node *n, struct recExtractCtrl *p,
158 char *b = n->u.data.data;
162 remain = n->u.data.len - (b - n->u.data.data);
164 map = (*p->map_chrs_input)(&b, remain);
169 char buf[GRS_MAX_WORD+1];
173 while (map && *map && **map == *CHR_SPACE)
175 remain = n->u.data.len - (b - n->u.data.data);
177 map = (*p->map_chrs_input)(&b, remain);
184 while (map && *map && **map != *CHR_SPACE)
188 while (i < GRS_MAX_WORD && *cp)
190 remain = n->u.data.len - (b - n->u.data.data);
192 map = (*p->map_chrs_input)(&b, remain);
199 (*p->init)(&wrd); /* set defaults */
200 wrd.which = Word_String;
203 wrd.attrSet = att->parent->ordinal;
204 wrd.attrUse = att->locals->local;
209 static void dumpkeys_phrase(data1_node *n, struct recExtractCtrl *p,
212 char *b = n->u.data.data;
213 char buf[GRS_MAX_WORD+1], **map = 0;
217 remain = n->u.data.len - (b - n->u.data.data);
219 map = (*p->map_chrs_input)(&b, remain);
221 while (remain > 0 && i < GRS_MAX_WORD)
223 while (map && *map && **map == *CHR_SPACE)
225 remain = n->u.data.len - (b - n->u.data.data);
227 map = (*p->map_chrs_input)(&b, remain);
234 if (i && i < GRS_MAX_WORD)
235 buf[i++] = *CHR_SPACE;
236 while (map && *map && **map != *CHR_SPACE)
240 if (i >= GRS_MAX_WORD)
242 while (i < GRS_MAX_WORD && *cp)
244 remain = n->u.data.len - (b - n->u.data.data);
246 map = (*p->map_chrs_input)(&b, remain);
255 wrd.which = Word_Phrase;
258 wrd.attrSet = att->parent->ordinal;
259 wrd.attrUse = att->locals->local;
263 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
265 for (; n; n = n->next)
267 if (p->flagShowRecords) /* display element description to user */
269 if (n->which == DATA1N_root)
271 printf("%*s", level * 4, "");
272 printf("Record type: '%s'\n", n->u.root.absyn->name);
274 else if (n->which == DATA1N_tag)
278 printf("%*s", level * 4, "");
279 if (!(e = n->u.tag.element))
280 printf("Local tag: '%s'\n", n->u.tag.tag);
283 printf("Elm: '%s' ", e->name);
286 data1_tag *t = e->tag;
288 printf("TagNam: '%s' ", t->names->name);
291 printf("%s[%d],", t->tagset->name, t->tagset->type);
294 if (t->which == DATA1T_numeric)
295 printf("%d)", t->value.numeric);
297 printf("'%s')", t->value.string);
305 if (dumpkeys(n->child, p, level + 1) < 0)
308 if (n->which == DATA1N_data)
310 data1_node *par = get_parent_tag(n);
311 data1_termlist *tlist = 0;
313 if (p->flagShowRecords)
315 printf("%*s", level * 4, "");
317 if (n->u.data.len > 20)
318 printf("'%.20s...'\n", n->u.data.data);
319 else if (n->u.data.len > 0)
320 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
328 * cycle up towards the root until we find a tag with an att..
329 * this has the effect of indexing locally defined tags with
330 * the attribute of their ancestor in the record.
333 while (!par->u.tag.element)
334 if (!par->parent || !(par = get_parent_tag(par->parent)))
338 else if (par->u.tag.element->termlists)
339 tlist = par->u.tag.element->termlists;
343 for (; tlist; tlist = tlist->next)
345 if (p->flagShowRecords)
347 printf("%*sIdx: [", (level + 1) * 4, "");
348 switch (tlist->structure)
350 case DATA1S_word: printf("w"); break;
351 case DATA1S_phrase: printf("p"); break;
352 default: printf("?"); break;
355 printf("%s:%s [%d]\n", tlist->att->parent->name,
356 tlist->att->name, tlist->att->value);
358 else switch (tlist->structure)
361 dumpkeys_word(n, p, tlist->att); break;
363 dumpkeys_phrase(n, p, tlist->att); break;
365 logf(LOG_FATAL, "Bad structure type in dumpkeys");
370 if (p->flagShowRecords && n->which == DATA1N_root)
372 printf("%*s-------------\n\n", level * 4, "");
378 static int grs_extract(struct recExtractCtrl *p)
381 NMEM mem = nmem_create();
382 struct grs_read_info gri;
385 gri.readf = p->readf;
386 gri.seekf = p->seekf;
389 gri.offset = p->offset;
392 n = read_grs_type (&gri, p->subType);
395 if (dumpkeys(n, p, 0) < 0)
406 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
408 static int process_comp(data1_node *n, Z_RecordComposition *c)
410 data1_esetname *eset;
416 case Z_RecordComp_simple:
417 if (c->u.simple->which != Z_ElementSetNames_generic)
418 return 26; /* only generic form supported. Fix this later */
419 if (!(eset = data1_getesetbyname(n->u.root.absyn,
420 c->u.simple->u.generic)))
422 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
423 return 25; /* invalid esetname */
425 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
426 c->u.simple->u.generic);
429 case Z_RecordComp_complex:
430 if (c->u.complex->generic)
432 /* insert check for schema */
433 if ((p = c->u.complex->generic->elementSpec))
436 case Z_ElementSpec_elementSetName:
437 if (!(eset = data1_getesetbyname(n->u.root.absyn,
438 p->u.elementSetName)))
440 logf(LOG_LOG, "Unknown esetname '%s'",
441 p->u.elementSetName);
442 return 25; /* invalid esetname */
444 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
445 p->u.elementSetName);
448 case Z_ElementSpec_externalSpec:
449 if (p->u.externalSpec->which == Z_External_espec1)
451 logf(LOG_DEBUG, "Got Espec-1");
452 espec = p->u.externalSpec-> u.espec1;
456 logf(LOG_LOG, "Unknown external espec.");
457 return 25; /* bad. what is proper diagnostic? */
466 return data1_doespec1(n, espec);
471 static int grs_retrieve(struct recRetrieveCtrl *p)
473 data1_node *node = 0, *onode = 0;
476 int res, selected = 0;
477 NMEM mem = nmem_create();
478 struct grs_read_info gri;
480 gri.readf = p->readf;
481 gri.seekf = p->seekf;
487 node = read_grs_type (&gri, p->subType);
488 /* node = data1_read_record(p->readf, p->fh, mem); */
494 if (p->score >= 0 && (new = data1_insert_taggeddata(node, node, "rank",
497 new->u.data.what = DATA1I_num;
498 new->u.data.data = new->u.data.lbuf;
499 sprintf(new->u.data.data, "%d", p->score);
500 new->u.data.len = strlen(new->u.data.data);
502 if ((new = data1_insert_taggeddata(node, node, "localControlNumber", mem)))
504 new->u.data.what = DATA1I_text;
505 new->u.data.data = new->u.data.lbuf;
506 sprintf(new->u.data.data, "%d", p->localno);
507 new->u.data.len = strlen(new->u.data.data);
509 if (p->input_format == VAL_GRS1 && node->u.root.absyn &&
510 node->u.root.absyn->reference != VAL_NONE)
515 oe.proto = PROTO_Z3950;
516 oe.oclass = CLASS_SCHEMA;
517 oe.value = node->u.root.absyn->reference;
519 if ((oid = oid_getoidbyent(&oe)))
525 for (ii = oid; *ii >= 0; ii++)
529 sprintf(p, "%d", *ii);
534 if ((new = data1_insert_taggeddata(node, node, "schemaIdentifier",
537 new->u.data.what = DATA1I_oid;
538 new->u.data.data = nmem_malloc(mem, p - tmp);
539 memcpy(new->u.data.data, tmp, p - tmp);
540 new->u.data.len = p - tmp;
546 * Does the requested format match a known schema-mapping? (this reflects
547 * the overlap of schema and formatting which is inherent in the MARC
549 * NOTE: This should look at the schema-specification in the compspec
552 for (map = node->u.root.absyn->maptabs; map; map = map->next)
553 if (map->target_absyn_ref == p->input_format)
556 if (!(node = data1_map_record(onode, map, mem)))
565 if (p->comp && (res = process_comp(node, p->comp)) > 0)
569 data1_free_tree(onode);
570 data1_free_tree(node);
574 else if (p->comp && !res)
577 switch (p->output_format = (p->input_format != VAL_NONE ?
578 p->input_format : VAL_SUTRS))
580 data1_marctab *marctab;
585 if (!(p->rec_buf = data1_nodetogr(node, selected, p->odr, &dummy)))
586 p->diagnostic = 2; /* this should be better specified */
591 if (!(p->rec_buf = data1_nodetoexplain(node, selected, p->odr)))
592 p->diagnostic = 2; /* this should be better specified */
597 if (!(p->rec_buf = data1_nodetosummary(node, selected, p->odr)))
603 if (!(p->rec_buf = data1_nodetobuf(node, selected,
611 if (!(p->rec_buf = data1_nodetosoif(node, selected,
619 for (marctab = node->u.root.absyn->marc; marctab;
620 marctab = marctab->next)
621 if (marctab->reference == p->input_format)
628 if (!(p->rec_buf = data1_nodetomarc(marctab, node, selected,
636 data1_free_tree(node);
638 data1_free_tree(onode);
643 static struct recType grs_type =
651 RecType recTypeGrs = &grs_type;