X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Flogic.c;h=2ee120341b2f034426f7d7b97aaad8d72e5794b1;hb=5c017d34db6ea43ee6a95b8c462b9b824c26fbb8;hp=7f23b2741d28343761bbf768d96634e17adda76a;hpb=d54caf29766776a237d07c93e8f2bc76f5dc9c5e;p=pazpar2-moved-to-github.git diff --git a/src/logic.c b/src/logic.c index 7f23b27..2ee1203 100644 --- a/src/logic.c +++ b/src/logic.c @@ -68,7 +68,6 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include "client.h" #include "settings.h" #include "normalize7bit.h" -#include "marcmap.h" #define TERMLIST_HIGH_SCORE 25 @@ -78,8 +77,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA struct parameters global_parameters = { 0, // dump_records - 0, // debug_mode - 100, + 0 // debug_mode }; static void log_xml_doc(xmlDoc *doc) @@ -147,14 +145,13 @@ 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, - TERMLIST_HIGH_SCORE); + = termlist_create(s->nmem, TERMLIST_HIGH_SCORE); s->num_termlists = i + 1; } termlist_insert(s->termlists[i].termlist, value); } -xmlDoc *record_to_xml(struct session_database *sdb, const char *rec) +static xmlDoc *record_to_xml(struct session_database *sdb, const char *rec) { struct database *db = sdb->database; xmlDoc *rdoc = 0; @@ -241,55 +238,32 @@ static void insert_settings_values(struct session_database *sdb, xmlDoc *doc, } } -xmlDoc *normalize_record(struct session_database *sdb, struct session *se, - const char *rec) +static xmlDoc *normalize_record(struct session_database *sdb, + struct session *se, + const char *rec) { - struct database_retrievalmap *m; xmlDoc *rdoc = record_to_xml(sdb, rec); + if (rdoc) { - for (m = sdb->map; m; m = m->next) + char *parms[MAX_XSLT_ARGS*2+1]; + + insert_settings_parameters(sdb, se, parms); + + if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms)) { - xmlDoc *new = 0; + yaz_log(YLOG_WARN, "Normalize failed from %s", sdb->database->url); + } + else + { + insert_settings_values(sdb, rdoc, se->service); + if (global_parameters.dump_records) { - xmlNodePtr root = 0; - char *parms[MAX_XSLT_ARGS*2+1]; - - insert_settings_parameters(sdb, se, parms); - - if (m->stylesheet) - { - new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms); - } - else if (m->marcmap) - { - new = marcmap_apply(m->marcmap, rdoc); - } - - 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; - } + yaz_log(YLOG_LOG, "Normalized record from %s", + sdb->database->url); + log_xml_doc(rdoc); } - - xmlFreeDoc(rdoc); - rdoc = new; - } - - insert_settings_values(sdb, rdoc, se->service); - - if (global_parameters.dump_records) - { - yaz_log(YLOG_LOG, "Normalized record from %s", - sdb->database->url); - log_xml_doc(rdoc); } } return rdoc; @@ -320,9 +294,6 @@ static int prepare_map(struct session *se, struct session_database *sdb) } if ((s = session_setting_oneval(sdb, PZ_XSLT))) { - char **stylesheets; - struct database_retrievalmap **m = &sdb->map; - int num, i; char auto_stylesheet[256]; if (!strcmp(s, "auto")) @@ -347,41 +318,11 @@ static int prepare_map(struct session *se, struct session_database *sdb) yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet"); } } - nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num); - for (i = 0; i < num; i++) - { - (*m) = nmem_malloc(se->session_nmem, sizeof(**m)); - (*m)->next = 0; - - // XSLT - if (!strcmp(&stylesheets[i][strlen(stylesheets[i])-4], ".xsl")) - { - (*m)->marcmap = NULL; - if (!((*m)->stylesheet = conf_load_stylesheet(se->service, stylesheets[i]))) - { - yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s", - stylesheets[i]); - return -1; - } - } - // marcmap - else if (!strcmp(&stylesheets[i][strlen(stylesheets[i])-5], ".mmap")) - { - (*m)->stylesheet = NULL; - if (!((*m)->marcmap = marcmap_load(stylesheets[i], se->session_nmem))) - { - yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load marcmap: %s", - stylesheets[i]); - return -1; - } - } - - m = &(*m)->next; - } + sdb->map = normalize_cache_get(se->normalize_cache, + se->service, s); + if (!sdb->map) + return -1; } - if (!sdb->map) - yaz_log(YLOG_WARN, "No Normalization stylesheet for target %s", - sdb->database->url); return 0; } @@ -500,10 +441,14 @@ static struct database_criterion *parse_filter(NMEM m, const char *buf) int subnum; int subi; struct database_criterion *new = nmem_malloc(m, sizeof(*new)); - char *eq = strchr(values[i], '='); - if (!eq) + char *eq; + if ((eq = strchr(values[i], '='))) + new->type = PAZPAR2_STRING_MATCH; + else if ((eq = strchr(values[i], '~'))) + new->type = PAZPAR2_SUBSTRING_MATCH; + else { - yaz_log(YLOG_WARN, "Missing equal-sign in filter"); + yaz_log(YLOG_WARN, "Missing equal-sign/tilde in filter"); return 0; } *(eq++) = '\0'; @@ -525,7 +470,9 @@ static struct database_criterion *parse_filter(NMEM m, const char *buf) } enum pazpar2_error_code search(struct session *se, - const char *query, const char *filter, + const char *query, + const char *maxrecs, + const char *filter, const char **addinfo) { int live_channels = 0; @@ -546,37 +493,37 @@ enum pazpar2_error_code search(struct session *se, live_channels = select_targets(se, criteria); if (live_channels) { - int maxrecs = live_channels * global_parameters.toget; // This is buggy!!! - se->reclist = reclist_create(se->nmem, maxrecs); - se->expected_maxrecs = maxrecs; + se->reclist = reclist_create(se->nmem); } else return PAZPAR2_NO_TARGETS; for (cl = se->clients; cl; cl = client_next_in_session(cl)) { + if (maxrecs) + client_set_maxrecs(cl, atoi(maxrecs)); if (prepare_session_database(se, client_get_database(cl)) < 0) - { - *addinfo = client_get_database(cl)->database->url; - return PAZPAR2_CONFIG_TARGET; - } + continue; // Parse query for target if (client_parse_query(cl, query) < 0) no_failed++; else { no_working++; - if (client_prep_connection(cl, se->service->z3950_connect_timeout, - se->service->z3950_session_timeout)) + if (client_prep_connection(cl, se->service->z3950_operation_timeout, + se->service->z3950_session_timeout)) client_start_search(cl); } } - - // If no queries could be mapped, we signal an error if (no_working == 0) { - *addinfo = "query"; - return PAZPAR2_MALFORMED_PARAMETER_VALUE; + if (no_failed > 0) + { + *addinfo = "query"; + return PAZPAR2_MALFORMED_PARAMETER_VALUE; + } + else + return PAZPAR2_NO_TARGETS; } return PAZPAR2_NO_ERROR; } @@ -609,10 +556,7 @@ static void session_init_databases_fun(void *context, struct database *db) // Doesn't free memory associated with sdb -- nmem takes care of that static void session_database_destroy(struct session_database *sdb) { - struct database_retrievalmap *m; - - for (m = sdb->map; m; m = m->next) - xsltFreeStylesheet(m->stylesheet); + sdb->map = 0; } // Initialize session_database list -- this represents this session's view @@ -683,10 +627,6 @@ void session_apply_setting(struct session *se, char *dbname, char *setting, 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; @@ -701,6 +641,7 @@ void destroy_session(struct session *s) client_destroy(s->clients); for (sdb = s->databases; sdb; sdb = sdb->next) session_database_destroy(sdb); + normalize_cache_destroy(s->normalize_cache); nmem_destroy(s->nmem); service_destroy(s->service); wrbuf_destroy(s->wrbuf); @@ -722,7 +663,6 @@ struct session *new_session(NMEM nmem, struct conf_service *service) session->num_termlists = 0; session->reclist = 0; session->clients = 0; - session->expected_maxrecs = 0; session->session_nmem = nmem; session->nmem = nmem_create(); session->wrbuf = wrbuf_alloc(); @@ -732,6 +672,8 @@ struct session *new_session(NMEM nmem, struct conf_service *service) session->watchlist[i].data = 0; session->watchlist[i].fun = 0; } + session->normalize_cache = normalize_cache_create(); + return session; } @@ -836,7 +778,7 @@ struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, } reclist_sort(s->reclist, sp); - *total = s->reclist->num_records; + *total = reclist_get_num_records(s->reclist); *sumhits = s->total_hits; for (i = 0; i < start; i++) @@ -940,15 +882,60 @@ static struct record_metadata *record_metadata_init( return rec_md; } +static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name, + struct conf_service *service, WRBUF norm_wr) +{ + xmlNode *n; + int no_found = 0; + for (n = root->children; n; n = n->next) + { + if (n->type != XML_ELEMENT_NODE) + continue; + if (!strcmp((const char *) n->name, "metadata")) + { + xmlChar *type = xmlGetProp(n, (xmlChar *) "type"); + if (!strcmp(name, (const char *) type)) + { + xmlChar *value = xmlNodeListGetString(doc, n->children, 1); + if (value) + { + const char *norm_str; + pp2_relevance_token_t prt = + pp2_relevance_tokenize( + service->mergekey_pct, + (const char *) value); + + wrbuf_puts(norm_wr, name); + wrbuf_puts(norm_wr, "="); + 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); + no_found++; + } + } + xmlFree(type); + } + } + return no_found; +} + static 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) */ + /* consider mergekey from XSL first */ xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey"); if (mergekey) { @@ -970,53 +957,25 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no, pp2_relevance_token_destroy(prt); xmlFree(mergekey); } - /* append (if any) mergekey=yes metadata values */ - for (n = root->children; n; n = n->next) + else { - if (n->type != XML_ELEMENT_NODE) - continue; - if (!strcmp((const char *) n->name, "metadata")) + /* no mergekey defined in XSL. Look for mergekey metadata instead */ + int field_id; + for (field_id = 0; field_id < service->num_metadata; field_id++) { - 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) + struct conf_metadata *ser_md = &service->metadata[field_id]; + if (ser_md->mergekey != Metadata_mergekey_no) { - ser_md = &service->metadata[md_field_id]; - if (ser_md->mergekey == Metadata_mergekey_yes) + int r = get_mergekey_from_doc(doc, root, ser_md->name, + service, norm_wr); + if (r == 0 && ser_md->mergekey == Metadata_mergekey_required) { - xmlChar *value = xmlNodeListGetString(doc, n->children, 1); - if (value) - { - const char *norm_str; - pp2_relevance_token_t prt = - pp2_relevance_tokenize( - service->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); - } + /* no mergekey on this one and it is required.. + Generate unique key instead */ + wrbuf_rewind(norm_wr); + break; } } - xmlFree(type); } } @@ -1059,19 +1018,27 @@ static int check_record_filter(xmlNode *root, struct session_database *sdb) if (!strcmp((const char *) n->name, "metadata")) { xmlChar *type = xmlGetProp(n, (xmlChar *) "type"); - - if (!type) - continue; - if (!strcmp((const char *) type, s)) + if (type) { - xmlChar *value = xmlNodeGetContent(n); - if (value && *value) + size_t len; + const char *eq = strchr(s, '='); + if (eq) + len = eq - s; + else + len = strlen(s); + if (len == strlen((const char *)type) && + !memcmp((const char *) type, s, len)) { + xmlChar *value = xmlNodeGetContent(n); + if (value && *value) + { + if (!eq || strstr((const char *) value, eq+1)) + match = 1; + } xmlFree(value); - match = 1; } + xmlFree(type); } - xmlFree(type); } } return match; @@ -1088,11 +1055,11 @@ struct record *ingest_record(struct client *cl, const char *rec, int record_no) { struct session_database *sdb = client_get_database(cl); - xmlDoc *xdoc = normalize_record(sdb, client_get_session(cl), rec); + struct session *se = client_get_session(cl); + xmlDoc *xdoc = normalize_record(sdb, se, rec); xmlNode *root, *n; struct record *record; struct record_cluster *cluster; - struct session *se = client_get_session(cl); const char *mergekey_norm; xmlChar *type = 0; xmlChar *value = 0; @@ -1126,15 +1093,15 @@ struct record *ingest_record(struct client *cl, const char *rec, service, record, (char *) mergekey_norm, &se->total_merged); - if (global_parameters.dump_records) - yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid, - client_get_database(cl)->database->url, record_no); if (!cluster) { /* no room for record */ xmlFreeDoc(xdoc); return 0; } + if (global_parameters.dump_records) + yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid, + sdb->database->url, record_no); relevance_newrec(se->relevance, cluster); // now parsing XML record and adding data to cluster or record metadata