2 * Copyright (C) 1994-1996, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.1 1996-10-11 10:57:25 adam
8 * New module recctrl. Used to manage records (extract/retrieval).
10 * Revision 1.29 1996/10/08 10:30:21 quinn
13 * Revision 1.28 1996/10/07 16:06:40 quinn
16 * Revision 1.27 1996/06/11 10:54:12 quinn
19 * Revision 1.26 1996/06/06 12:08:45 quinn
20 * Added showRecord function
22 * Revision 1.25 1996/06/04 14:18:53 quinn
25 * Revision 1.24 1996/06/04 13:27:54 quinn
26 * More work on charmapping
28 * Revision 1.23 1996/06/04 10:19:01 adam
29 * Minor changes - removed include of ctype.h.
31 * Revision 1.22 1996/06/03 10:15:27 quinn
32 * Various character-mapping.
34 * Revision 1.21 1996/05/31 13:27:24 quinn
35 * Character-conversion in phrases, too.
37 * Revision 1.19 1996/05/16 15:31:14 quinn
40 * Revision 1.18 1996/05/09 07:28:56 quinn
41 * Work towards phrases and multiple registers
43 * Revision 1.17 1996/05/01 13:46:37 adam
44 * First work on multiple records in one file.
45 * New option, -offset, to the "unread" command in the filter module.
47 * Revision 1.16 1996/01/17 14:57:54 adam
48 * Prototype changed for reader functions in extract/retrieve. File
49 * is identified by 'void *' instead of 'int.
51 * Revision 1.15 1996/01/08 19:15:47 adam
52 * New input filter that works!
54 * Revision 1.14 1995/12/15 12:36:11 adam
55 * Retrieval calls data1_read_regx when subType is specified.
57 * Revision 1.13 1995/12/15 12:24:43 quinn
58 * *** empty log message ***
60 * Revision 1.12 1995/12/15 12:20:28 quinn
61 * *** empty log message ***
63 * Revision 1.11 1995/12/15 12:07:57 quinn
64 * Changed extraction strategy.
66 * Revision 1.10 1995/12/14 11:10:48 quinn
69 * Revision 1.9 1995/12/13 17:14:05 quinn
70 * *** empty log message ***
72 * Revision 1.8 1995/12/13 15:33:18 quinn
73 * *** empty log message ***
75 * Revision 1.7 1995/12/13 13:45:39 quinn
76 * Changed data1 to use nmem.
78 * Revision 1.6 1995/12/04 14:22:30 adam
79 * Extra arg to recType_byName.
80 * Started work on new regular expression parsed input to
83 * Revision 1.5 1995/11/28 14:18:37 quinn
86 * Revision 1.4 1995/11/21 13:14:49 quinn
87 * Fixed end-of-data-field problem (maybe).
89 * Revision 1.3 1995/11/15 19:13:09 adam
90 * Work on record management.
96 #include <sys/types.h>
106 #define GRS_MAX_WORD 512
108 static int seqno = 0;
110 static data1_node *read_grs_type (struct grs_read_info *p, const char *type)
114 data1_node *(*func)(struct grs_read_info *p);
116 { "sgml", grs_read_sgml },
117 { "regx", grs_read_regx },
120 const char *cp = strchr (type, '.');
123 if (cp == NULL || cp == type)
125 cp = strlen(type) + type;
129 strcpy (p->type, cp+1);
130 for (i=0; tab[i].type; i++)
132 if (!memcmp (type, tab[i].type, cp-type))
133 return (tab[i].func)(p);
138 static void grs_init(void)
142 static void dumpkeys_word(data1_node *n, struct recExtractCtrl *p,
145 char *b = n->u.data.data;
149 remain = n->u.data.len - (b - n->u.data.data);
151 map = (*p->map_chrs_input)(&b, remain);
156 char buf[GRS_MAX_WORD+1];
160 while (map && *map && **map == *CHR_SPACE)
162 remain = n->u.data.len - (b - n->u.data.data);
164 map = (*p->map_chrs_input)(&b, remain);
171 while (map && *map && **map != *CHR_SPACE)
175 while (i < GRS_MAX_WORD && *cp)
177 remain = n->u.data.len - (b - n->u.data.data);
179 map = (*p->map_chrs_input)(&b, remain);
186 (*p->init)(&wrd); /* set defaults */
187 wrd.which = Word_String;
190 wrd.attrSet = att->parent->ordinal;
191 wrd.attrUse = att->locals->local;
196 static void dumpkeys_phrase(data1_node *n, struct recExtractCtrl *p,
199 char *b = n->u.data.data;
200 char buf[GRS_MAX_WORD+1], **map = 0;
204 remain = n->u.data.len - (b - n->u.data.data);
206 map = (*p->map_chrs_input)(&b, remain);
208 while (remain > 0 && i < GRS_MAX_WORD)
210 while (map && *map && **map == *CHR_SPACE)
212 remain = n->u.data.len - (b - n->u.data.data);
214 map = (*p->map_chrs_input)(&b, remain);
221 if (i && i < GRS_MAX_WORD)
222 buf[i++] = *CHR_SPACE;
223 while (map && *map && **map != *CHR_SPACE)
227 if (i >= GRS_MAX_WORD)
229 while (i < GRS_MAX_WORD && *cp)
231 remain = n->u.data.len - (b - n->u.data.data);
233 map = (*p->map_chrs_input)(&b, remain);
242 wrd.which = Word_Phrase;
245 wrd.attrSet = att->parent->ordinal;
246 wrd.attrUse = att->locals->local;
250 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level)
252 for (; n; n = n->next)
254 if (p->flagShowRecords) /* display element description to user */
256 if (n->which == DATA1N_root)
258 printf("%*s", level * 4, "");
259 printf("Record type: '%s'\n", n->u.root.absyn->name);
261 else if (n->which == DATA1N_tag)
265 printf("%*s", level * 4, "");
266 if (!(e = n->u.tag.element))
267 printf("Local tag: '%s'\n", n->u.tag.tag);
270 printf("Elm: '%s' ", e->name);
273 data1_tag *t = e->tag;
275 printf("TagNam: '%s' ", t->names->name);
278 printf("%s[%d],", t->tagset->name, t->tagset->type);
281 if (t->which == DATA1T_numeric)
282 printf("%d)", t->value.numeric);
284 printf("'%s')", t->value.string);
292 if (dumpkeys(n->child, p, level + 1) < 0)
295 if (n->which == DATA1N_data)
297 data1_node *par = get_parent_tag(n);
298 data1_termlist *tlist = 0;
300 if (p->flagShowRecords)
302 printf("%*s", level * 4, "");
304 if (n->u.data.len > 20)
305 printf("'%.20s...'\n", n->u.data.data);
306 else if (n->u.data.len > 0)
307 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
315 * cycle up towards the root until we find a tag with an att..
316 * this has the effect of indexing locally defined tags with
317 * the attribute of their ancestor in the record.
320 while (!par->u.tag.element)
321 if (!par->parent || !(par = get_parent_tag(par->parent)))
325 else if (par->u.tag.element->termlists)
326 tlist = par->u.tag.element->termlists;
330 for (; tlist; tlist = tlist->next)
332 if (p->flagShowRecords)
334 printf("%*sIdx: [", (level + 1) * 4, "");
335 switch (tlist->structure)
337 case DATA1S_word: printf("w"); break;
338 case DATA1S_phrase: printf("p"); break;
339 default: printf("?"); break;
342 printf("%s:%s [%d]\n", tlist->att->parent->name,
343 tlist->att->name, tlist->att->value);
345 else switch (tlist->structure)
348 dumpkeys_word(n, p, tlist->att); break;
350 dumpkeys_phrase(n, p, tlist->att); break;
352 logf(LOG_FATAL, "Bad structure type in dumpkeys");
357 if (p->flagShowRecords && n->which == DATA1N_root)
359 printf("%*s-------------\n\n", level * 4, "");
365 static int grs_extract(struct recExtractCtrl *p)
368 NMEM mem = nmem_create();
369 struct grs_read_info gri;
372 gri.readf = p->readf;
373 gri.seekf = p->seekf;
376 gri.offset = p->offset;
379 n = read_grs_type (&gri, p->subType);
382 if (dumpkeys(n, p, 0) < 0)
393 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
395 static int process_comp(data1_node *n, Z_RecordComposition *c)
397 data1_esetname *eset;
403 case Z_RecordComp_simple:
404 if (c->u.simple->which != Z_ElementSetNames_generic)
405 return 26; /* only generic form supported. Fix this later */
406 if (!(eset = data1_getesetbyname(n->u.root.absyn,
407 c->u.simple->u.generic)))
409 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
410 return 25; /* invalid esetname */
412 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
413 c->u.simple->u.generic);
416 case Z_RecordComp_complex:
417 if (c->u.complex->generic)
419 /* insert check for schema */
420 if ((p = c->u.complex->generic->elementSpec))
423 case Z_ElementSpec_elementSetName:
424 if (!(eset = data1_getesetbyname(n->u.root.absyn,
425 p->u.elementSetName)))
427 logf(LOG_LOG, "Unknown esetname '%s'",
428 p->u.elementSetName);
429 return 25; /* invalid esetname */
431 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
432 p->u.elementSetName);
435 case Z_ElementSpec_externalSpec:
436 if (p->u.externalSpec->which == Z_External_espec1)
438 logf(LOG_DEBUG, "Got Espec-1");
439 espec = p->u.externalSpec-> u.espec1;
443 logf(LOG_LOG, "Unknown external espec.");
444 return 25; /* bad. what is proper diagnostic? */
453 return data1_doespec1(n, espec);
458 static int grs_retrieve(struct recRetrieveCtrl *p)
460 data1_node *node = 0, *onode = 0;
463 int res, selected = 0;
464 NMEM mem = nmem_create();
465 struct grs_read_info gri;
467 gri.readf = p->readf;
468 gri.seekf = p->seekf;
474 node = read_grs_type (&gri, p->subType);
475 /* node = data1_read_record(p->readf, p->fh, mem); */
481 if (p->score >= 0 && (new = data1_insert_taggeddata(node, node, "rank",
484 new->u.data.what = DATA1I_num;
485 new->u.data.data = new->u.data.lbuf;
486 sprintf(new->u.data.data, "%d", p->score);
487 new->u.data.len = strlen(new->u.data.data);
489 if ((new = data1_insert_taggeddata(node, node, "localControlNumber", mem)))
491 new->u.data.what = DATA1I_text;
492 new->u.data.data = new->u.data.lbuf;
493 sprintf(new->u.data.data, "%d", p->localno);
494 new->u.data.len = strlen(new->u.data.data);
496 if (p->input_format == VAL_GRS1 && node->u.root.absyn &&
497 node->u.root.absyn->reference != VAL_NONE)
502 oe.proto = PROTO_Z3950;
503 oe.oclass = CLASS_SCHEMA;
504 oe.value = node->u.root.absyn->reference;
506 if ((oid = oid_getoidbyent(&oe)))
512 for (ii = oid; *ii >= 0; ii++)
516 sprintf(p, "%d", *ii);
521 if ((new = data1_insert_taggeddata(node, node, "schemaIdentifier",
524 new->u.data.what = DATA1I_oid;
525 new->u.data.data = nmem_malloc(mem, p - tmp);
526 memcpy(new->u.data.data, tmp, p - tmp);
527 new->u.data.len = p - tmp;
533 * Does the requested format match a known schema-mapping? (this reflects
534 * the overlap of schema and formatting which is inherent in the MARC
536 * NOTE: This should look at the schema-specification in the compspec
539 for (map = node->u.root.absyn->maptabs; map; map = map->next)
540 if (map->target_absyn_ref == p->input_format)
543 if (!(node = data1_map_record(onode, map, mem)))
552 if (p->comp && (res = process_comp(node, p->comp)) > 0)
556 data1_free_tree(onode);
557 data1_free_tree(node);
561 else if (p->comp && !res)
564 switch (p->output_format = (p->input_format != VAL_NONE ?
565 p->input_format : VAL_SUTRS))
567 data1_marctab *marctab;
570 if (!(p->rec_buf = data1_nodetogr(node, selected, p->odr)))
571 p->diagnostic = 2; /* this should be better specified */
576 if (!(p->rec_buf = data1_nodetoexplain(node, selected, p->odr)))
577 p->diagnostic = 2; /* this should be better specified */
582 if (!(p->rec_buf = data1_nodetosummary(node, selected, p->odr)))
588 if (!(p->rec_buf = data1_nodetobuf(node, selected,
596 if (!(p->rec_buf = data1_nodetosoif(node, selected,
604 for (marctab = node->u.root.absyn->marc; marctab;
605 marctab = marctab->next)
606 if (marctab->reference == p->input_format)
613 if (!(p->rec_buf = data1_nodetomarc(marctab, node, selected,
621 data1_free_tree(node);
623 data1_free_tree(onode);
628 static struct recType grs_type =
636 RecType recTypeGrs = &grs_type;