X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Flogic.c;h=a1de74816751790be1375517159873eb5c3475f1;hb=9c5d5d176df284a6ec614456306e87cb3a8d2f12;hp=f8a8f4ded7d0742905fbdb8c1fbbd4b4d3168208;hpb=d97bb5af15e77bfec1e8461afe475b5fe9d41106;p=pazpar2-moved-to-github.git diff --git a/src/logic.c b/src/logic.c index f8a8f4d..a1de748 100644 --- a/src/logic.c +++ b/src/logic.c @@ -1,23 +1,23 @@ -/* $Id: logic.c,v 1.52 2007-07-16 09:00:22 adam Exp $ +/* $Id: logic.c,v 1.67 2007-10-02 07:50:12 adam Exp $ Copyright (c) 2006-2007, Index Data. -This file is part of Pazpar2. + This file is part of Pazpar2. -Pazpar2 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. + Pazpar2 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. -Pazpar2 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. + Pazpar2 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 Pazpar2; see the file LICENSE. If not, write to the -Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA -02111-1307, USA. - */ + You should have received a copy of the GNU General Public License + along with Pazpar2; see the file LICENSE. If not, write to the + Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA + 02111-1307, USA. +*/ /** \file logic.c \brief high-level logic; mostly user sessions and settings @@ -46,6 +46,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include #include +#include #if HAVE_CONFIG_H #include "cconfig.h" @@ -92,10 +93,11 @@ struct parameters global_parameters = 100, MAX_CHUNK, 0, - 0 + 0, + 180, + 15 }; - // Recursively traverse query structure to extract terms. void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) { @@ -105,20 +107,20 @@ void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) switch (n->kind) { - case CCL_RPN_AND: - case CCL_RPN_OR: - case CCL_RPN_NOT: - case CCL_RPN_PROX: - pull_terms(nmem, n->u.p[0], termlist, num); - pull_terms(nmem, n->u.p[1], termlist, num); - break; - case CCL_RPN_TERM: - nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords); - for (i = 0; i < numwords; i++) - termlist[(*num)++] = words[i]; - break; - default: // NOOP - break; + case CCL_RPN_AND: + case CCL_RPN_OR: + case CCL_RPN_NOT: + case CCL_RPN_PROX: + pull_terms(nmem, n->u.p[0], termlist, num); + pull_terms(nmem, n->u.p[1], termlist, num); + break; + case CCL_RPN_TERM: + nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords); + for (i = 0; i < numwords; i++) + termlist[(*num)++] = words[i]; + break; + default: // NOOP + break; } } @@ -189,7 +191,18 @@ xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec) rdoc = xmlParseMemory((char*) wrbuf_buf(wrbuf_opac), wrbuf_len(wrbuf_opac)); if (!rdoc) + { yaz_log(YLOG_WARN, "Unable to parse OPAC XML"); + /* Was used to debug bug #1348 */ +#if 0 + FILE *f = fopen("/tmp/opac.xml.txt", "wb"); + if (f) + { + fwrite(wrbuf_buf(wrbuf_opac), 1, wrbuf_len(wrbuf_opac), f); + fclose(f); + } +#endif + } wrbuf_destroy(wrbuf_opac); } } @@ -221,7 +234,6 @@ xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec) return 0; } - yaz_marc_write_using_libxml2(sdb->yaz_marc, 1); if (yaz_marc_write_xml(sdb->yaz_marc, &res, "http://www.loc.gov/MARC21/slim", 0, 0) < 0) { @@ -248,7 +260,7 @@ xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec) FILE *lf = yaz_log_file(); if (lf) { - yaz_log(YLOG_LOG, "Normalized record from %s", db->url); + yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url); #if LIBXML_VERSION >= 20600 xmlDocFormatDump(lf, rdoc, 1); #else @@ -260,18 +272,88 @@ xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec) return rdoc; } -xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec) +#define MAX_XSLT_ARGS 16 + +// Add static values from session database settings if applicable +static void insert_settings_parameters(struct session_database *sdb, + struct session *se, char **parms) +{ + struct conf_service *service = global_parameters.server->service; + int i; + int nparms = 0; + int offset = 0; + + for (i = 0; i < service->num_metadata; i++) + { + struct conf_metadata *md = &service->metadata[i]; + int setting; + + if (md->setting == Metadata_setting_parameter && + (setting = settings_offset(md->name)) > 0) + { + const char *val = session_setting_oneval(sdb, setting); + if (val && nparms < MAX_XSLT_ARGS) + { + char *buf; + int len = strlen(val); + buf = nmem_malloc(se->nmem, len + 3); + buf[0] = '\''; + strcpy(buf + 1, val); + buf[len+1] = '\''; + buf[len+2] = '\0'; + parms[offset++] = md->name; + parms[offset++] = buf; + nparms++; + } + } + } + parms[offset] = 0; +} + +// Add static values from session database settings if applicable +static void insert_settings_values(struct session_database *sdb, xmlDoc *doc) +{ + struct conf_service *service = global_parameters.server->service; + int i; + + for (i = 0; i < service->num_metadata; i++) + { + struct conf_metadata *md = &service->metadata[i]; + int offset; + + if (md->setting == Metadata_setting_postproc && + (offset = settings_offset(md->name)) > 0) + { + const char *val = session_setting_oneval(sdb, offset); + if (val) + { + xmlNode *r = xmlDocGetRootElement(doc); + xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata", + (xmlChar *) val); + xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name); + } + } + } +} + +xmlDoc *normalize_record(struct session_database *sdb, struct session *se, + Z_External *rec) { struct database_retrievalmap *m; xmlDoc *rdoc = record_to_xml(sdb, rec); if (rdoc) { - for (m = sdb->map; m; m = m->next){ + for (m = sdb->map; m; m = m->next) + { xmlDoc *new = 0; { xmlNodePtr root = 0; - new = xsltApplyStylesheet(m->stylesheet, rdoc, 0); + char *parms[MAX_XSLT_ARGS*2+1]; + + insert_settings_parameters(sdb, se, parms); + + new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms); root= xmlDocGetRootElement(new); if (!new || !root || !(root->children)) { @@ -286,15 +368,24 @@ xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec) xmlFreeDoc(rdoc); rdoc = new; } + + insert_settings_values(sdb, rdoc); + if (global_parameters.dump_records) { - fprintf(stderr, "Record from %s\n----------------\n", - sdb->database->url); + FILE *lf = yaz_log_file(); + + if (lf) + { + yaz_log(YLOG_LOG, "Normalized record from %s", + sdb->database->url); #if LIBXML_VERSION >= 20600 - xmlDocFormatDump(stderr, rdoc, 1); + xmlDocFormatDump(lf, rdoc, 1); #else - xmlDocDump(stderr, rdoc); + xmlDocDump(lf, rdoc); #endif + fprintf(lf, "\n"); + } } } return rdoc; @@ -302,7 +393,7 @@ xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec) // Retrieve first defined value for 'name' for given database. // Will be extended to take into account user associated with session -char *session_setting_oneval(struct session_database *db, int offset) +const char *session_setting_oneval(struct session_database *db, int offset) { if (!db->settings[offset]) return ""; @@ -314,7 +405,7 @@ char *session_setting_oneval(struct session_database *db, int offset) // Initialize YAZ Map structures for MARC-based targets static int prepare_yazmarc(struct session_database *sdb) { - char *s; + const char *s; if (!sdb->settings) { @@ -354,7 +445,7 @@ static int prepare_yazmarc(struct session_database *sdb) // setting. However, this is not a realistic use scenario. static int prepare_map(struct session *se, struct session_database *sdb) { - char *s; + const char *s; if (!sdb->settings) { @@ -366,7 +457,30 @@ static int prepare_map(struct session *se, struct session_database *sdb) char **stylesheets; struct database_retrievalmap **m = &sdb->map; int num, i; + char auto_stylesheet[256]; + if (!strcmp(s, "auto")) + { + const char *request_syntax = session_setting_oneval( + sdb, PZ_REQUESTSYNTAX); + if (request_syntax) + { + char *cp; + yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet), + "%s.xsl", request_syntax); + for (cp = auto_stylesheet; *cp; cp++) + { + /* deliberately only consider ASCII */ + if (*cp > 32 && *cp < 127) + *cp = tolower(*cp); + } + s = auto_stylesheet; + } + else + { + yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet"); + } + } nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num); for (i = 0; i < num; i++) { @@ -437,13 +551,21 @@ int session_set_watch(struct session *s, int what, void session_alert_watch(struct session *s, int what) { - if (!s->watchlist[what].fun) - return; - http_remove_observer(s->watchlist[what].obs); - (*s->watchlist[what].fun)(s->watchlist[what].data); - s->watchlist[what].fun = 0; - s->watchlist[what].data = 0; - s->watchlist[what].obs = 0; + if (s->watchlist[what].fun) + { + /* our watch is no longer associated with http_channel */ + http_remove_observer(s->watchlist[what].obs); + session_watchfun fun = s->watchlist[what].fun; + void *data = s->watchlist[what].data; + + /* reset watch before fun is invoked - in case fun wants to set + it again */ + s->watchlist[what].fun = 0; + s->watchlist[what].data = 0; + s->watchlist[what].obs = 0; + + fun(data); + } } //callback for grep_databases @@ -533,23 +655,22 @@ enum pazpar2_error_code search(struct session *se, *addinfo = 0; nmem_reset(se->nmem); + se->relevance = 0; + se->total_records = se->total_hits = se->total_merged = 0; + se->reclist = 0; + se->num_termlists = 0; criteria = parse_filter(se->nmem, filter); se->requestid++; live_channels = select_targets(se, criteria); if (live_channels) { int maxrecs = live_channels * global_parameters.toget; - se->num_termlists = 0; se->reclist = reclist_create(se->nmem, maxrecs); - // This will be initialized in send_search() - se->total_records = se->total_hits = se->total_merged = 0; se->expected_maxrecs = maxrecs; } else return PAZPAR2_NO_TARGETS; - se->relevance = 0; - for (cl = se->clients; cl; cl = client_next_in_session(cl)) { if (prepare_session_database(se, client_get_database(cl)) < 0) @@ -587,15 +708,6 @@ static void session_init_databases_fun(void *context, struct database *db) new->database = db; new->yaz_marc = 0; -#ifdef HAVE_ICU - if (global_parameters.server && global_parameters.server->icu_chn) - new->pct = pp2_charset_create(global_parameters.server->icu_chn); - else - new->pct = pp2_charset_create(0); -#else // HAVE_ICU - new->pct = pp2_charset_create(0); -#endif // HAVE_ICU - new->map = 0; new->settings = nmem_malloc(se->session_nmem, sizeof(struct settings *) * num); @@ -619,8 +731,6 @@ static void session_database_destroy(struct session_database *sdb) xsltFreeStylesheet(m->stylesheet); if (sdb->yaz_marc) yaz_marc_destroy(sdb->yaz_marc); - if (sdb->pct) - pp2_charset_destroy(sdb->pct); } // Initialize session_database list -- this represents this session's view @@ -670,10 +780,10 @@ void session_apply_setting(struct session *se, char *dbname, char *setting, } // Jakub: This breaks the filter setting. /*if (offset == PZ_ID) - { - yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring"); - return; - }*/ + { + yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring"); + return; + }*/ new->precedence = 0; new->target = dbname; new->name = setting; @@ -685,23 +795,23 @@ void session_apply_setting(struct session *se, char *dbname, char *setting, // (happens when a search starts and client connections are prepared) switch (offset) { - case PZ_NATIVESYNTAX: - if (sdb->yaz_marc) - { - yaz_marc_destroy(sdb->yaz_marc); - sdb->yaz_marc = 0; - } - break; - case PZ_XSLT: - if (sdb->map) - { - struct database_retrievalmap *m; - // We don't worry about the map structure -- it's in nmem - for (m = sdb->map; m; m = m->next) - xsltFreeStylesheet(m->stylesheet); - sdb->map = 0; - } - break; + case PZ_NATIVESYNTAX: + if (sdb->yaz_marc) + { + yaz_marc_destroy(sdb->yaz_marc); + sdb->yaz_marc = 0; + } + break; + case PZ_XSLT: + if (sdb->map) + { + struct database_retrievalmap *m; + // We don't worry about the map structure -- it's in nmem + for (m = sdb->map; m; m = m->next) + xsltFreeStylesheet(m->stylesheet); + sdb->map = 0; + } + break; } } @@ -741,19 +851,24 @@ struct session *new_session(NMEM nmem) session->watchlist[i].data = 0; session->watchlist[i].fun = 0; } - return session; } -struct hitsbytarget *hitsbytarget(struct session *se, int *count) +struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem) { - static struct hitsbytarget res[1000]; // FIXME MM + struct hitsbytarget *res = 0; struct client *cl; + size_t sz = 0; + for (cl = se->clients; cl; cl = client_next_in_session(cl)) + sz++; + + res = nmem_malloc(nmem, sizeof(*res) * sz); *count = 0; for (cl = se->clients; cl; cl = client_next_in_session(cl)) { - char *name = session_setting_oneval(client_get_database(cl), PZ_NAME); + const char *name = session_setting_oneval(client_get_database(cl), + PZ_NAME); res[*count].id = client_get_database(cl)->database->url; res[*count].name = *name ? name : "Unknown"; @@ -764,7 +879,6 @@ struct hitsbytarget *hitsbytarget(struct session *se, int *count) res[*count].connected = client_get_connection(cl) ? 1 : 0; (*count)++; } - return res; } @@ -791,13 +905,13 @@ void report_nmem_stats(void) } #endif -struct record_cluster *show_single(struct session *s, int id) +struct record_cluster *show_single(struct session *s, const char *id) { struct record_cluster *r; reclist_rewind(s->reclist); while ((r = reclist_read_record(s->reclist))) - if (r->recid == id) + if (!strcmp(r->recid, id)) return r; return 0; } @@ -807,7 +921,7 @@ struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, NMEM nmem_show) { struct record_cluster **recs = nmem_malloc(nmem_show, *num - * sizeof(struct record_cluster *)); + * sizeof(struct record_cluster *)); struct reclist_sortparms *spp; int i; #if USE_TIMING @@ -875,14 +989,14 @@ void statistics(struct session *se, struct statistics *stat) stat->num_no_connection++; switch (client_get_state(cl)) { - case Client_Connecting: stat->num_connecting++; break; - case Client_Initializing: stat->num_initializing++; break; - case Client_Searching: stat->num_searching++; break; - case Client_Presenting: stat->num_presenting++; break; - case Client_Idle: stat->num_idle++; break; - case Client_Failed: stat->num_failed++; break; - case Client_Error: stat->num_error++; break; - default: break; + case Client_Connecting: stat->num_connecting++; break; + case Client_Initializing: stat->num_initializing++; break; + case Client_Searching: stat->num_searching++; break; + case Client_Presenting: stat->num_presenting++; break; + case Client_Idle: stat->num_idle++; break; + case Client_Failed: stat->num_failed++; break; + case Client_Error: stat->num_error++; break; + default: break; } count++; } @@ -958,7 +1072,8 @@ static struct record_metadata *record_metadata_init( char * p = value; p = normalize7bit_generic(p, " ,/.:(["); - rec_md->data.text = nmem_strdup(nmem, p); + rec_md->data.text.disp = nmem_strdup(nmem, p); + rec_md->data.text.sort = 0; } else if (type == Metadata_type_year) { @@ -976,7 +1091,8 @@ static struct record_metadata *record_metadata_init( struct record *ingest_record(struct client *cl, Z_External *rec, int record_no) { - xmlDoc *xdoc = normalize_record(client_get_database(cl), rec); + xmlDoc *xdoc = normalize_record(client_get_database(cl), + client_get_session(cl), rec); xmlNode *root, *n; struct record *record; struct record_cluster *cluster; @@ -985,6 +1101,9 @@ struct record *ingest_record(struct client *cl, Z_External *rec, xmlChar *type = 0; xmlChar *value = 0; struct conf_service *service = global_parameters.server->service; + const char *norm_str = 0; + pp2_relevance_token_t prt = 0; + WRBUF norm_wr = 0; if (!xdoc) return 0; @@ -996,21 +1115,40 @@ struct record *ingest_record(struct client *cl, Z_External *rec, xmlFreeDoc(xdoc); return 0; } - + record = record_create(se->nmem, service->num_metadata, service->num_sortkeys, cl, record_no); - mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey); - xmlFree(mergekey); - normalize7bit_mergekey((char *) mergekey_norm, 0); + prt = pp2_relevance_tokenize( + global_parameters.server->mergekey_pct, (const char *) mergekey); + + + norm_wr = wrbuf_alloc(); + + while ((norm_str = pp2_relevance_token_next(prt))) + { + if (*norm_str) + { + if (wrbuf_len(norm_wr)) + wrbuf_puts(norm_wr, " "); + wrbuf_puts(norm_wr, norm_str); + } + } + + mergekey_norm = (xmlChar *)nmem_strdup(se->nmem, wrbuf_cstr(norm_wr)); + wrbuf_destroy(norm_wr); + + pp2_relevance_token_destroy(prt); + xmlFree(mergekey); + cluster = reclist_insert(se->reclist, global_parameters.server->service, record, (char *) mergekey_norm, &se->total_merged); if (global_parameters.dump_records) - yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid, + yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid, client_get_database(cl)->database->url, record_no); if (!cluster) { @@ -1019,99 +1157,119 @@ struct record *ingest_record(struct client *cl, Z_External *rec, return 0; } relevance_newrec(se->relevance, cluster); - - - // now parsing XML record and adding data to cluster or record metadata - for (n = root->children; n; n = n->next) - { - if (type) - xmlFree(type); - if (value) - xmlFree(value); - type = value = 0; - - if (n->type != XML_ELEMENT_NODE) - continue; - if (!strcmp((const char *) n->name, "metadata")) - { - struct conf_metadata *ser_md = 0; - struct conf_sortkey *ser_sk = 0; - struct record_metadata **wheretoput = 0; - struct record_metadata *rec_md = 0; - int md_field_id = -1; - int sk_field_id = -1; - - type = xmlGetProp(n, (xmlChar *) "type"); - value = xmlNodeListGetString(xdoc, n->children, 1); - - if (!type || !value) - continue; - - md_field_id - = conf_service_metadata_field_id(service, (const char *) type); - if (md_field_id < 0) - { - yaz_log(YLOG_WARN, - "Ignoring unknown metadata element: %s", type); - continue; - } - - ser_md = &service->metadata[md_field_id]; - - if (ser_md->sortkey_offset >= 0){ - sk_field_id = ser_md->sortkey_offset; - ser_sk = &service->sortkeys[sk_field_id]; - } - - // non-merged metadata - rec_md = record_metadata_init(se->nmem, (char *) value, - ser_md->type); - if (!rec_md) - { - yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'", - value, type); - continue; - } - rec_md->next = record->metadata[md_field_id]; - record->metadata[md_field_id] = rec_md; - - // merged metadata - rec_md = record_metadata_init(se->nmem, (char *) value, - ser_md->type); - wheretoput = &cluster->metadata[md_field_id]; - - // and polulate with data: - // assign cluster or record based on merge action - if (ser_md->merge == Metadata_merge_unique) - { - struct record_metadata *mnode; - for (mnode = *wheretoput; mnode; mnode = mnode->next) - if (!strcmp((const char *) mnode->data.text, - rec_md->data.text)) - break; - if (!mnode) - { - rec_md->next = *wheretoput; - *wheretoput = rec_md; - } - } - else if (ser_md->merge == Metadata_merge_longest) - { - if (!*wheretoput - || strlen(rec_md->data.text) - > strlen((*wheretoput)->data.text)) - { - *wheretoput = rec_md; - if (ser_sk) - { - char *s = nmem_strdup(se->nmem, rec_md->data.text); - if (!cluster->sortkeys[sk_field_id]) - cluster->sortkeys[sk_field_id] = - nmem_malloc(se->nmem, - sizeof(union data_types)); - normalize7bit_mergekey(s, - (ser_sk->type == Metadata_sortkey_skiparticle)); - cluster->sortkeys[sk_field_id]->text = s; + + + // now parsing XML record and adding data to cluster or record metadata + for (n = root->children; n; n = n->next) + { + if (type) + xmlFree(type); + if (value) + xmlFree(value); + type = value = 0; + + if (n->type != XML_ELEMENT_NODE) + continue; + if (!strcmp((const char *) n->name, "metadata")) + { + struct conf_metadata *ser_md = 0; + struct conf_sortkey *ser_sk = 0; + struct record_metadata **wheretoput = 0; + struct record_metadata *rec_md = 0; + int md_field_id = -1; + int sk_field_id = -1; + + type = xmlGetProp(n, (xmlChar *) "type"); + value = xmlNodeListGetString(xdoc, n->children, 1); + + if (!type || !value || !*value) + continue; + + md_field_id + = conf_service_metadata_field_id(service, (const char *) type); + if (md_field_id < 0) + { + yaz_log(YLOG_WARN, + "Ignoring unknown metadata element: %s", type); + continue; + } + + ser_md = &service->metadata[md_field_id]; + + if (ser_md->sortkey_offset >= 0){ + sk_field_id = ser_md->sortkey_offset; + ser_sk = &service->sortkeys[sk_field_id]; + } + + // non-merged metadata + rec_md = record_metadata_init(se->nmem, (char *) value, + ser_md->type); + if (!rec_md) + { + yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'", + value, type); + continue; + } + rec_md->next = record->metadata[md_field_id]; + record->metadata[md_field_id] = rec_md; + + // merged metadata + rec_md = record_metadata_init(se->nmem, (char *) value, + ser_md->type); + wheretoput = &cluster->metadata[md_field_id]; + + // and polulate with data: + // assign cluster or record based on merge action + if (ser_md->merge == Metadata_merge_unique) + { + struct record_metadata *mnode; + for (mnode = *wheretoput; mnode; mnode = mnode->next) + if (!strcmp((const char *) mnode->data.text.disp, + rec_md->data.text.disp)) + break; + if (!mnode) + { + rec_md->next = *wheretoput; + *wheretoput = rec_md; + } + } + else if (ser_md->merge == Metadata_merge_longest) + { + if (!*wheretoput + || strlen(rec_md->data.text.disp) + > strlen((*wheretoput)->data.text.disp)) + { + *wheretoput = rec_md; + if (ser_sk) + { + const char *sort_str = 0; + int skip_article = + ser_sk->type == Metadata_sortkey_skiparticle; + + if (!cluster->sortkeys[sk_field_id]) + cluster->sortkeys[sk_field_id] = + nmem_malloc(se->nmem, + sizeof(union data_types)); + + prt = pp2_relevance_tokenize( + global_parameters.server->sort_pct, + rec_md->data.text.disp); + + pp2_relevance_token_next(prt); + + sort_str = pp2_get_sort(prt, skip_article); + + cluster->sortkeys[sk_field_id]->text.disp = + rec_md->data.text.disp; + cluster->sortkeys[sk_field_id]->text.sort = + nmem_strdup(se->nmem, sort_str); +#if 0 + yaz_log(YLOG_LOG, "text disp=%s", + cluster->sortkeys[sk_field_id]->text.disp); + yaz_log(YLOG_LOG, "text sort=%s", + cluster->sortkeys[sk_field_id]->text.sort); +#endif + pp2_relevance_token_destroy(prt); } } }