2 * Copyright (C) 1994-2002, Index Data
5 * $Id: recgrs.c,v 1.56 2002-08-01 08:53:35 adam Exp $
10 #include <sys/types.h>
21 #define GRS_MAX_WORD 512
27 struct grs_handler *next;
31 struct grs_handler *handlers;
34 static int read_grs_type (struct grs_handlers *h,
35 struct grs_read_info *p, const char *type,
38 struct grs_handler *gh = h->handlers;
39 const char *cp = strchr (type, '.');
41 if (cp == NULL || cp == type)
43 cp = strlen(type) + type;
47 strcpy (p->type, cp+1);
48 for (gh = h->handlers; gh; gh = gh->next)
50 if (!memcmp (type, gh->type->type, cp-type))
55 gh->clientData = (*gh->type->init)();
57 p->clientData = gh->clientData;
58 *root = (gh->type->read)(p);
59 gh->clientData = p->clientData;
66 static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t)
68 struct grs_handler *gh = (struct grs_handler *) xmalloc (sizeof(*gh));
69 gh->next = h->handlers;
76 static void *grs_init(RecType recType)
78 struct grs_handlers *h = (struct grs_handlers *) xmalloc (sizeof(*h));
81 grs_add_handler (h, recTypeGrs_sgml);
82 grs_add_handler (h, recTypeGrs_regx);
84 grs_add_handler (h, recTypeGrs_tcl);
86 grs_add_handler (h, recTypeGrs_marc);
88 grs_add_handler (h, recTypeGrs_xml);
93 static void grs_destroy(void *clientData)
95 struct grs_handlers *h = (struct grs_handlers *) clientData;
96 struct grs_handler *gh = h->handlers, *gh_next;
101 (*gh->type->destroy)(gh->clientData);
109 1 start element (tag)
111 3 start attr (and attr-exact)
118 static void index_xpath (data1_node *n, struct recExtractCtrl *p,
119 int level, RecWord *wrd, int use)
122 char tag_path_full[1024];
130 wrd->string = n->u.data.data;
131 wrd->length = n->u.data.len;
132 wrd->attrSet = VAL_IDXPATH,
134 if (p->flagShowRecords)
136 printf("%*s data=", (level + 1) * 4, "");
137 for (i = 0; i<wrd->length && i < 8; i++)
138 fputc (wrd->string[i], stdout);
147 for (nn = n; nn; nn = nn->parent)
149 if (nn->which == DATA1N_tag)
151 size_t tlen = strlen(nn->u.tag.tag);
152 if (tlen + flen > (sizeof(tag_path_full)-2))
154 memcpy (tag_path_full + flen, nn->u.tag.tag, tlen);
156 tag_path_full[flen++] = '/';
158 else if (nn->which == DATA1N_root)
162 wrd->string = tag_path_full;
164 wrd->attrSet = VAL_IDXPATH;
166 if (p->flagShowRecords)
168 printf("%*s tag=", (level + 1) * 4, "");
169 for (i = 0; i<wrd->length && i < 40; i++)
170 fputc (wrd->string[i], stdout);
180 for (xp = n->u.tag.attributes; xp; xp = xp->next)
185 { /* attribute start */
188 wrd->string = xp->name;
189 wrd->length = strlen(xp->name);
195 if (use == 1 && xp->value &&
196 strlen(xp->name) + strlen(xp->value) < sizeof(comb)-2)
198 /* attribute value exact */
199 strcpy (comb, xp->name);
201 strcat (comb, xp->value);
206 wrd->length = strlen(comb);
211 /* attribute value phrase */
215 wrd->string = xp->value;
216 wrd->length = strlen(xp->value);
224 wrd->string = xp->name;
225 wrd->length = strlen(xp->name);
235 static void index_termlist (data1_node *par, data1_node *n,
236 struct recExtractCtrl *p, int level, RecWord *wrd)
238 data1_termlist *tlist = 0;
239 data1_datatype dtype = DATA1K_string;
241 * cycle up towards the root until we find a tag with an att..
242 * this has the effect of indexing locally defined tags with
243 * the attribute of their ancestor in the record.
246 while (!par->u.tag.element)
247 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
249 if (!par || !(tlist = par->u.tag.element->termlists))
251 if (par->u.tag.element->tag)
252 dtype = par->u.tag.element->tag->kind;
254 for (; tlist; tlist = tlist->next)
257 /* consider source */
260 if (!strcmp (tlist->source, "data") && n->which == DATA1N_data)
262 wrd->string = n->u.data.data;
263 wrd->length = n->u.data.len;
265 else if (!strcmp (tlist->source, "tag") && n->which == DATA1N_tag)
267 wrd->string = n->u.tag.tag;
268 wrd->length = strlen(n->u.tag.tag);
270 else if (sscanf (tlist->source, "attr(%511[^)])", xattr) == 1 &&
271 n->which == DATA1N_tag)
273 data1_xattr *p = n->u.tag.attributes;
274 while (p && strcmp (p->name, xattr))
278 wrd->string = p->value;
279 wrd->length = strlen(p->value);
284 if (p->flagShowRecords)
287 printf("%*sIdx: [%s]", (level + 1) * 4, "",
289 printf("%s:%s [%d] %s",
290 tlist->att->parent->name,
291 tlist->att->name, tlist->att->value,
294 for (i = 0; i<wrd->length && i < 8; i++)
295 fputc (wrd->string[i], stdout);
299 fputc ('\n', stdout);
303 wrd->reg_type = *tlist->structure;
304 wrd->attrSet = (int) (tlist->att->parent->reference);
305 wrd->attrUse = tlist->att->locals->local;
312 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level,
315 for (; n; n = n->next)
317 if (p->flagShowRecords) /* display element description to user */
319 if (n->which == DATA1N_root)
321 printf("%*s", level * 4, "");
322 printf("Record type: '%s'\n", n->u.root.type);
324 else if (n->which == DATA1N_tag)
328 printf("%*s", level * 4, "");
329 if (!(e = n->u.tag.element))
330 printf("Local tag: '%s'\n", n->u.tag.tag);
333 printf("Elm: '%s' ", e->name);
336 data1_tag *t = e->tag;
338 printf("TagNam: '%s' ", t->names->name);
341 printf("%s[%d],", t->tagset->name, t->tagset->type);
344 if (t->which == DATA1T_numeric)
345 printf("%d)", t->value.numeric);
347 printf("'%s')", t->value.string);
354 if (n->which == DATA1N_tag)
356 index_termlist (n, n, p, level, wrd);
357 /* index start tag */
358 if (!n->root->u.root.absyn)
359 index_xpath (n, p, level, wrd, 1);
363 if (dumpkeys(n->child, p, level + 1, wrd) < 0)
367 if (n->which == DATA1N_data)
369 data1_node *par = get_parent_tag(p->dh, n);
371 if (p->flagShowRecords)
373 printf("%*s", level * 4, "");
375 if (n->u.data.len > 32)
376 printf("'%.24s ... %.6s'\n", n->u.data.data,
377 n->u.data.data + n->u.data.len-6);
378 else if (n->u.data.len > 0)
379 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
385 index_termlist (par, n, p, level, wrd);
386 if (!n->root->u.root.absyn)
387 index_xpath (n, p, level, wrd, 1016);
391 if (n->which == DATA1N_tag)
394 if (!n->root->u.root.absyn)
395 index_xpath (n, p, level, wrd, 2);
399 if (p->flagShowRecords && n->which == DATA1N_root)
401 printf("%*s-------------\n\n", level * 4, "");
407 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
410 int oidtmp[OID_SIZE];
413 oe.proto = PROTO_Z3950;
414 oe.oclass = CLASS_SCHEMA;
417 oe.value = n->u.root.absyn->reference;
419 if ((oid_ent_to_oid (&oe, oidtmp)))
420 (*p->schemaAdd)(p, oidtmp);
424 return dumpkeys(n, p, 0, &wrd);
427 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
431 struct grs_read_info gri;
433 int oidtmp[OID_SIZE];
436 gri.readf = p->readf;
437 gri.seekf = p->seekf;
438 gri.tellf = p->tellf;
441 gri.offset = p->offset;
445 if (read_grs_type (h, &gri, p->subType, &n))
446 return RECCTRL_EXTRACT_ERROR;
448 return RECCTRL_EXTRACT_EOF;
449 oe.proto = PROTO_Z3950;
450 oe.oclass = CLASS_SCHEMA;
452 if (!n->u.root.absyn)
453 return RECCTRL_EXTRACT_ERROR;
457 oe.value = n->u.root.absyn->reference;
458 if ((oid_ent_to_oid (&oe, oidtmp)))
459 (*p->schemaAdd)(p, oidtmp);
462 /* ensure our data1 tree is UTF-8 */
463 data1_iconv (p->dh, mem, n, "UTF-8", data1_get_encoding(p->dh, n));
466 data1_pr_tree (p->dh, n, stdout);
470 if (dumpkeys(n, p, 0, &wrd) < 0)
472 data1_free_tree(p->dh, n);
473 return RECCTRL_EXTRACT_ERROR;
475 data1_free_tree(p->dh, n);
476 return RECCTRL_EXTRACT_OK;
479 static int grs_extract(void *clientData, struct recExtractCtrl *p)
482 NMEM mem = nmem_create ();
483 struct grs_handlers *h = (struct grs_handlers *) clientData;
485 ret = grs_extract_sub(h, p, mem);
491 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
493 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
495 data1_esetname *eset;
501 case Z_RecordComp_simple:
502 if (c->u.simple->which != Z_ElementSetNames_generic)
503 return 26; /* only generic form supported. Fix this later */
504 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
505 c->u.simple->u.generic)))
507 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
508 return 25; /* invalid esetname */
510 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
511 c->u.simple->u.generic);
514 case Z_RecordComp_complex:
515 if (c->u.complex->generic)
517 /* insert check for schema */
518 if ((p = c->u.complex->generic->elementSpec))
522 case Z_ElementSpec_elementSetName:
524 data1_getesetbyname(dh, n->u.root.absyn,
525 p->u.elementSetName)))
527 logf(LOG_LOG, "Unknown esetname '%s'",
528 p->u.elementSetName);
529 return 25; /* invalid esetname */
531 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
532 p->u.elementSetName);
535 case Z_ElementSpec_externalSpec:
536 if (p->u.externalSpec->which == Z_External_espec1)
538 logf(LOG_DEBUG, "Got Espec-1");
539 espec = p->u.externalSpec-> u.espec1;
543 logf(LOG_LOG, "Unknown external espec.");
544 return 25; /* bad. what is proper diagnostic? */
555 logf (LOG_DEBUG, "Element: Espec-1 match");
556 return data1_doespec1(dh, n, espec);
560 logf (LOG_DEBUG, "Element: all match");
565 static void add_idzebra_info (struct recRetrieveCtrl *p, data1_node *top,
568 const char *idzebra_ns[7];
570 idzebra_ns[0] = "xmlns:idzebra";
571 idzebra_ns[1] = "http://www.indexdata.dk/zebra/";
574 data1_tag_add_attr (p->dh, mem, top, idzebra_ns);
576 data1_mk_tag_data_int (p->dh, top, "idzebra:size", p->recordSize,
579 data1_mk_tag_data_int (p->dh, top, "idzebra:score",
582 data1_mk_tag_data_int (p->dh, top, "idzebra:localnumber", p->localno,
585 data1_mk_tag_data_text(p->dh, top, "idzebra:filename",
589 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
591 data1_node *node = 0, *onode = 0, *top;
594 int res, selected = 0;
596 struct grs_read_info gri;
598 struct grs_handlers *h = (struct grs_handlers *) clientData;
599 int requested_schema = VAL_NONE;
600 data1_marctab *marctab;
604 gri.readf = p->readf;
605 gri.seekf = p->seekf;
606 gri.tellf = p->tellf;
613 logf (LOG_DEBUG, "grs_retrieve");
614 if (read_grs_type (h, &gri, p->subType, &node))
626 /* ensure our data1 tree is UTF-8 */
627 data1_iconv (p->dh, mem, node, "UTF-8", data1_get_encoding(p->dh, node));
630 data1_pr_tree (p->dh, node, stdout);
632 top = data1_get_root_tag (p->dh, node);
634 logf (LOG_DEBUG, "grs_retrieve: size");
635 if ((dnew = data1_mk_tag_data_wd(p->dh, top, "size", mem)))
637 dnew->u.data.what = DATA1I_text;
638 dnew->u.data.data = dnew->lbuf;
639 sprintf(dnew->u.data.data, "%d", p->recordSize);
640 dnew->u.data.len = strlen(dnew->u.data.data);
643 tagname = res_get_def(p->res, "tagrank", "rank");
644 if (strcmp(tagname, "0") && p->score >= 0 &&
645 (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
647 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
648 dnew->u.data.what = DATA1I_num;
649 dnew->u.data.data = dnew->lbuf;
650 sprintf(dnew->u.data.data, "%d", p->score);
651 dnew->u.data.len = strlen(dnew->u.data.data);
654 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
655 if (strcmp(tagname, "0") && p->localno > 0 &&
656 (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
658 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
659 dnew->u.data.what = DATA1I_text;
660 dnew->u.data.data = dnew->lbuf;
662 sprintf(dnew->u.data.data, "%d", p->localno);
663 dnew->u.data.len = strlen(dnew->u.data.data);
666 data1_pr_tree (p->dh, node, stdout);
668 if (p->comp && p->comp->which == Z_RecordComp_complex &&
669 p->comp->u.complex->generic &&
670 p->comp->u.complex->generic->schema)
672 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
674 requested_schema = oe->value;
677 /* If schema has been specified, map if possible, then check that
678 * we got the right one
680 if (requested_schema != VAL_NONE)
682 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
683 for (map = node->u.root.absyn->maptabs; map; map = map->next)
685 if (map->target_absyn_ref == requested_schema)
688 if (!(node = data1_map_record(p->dh, onode, map, mem)))
697 if (node->u.root.absyn &&
698 requested_schema != node->u.root.absyn->reference)
706 * Does the requested format match a known syntax-mapping? (this reflects
707 * the overlap of schema and formatting which is inherent in the MARC
710 yaz_log (LOG_DEBUG, "grs_retrieve: syntax mapping");
711 if (node->u.root.absyn)
712 for (map = node->u.root.absyn->maptabs; map; map = map->next)
714 if (map->target_absyn_ref == p->input_format)
717 if (!(node = data1_map_record(p->dh, onode, map, mem)))
726 yaz_log (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
727 if (node->u.root.absyn &&
728 node->u.root.absyn->reference != VAL_NONE &&
729 p->input_format == VAL_GRS1)
733 int oidtmp[OID_SIZE];
735 oe.proto = PROTO_Z3950;
736 oe.oclass = CLASS_SCHEMA;
737 oe.value = node->u.root.absyn->reference;
739 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
742 data1_handle dh = p->dh;
746 for (ii = oid; *ii >= 0; ii++)
750 sprintf(p, "%d", *ii);
755 if ((dnew = data1_mk_tag_data_wd(dh, node,
756 "schemaIdentifier", mem)))
758 dnew->u.data.what = DATA1I_oid;
759 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
760 memcpy(dnew->u.data.data, tmp, p - tmp);
761 dnew->u.data.len = p - tmp;
766 logf (LOG_DEBUG, "grs_retrieve: element spec");
767 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
771 data1_free_tree(p->dh, onode);
772 data1_free_tree(p->dh, node);
776 else if (p->comp && !res)
780 data1_pr_tree (p->dh, node, stdout);
782 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
783 switch (p->output_format = (p->input_format != VAL_NONE ?
784 p->input_format : VAL_SUTRS))
787 add_idzebra_info (p, top, mem);
790 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
792 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
797 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
798 memcpy (new_buf, p->rec_buf, p->rec_len);
799 p->rec_buf = new_buf;
804 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
806 p->diagnostic = 238; /* not available in requested syntax */
808 p->rec_len = (size_t) (-1);
811 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
815 p->rec_len = (size_t) (-1);
818 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
822 p->rec_len = (size_t) (-1);
826 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
827 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
832 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
833 memcpy (new_buf, p->rec_buf, p->rec_len);
834 p->rec_buf = new_buf;
838 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
843 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
844 memcpy (new_buf, p->rec_buf, p->rec_len);
845 p->rec_buf = new_buf;
849 if (!node->u.root.absyn)
854 for (marctab = node->u.root.absyn->marc; marctab;
855 marctab = marctab->next)
856 if (marctab->reference == p->input_format)
864 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
865 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
866 selected, &p->rec_len)))
870 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
871 memcpy (new_buf, p->rec_buf, p->rec_len);
872 p->rec_buf = new_buf;
876 data1_free_tree(p->dh, node);
878 data1_free_tree(p->dh, onode);
883 static struct recType grs_type =
892 RecType recTypeGrs = &grs_type;