X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Flogic.c;h=02a133d33e2a2e7b9632576b2fca20dd30163463;hb=3d99aecedf9d6e6ebeadcedbee081049b03dbbf8;hp=167c2ae35c0c0511d639818483a0c27115969ea7;hpb=98748934f357929917e4884ea656b1a003d4c775;p=pazpar2-moved-to-github.git diff --git a/src/logic.c b/src/logic.c index 167c2ae..02a133d 100644 --- a/src/logic.c +++ b/src/logic.c @@ -1,5 +1,5 @@ /* This file is part of Pazpar2. - Copyright (C) 2006-2008 Index Data + Copyright (C) 2006-2009 Index Data 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 @@ -34,12 +34,6 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #if HAVE_UNISTD_H #include #endif -#if HAVE_SYS_SOCKET_H -#include -#endif -#if HAVE_NETDB_H -#include -#endif #include #include #include @@ -63,9 +57,6 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #endif -#if HAVE_NETINET_IN_H -#include -#endif #include "pazpar2.h" #include "eventl.h" @@ -104,6 +95,24 @@ struct parameters global_parameters = 15 // Connect timeout }; +static void log_xml_doc(xmlDoc *doc) +{ + FILE *lf = yaz_log_file(); + xmlChar *result = 0; + int len = 0; +#if LIBXML_VERSION >= 20600 + xmlDocDumpFormatMemory(doc, &result, &len, 1); +#else + xmlDocDumpMemory(doc, &result, &len); +#endif + if (lf && len) + { + fwrite(result, 1, len, lf); + fprintf(lf, "\n"); + } + xmlFree(result); +} + // Recursively traverse query structure to extract terms. void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) { @@ -174,17 +183,8 @@ xmlDoc *record_to_xml(struct session_database *sdb, const char *rec) if (global_parameters.dump_records) { - FILE *lf = yaz_log_file(); - if (lf) - { - yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url); -#if LIBXML_VERSION >= 20600 - xmlDocFormatDump(lf, rdoc, 1); -#else - xmlDocDump(lf, rdoc); -#endif - fprintf(lf, "\n"); - } + yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url); + log_xml_doc(rdoc); } return rdoc; @@ -291,19 +291,9 @@ xmlDoc *normalize_record(struct session_database *sdb, struct session *se, if (global_parameters.dump_records) { - FILE *lf = yaz_log_file(); - - if (lf) - { - yaz_log(YLOG_LOG, "Normalized record from %s", - sdb->database->url); -#if LIBXML_VERSION >= 20600 - xmlDocFormatDump(lf, rdoc, 1); -#else - xmlDocDump(lf, rdoc); -#endif - fprintf(lf, "\n"); - } + yaz_log(YLOG_LOG, "Normalized record from %s", + sdb->database->url); + log_xml_doc(rdoc); } } return rdoc; @@ -539,7 +529,6 @@ enum pazpar2_error_code search(struct session *se, se->reclist = 0; se->num_termlists = 0; criteria = parse_filter(se->nmem, filter); - se->requestid++; live_channels = select_targets(se, criteria); if (live_channels) { @@ -708,9 +697,9 @@ struct session *new_session(NMEM nmem) session->total_hits = 0; session->total_records = 0; session->number_of_warnings_unknown_elements = 0; + session->number_of_warnings_unknown_metadata = 0; session->num_termlists = 0; session->reclist = 0; - session->requestid = -1; session->clients = 0; session->expected_maxrecs = 0; session->session_nmem = nmem; @@ -962,6 +951,106 @@ static struct record_metadata *record_metadata_init( return rec_md; } +const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no, + struct conf_service *service, NMEM nmem) +{ + char *mergekey_norm = 0; + xmlNode *root = xmlDocGetRootElement(doc); + WRBUF norm_wr = wrbuf_alloc(); + xmlNode *n; + + /* create mergekey based on mergekey attribute from XSL (if any) */ + xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey"); + if (mergekey) + { + const char *norm_str; + pp2_relevance_token_t prt = + pp2_relevance_tokenize( + global_parameters.server->mergekey_pct, + (const char *) mergekey); + + 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); + } + } + pp2_relevance_token_destroy(prt); + xmlFree(mergekey); + } + /* append (if any) mergekey=yes metadata values */ + for (n = root->children; n; n = n->next) + { + if (n->type != XML_ELEMENT_NODE) + continue; + if (!strcmp((const char *) n->name, "metadata")) + { + struct conf_metadata *ser_md = 0; + int md_field_id = -1; + + xmlChar *type = xmlGetProp(n, (xmlChar *) "type"); + + if (!type) + continue; + + md_field_id + = conf_service_metadata_field_id(service, + (const char *) type); + if (md_field_id >= 0) + { + ser_md = &service->metadata[md_field_id]; + if (ser_md->mergekey == Metadata_mergekey_yes) + { + xmlChar *value = xmlNodeListGetString(doc, n->children, 1); + if (value) + { + const char *norm_str; + pp2_relevance_token_t prt = + pp2_relevance_tokenize( + global_parameters.server->mergekey_pct, + (const char *) value); + + 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); + } + } + xmlFree(value); + pp2_relevance_token_destroy(prt); + } + } + } + xmlFree(type); + } + } + + /* generate unique key if none is not generated already or is empty */ + if (wrbuf_len(norm_wr) == 0) + { + wrbuf_printf(norm_wr, "%s-%d", + client_get_database(cl)->database->url, record_no); + } + if (wrbuf_len(norm_wr) > 0) + mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr)); + wrbuf_destroy(norm_wr); + return mergekey_norm; +} + + + +/** \brief ingest XML record + \param cl client holds the result set for record + \param rec record buffer (0 terminated) + \param record_no record position (1, 2, ..) + \returns resulting record or NULL on failure +*/ struct record *ingest_record(struct client *cl, const char *rec, int record_no) { @@ -971,52 +1060,27 @@ struct record *ingest_record(struct client *cl, const char *rec, struct record *record; struct record_cluster *cluster; struct session *se = client_get_session(cl); - xmlChar *mergekey, *mergekey_norm; + const char *mergekey_norm; 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; root = xmlDocGetRootElement(xdoc); - if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey"))) + + mergekey_norm = get_mergekey(xdoc, cl, record_no, service, se->nmem); + if (!mergekey_norm) { - yaz_log(YLOG_WARN, "No mergekey found in record"); + yaz_log(YLOG_WARN, "Got no mergekey"); xmlFreeDoc(xdoc); return 0; } - record = record_create(se->nmem, service->num_metadata, service->num_sortkeys, cl, record_no); - 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, @@ -1031,133 +1095,131 @@ struct record *ingest_record(struct client *cl, const char *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 || !*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; - } - wheretoput = &record->metadata[md_field_id]; - while (*wheretoput) - wheretoput = &(*wheretoput)->next; - *wheretoput = 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_no) - { - while (*wheretoput) - wheretoput = &(*wheretoput)->next; - *wheretoput = rec_md; - } - 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)); + + // now parsing XML record and adding data to cluster or record metadata + for (n = root->children; n; n = n->next) + { + pp2_relevance_token_t prt; + 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) + { + if (se->number_of_warnings_unknown_metadata == 0) + { + yaz_log(YLOG_WARN, + "Ignoring unknown metadata element: %s", type); + } + se->number_of_warnings_unknown_metadata++; + 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; + } + wheretoput = &record->metadata[md_field_id]; + while (*wheretoput) + wheretoput = &(*wheretoput)->next; + *wheretoput = 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); + prt = pp2_relevance_tokenize( + global_parameters.server->sort_pct, + rec_md->data.text.disp); - pp2_relevance_token_next(prt); + pp2_relevance_token_next(prt); - sort_str = pp2_get_sort(prt, skip_article); + sort_str = pp2_get_sort(prt, skip_article); - cluster->sortkeys[sk_field_id]->text.disp = - rec_md->data.text.disp; - if (!sort_str) - { - sort_str = rec_md->data.text.disp; - yaz_log(YLOG_WARN, - "Could not make sortkey. Bug #1858"); - } - cluster->sortkeys[sk_field_id]->text.sort = - nmem_strdup(se->nmem, sort_str); + cluster->sortkeys[sk_field_id]->text.disp = + rec_md->data.text.disp; + if (!sort_str) + { + sort_str = rec_md->data.text.disp; + yaz_log(YLOG_WARN, + "Could not make sortkey. Bug #1858"); + } + 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); + 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); + pp2_relevance_token_destroy(prt); } } } @@ -1215,13 +1277,13 @@ struct record *ingest_record(struct client *cl, const char *rec, xmlFree(value); type = value = 0; } - else - { - if (se->number_of_warnings_unknown_elements == 0) - yaz_log(YLOG_WARN, - "Unexpected element in internal record: %s", n->name); - se->number_of_warnings_unknown_elements++; - } + else + { + if (se->number_of_warnings_unknown_elements == 0) + yaz_log(YLOG_WARN, + "Unexpected element in internal record: %s", n->name); + se->number_of_warnings_unknown_elements++; + } } if (type) xmlFree(type); @@ -1241,7 +1303,9 @@ struct record *ingest_record(struct client *cl, const char *rec, /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab */ +