X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Flogic.c;h=f8a8f4ded7d0742905fbdb8c1fbbd4b4d3168208;hb=d97bb5af15e77bfec1e8461afe475b5fe9d41106;hp=b484279429d0d5e6eb6bd77d550a988c64df81a8;hpb=69ae0b0ac176c892d6fd9d3b1e0c9d0e4a5c6b64;p=pazpar2-moved-to-github.git diff --git a/src/logic.c b/src/logic.c index b484279..f8a8f4d 100644 --- a/src/logic.c +++ b/src/logic.c @@ -1,4 +1,4 @@ -/* $Id: logic.c,v 1.28 2007-05-14 08:01:39 marc Exp $ +/* $Id: logic.c,v 1.52 2007-07-16 09:00:22 adam Exp $ Copyright (c) 2006-2007, Index Data. This file is part of Pazpar2. @@ -45,9 +45,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include #include -#if YAZ_VERSIONL >= 0x020163 #include -#endif #if HAVE_CONFIG_H #include "cconfig.h" @@ -72,6 +70,8 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include "settings.h" #include "normalize7bit.h" +#define TERMLIST_HIGH_SCORE 25 + #define MAX_CHUNK 15 // Note: Some things in this structure will eventually move to configuration @@ -79,10 +79,10 @@ struct parameters global_parameters = { "", "", - "", - "", - 0, + "", 0, + 0, /* dump_records */ + 0, /* debug_mode */ 30, "81", "Index Data PazPar2", @@ -143,98 +143,159 @@ static void add_facet(struct session *s, const char *type, const char *value) s->termlists[i].name = nmem_strdup(s->nmem, type); s->termlists[i].termlist - = termlist_create(s->nmem, s->expected_maxrecs, 15); + = termlist_create(s->nmem, s->expected_maxrecs, + TERMLIST_HIGH_SCORE); s->num_termlists = i + 1; } termlist_insert(s->termlists[i].termlist, value); } -xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec) +xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec) { - struct database_retrievalmap *m; struct database *db = sdb->database; - xmlNode *res; - xmlDoc *rdoc; + xmlDoc *rdoc = 0; + const Odr_oid *oid = rec->direct_reference; - // First normalize to XML - if (sdb->yaz_marc) + /* convert response record to XML somehow */ + if (rec->which == Z_External_octet && oid + && !oid_oidcmp(oid, yaz_oid_recsyn_xml)) { - char *buf; - int len; - if (rec->which != Z_External_octet) + /* xml already */ + rdoc = xmlParseMemory((char*) rec->u.octet_aligned->buf, + rec->u.octet_aligned->len); + if (!rdoc) { - yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s", + yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s", db->url); return 0; } - buf = (char*) rec->u.octet_aligned->buf; - len = rec->u.octet_aligned->len; - if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0) + } + else if (rec->which == Z_External_OPAC) + { + if (!sdb->yaz_marc) { - yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url); + yaz_log(YLOG_WARN, "MARC decoding not configured"); 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) + else { - yaz_log(YLOG_WARN, "Failed to encode as XML %s", - db->url); + /* OPAC gets converted to XML too */ + WRBUF wrbuf_opac = wrbuf_alloc(); + /* MARCXML inside the OPAC XML. Charset is in effect because we + use the yaz_marc handle */ + yaz_marc_xml(sdb->yaz_marc, YAZ_MARC_MARCXML); + yaz_opac_decode_wrbuf(sdb->yaz_marc, rec->u.opac, wrbuf_opac); + + rdoc = xmlParseMemory((char*) wrbuf_buf(wrbuf_opac), + wrbuf_len(wrbuf_opac)); + if (!rdoc) + yaz_log(YLOG_WARN, "Unable to parse OPAC XML"); + wrbuf_destroy(wrbuf_opac); + } + } + else if (oid && yaz_oid_is_iso2709(oid)) + { + /* ISO2709 gets converted to MARCXML */ + if (!sdb->yaz_marc) + { + yaz_log(YLOG_WARN, "MARC decoding not configured"); return 0; } - rdoc = xmlNewDoc((xmlChar *) "1.0"); - xmlDocSetRootElement(rdoc, res); - + else + { + xmlNode *res; + char *buf; + int len; + + if (rec->which != Z_External_octet) + { + yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s", + db->url); + return 0; + } + buf = (char*) rec->u.octet_aligned->buf; + len = rec->u.octet_aligned->len; + if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0) + { + yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url); + 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) + { + yaz_log(YLOG_WARN, "Failed to encode as XML %s", + db->url); + return 0; + } + rdoc = xmlNewDoc((xmlChar *) "1.0"); + xmlDocSetRootElement(rdoc, res); + } } else { + char oid_name_buf[OID_STR_MAX]; + const char *oid_name = yaz_oid_to_string_buf(oid, 0, oid_name_buf); yaz_log(YLOG_FATAL, - "Unknown native_syntax in normalize_record from %s", - db->url); + "Unable to handle record of type %s from %s", + oid_name, db->url); return 0; } - - if (global_parameters.dump_records){ - fprintf(stderr, - "Input Record (normalized) from %s\n----------------\n", - db->url); + + if (global_parameters.dump_records) + { + FILE *lf = yaz_log_file(); + if (lf) + { + yaz_log(YLOG_LOG, "Normalized record from %s", db->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; +} - for (m = sdb->map; m; m = m->next){ - xmlDoc *new = 0; - - { - xmlNodePtr root = 0; - new = xsltApplyStylesheet(m->stylesheet, rdoc, 0); - root= xmlDocGetRootElement(new); - if (!new || !root || !(root->children)) - { - yaz_log(YLOG_WARN, "XSLT transformation failed from %s", - sdb->database->url); - xmlFreeDoc(new); +xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec) +{ + struct database_retrievalmap *m; + xmlDoc *rdoc = record_to_xml(sdb, rec); + if (rdoc) + { + for (m = sdb->map; m; m = m->next){ + xmlDoc *new = 0; + + { + xmlNodePtr root = 0; + new = xsltApplyStylesheet(m->stylesheet, rdoc, 0); + root= xmlDocGetRootElement(new); + if (!new || !root || !(root->children)) + { + yaz_log(YLOG_WARN, "XSLT transformation failed from %s", + sdb->database->url); + xmlFreeDoc(new); + xmlFreeDoc(rdoc); + return 0; + } + } + xmlFreeDoc(rdoc); - return 0; - } + rdoc = new; } - - xmlFreeDoc(rdoc); - rdoc = new; - } - if (global_parameters.dump_records) - { - fprintf(stderr, "Record from %s\n----------------\n", - sdb->database->url); + if (global_parameters.dump_records) + { + fprintf(stderr, "Record from %s\n----------------\n", + sdb->database->url); #if LIBXML_VERSION >= 20600 - xmlDocFormatDump(stderr, rdoc, 1); + xmlDocFormatDump(stderr, rdoc, 1); #else - xmlDocDump(stderr, rdoc); + xmlDocDump(stderr, rdoc); #endif + } } return rdoc; } @@ -313,7 +374,7 @@ static int prepare_map(struct session *se, struct session_database *sdb) (*m)->next = 0; if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i]))) { - yaz_log(YLOG_FATAL, "Unable to load stylesheet: %s", + yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s", stylesheets[i]); return -1; } @@ -350,21 +411,39 @@ static int prepare_session_database(struct session *se, return 0; } +// called if watch should be removed because http_channel is to be destroyed +static void session_watch_cancel(void *data, struct http_channel *c) +{ + struct session_watchentry *ent = data; -void session_set_watch(struct session *s, int what, - session_watchfun fun, void *data) + ent->fun = 0; + ent->data = 0; + ent->obs = 0; +} + +// set watch. Returns 0=OK, -1 if watch is already set +int session_set_watch(struct session *s, int what, + session_watchfun fun, void *data, + struct http_channel *chan) { + if (s->watchlist[what].fun) + return -1; s->watchlist[what].fun = fun; s->watchlist[what].data = data; + s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what], + session_watch_cancel); + return 0; } 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; } //callback for grep_databases @@ -440,14 +519,19 @@ static struct database_criterion *parse_filter(NMEM m, const char *buf) return res; } -char *search(struct session *se, char *query, char *filter) +enum pazpar2_error_code search(struct session *se, + char *query, char *filter, + const char **addinfo) { int live_channels = 0; + int no_working = 0; + int no_failed = 0; struct client *cl; struct database_criterion *criteria; yaz_log(YLOG_DEBUG, "Search"); + *addinfo = 0; nmem_reset(se->nmem); criteria = parse_filter(se->nmem, filter); se->requestid++; @@ -462,25 +546,34 @@ char *search(struct session *se, char *query, char *filter) se->expected_maxrecs = maxrecs; } else - return "NOTARGETS"; + 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) - return "CONFIG_ERROR"; - // Query must parse for all targets - if (client_parse_query(cl, query) < 0) - return "QUERY"; + { + *addinfo = client_get_database(cl)->database->url; + return PAZPAR2_CONFIG_TARGET; + } + // Parse query for target + if (client_parse_query(cl, query) < 0) + no_failed++; + else + { + no_working++; + client_prep_connection(cl); + } } - for (cl = se->clients; cl; cl = client_next_in_session(cl)) + // If no queries could be mapped, we signal an error + if (no_working == 0) { - client_prep_connection(cl); + *addinfo = "query"; + return PAZPAR2_MALFORMED_PARAMETER_VALUE; } - - return 0; + return PAZPAR2_NO_ERROR; } // Creates a new session_database object for a database @@ -493,7 +586,16 @@ static void session_init_databases_fun(void *context, struct database *db) new->database = db; new->yaz_marc = 0; - new->pct = pp2_charset_create(); + +#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); @@ -526,7 +628,7 @@ static void session_database_destroy(struct session_database *sdb) void session_init_databases(struct session *se) { se->databases = 0; - grep_databases(se, 0, session_init_databases_fun); + predef_grep_databases(se, 0, session_init_databases_fun); } // Probably session_init_databases_fun should be refactored instead of @@ -559,13 +661,19 @@ void session_apply_setting(struct session *se, char *dbname, char *setting, { struct session_database *sdb = find_session_database(se, dbname); struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new)); - int offset = settings_offset(setting); + int offset = settings_offset_cprefix(setting); if (offset < 0) { yaz_log(YLOG_WARN, "Unknown setting %s", setting); return; } + // Jakub: This breaks the filter setting. + /*if (offset == PZ_ID) + { + yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring"); + return; + }*/ new->precedence = 0; new->target = dbname; new->name = setting; @@ -601,7 +709,6 @@ void destroy_session(struct session *s) { struct session_database *sdb; - yaz_log(YLOG_LOG, "Destroying session"); while (s->clients) client_destroy(s->clients); for (sdb = s->databases; sdb; sdb = sdb->next) @@ -617,6 +724,7 @@ struct session *new_session(NMEM nmem) yaz_log(YLOG_DEBUG, "New Pazpar2 session"); + session->relevance = 0; session->total_hits = 0; session->total_records = 0; session->num_termlists = 0; @@ -627,7 +735,7 @@ struct session *new_session(NMEM nmem) session->session_nmem = nmem; session->nmem = nmem_create(); session->wrbuf = wrbuf_alloc(); - session_init_databases(session); + session->databases = 0; for (i = 0; i <= SESSION_WATCH_MAX; i++) { session->watchlist[i].data = 0; @@ -706,34 +814,44 @@ struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, yaz_timing_t t = yaz_timing_create(); #endif - for (spp = sp; spp; spp = spp->next) - if (spp->type == Metadata_sortkey_relevance) - { - relevance_prepare_read(s->relevance, s->reclist); - break; - } - reclist_sort(s->reclist, sp); - - *total = s->reclist->num_records; - *sumhits = s->total_hits; - - for (i = 0; i < start; i++) - if (!reclist_read_record(s->reclist)) - { - *num = 0; - recs = 0; - break; - } - - for (i = 0; i < *num; i++) + if (!s->relevance) { - struct record_cluster *r = reclist_read_record(s->reclist); - if (!r) + *num = 0; + *total = 0; + *sumhits = 0; + recs = 0; + } + else + { + for (spp = sp; spp; spp = spp->next) + if (spp->type == Metadata_sortkey_relevance) + { + relevance_prepare_read(s->relevance, s->reclist); + break; + } + reclist_sort(s->reclist, sp); + + *total = s->reclist->num_records; + *sumhits = s->total_hits; + + for (i = 0; i < start; i++) + if (!reclist_read_record(s->reclist)) + { + *num = 0; + recs = 0; + break; + } + + for (i = 0; i < *num; i++) { - *num = i; - break; + struct record_cluster *r = reclist_read_record(s->reclist); + if (!r) + { + *num = i; + break; + } + recs[i] = r; } - recs[i] = r; } #if USE_TIMING yaz_timing_stop(t); @@ -817,38 +935,6 @@ void start_proxy(void) http_set_proxyaddr(hp, ser->myurl ? ser->myurl : ""); } -void start_zproxy(void) -{ - struct conf_server *ser = global_parameters.server; - - if (*global_parameters.zproxy_override){ - yaz_log(YLOG_LOG, "Z39.50 proxy %s", - global_parameters.zproxy_override); - return; - } - - else if (ser->zproxy_host || ser->zproxy_port) - { - char hp[128] = ""; - - strcpy(hp, ser->zproxy_host ? ser->zproxy_host : ""); - if (ser->zproxy_port) - { - if (*hp) - strcat(hp, ":"); - else - strcat(hp, "@:"); - - sprintf(hp + strlen(hp), "%d", ser->zproxy_port); - } - strcpy(global_parameters.zproxy_override, hp); - yaz_log(YLOG_LOG, "Z39.50 proxy %s", - global_parameters.zproxy_override); - - } - else - return; -} // Master list of connections we're handling events to static IOCHAN channel_list = 0; @@ -863,6 +949,30 @@ void pazpar2_event_loop() event_loop(&channel_list); } +static struct record_metadata *record_metadata_init( + NMEM nmem, char *value, enum conf_metadata_type type) +{ + struct record_metadata *rec_md = record_metadata_create(nmem); + if (type == Metadata_type_generic) + { + char * p = value; + p = normalize7bit_generic(p, " ,/.:(["); + + rec_md->data.text = nmem_strdup(nmem, p); + } + else if (type == Metadata_type_year) + { + int first, last; + if (extract7bit_years((char *) value, &first, &last) < 0) + return 0; + rec_md->data.number.min = first; + rec_md->data.number.max = last; + } + else + return 0; + return rec_md; +} + struct record *ingest_record(struct client *cl, Z_External *rec, int record_no) { @@ -888,8 +998,8 @@ struct record *ingest_record(struct client *cl, Z_External *rec, } record = record_create(se->nmem, - service->num_metadata, service->num_sortkeys); - record_assign_client(record, cl); + service->num_metadata, service->num_sortkeys, cl, + record_no); mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey); xmlFree(mergekey); @@ -930,10 +1040,9 @@ struct record *ingest_record(struct client *cl, Z_External *rec, struct record_metadata *rec_md = 0; int md_field_id = -1; int sk_field_id = -1; - int first, last; type = xmlGetProp(n, (xmlChar *) "type"); - value = xmlNodeListGetString(xdoc, n->children, 0); + value = xmlNodeListGetString(xdoc, n->children, 1); if (!type || !value) continue; @@ -954,37 +1063,22 @@ struct record *ingest_record(struct client *cl, Z_External *rec, ser_sk = &service->sortkeys[sk_field_id]; } - // Find out where we are putting it - based on merge or not - if (ser_md->merge == Metadata_merge_no) - wheretoput = &record->metadata[md_field_id]; - else - wheretoput = &cluster->metadata[md_field_id]; - - // create new record_metadata - rec_md = record_metadata_create(se->nmem); - - // and polulate with data: - // type based charmapping decisions follow here - if (ser_md->type == Metadata_type_generic) + // non-merged metadata + rec_md = record_metadata_init(se->nmem, (char *) value, + ser_md->type); + if (!rec_md) { - - char * p = (char *) value; - p = normalize7bit_generic(p, " ,/.:(["); - - rec_md->data.text = nmem_strdup(se->nmem, p); - - } - else if (ser_md->type == Metadata_type_year) - { - if (extract7bit_years((char *) value, &first, &last) < 0) - continue; - } - else - { - yaz_log(YLOG_WARN, - "Unknown type in metadata element %s", type); + 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 @@ -1021,8 +1115,7 @@ struct record *ingest_record(struct client *cl, Z_External *rec, } } } - else if (ser_md->merge == Metadata_merge_all - || ser_md->merge == Metadata_merge_no) + else if (ser_md->merge == Metadata_merge_all) { rec_md->next = *wheretoput; *wheretoput = rec_md; @@ -1032,18 +1125,18 @@ struct record *ingest_record(struct client *cl, Z_External *rec, if (!*wheretoput) { *wheretoput = rec_md; - (*wheretoput)->data.number.min = first; - (*wheretoput)->data.number.max = last; if (ser_sk) cluster->sortkeys[sk_field_id] = &rec_md->data; } else { - if (first < (*wheretoput)->data.number.min) - (*wheretoput)->data.number.min = first; - if (last > (*wheretoput)->data.number.max) - (*wheretoput)->data.number.max = last; + int this_min = rec_md->data.number.min; + int this_max = rec_md->data.number.max; + if (this_min < (*wheretoput)->data.number.min) + (*wheretoput)->data.number.min = this_min; + if (this_max > (*wheretoput)->data.number.max) + (*wheretoput)->data.number.max = this_max; } #ifdef GAGA if (ser_sk) @@ -1068,11 +1161,11 @@ struct record *ingest_record(struct client *cl, Z_External *rec, if (ser_md->type == Metadata_type_year) { char year[64]; - sprintf(year, "%d", last); + sprintf(year, "%d", rec_md->data.number.max); add_facet(se, (char *) type, year); - if (first != last) + if (rec_md->data.number.max != rec_md->data.number.min) { - sprintf(year, "%d", first); + sprintf(year, "%d", rec_md->data.number.min); add_facet(se, (char *) type, year); } }