2 * Copyright (C) 1994-2002, Index Data
5 * $Id: recgrs.c,v 1.57 2002-08-01 09:37:44 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);
181 for (xp = n->u.tag.attributes; xp; xp = xp->next)
184 { /* attribute (no value) */
187 wrd->string = xp->name;
188 wrd->length = strlen(xp->name);
195 for (xp = n->u.tag.attributes; xp; xp = xp->next)
200 { /* attribute start */
203 wrd->string = xp->name;
204 wrd->length = strlen(xp->name);
210 if (use == 1 && xp->value &&
211 strlen(xp->name) + strlen(xp->value) < sizeof(comb)-2)
213 /* attribute value exact */
214 strcpy (comb, xp->name);
216 strcat (comb, xp->value);
221 wrd->length = strlen(comb);
226 /* attribute value phrase */
230 wrd->string = xp->value;
231 wrd->length = strlen(xp->value);
239 wrd->string = xp->name;
240 wrd->length = strlen(xp->name);
251 static void index_termlist (data1_node *par, data1_node *n,
252 struct recExtractCtrl *p, int level, RecWord *wrd)
254 data1_termlist *tlist = 0;
255 data1_datatype dtype = DATA1K_string;
257 * cycle up towards the root until we find a tag with an att..
258 * this has the effect of indexing locally defined tags with
259 * the attribute of their ancestor in the record.
262 while (!par->u.tag.element)
263 if (!par->parent || !(par=get_parent_tag(p->dh, par->parent)))
265 if (!par || !(tlist = par->u.tag.element->termlists))
267 if (par->u.tag.element->tag)
268 dtype = par->u.tag.element->tag->kind;
270 for (; tlist; tlist = tlist->next)
273 /* consider source */
276 if (!strcmp (tlist->source, "data") && n->which == DATA1N_data)
278 wrd->string = n->u.data.data;
279 wrd->length = n->u.data.len;
281 else if (!strcmp (tlist->source, "tag") && n->which == DATA1N_tag)
283 wrd->string = n->u.tag.tag;
284 wrd->length = strlen(n->u.tag.tag);
286 else if (sscanf (tlist->source, "attr(%511[^)])", xattr) == 1 &&
287 n->which == DATA1N_tag)
289 data1_xattr *p = n->u.tag.attributes;
290 while (p && strcmp (p->name, xattr))
294 wrd->string = p->value;
295 wrd->length = strlen(p->value);
300 if (p->flagShowRecords)
303 printf("%*sIdx: [%s]", (level + 1) * 4, "",
305 printf("%s:%s [%d] %s",
306 tlist->att->parent->name,
307 tlist->att->name, tlist->att->value,
310 for (i = 0; i<wrd->length && i < 8; i++)
311 fputc (wrd->string[i], stdout);
315 fputc ('\n', stdout);
319 wrd->reg_type = *tlist->structure;
320 wrd->attrSet = (int) (tlist->att->parent->reference);
321 wrd->attrUse = tlist->att->locals->local;
328 static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level,
331 for (; n; n = n->next)
333 if (p->flagShowRecords) /* display element description to user */
335 if (n->which == DATA1N_root)
337 printf("%*s", level * 4, "");
338 printf("Record type: '%s'\n", n->u.root.type);
340 else if (n->which == DATA1N_tag)
344 printf("%*s", level * 4, "");
345 if (!(e = n->u.tag.element))
346 printf("Local tag: '%s'\n", n->u.tag.tag);
349 printf("Elm: '%s' ", e->name);
352 data1_tag *t = e->tag;
354 printf("TagNam: '%s' ", t->names->name);
357 printf("%s[%d],", t->tagset->name, t->tagset->type);
360 if (t->which == DATA1T_numeric)
361 printf("%d)", t->value.numeric);
363 printf("'%s')", t->value.string);
370 if (n->which == DATA1N_tag)
372 index_termlist (n, n, p, level, wrd);
373 /* index start tag */
374 if (!n->root->u.root.absyn)
375 index_xpath (n, p, level, wrd, 1);
379 if (dumpkeys(n->child, p, level + 1, wrd) < 0)
383 if (n->which == DATA1N_data)
385 data1_node *par = get_parent_tag(p->dh, n);
387 if (p->flagShowRecords)
389 printf("%*s", level * 4, "");
391 if (n->u.data.len > 32)
392 printf("'%.24s ... %.6s'\n", n->u.data.data,
393 n->u.data.data + n->u.data.len-6);
394 else if (n->u.data.len > 0)
395 printf("'%.*s'\n", n->u.data.len, n->u.data.data);
401 index_termlist (par, n, p, level, wrd);
402 if (!n->root->u.root.absyn)
403 index_xpath (n, p, level, wrd, 1016);
407 if (n->which == DATA1N_tag)
410 if (!n->root->u.root.absyn)
411 index_xpath (n, p, level, wrd, 2);
415 if (p->flagShowRecords && n->which == DATA1N_root)
417 printf("%*s-------------\n\n", level * 4, "");
423 int grs_extract_tree(struct recExtractCtrl *p, data1_node *n)
426 int oidtmp[OID_SIZE];
429 oe.proto = PROTO_Z3950;
430 oe.oclass = CLASS_SCHEMA;
433 oe.value = n->u.root.absyn->reference;
435 if ((oid_ent_to_oid (&oe, oidtmp)))
436 (*p->schemaAdd)(p, oidtmp);
440 return dumpkeys(n, p, 0, &wrd);
443 static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p,
447 struct grs_read_info gri;
449 int oidtmp[OID_SIZE];
452 gri.readf = p->readf;
453 gri.seekf = p->seekf;
454 gri.tellf = p->tellf;
457 gri.offset = p->offset;
461 if (read_grs_type (h, &gri, p->subType, &n))
462 return RECCTRL_EXTRACT_ERROR;
464 return RECCTRL_EXTRACT_EOF;
465 oe.proto = PROTO_Z3950;
466 oe.oclass = CLASS_SCHEMA;
468 if (!n->u.root.absyn)
469 return RECCTRL_EXTRACT_ERROR;
473 oe.value = n->u.root.absyn->reference;
474 if ((oid_ent_to_oid (&oe, oidtmp)))
475 (*p->schemaAdd)(p, oidtmp);
478 /* ensure our data1 tree is UTF-8 */
479 data1_iconv (p->dh, mem, n, "UTF-8", data1_get_encoding(p->dh, n));
482 data1_pr_tree (p->dh, n, stdout);
486 if (dumpkeys(n, p, 0, &wrd) < 0)
488 data1_free_tree(p->dh, n);
489 return RECCTRL_EXTRACT_ERROR;
491 data1_free_tree(p->dh, n);
492 return RECCTRL_EXTRACT_OK;
495 static int grs_extract(void *clientData, struct recExtractCtrl *p)
498 NMEM mem = nmem_create ();
499 struct grs_handlers *h = (struct grs_handlers *) clientData;
501 ret = grs_extract_sub(h, p, mem);
507 * Return: -1: Nothing done. 0: Ok. >0: Bib-1 diagnostic.
509 static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c)
511 data1_esetname *eset;
517 case Z_RecordComp_simple:
518 if (c->u.simple->which != Z_ElementSetNames_generic)
519 return 26; /* only generic form supported. Fix this later */
520 if (!(eset = data1_getesetbyname(dh, n->u.root.absyn,
521 c->u.simple->u.generic)))
523 logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic);
524 return 25; /* invalid esetname */
526 logf(LOG_DEBUG, "Esetname '%s' in simple compspec",
527 c->u.simple->u.generic);
530 case Z_RecordComp_complex:
531 if (c->u.complex->generic)
533 /* insert check for schema */
534 if ((p = c->u.complex->generic->elementSpec))
538 case Z_ElementSpec_elementSetName:
540 data1_getesetbyname(dh, n->u.root.absyn,
541 p->u.elementSetName)))
543 logf(LOG_LOG, "Unknown esetname '%s'",
544 p->u.elementSetName);
545 return 25; /* invalid esetname */
547 logf(LOG_DEBUG, "Esetname '%s' in complex compspec",
548 p->u.elementSetName);
551 case Z_ElementSpec_externalSpec:
552 if (p->u.externalSpec->which == Z_External_espec1)
554 logf(LOG_DEBUG, "Got Espec-1");
555 espec = p->u.externalSpec-> u.espec1;
559 logf(LOG_LOG, "Unknown external espec.");
560 return 25; /* bad. what is proper diagnostic? */
571 logf (LOG_DEBUG, "Element: Espec-1 match");
572 return data1_doespec1(dh, n, espec);
576 logf (LOG_DEBUG, "Element: all match");
581 static void add_idzebra_info (struct recRetrieveCtrl *p, data1_node *top,
584 const char *idzebra_ns[7];
586 idzebra_ns[0] = "xmlns:idzebra";
587 idzebra_ns[1] = "http://www.indexdata.dk/zebra/";
590 data1_tag_add_attr (p->dh, mem, top, idzebra_ns);
592 data1_mk_tag_data_int (p->dh, top, "idzebra:size", p->recordSize,
595 data1_mk_tag_data_int (p->dh, top, "idzebra:score",
598 data1_mk_tag_data_int (p->dh, top, "idzebra:localnumber", p->localno,
601 data1_mk_tag_data_text(p->dh, top, "idzebra:filename",
605 static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p)
607 data1_node *node = 0, *onode = 0, *top;
610 int res, selected = 0;
612 struct grs_read_info gri;
614 struct grs_handlers *h = (struct grs_handlers *) clientData;
615 int requested_schema = VAL_NONE;
616 data1_marctab *marctab;
620 gri.readf = p->readf;
621 gri.seekf = p->seekf;
622 gri.tellf = p->tellf;
629 logf (LOG_DEBUG, "grs_retrieve");
630 if (read_grs_type (h, &gri, p->subType, &node))
642 /* ensure our data1 tree is UTF-8 */
643 data1_iconv (p->dh, mem, node, "UTF-8", data1_get_encoding(p->dh, node));
646 data1_pr_tree (p->dh, node, stdout);
648 top = data1_get_root_tag (p->dh, node);
650 logf (LOG_DEBUG, "grs_retrieve: size");
651 if ((dnew = data1_mk_tag_data_wd(p->dh, top, "size", mem)))
653 dnew->u.data.what = DATA1I_text;
654 dnew->u.data.data = dnew->lbuf;
655 sprintf(dnew->u.data.data, "%d", p->recordSize);
656 dnew->u.data.len = strlen(dnew->u.data.data);
659 tagname = res_get_def(p->res, "tagrank", "rank");
660 if (strcmp(tagname, "0") && p->score >= 0 &&
661 (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
663 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
664 dnew->u.data.what = DATA1I_num;
665 dnew->u.data.data = dnew->lbuf;
666 sprintf(dnew->u.data.data, "%d", p->score);
667 dnew->u.data.len = strlen(dnew->u.data.data);
670 tagname = res_get_def(p->res, "tagsysno", "localControlNumber");
671 if (strcmp(tagname, "0") && p->localno > 0 &&
672 (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem)))
674 logf (LOG_DEBUG, "grs_retrieve: %s", tagname);
675 dnew->u.data.what = DATA1I_text;
676 dnew->u.data.data = dnew->lbuf;
678 sprintf(dnew->u.data.data, "%d", p->localno);
679 dnew->u.data.len = strlen(dnew->u.data.data);
682 data1_pr_tree (p->dh, node, stdout);
684 if (p->comp && p->comp->which == Z_RecordComp_complex &&
685 p->comp->u.complex->generic &&
686 p->comp->u.complex->generic->schema)
688 oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema);
690 requested_schema = oe->value;
693 /* If schema has been specified, map if possible, then check that
694 * we got the right one
696 if (requested_schema != VAL_NONE)
698 logf (LOG_DEBUG, "grs_retrieve: schema mapping");
699 for (map = node->u.root.absyn->maptabs; map; map = map->next)
701 if (map->target_absyn_ref == requested_schema)
704 if (!(node = data1_map_record(p->dh, onode, map, mem)))
713 if (node->u.root.absyn &&
714 requested_schema != node->u.root.absyn->reference)
722 * Does the requested format match a known syntax-mapping? (this reflects
723 * the overlap of schema and formatting which is inherent in the MARC
726 yaz_log (LOG_DEBUG, "grs_retrieve: syntax mapping");
727 if (node->u.root.absyn)
728 for (map = node->u.root.absyn->maptabs; map; map = map->next)
730 if (map->target_absyn_ref == p->input_format)
733 if (!(node = data1_map_record(p->dh, onode, map, mem)))
742 yaz_log (LOG_DEBUG, "grs_retrieve: schemaIdentifier");
743 if (node->u.root.absyn &&
744 node->u.root.absyn->reference != VAL_NONE &&
745 p->input_format == VAL_GRS1)
749 int oidtmp[OID_SIZE];
751 oe.proto = PROTO_Z3950;
752 oe.oclass = CLASS_SCHEMA;
753 oe.value = node->u.root.absyn->reference;
755 if ((oid = oid_ent_to_oid (&oe, oidtmp)))
758 data1_handle dh = p->dh;
762 for (ii = oid; *ii >= 0; ii++)
766 sprintf(p, "%d", *ii);
771 if ((dnew = data1_mk_tag_data_wd(dh, node,
772 "schemaIdentifier", mem)))
774 dnew->u.data.what = DATA1I_oid;
775 dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp);
776 memcpy(dnew->u.data.data, tmp, p - tmp);
777 dnew->u.data.len = p - tmp;
782 logf (LOG_DEBUG, "grs_retrieve: element spec");
783 if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0)
787 data1_free_tree(p->dh, onode);
788 data1_free_tree(p->dh, node);
792 else if (p->comp && !res)
796 data1_pr_tree (p->dh, node, stdout);
798 logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping");
799 switch (p->output_format = (p->input_format != VAL_NONE ?
800 p->input_format : VAL_SUTRS))
803 add_idzebra_info (p, top, mem);
806 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
808 if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected,
813 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
814 memcpy (new_buf, p->rec_buf, p->rec_len);
815 p->rec_buf = new_buf;
820 if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected,
822 p->diagnostic = 238; /* not available in requested syntax */
824 p->rec_len = (size_t) (-1);
827 if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected,
831 p->rec_len = (size_t) (-1);
834 if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected,
838 p->rec_len = (size_t) (-1);
842 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
843 if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected,
848 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
849 memcpy (new_buf, p->rec_buf, p->rec_len);
850 p->rec_buf = new_buf;
854 if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected,
859 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
860 memcpy (new_buf, p->rec_buf, p->rec_len);
861 p->rec_buf = new_buf;
865 if (!node->u.root.absyn)
870 for (marctab = node->u.root.absyn->marc; marctab;
871 marctab = marctab->next)
872 if (marctab->reference == p->input_format)
880 data1_iconv (p->dh, mem, node, p->encoding, "UTF-8");
881 if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node,
882 selected, &p->rec_len)))
886 char *new_buf = (char*) odr_malloc (p->odr, p->rec_len);
887 memcpy (new_buf, p->rec_buf, p->rec_len);
888 p->rec_buf = new_buf;
892 data1_free_tree(p->dh, node);
894 data1_free_tree(p->dh, onode);
899 static struct recType grs_type =
908 RecType recTypeGrs = &grs_type;