X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=recctrl%2Frecgrs.c;h=ae5a782b93e8b1200006ee25376e037bb64fad3e;hb=26a5a4b96b90361859cf856d6c41852b03ca2f84;hp=f1f41791c3ef891764776b85ba8fc9a0ff78df31;hpb=392b67432545d96cb2f9b3264a9427582fc1934f;p=idzebra-moved-to-github.git diff --git a/recctrl/recgrs.c b/recctrl/recgrs.c index f1f4179..ae5a782 100644 --- a/recctrl/recgrs.c +++ b/recctrl/recgrs.c @@ -1,196 +1,57 @@ -/* - * Copyright (C) 1994-1998, Index Data - * All rights reserved. - * Sebastian Hammer, Adam Dickmeiss - * - * $Log: recgrs.c,v $ - * Revision 1.23 1998-10-18 07:51:10 adam - * Changed one logf call. - * - * Revision 1.22 1998/10/16 08:14:37 adam - * Updated record control system. - * - * Revision 1.21 1998/07/01 09:16:10 adam - * Element localno only added when it's greater than 0. - * - * Revision 1.20 1998/05/20 10:12:26 adam - * Implemented automatic EXPLAIN database maintenance. - * Modified Zebra to work with ASN.1 compiled version of YAZ. - * - * Revision 1.19 1998/03/11 11:19:05 adam - * Changed the way sequence numbers are generated. - * - * Revision 1.18 1998/03/05 08:41:31 adam - * Minor changes. - * - * Revision 1.17 1998/02/10 12:03:06 adam - * Implemented Sort. - * - * Revision 1.16 1998/01/29 13:38:17 adam - * Fixed problem with mapping to record with unknown schema. - * - * Revision 1.15 1998/01/26 10:37:57 adam - * Better diagnostics. - * - * Revision 1.14 1997/11/06 11:41:01 adam - * Implemented "begin variant" for the sgml.regx filter. - * - * Revision 1.13 1997/10/31 12:35:44 adam - * Added a few log statements. - * - * Revision 1.12 1997/10/29 12:02:22 adam - * Using oid_ent_to_oid used instead of the non thread-safe oid_getoidbyent. - * - * Revision 1.11 1997/10/27 14:34:00 adam - * Work on generic character mapping depending on "structure" field - * in abstract syntax file. - * - * Revision 1.10 1997/09/18 08:59:21 adam - * Extra generic handle for the character mapping routines. - * - * Revision 1.9 1997/09/17 12:19:21 adam - * Zebra version corresponds to YAZ version 1.4. - * Changed Zebra server so that it doesn't depend on global common_resource. - * - * Revision 1.8 1997/09/09 13:38:14 adam - * Partial port to WIN95/NT. - * - * Revision 1.7 1997/09/05 15:30:10 adam - * Changed prototype for chr_map_input - added const. - * Added support for C++, headers uses extern "C" for public definitions. - * - * Revision 1.6 1997/09/04 13:54:40 adam - * Added MARC filter - type grs.marc. where syntax refers - * to abstract syntax. New method tellf in retrieve/extract method. - * - * Revision 1.5 1997/07/15 16:29:03 adam - * Initialized dummy variable to keep checker gcc happy. - * - * Revision 1.4 1997/04/30 08:56:08 quinn - * null - * - * Revision 1.2 1996/10/11 16:06:43 quinn - * Revision 1.3 1997/02/24 10:41:50 adam - * Cleanup of code and commented out the "end element-end-record" code. - * - * Revision 1.2 1996/10/11 16:06:43 quinn - * Fixed arguments to nodetogr - * - * Revision 1.1 1996/10/11 10:57:25 adam - * New module recctrl. Used to manage records (extract/retrieval). - * - * Revision 1.29 1996/10/08 10:30:21 quinn - * Fixed type mismatch - * - * Revision 1.28 1996/10/07 16:06:40 quinn - * Added SOIF support - * - * Revision 1.27 1996/06/11 10:54:12 quinn - * Relevance work - * - * Revision 1.26 1996/06/06 12:08:45 quinn - * Added showRecord function - * - * Revision 1.25 1996/06/04 14:18:53 quinn - * Charmap work - * - * Revision 1.24 1996/06/04 13:27:54 quinn - * More work on charmapping - * - * Revision 1.23 1996/06/04 10:19:01 adam - * Minor changes - removed include of ctype.h. - * - * Revision 1.22 1996/06/03 10:15:27 quinn - * Various character-mapping. - * - * Revision 1.21 1996/05/31 13:27:24 quinn - * Character-conversion in phrases, too. - * - * Revision 1.19 1996/05/16 15:31:14 quinn - * a7 - * - * Revision 1.18 1996/05/09 07:28:56 quinn - * Work towards phrases and multiple registers - * - * Revision 1.17 1996/05/01 13:46:37 adam - * First work on multiple records in one file. - * New option, -offset, to the "unread" command in the filter module. - * - * Revision 1.16 1996/01/17 14:57:54 adam - * Prototype changed for reader functions in extract/retrieve. File - * is identified by 'void *' instead of 'int. - * - * Revision 1.15 1996/01/08 19:15:47 adam - * New input filter that works! - * - * Revision 1.14 1995/12/15 12:36:11 adam - * Retrieval calls data1_read_regx when subType is specified. - * - * Revision 1.13 1995/12/15 12:24:43 quinn - * *** empty log message *** - * - * Revision 1.12 1995/12/15 12:20:28 quinn - * *** empty log message *** - * - * Revision 1.11 1995/12/15 12:07:57 quinn - * Changed extraction strategy. - * - * Revision 1.10 1995/12/14 11:10:48 quinn - * Explain work - * - * Revision 1.9 1995/12/13 17:14:05 quinn - * *** empty log message *** - * - * Revision 1.8 1995/12/13 15:33:18 quinn - * *** empty log message *** - * - * Revision 1.7 1995/12/13 13:45:39 quinn - * Changed data1 to use nmem. - * - * Revision 1.6 1995/12/04 14:22:30 adam - * Extra arg to recType_byName. - * Started work on new regular expression parsed input to - * structured records. - * - * Revision 1.5 1995/11/28 14:18:37 quinn - * Set output_format. - * - * Revision 1.4 1995/11/21 13:14:49 quinn - * Fixed end-of-data-field problem (maybe). - * - * Revision 1.3 1995/11/15 19:13:09 adam - * Work on record management. - * - */ +/* $Id: recgrs.c,v 1.72 2003-02-04 12:06:47 pop Exp $ + Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002 + Index Data Aps + +This file is part of the Zebra server. + +Zebra is free software; you can redistribute it and/or modify it under +the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2, or (at your option) any later +version. + +Zebra is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received a copy of the GNU General Public License +along with Zebra; see the file LICENSE.zebra. If not, write to the +Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA +02111-1307, USA. +*/ #include #include #include -#ifndef WINDOWS +#ifndef WIN32 #include #endif -#include -#include +#include +#include #include #include "grsread.h" #define GRS_MAX_WORD 512 -static data1_node *read_grs_type (struct grs_read_info *p, const char *type) +struct grs_handler { + RecTypeGrs type; + void *clientData; + int initFlag; + struct grs_handler *next; +}; + +struct grs_handlers { + struct grs_handler *handlers; +}; + +static int read_grs_type (struct grs_handlers *h, + struct grs_read_info *p, const char *type, + data1_node **root) { - static struct { - char *type; - data1_node *(*func)(struct grs_read_info *p); - } tab[] = { - { "sgml", grs_read_sgml }, - { "regx", grs_read_regx }, - { "marc", grs_read_marc }, - { NULL, NULL } - }; + struct grs_handler *gh = h->handlers; const char *cp = strchr (type, '.'); - int i; if (cp == NULL || cp == type) { @@ -199,26 +60,528 @@ static data1_node *read_grs_type (struct grs_read_info *p, const char *type) } else strcpy (p->type, cp+1); - for (i=0; tab[i].type; i++) + for (gh = h->handlers; gh; gh = gh->next) { - if (!memcmp (type, tab[i].type, cp-type)) - return (tab[i].func)(p); + if (!memcmp (type, gh->type->type, cp-type)) + { + if (!gh->initFlag) + { + gh->initFlag = 1; + gh->clientData = (*gh->type->init)(); + } + p->clientData = gh->clientData; + *root = (gh->type->read)(p); + gh->clientData = p->clientData; + return 0; + } } - return NULL; + return 1; } -static void grs_init(RecType recType) +static void grs_add_handler (struct grs_handlers *h, RecTypeGrs t) { + struct grs_handler *gh = (struct grs_handler *) xmalloc (sizeof(*gh)); + gh->next = h->handlers; + h->handlers = gh; + gh->initFlag = 0; + gh->clientData = 0; + gh->type = t; } -static void grs_destroy(RecType recType) +static void *grs_init(RecType recType) { + struct grs_handlers *h = (struct grs_handlers *) xmalloc (sizeof(*h)); + h->handlers = 0; + + grs_add_handler (h, recTypeGrs_sgml); + grs_add_handler (h, recTypeGrs_regx); +#if HAVE_TCL_H + grs_add_handler (h, recTypeGrs_tcl); +#endif + grs_add_handler (h, recTypeGrs_marc); +#if HAVE_EXPAT_H + grs_add_handler (h, recTypeGrs_xml); +#endif +#if HAVE_PERL + grs_add_handler (h, recTypeGrs_perl); +#endif + return h; } -static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level) +static void grs_destroy(void *clientData) +{ + struct grs_handlers *h = (struct grs_handlers *) clientData; + struct grs_handler *gh = h->handlers, *gh_next; + while (gh) + { + gh_next = gh->next; + if (gh->initFlag) + (*gh->type->destroy)(gh->clientData); + xfree (gh); + gh = gh_next; + } + xfree (h); +} + +int d1_check_xpath_predicate(data1_node *n, struct xpath_predicate *p) { + int res = 1; + char *attname; + data1_xattr *attr; + + if (!p) { + return (1); + } else { + if (p->which == XPATH_PREDICATE_RELATION) { + if (p->u.relation.name[0]) { + if (*p->u.relation.name != '@') { + logf(LOG_WARN, + " Only attributes (@) are supported in xelm xpath predicates"); + logf(LOG_WARN, "predicate %s ignored", p->u.relation.name); + return (1); + } + attname = p->u.relation.name + 1; + res = 0; + /* looking for the attribute with a specified name */ + for (attr = n->u.tag.attributes; attr; attr = attr->next) { + logf(LOG_DEBUG," - attribute %s <-> %s", attname, attr->name ); + + if (!strcmp(attr->name, attname)) { + if (p->u.relation.op[0]) { + if (*p->u.relation.op != '=') { + logf(LOG_WARN, + "Only '=' relation is supported (%s)",p->u.relation.op); + logf(LOG_WARN, "predicate %s ignored", p->u.relation.name); + res = 1; break; + } else { + logf(LOG_DEBUG," - value %s <-> %s", + p->u.relation.value, attr->value ); + if (!strcmp(attr->value, p->u.relation.value)) { + res = 1; break; + } + } + } else { + /* attribute exists, no value specified */ + res = 1; break; + } + } + } + return (res); + } else { + return (1); + } + } + else if (p->which == XPATH_PREDICATE_BOOLEAN) { + if (!strcmp(p->u.boolean.op,"and")) { + return (d1_check_xpath_predicate(n, p->u.boolean.left) + && d1_check_xpath_predicate(n, p->u.boolean.right)); + } + else if (!strcmp(p->u.boolean.op,"or")) { + return (d1_check_xpath_predicate(n, p->u.boolean.left) + || d1_check_xpath_predicate(n, p->u.boolean.right)); + } else { + logf(LOG_WARN, "Unknown boolean relation %s, ignored",p->u.boolean.op); + return (1); + } + } + } + return 0; +} + + +/* *ostrich* + + New function, looking for xpath "element" definitions in abs, by + tagpath, using a kind of ugly regxp search.The DFA was built while + parsing abs, so here we just go trough them and try to match + against the given tagpath. The first matching entry is returned. + + pop, 2002-12-13 + + Added support for enhanced xelm. Now [] predicates are considered + as well, when selecting indexing rules... (why the hell it's called + termlist???) + + pop, 2003-01-17 + + */ + +data1_termlist *xpath_termlist_by_tagpath(char *tagpath, data1_node *n) +{ + data1_absyn *abs = n->root->u.root.absyn; + data1_xpelement *xpe = abs->xp_elements; + data1_node *nn; +#ifdef ENHANCED_XELM + struct xpath_location_step *xp; + +#endif + char *pexpr = malloc(strlen(tagpath)+2); + int ok = 0; + + sprintf (pexpr, "%s\n", tagpath); + while (xpe) + { + struct DFA_state **dfaar = xpe->dfa->states; + struct DFA_state *s=dfaar[0]; + struct DFA_tran *t; + const char *p; + int i; + unsigned char c; + int start_line = 1; + + c = *pexpr++; t = s->trans; i = s->tran_no; + if (c >= t->ch[0] && c <= t->ch[1]) { + p = pexpr; + do { + if ((s = dfaar[t->to])->rule_no && + (start_line || s->rule_nno)) { + ok = 1; + break; + } + for (t=s->trans, i=s->tran_no; --i >= 0; t++) { + if ((unsigned) *p >= t->ch[0] && (unsigned) *p <= t->ch[1]) + break; + } + p++; + } while (i >= 0); + } + pexpr--; + if (ok) { +#ifdef ENHANCED_XELM + /* we have to check the perdicates up to the root node */ + xp = xpe->xpath; + + /* find the first tag up in the node structure */ + nn = n; while (nn && nn->which != DATA1N_tag) { + nn = nn->parent; + } + + /* go from inside out in the node structure, while going + backwards trough xpath location steps ... */ + for (i=xpe->xpath_len - 1; i>0; i--) { + + logf(LOG_DEBUG,"Checking step %d: %s on tag %s", + i,xp[i].part,nn->u.tag.tag); + + if (!d1_check_xpath_predicate(nn, xp[i].predicate)) { + logf(LOG_DEBUG," Predicates didn't match"); + ok = 0; + break; + } + + if (nn->which == DATA1N_tag) { + nn = nn->parent; + } + } +#endif + if (ok) { + break; + } + } + xpe = xpe->next; + } + + if (ok) { + logf(LOG_DEBUG,"Got it"); + return xpe->termlists; + } else { + return NULL; + } +} + +/* use + 1 start element (tag) + 2 end element + 3 start attr (and attr-exact) + 4 end attr + + 1016 cdata + 1015 attr data + + *ostrich* + + Now, if there is a matching xelm described in abs, for the + indexed element or the attribute, then the data is handled according + to those definitions... + + modified by pop, 2002-12-13 +*/ + +static void index_xpath (data1_node *n, struct recExtractCtrl *p, + int level, RecWord *wrd, int use) +{ + int i; + char tag_path_full[1024]; + size_t flen = 0; + data1_node *nn; + + switch (n->which) + { + case DATA1N_data: + wrd->string = n->u.data.data; + wrd->length = n->u.data.len; + if (p->flagShowRecords) + { + printf("%*s data=", (level + 1) * 4, ""); + for (i = 0; ilength && i < 8; i++) + fputc (wrd->string[i], stdout); + printf("\n"); + } + else { + data1_termlist *tl; + int xpdone = 0; + flen = 0; + + /* we have to fetch the whole path to the data tag */ + for (nn = n; nn; nn = nn->parent) { + if (nn->which == DATA1N_tag) { + size_t tlen = strlen(nn->u.tag.tag); + if (tlen + flen > (sizeof(tag_path_full)-2)) return; + memcpy (tag_path_full + flen, nn->u.tag.tag, tlen); + flen += tlen; + tag_path_full[flen++] = '/'; + } + else if (nn->which == DATA1N_root) break; + } + + tag_path_full[flen] = 0; + + /* If we have a matching termlist... */ + if ((tl = xpath_termlist_by_tagpath(tag_path_full, n))) { + for (; tl; tl = tl->next) { + wrd->reg_type = *tl->structure; + /* this is the ! case, so structure is for the xpath index */ + if (!tl->att) { + wrd->attrSet = VAL_IDXPATH; + wrd->attrUse = use; + (*p->tokenAdd)(wrd); + xpdone = 1; + /* this is just the old fashioned attribute based index */ + } else { + wrd->attrSet = (int) (tl->att->parent->reference); + wrd->attrUse = tl->att->locals->local; + (*p->tokenAdd)(wrd); + } + } + } + /* xpath indexing is done, if there was no termlist given, + or no ! attribute... */ + if (!xpdone) { + wrd->attrSet = VAL_IDXPATH; + wrd->attrUse = use; + wrd->reg_type = 'w'; + (*p->tokenAdd)(wrd); + } + } + break; + case DATA1N_tag: + flen = 0; + for (nn = n; nn; nn = nn->parent) + { + if (nn->which == DATA1N_tag) + { + size_t tlen = strlen(nn->u.tag.tag); + if (tlen + flen > (sizeof(tag_path_full)-2)) + return; + memcpy (tag_path_full + flen, nn->u.tag.tag, tlen); + flen += tlen; + tag_path_full[flen++] = '/'; + } + else if (nn->which == DATA1N_root) + break; + } + + + wrd->reg_type = '0'; + wrd->string = tag_path_full; + wrd->length = flen; + wrd->attrSet = VAL_IDXPATH; + wrd->attrUse = use; + if (p->flagShowRecords) + { + printf("%*s tag=", (level + 1) * 4, ""); + for (i = 0; ilength && i < 40; i++) + fputc (wrd->string[i], stdout); + if (i == 40) + printf (" .."); + printf("\n"); + } + else + { + data1_xattr *xp; + (*p->tokenAdd)(wrd); /* index element pag (AKA tag path) */ + if (use == 1) + { + for (xp = n->u.tag.attributes; xp; xp = xp->next) + { + char comb[512]; + /* attribute (no value) */ + wrd->reg_type = '0'; + wrd->attrUse = 3; + wrd->string = xp->name; + wrd->length = strlen(xp->name); + + wrd->seqno--; + (*p->tokenAdd)(wrd); + + if (xp->value && + strlen(xp->name) + strlen(xp->value) < sizeof(comb)-2) + { + /* attribute value exact */ + strcpy (comb, xp->name); + strcat (comb, "="); + strcat (comb, xp->value); + + wrd->attrUse = 3; + wrd->reg_type = '0'; + wrd->string = comb; + wrd->length = strlen(comb); + wrd->seqno--; + + (*p->tokenAdd)(wrd); + } + } + for (xp = n->u.tag.attributes; xp; xp = xp->next) + { + char attr_tag_path_full[1024]; + int int_len = flen; + + sprintf (attr_tag_path_full, "@%s/%.*s", + xp->name, int_len, tag_path_full); + wrd->reg_type = '0'; + wrd->attrUse = 1; + wrd->string = attr_tag_path_full; + wrd->length = strlen(attr_tag_path_full); + (*p->tokenAdd)(wrd); + + if (xp->value) + { + /* the same jokes, as with the data nodes ... */ + data1_termlist *tl; + int xpdone = 0; + + wrd->string = xp->value; + wrd->length = strlen(xp->value); + wrd->reg_type = 'w'; + + if ((tl = xpath_termlist_by_tagpath(attr_tag_path_full, + n))) { + for (; tl; tl = tl->next) { + wrd->reg_type = *tl->structure; + if (!tl->att) { + wrd->attrSet = VAL_IDXPATH; + wrd->attrUse = 1015; + (*p->tokenAdd)(wrd); + xpdone = 1; + } else { + wrd->attrSet = (int) (tl->att->parent->reference); + wrd->attrUse = tl->att->locals->local; + (*p->tokenAdd)(wrd); + } + } + + } + if (!xpdone) { + wrd->attrSet = VAL_IDXPATH; + wrd->attrUse = 1015; + wrd->reg_type = 'w'; + (*p->tokenAdd)(wrd); + } + } + + wrd->attrSet = VAL_IDXPATH; + wrd->reg_type = '0'; + wrd->attrUse = 2; + wrd->string = attr_tag_path_full; + wrd->length = strlen(attr_tag_path_full); + (*p->tokenAdd)(wrd); + } + } + } + } +} + +static void index_termlist (data1_node *par, data1_node *n, + struct recExtractCtrl *p, int level, RecWord *wrd) +{ + data1_termlist *tlist = 0; + data1_datatype dtype = DATA1K_string; + + /* + * cycle up towards the root until we find a tag with an att.. + * this has the effect of indexing locally defined tags with + * the attribute of their ancestor in the record. + */ + + while (!par->u.tag.element) + if (!par->parent || !(par=get_parent_tag(p->dh, par->parent))) + break; + if (!par || !(tlist = par->u.tag.element->termlists)) + return; + if (par->u.tag.element->tag) + dtype = par->u.tag.element->tag->kind; + + for (; tlist; tlist = tlist->next) + { + + char xattr[512]; + /* consider source */ + wrd->string = 0; + + if (!strcmp (tlist->source, "data") && n->which == DATA1N_data) + { + wrd->string = n->u.data.data; + wrd->length = n->u.data.len; + } + else if (!strcmp (tlist->source, "tag") && n->which == DATA1N_tag) + { + wrd->string = n->u.tag.tag; + wrd->length = strlen(n->u.tag.tag); + } + else if (sscanf (tlist->source, "attr(%511[^)])", xattr) == 1 && + n->which == DATA1N_tag) + { + data1_xattr *p = n->u.tag.attributes; + while (p && strcmp (p->name, xattr)) + p = p->next; + if (p) + { + wrd->string = p->value; + wrd->length = strlen(p->value); + } + } + if (wrd->string) + { + if (p->flagShowRecords) + { + int i; + printf("%*sIdx: [%s]", (level + 1) * 4, "", + tlist->structure); + printf("%s:%s [%d] %s", + tlist->att->parent->name, + tlist->att->name, tlist->att->value, + tlist->source); + printf (" data=\""); + for (i = 0; ilength && i < 8; i++) + fputc (wrd->string[i], stdout); + fputc ('"', stdout); + if (wrd->length > 8) + printf (" ..."); + fputc ('\n', stdout); + } + else + { + wrd->reg_type = *tlist->structure; + wrd->attrSet = (int) (tlist->att->parent->reference); + wrd->attrUse = tlist->att->locals->local; + (*p->tokenAdd)(wrd); + } + } + } +} + +static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level, + RecWord *wrd) { - RecWord wrd; - (*p->init)(p, &wrd); /* set defaults */ for (; n; n = n->next) { if (p->flagShowRecords) /* display element description to user */ @@ -226,7 +589,7 @@ static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level) if (n->which == DATA1N_root) { printf("%*s", level * 4, ""); - printf("Record type: '%s'\n", n->u.root.absyn->name); + printf("Record type: '%s'\n", n->u.root.type); } else if (n->which == DATA1N_tag) { @@ -258,22 +621,33 @@ static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level) } } + if (n->which == DATA1N_tag) + { + index_termlist (n, n, p, level, wrd); + /* index start tag */ + assert (n->root->u.root.absyn); + + if (!n->root->u.root.absyn) + index_xpath (n, p, level, wrd, 1); + else if (n->root->u.root.absyn->enable_xpath_indexing) + index_xpath (n, p, level, wrd, 1); + } + if (n->child) - if (dumpkeys(n->child, p, level + 1) < 0) + if (dumpkeys(n->child, p, level + 1, wrd) < 0) return -1; + if (n->which == DATA1N_data) { data1_node *par = get_parent_tag(p->dh, n); - data1_termlist *tlist = 0; - data1_datatype dtype = DATA1K_string; if (p->flagShowRecords) { printf("%*s", level * 4, ""); printf("Data: "); - if (n->u.data.len > 32) - printf("'%.24s ... %.6s'\n", n->u.data.data, + if (n->u.data.len > 256) + printf("'%.240s ... %.6s'\n", n->u.data.data, n->u.data.data + n->u.data.len-6); else if (n->u.data.len > 0) printf("'%.*s'\n", n->u.data.len, n->u.data.data); @@ -281,42 +655,23 @@ static int dumpkeys(data1_node *n, struct recExtractCtrl *p, int level) printf("NULL\n"); } - assert(par); - - /* - * cycle up towards the root until we find a tag with an att.. - * this has the effect of indexing locally defined tags with - * the attribute of their ancestor in the record. - */ + if (par) + index_termlist (par, n, p, level, wrd); + if (!n->root->u.root.absyn) + index_xpath (n, p, level, wrd, 1016); + else if (n->root->u.root.absyn->enable_xpath_indexing) + index_xpath (n, p, level, wrd, 1016); + } - while (!par->u.tag.element) - if (!par->parent || !(par=get_parent_tag(p->dh, par->parent))) - break; - if (!par || !(tlist = par->u.tag.element->termlists)) - continue; - if (par->u.tag.element->tag) - dtype = par->u.tag.element->tag->kind; - for (; tlist; tlist = tlist->next) - { - if (p->flagShowRecords) - { - printf("%*sIdx: [%s]", (level + 1) * 4, "", - tlist->structure); - printf("%s:%s [%d]\n", - tlist->att->parent->name, - tlist->att->name, tlist->att->value); - } - else - { - wrd.reg_type = *tlist->structure; - wrd.string = n->u.data.data; - wrd.length = n->u.data.len; - wrd.attrSet = (int) (tlist->att->parent->reference); - wrd.attrUse = tlist->att->locals->local; - (*p->addWord)(&wrd); - } - } + if (n->which == DATA1N_tag) + { + /* index end tag */ + if (!n->root->u.root.absyn) + index_xpath (n, p, level, wrd, 2); + else if (n->root->u.root.absyn->enable_xpath_indexing) + index_xpath (n, p, level, wrd, 2); } + if (p->flagShowRecords && n->which == DATA1N_root) { printf("%*s-------------\n\n", level * 4, ""); @@ -329,26 +684,31 @@ int grs_extract_tree(struct recExtractCtrl *p, data1_node *n) { oident oe; int oidtmp[OID_SIZE]; + RecWord wrd; oe.proto = PROTO_Z3950; oe.oclass = CLASS_SCHEMA; - oe.value = n->u.root.absyn->reference; - - if ((oid_ent_to_oid (&oe, oidtmp))) - (*p->addSchema)(p, oidtmp); + if (n->u.root.absyn) + { + oe.value = n->u.root.absyn->reference; + + if ((oid_ent_to_oid (&oe, oidtmp))) + (*p->schemaAdd)(p, oidtmp); + } + (*p->init)(p, &wrd); - return dumpkeys(n, p, 0); + return dumpkeys(n, p, 0, &wrd); } -static int grs_extract(struct recExtractCtrl *p) +static int grs_extract_sub(struct grs_handlers *h, struct recExtractCtrl *p, + NMEM mem) { data1_node *n; - NMEM mem; struct grs_read_info gri; oident oe; int oidtmp[OID_SIZE]; + RecWord wrd; - mem = nmem_create (); gri.readf = p->readf; gri.seekf = p->seekf; gri.tellf = p->tellf; @@ -358,24 +718,49 @@ static int grs_extract(struct recExtractCtrl *p) gri.mem = mem; gri.dh = p->dh; - n = read_grs_type (&gri, p->subType); + if (read_grs_type (h, &gri, p->subType, &n)) + return RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER; if (!n) - return -1; - + return RECCTRL_EXTRACT_EOF; oe.proto = PROTO_Z3950; oe.oclass = CLASS_SCHEMA; - oe.value = n->u.root.absyn->reference; - if ((oid_ent_to_oid (&oe, oidtmp))) - (*p->addSchema)(p, oidtmp); +#if 0 + if (!n->u.root.absyn) + return RECCTRL_EXTRACT_ERROR; +#endif + if (n->u.root.absyn) + { + oe.value = n->u.root.absyn->reference; + if ((oid_ent_to_oid (&oe, oidtmp))) + (*p->schemaAdd)(p, oidtmp); + } + + /* ensure our data1 tree is UTF-8 */ + data1_iconv (p->dh, mem, n, "UTF-8", data1_get_encoding(p->dh, n)); + +#if 0 + data1_pr_tree (p->dh, n, stdout); +#endif - if (dumpkeys(n, p, 0) < 0) + (*p->init)(p, &wrd); + if (dumpkeys(n, p, 0, &wrd) < 0) { data1_free_tree(p->dh, n); - return -2; + return RECCTRL_EXTRACT_ERROR_GENERIC; } data1_free_tree(p->dh, n); + return RECCTRL_EXTRACT_OK; +} + +static int grs_extract(void *clientData, struct recExtractCtrl *p) +{ + int ret; + NMEM mem = nmem_create (); + struct grs_handlers *h = (struct grs_handlers *) clientData; + + ret = grs_extract_sub(h, p, mem); nmem_destroy(mem); - return 0; + return ret; } /* @@ -389,56 +774,57 @@ static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c) switch (c->which) { - case Z_RecordComp_simple: - if (c->u.simple->which != Z_ElementSetNames_generic) - return 26; /* only generic form supported. Fix this later */ - if (!(eset = data1_getesetbyname(dh, n->u.root.absyn, - c->u.simple->u.generic))) - { - logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic); - return 25; /* invalid esetname */ - } - logf(LOG_DEBUG, "Esetname '%s' in simple compspec", - c->u.simple->u.generic); - espec = eset->spec; - break; - case Z_RecordComp_complex: - if (c->u.complex->generic) + case Z_RecordComp_simple: + if (c->u.simple->which != Z_ElementSetNames_generic) + return 26; /* only generic form supported. Fix this later */ + if (!(eset = data1_getesetbyname(dh, n->u.root.absyn, + c->u.simple->u.generic))) + { + logf(LOG_LOG, "Unknown esetname '%s'", c->u.simple->u.generic); + return 25; /* invalid esetname */ + } + logf(LOG_DEBUG, "Esetname '%s' in simple compspec", + c->u.simple->u.generic); + espec = eset->spec; + break; + case Z_RecordComp_complex: + if (c->u.complex->generic) + { + /* insert check for schema */ + if ((p = c->u.complex->generic->elementSpec)) { - /* insert check for schema */ - if ((p = c->u.complex->generic->elementSpec)) - switch (p->which) + switch (p->which) + { + case Z_ElementSpec_elementSetName: + if (!(eset = + data1_getesetbyname(dh, n->u.root.absyn, + p->u.elementSetName))) { - case Z_ElementSpec_elementSetName: - if (!(eset = - data1_getesetbyname(dh, - n->u.root.absyn, - p->u.elementSetName))) - { - logf(LOG_LOG, "Unknown esetname '%s'", - p->u.elementSetName); - return 25; /* invalid esetname */ - } - logf(LOG_DEBUG, "Esetname '%s' in complex compspec", - p->u.elementSetName); - espec = eset->spec; - break; - case Z_ElementSpec_externalSpec: - if (p->u.externalSpec->which == Z_External_espec1) - { - logf(LOG_DEBUG, "Got Espec-1"); - espec = p->u.externalSpec-> u.espec1; - } - else - { - logf(LOG_LOG, "Unknown external espec."); - return 25; /* bad. what is proper diagnostic? */ - } - break; + logf(LOG_LOG, "Unknown esetname '%s'", + p->u.elementSetName); + return 25; /* invalid esetname */ } + logf(LOG_DEBUG, "Esetname '%s' in complex compspec", + p->u.elementSetName); + espec = eset->spec; + break; + case Z_ElementSpec_externalSpec: + if (p->u.externalSpec->which == Z_External_espec1) + { + logf(LOG_DEBUG, "Got Espec-1"); + espec = p->u.externalSpec-> u.espec1; + } + else + { + logf(LOG_LOG, "Unknown external espec."); + return 25; /* bad. what is proper diagnostic? */ + } + break; + } } - else - return 26; /* fix */ + } + else + return 26; /* fix */ } if (espec) { @@ -452,14 +838,67 @@ static int process_comp(data1_handle dh, data1_node *n, Z_RecordComposition *c) } } -static int grs_retrieve(struct recRetrieveCtrl *p) +/* Add Zebra info in separate namespace ... + + 359 + 447 + records/genera.xml + + +*/ + +static void zebra_xml_metadata (struct recRetrieveCtrl *p, data1_node *top, + NMEM mem) +{ + const char *idzebra_ns[3]; + const char *i2 = "\n "; + const char *i4 = "\n "; + data1_node *n; + + idzebra_ns[0] = "xmlns"; + idzebra_ns[1] = "http://www.indexdata.dk/zebra/"; + idzebra_ns[2] = 0; + + data1_mk_text (p->dh, mem, i2, top); + + n = data1_mk_tag (p->dh, mem, "idzebra", idzebra_ns, top); + + data1_mk_text (p->dh, mem, "\n", top); + + data1_mk_text (p->dh, mem, i4, n); + + data1_mk_tag_data_int (p->dh, n, "size", p->recordSize, mem); + + if (p->score != -1) + { + data1_mk_text (p->dh, mem, i4, n); + data1_mk_tag_data_int (p->dh, n, "score", p->score, mem); + } + data1_mk_text (p->dh, mem, i4, n); + data1_mk_tag_data_int (p->dh, n, "localnumber", p->localno, mem); + if (p->fname) + { + data1_mk_text (p->dh, mem, i4, n); + data1_mk_tag_data_text(p->dh, n, "filename", p->fname, mem); + } + data1_mk_text (p->dh, mem, i2, n); +} + +static int grs_retrieve(void *clientData, struct recRetrieveCtrl *p) { - data1_node *node = 0, *onode = 0; + data1_node *node = 0, *onode = 0, *top; data1_node *dnew; data1_maptab *map; int res, selected = 0; NMEM mem; struct grs_read_info gri; + const char *tagname; + struct grs_handlers *h = (struct grs_handlers *) clientData; + int requested_schema = VAL_NONE; + data1_marctab *marctab; + int dummy; mem = nmem_create(); gri.readf = p->readf; @@ -472,105 +911,159 @@ static int grs_retrieve(struct recRetrieveCtrl *p) gri.dh = p->dh; logf (LOG_DEBUG, "grs_retrieve"); - node = read_grs_type (&gri, p->subType); + if (read_grs_type (h, &gri, p->subType, &node)) + { + p->diagnostic = 14; + nmem_destroy (mem); + return 0; + } if (!node) { p->diagnostic = 14; nmem_destroy (mem); return 0; } + /* ensure our data1 tree is UTF-8 */ + data1_iconv (p->dh, mem, node, "UTF-8", data1_get_encoding(p->dh, node)); + +#if 0 + data1_pr_tree (p->dh, node, stdout); +#endif + top = data1_get_root_tag (p->dh, node); + logf (LOG_DEBUG, "grs_retrieve: size"); - if ((dnew = data1_insert_taggeddata(p->dh, node, node, - "size", mem))) + tagname = data1_systag_lookup(node->u.root.absyn, "size", "size"); + if (tagname && + (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem))) { dnew->u.data.what = DATA1I_text; dnew->u.data.data = dnew->lbuf; sprintf(dnew->u.data.data, "%d", p->recordSize); dnew->u.data.len = strlen(dnew->u.data.data); } - - logf (LOG_DEBUG, "grs_retrieve: score"); - if (p->score >= 0 && (dnew = - data1_insert_taggeddata(p->dh, node, - node, "rank", - mem))) + + tagname = data1_systag_lookup(node->u.root.absyn, "rank", "rank"); + if (tagname && p->score >= 0 && + (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem))) { + logf (LOG_DEBUG, "grs_retrieve: %s", tagname); dnew->u.data.what = DATA1I_num; dnew->u.data.data = dnew->lbuf; sprintf(dnew->u.data.data, "%d", p->score); dnew->u.data.len = strlen(dnew->u.data.data); } - logf (LOG_DEBUG, "grs_retrieve: localControlNumber"); - if (p->localno > 0 && (dnew = data1_insert_taggeddata(p->dh, node, node, - "localControlNumber", mem))) + tagname = data1_systag_lookup(node->u.root.absyn, "sysno", + "localControlNumber"); + if (tagname && p->localno > 0 && + (dnew = data1_mk_tag_data_wd(p->dh, top, tagname, mem))) { + logf (LOG_DEBUG, "grs_retrieve: %s", tagname); dnew->u.data.what = DATA1I_text; dnew->u.data.data = dnew->lbuf; + sprintf(dnew->u.data.data, "%d", p->localno); dnew->u.data.len = strlen(dnew->u.data.data); } +#if 0 + data1_pr_tree (p->dh, node, stdout); +#endif + if (p->comp && p->comp->which == Z_RecordComp_complex && + p->comp->u.complex->generic && + p->comp->u.complex->generic->schema) + { + oident *oe = oid_getentbyoid (p->comp->u.complex->generic->schema); + if (oe) + requested_schema = oe->value; + } - logf (LOG_DEBUG, "grs_retrieve: schemaIdentifier"); - if (p->input_format == VAL_GRS1 && node->u.root.absyn && - node->u.root.absyn->reference != VAL_NONE) + /* If schema has been specified, map if possible, then check that + * we got the right one + */ + if (requested_schema != VAL_NONE) + { + logf (LOG_DEBUG, "grs_retrieve: schema mapping"); + for (map = node->u.root.absyn->maptabs; map; map = map->next) + { + if (map->target_absyn_ref == requested_schema) + { + onode = node; + if (!(node = data1_map_record(p->dh, onode, map, mem))) + { + p->diagnostic = 14; + nmem_destroy (mem); + return 0; + } + break; + } + } + if (node->u.root.absyn && + requested_schema != node->u.root.absyn->reference) + { + p->diagnostic = 238; + nmem_destroy (mem); + return 0; + } + } + /* + * Does the requested format match a known syntax-mapping? (this reflects + * the overlap of schema and formatting which is inherent in the MARC + * family) + */ + yaz_log (LOG_DEBUG, "grs_retrieve: syntax mapping"); + if (node->u.root.absyn) + for (map = node->u.root.absyn->maptabs; map; map = map->next) + { + if (map->target_absyn_ref == p->input_format) + { + onode = node; + if (!(node = data1_map_record(p->dh, onode, map, mem))) + { + p->diagnostic = 14; + nmem_destroy (mem); + return 0; + } + break; + } + } + yaz_log (LOG_DEBUG, "grs_retrieve: schemaIdentifier"); + if (node->u.root.absyn && + node->u.root.absyn->reference != VAL_NONE && + p->input_format == VAL_GRS1) { oident oe; Odr_oid *oid; int oidtmp[OID_SIZE]; - + oe.proto = PROTO_Z3950; oe.oclass = CLASS_SCHEMA; oe.value = node->u.root.absyn->reference; - + if ((oid = oid_ent_to_oid (&oe, oidtmp))) { char tmp[128]; data1_handle dh = p->dh; char *p = tmp; int *ii; - + for (ii = oid; *ii >= 0; ii++) { if (p != tmp) - *(p++) = '.'; + *(p++) = '.'; sprintf(p, "%d", *ii); p += strlen(p); } - *(p++) = '\0'; - - if ((dnew = data1_insert_taggeddata(dh, node, node, - "schemaIdentifier", mem))) + if ((dnew = data1_mk_tag_data_wd(dh, top, + "schemaIdentifier", mem))) { dnew->u.data.what = DATA1I_oid; - dnew->u.data.data = nmem_malloc(mem, p - tmp); + dnew->u.data.data = (char *) nmem_malloc(mem, p - tmp); memcpy(dnew->u.data.data, tmp, p - tmp); dnew->u.data.len = p - tmp; } } } - logf (LOG_DEBUG, "grs_retrieve: schema mapping"); - /* - * Does the requested format match a known schema-mapping? (this reflects - * the overlap of schema and formatting which is inherent in the MARC - * family) - * NOTE: This should look at the schema-specification in the compspec - * as well. - */ - for (map = node->u.root.absyn->maptabs; map; map = map->next) - if (map->target_absyn_ref == p->input_format) - { - onode = node; - if (!(node = data1_map_record(p->dh, onode, map, mem))) - { - p->diagnostic = 14; - nmem_destroy (mem); - return 0; - } - break; - } - logf (LOG_DEBUG, "grs_retrieve: element spec"); if (p->comp && (res = process_comp(p->dh, node, p->comp)) > 0) { @@ -584,67 +1077,105 @@ static int grs_retrieve(struct recRetrieveCtrl *p) else if (p->comp && !res) selected = 1; +#if 0 + data1_pr_tree (p->dh, node, stdout); +#endif logf (LOG_DEBUG, "grs_retrieve: transfer syntax mapping"); switch (p->output_format = (p->input_format != VAL_NONE ? - p->input_format : VAL_SUTRS)) + p->input_format : VAL_SUTRS)) { - data1_marctab *marctab; - int dummy; - - case VAL_GRS1: - dummy = 0; - if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected, - p->odr, &dummy))) - p->diagnostic = 238; /* not available in requested syntax */ - else - p->rec_len = -1; - break; - case VAL_EXPLAIN: - if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected, - p->odr))) - p->diagnostic = 238; - else - p->rec_len = -1; - break; - case VAL_SUMMARY: - if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected, - p->odr))) - p->diagnostic = 238; - else - p->rec_len = -1; - break; - case VAL_SUTRS: - if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected, - (int*)&p->rec_len))) - p->diagnostic = 238; - break; - case VAL_SOIF: - if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected, - (int*)&p->rec_len))) - p->diagnostic = 238; + case VAL_TEXT_XML: + zebra_xml_metadata (p, top, mem); + +#if 0 + data1_pr_tree (p->dh, node, stdout); +#endif + + if (p->encoding) + data1_iconv (p->dh, mem, node, p->encoding, "UTF-8"); + + if (!(p->rec_buf = data1_nodetoidsgml(p->dh, node, selected, + &p->rec_len))) + p->diagnostic = 238; + else + { + char *new_buf = (char*) odr_malloc (p->odr, p->rec_len); + memcpy (new_buf, p->rec_buf, p->rec_len); + p->rec_buf = new_buf; + } + break; + case VAL_GRS1: + dummy = 0; + if (!(p->rec_buf = data1_nodetogr(p->dh, node, selected, + p->odr, &dummy))) + p->diagnostic = 238; /* not available in requested syntax */ + else + p->rec_len = (size_t) (-1); + break; + case VAL_EXPLAIN: + if (!(p->rec_buf = data1_nodetoexplain(p->dh, node, selected, + p->odr))) + p->diagnostic = 238; + else + p->rec_len = (size_t) (-1); + break; + case VAL_SUMMARY: + if (!(p->rec_buf = data1_nodetosummary(p->dh, node, selected, + p->odr))) + p->diagnostic = 238; + else + p->rec_len = (size_t) (-1); + break; + case VAL_SUTRS: + if (p->encoding) + data1_iconv (p->dh, mem, node, p->encoding, "UTF-8"); + if (!(p->rec_buf = data1_nodetobuf(p->dh, node, selected, + &p->rec_len))) + p->diagnostic = 238; + else + { + char *new_buf = (char*) odr_malloc (p->odr, p->rec_len); + memcpy (new_buf, p->rec_buf, p->rec_len); + p->rec_buf = new_buf; + } + break; + case VAL_SOIF: + if (!(p->rec_buf = data1_nodetosoif(p->dh, node, selected, + &p->rec_len))) + p->diagnostic = 238; + else + { + char *new_buf = (char*) odr_malloc (p->odr, p->rec_len); + memcpy (new_buf, p->rec_buf, p->rec_len); + p->rec_buf = new_buf; + } + break; + default: + if (!node->u.root.absyn) + { + p->diagnostic = 238; break; - default: - if (!node->u.root.absyn) - { - p->diagnostic = 238; - break; - } - for (marctab = node->u.root.absyn->marc; marctab; - marctab = marctab->next) - if (marctab->reference == p->input_format) - break; - if (!marctab) - { - p->diagnostic = 238; - break; - } - if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node, - selected, - (int*)&p->rec_len))) - { - p->diagnostic = 238; + } + for (marctab = node->u.root.absyn->marc; marctab; + marctab = marctab->next) + if (marctab->reference == p->input_format) break; - } + if (!marctab) + { + p->diagnostic = 238; + break; + } + if (p->encoding) + data1_iconv (p->dh, mem, node, p->encoding, "UTF-8"); + if (!(p->rec_buf = data1_nodetomarc(p->dh, marctab, node, + selected, &p->rec_len))) + p->diagnostic = 238; + else + { + char *new_buf = (char*) odr_malloc (p->odr, p->rec_len); + memcpy (new_buf, p->rec_buf, p->rec_len); + p->rec_buf = new_buf; + } } if (node) data1_free_tree(p->dh, node);