X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fclient.c;h=318f64580a8c164bff865c21ede97c814c6e08dc;hb=a989048e5451df8a25201757ba09a4ae50fa6568;hp=98a2916f3d3fbbbd78e45652dadb54cb1838f338;hpb=29891cb3f3ed0507d9c0c86414b61adc23ca623f;p=pazpar2-moved-to-github.git diff --git a/src/client.c b/src/client.c index 98a2916..2b270ae 100644 --- a/src/client.c +++ b/src/client.c @@ -1,5 +1,5 @@ /* This file is part of Pazpar2. - Copyright (C) 2006-2008 Index Data + Copyright (C) 2006-2012 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 @@ -24,7 +24,6 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #if HAVE_CONFIG_H #include #endif - #include #include #include @@ -34,14 +33,7 @@ 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 #include @@ -58,21 +50,56 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include +#include +#include #define USE_TIMING 0 #if USE_TIMING #include #endif -#if HAVE_NETINET_IN_H -#include -#endif - -#include "pazpar2.h" - +#include "ppmutex.h" +#include "session.h" +#include "parameters.h" #include "client.h" #include "connection.h" #include "settings.h" +#include "relevance.h" +#include "incref.h" + +static YAZ_MUTEX g_mutex = 0; +static int no_clients = 0; +static int no_clients_total = 0; + +static int client_use(int delta) +{ + int clients; + if (!g_mutex) + yaz_mutex_create(&g_mutex); + yaz_mutex_enter(g_mutex); + no_clients += delta; + if (delta > 0) + no_clients_total += delta; + clients = no_clients; + yaz_mutex_leave(g_mutex); + yaz_log(YLOG_DEBUG, "%s clients=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), clients); + return clients; +} + +int clients_count(void) { + return client_use(0); +} + +int clients_count_total(void) { + int total = 0; + if (!g_mutex) + return 0; + yaz_mutex_enter(g_mutex); + total = no_clients_total; + yaz_mutex_leave(g_mutex); + return total; +} + /** \brief Represents client state for a connection to one search target */ struct client { @@ -80,14 +107,31 @@ struct client { struct connection *connection; struct session *session; char *pquery; // Current search - int hits; - int records; - int setno; - int requestid; // ID of current outstanding request + char *cqlquery; // used for SRU targets only + char *addinfo; // diagnostic info for most resent error + Odr_int hits; + int record_offset; + int filtered; // When using local:, this will count the number of filtered records. + int maxrecs; + int startrecs; int diagnostic; + int preferred; + struct suggestions *suggestions; enum client_state state; struct show_raw *show_raw; - struct client *next; // next client in session or next in free list + ZOOM_resultset resultset; + YAZ_MUTEX mutex; + int ref_count; + char *id; + facet_limits_t facet_limits; +}; + +struct suggestions { + NMEM nmem; + int num; + char **misspelled; + char **suggest; + char *passthrough; }; struct show_raw { @@ -96,6 +140,7 @@ struct show_raw { int binary; char *syntax; char *esn; + char *nativesyntax; void (*error_handler)(void *data, const char *addinfo); void (*record_handler)(void *data, const char *buf, size_t sz); void *data; @@ -104,20 +149,13 @@ struct show_raw { static const char *client_states[] = { "Client_Connecting", - "Client_Connected", "Client_Idle", - "Client_Initializing", - "Client_Searching", - "Client_Presenting", + "Client_Working", "Client_Error", "Client_Failed", - "Client_Disconnected", - "Client_Stopped", - "Client_Continue" + "Client_Disconnected" }; -static struct client *client_freelist = 0; - const char *client_get_state_str(struct client *cl) { return client_states[cl->state]; @@ -128,28 +166,37 @@ enum client_state client_get_state(struct client *cl) return cl->state; } +void client_set_state_nb(struct client *cl, enum client_state st) +{ + cl->state = st; +} + void client_set_state(struct client *cl, enum client_state st) { + int was_active = 0; + if (client_is_active(cl)) + was_active = 1; cl->state = st; - if (cl->session) + /* If client is going from being active to inactive and all clients + are now idle we fire a watch for the session . The assumption is + that session is not mutex locked if client is already active */ + if (was_active && !client_is_active(cl) && cl->session) { + int no_active = session_active_clients(cl->session); - if (no_active == 0) + yaz_log(YLOG_DEBUG, "%s: releasing watches on zero active: %d", + client_get_id(cl), no_active); + if (no_active == 0) { session_alert_watch(cl->session, SESSION_WATCH_SHOW); + session_alert_watch(cl->session, SESSION_WATCH_BYTARGET); + session_alert_watch(cl->session, SESSION_WATCH_TERMLIST); + session_alert_watch(cl->session, SESSION_WATCH_SHOW_PREF); + } } } static void client_show_raw_error(struct client *cl, const char *addinfo); -// Close connection and set state to error -void client_fatal(struct client *cl) -{ - //client_show_raw_error(cl, "client connection failure"); - yaz_log(YLOG_WARN, "Fatal error from %s", client_get_url(cl)); - connection_destroy(cl->connection); - client_set_state(cl, Client_Error); -} - struct connection *client_get_connection(struct client *cl) { return cl->connection; @@ -170,13 +217,43 @@ const char *client_get_pquery(struct client *cl) return cl->pquery; } -void client_set_requestid(struct client *cl, int id) +static void client_send_raw_present(struct client *cl); +static int nativesyntax_to_type(const char *s, char *type, ZOOM_record rec); + +static void client_show_immediate( + ZOOM_resultset resultset, struct session_database *sdb, int position, + void *data, + void (*error_handler)(void *data, const char *addinfo), + void (*record_handler)(void *data, const char *buf, size_t sz), + int binary, + const char *nativesyntax) { - cl->requestid = id; -} + ZOOM_record rec = 0; + char type[80]; + const char *buf; + int len; + if (!resultset) + { + error_handler(data, "no resultset"); + return; + } + rec = ZOOM_resultset_record_immediate(resultset, position-1); + if (!rec) + { + error_handler(data, "no record"); + return; + } + nativesyntax_to_type(nativesyntax, type, rec); + buf = ZOOM_record_get(rec, type, &len); + if (!buf) + { + error_handler(data, "no record"); + return; + } + record_handler(data, buf, len); +} -static void client_send_raw_present(struct client *cl); int client_show_raw_begin(struct client *cl, int position, const char *syntax, const char *esn, @@ -184,51 +261,82 @@ int client_show_raw_begin(struct client *cl, int position, void (*error_handler)(void *data, const char *addinfo), void (*record_handler)(void *data, const char *buf, size_t sz), - void **data2, - int binary) + int binary, + const char *nativesyntax) { - struct show_raw *rr, **rrp; - if (!cl->connection) - { /* the client has no connection */ - return -1; - } - rr = xmalloc(sizeof(*rr)); - *data2 = rr; - rr->position = position; - rr->active = 0; - rr->data = data; - rr->error_handler = error_handler; - rr->record_handler = record_handler; - rr->binary = binary; - if (syntax) - rr->syntax = xstrdup(syntax); - else - rr->syntax = 0; - if (esn) - rr->esn = xstrdup(esn); - else - rr->esn = 0; - rr->next = 0; - - for (rrp = &cl->show_raw; *rrp; rrp = &(*rrp)->next) - ; - *rrp = rr; - - if (cl->state == Client_Failed) - { - client_show_raw_error(cl, "client failed"); - } - else if (cl->state == Client_Disconnected) + if (!nativesyntax) { - client_show_raw_error(cl, "client disconnected"); + if (binary) + nativesyntax = "raw"; + else + { + struct session_database *sdb = client_get_database(cl); + nativesyntax = session_setting_oneval(sdb, PZ_NATIVESYNTAX); + } } + + if (syntax == 0 && esn == 0) + client_show_immediate(cl->resultset, client_get_database(cl), + position, data, + error_handler, record_handler, + binary, nativesyntax); else { - client_send_raw_present(cl); + struct show_raw *rr, **rrp; + + if (!cl->connection) + return -1; + + + rr = xmalloc(sizeof(*rr)); + rr->position = position; + rr->active = 0; + rr->data = data; + rr->error_handler = error_handler; + rr->record_handler = record_handler; + rr->binary = binary; + if (syntax) + rr->syntax = xstrdup(syntax); + else + rr->syntax = 0; + if (esn) + rr->esn = xstrdup(esn); + else + rr->esn = 0; + + assert(nativesyntax); + rr->nativesyntax = xstrdup(nativesyntax); + + rr->next = 0; + + for (rrp = &cl->show_raw; *rrp; rrp = &(*rrp)->next) + ; + *rrp = rr; + + if (cl->state == Client_Failed) + { + client_show_raw_error(cl, "client failed"); + } + else if (cl->state == Client_Disconnected) + { + client_show_raw_error(cl, "client disconnected"); + } + else + { + client_send_raw_present(cl); + } } return 0; } +static void client_show_raw_delete(struct show_raw *r) +{ + xfree(r->syntax); + xfree(r->esn); + xfree(r->nativesyntax); + xfree(r); +} + void client_show_raw_remove(struct client *cl, void *data) { struct show_raw *rr = data; @@ -238,7 +346,7 @@ void client_show_raw_remove(struct client *cl, void *data) if (*rrp) { *rrp = rr->next; - xfree(rr); + client_show_raw_delete(rr); } } @@ -247,7 +355,7 @@ void client_show_raw_dequeue(struct client *cl) struct show_raw *rr = cl->show_raw; cl->show_raw = rr->next; - xfree(rr); + client_show_raw_delete(rr); } static void client_show_raw_error(struct client *cl, const char *addinfo) @@ -263,7 +371,7 @@ static void client_send_raw_present(struct client *cl) { struct session_database *sdb = client_get_database(cl); struct connection *co = client_get_connection(cl); - ZOOM_resultset set = connection_get_resultset(co); + ZOOM_resultset set = cl->resultset; int offset = cl->show_raw->position; const char *syntax = 0; @@ -273,7 +381,7 @@ static void client_send_raw_present(struct client *cl) assert(set); yaz_log(YLOG_DEBUG, "%s: trying to present %d record(s) from %d", - client_get_url(cl), 1, offset); + client_get_id(cl), 1, offset); if (cl->show_raw->syntax) syntax = cl->show_raw->syntax; @@ -294,10 +402,9 @@ static void client_send_raw_present(struct client *cl) connection_continue(co); } -static int nativesyntax_to_type(struct session_database *sdb, char *type) +static int nativesyntax_to_type(const char *s, char *type, + ZOOM_record rec) { - const char *s = session_setting_oneval(sdb, PZ_NATIVESYNTAX); - if (s && *s) { if (!strncmp(s, "iso2709", 7)) @@ -305,16 +412,82 @@ static int nativesyntax_to_type(struct session_database *sdb, char *type) const char *cp = strchr(s, ';'); yaz_snprintf(type, 80, "xml; charset=%s", cp ? cp+1 : "marc-8s"); } - else if (!strncmp(s, "xml", 3)) + else if (!strncmp(s, "txml", 4)) { - strcpy(type, "xml"); + const char *cp = strchr(s, ';'); + yaz_snprintf(type, 80, "txml; charset=%s", cp ? cp+1 : "marc-8s"); } - else - return -1; - yaz_log(YLOG_LOG, "Returned type %s", type); + else /* pass verbatim to ZOOM - including "xml" */ + strcpy(type, s); return 0; } - return -1; + else /* attempt to deduce structure */ + { + const char *syntax = ZOOM_record_get(rec, "syntax", NULL); + if (syntax) + { + if (!strcmp(syntax, "XML")) + { + strcpy(type, "xml"); + return 0; + } + else if (!strcmp(syntax, "USmarc") || !strcmp(syntax, "MARC21")) + { + strcpy(type, "xml; charset=marc8-s"); + return 0; + } + else return -1; + } + else return -1; + } +} + +/** + * TODO Consider thread safety!!! + * + */ +void client_report_facets(struct client *cl, ZOOM_resultset rs) +{ + struct session_database *sdb = client_get_database(cl); + ZOOM_facet_field *facets = ZOOM_resultset_facets(rs); + + if (sdb && facets) + { + struct session *se = client_get_session(cl); + int facet_num = ZOOM_resultset_facets_size(rs); + struct setting *s; + + for (s = sdb->settings[PZ_FACETMAP]; s; s = s->next) + { + const char *p = strchr(s->name + 3, ':'); + if (p && p[1] && s->value && s->value[0]) + { + int facet_idx; + p++; /* p now holds logical facet name */ + for (facet_idx = 0; facet_idx < facet_num; facet_idx++) + { + const char *native_name = + ZOOM_facet_field_name(facets[facet_idx]); + if (native_name && !strcmp(s->value, native_name)) + { + size_t term_idx; + size_t term_num = + ZOOM_facet_field_term_count(facets[facet_idx]); + for (term_idx = 0; term_idx < term_num; term_idx++ ) + { + int freq; + const char *term = + ZOOM_facet_field_get_term(facets[facet_idx], + term_idx, &freq); + if (term) + add_facet(se, p, term, freq); + } + break; + } + } + } + } + } } static void ingest_raw_record(struct client *cl, ZOOM_record rec) @@ -323,108 +496,244 @@ static void ingest_raw_record(struct client *cl, ZOOM_record rec) int len; char type[80]; - if (cl->show_raw->binary) - strcpy(type, "raw"); - else - { - struct session_database *sdb = client_get_database(cl); - nativesyntax_to_type(sdb, type); - } - + nativesyntax_to_type(cl->show_raw->nativesyntax, type, rec); buf = ZOOM_record_get(rec, type, &len); cl->show_raw->record_handler(cl->show_raw->data, buf, len); client_show_raw_dequeue(cl); } +void client_check_preferred_watch(struct client *cl) +{ + struct session *se = cl->session; + yaz_log(YLOG_DEBUG, "client_check_preferred_watch: %s ", client_get_id(cl)); + if (se) + { + client_unlock(cl); + /* TODO possible threading issue. Session can have been destroyed */ + if (session_is_preferred_clients_ready(se)) { + session_alert_watch(se, SESSION_WATCH_SHOW_PREF); + } + else + yaz_log(YLOG_DEBUG, "client_check_preferred_watch: Still locked on preferred targets."); + + client_lock(cl); + } + else + yaz_log(YLOG_WARN, "client_check_preferred_watch: %s. No session!", client_get_id(cl)); + +} + +struct suggestions* client_suggestions_create(const char* suggestions_string); +static void client_suggestions_destroy(struct client *cl); + void client_search_response(struct client *cl) { struct connection *co = cl->connection; - struct session *se = cl->session; ZOOM_connection link = connection_get_link(co); - ZOOM_resultset resultset = connection_get_resultset(co); - const char *error, *addinfo; + ZOOM_resultset resultset = cl->resultset; + const char *error, *addinfo = 0; + if (ZOOM_connection_error(link, &error, &addinfo)) { cl->hits = 0; - cl->state = Client_Error; + client_set_state(cl, Client_Error); yaz_log(YLOG_WARN, "Search error %s (%s): %s", - error, addinfo, client_get_url(cl)); + error, addinfo, client_get_id(cl)); } else { + client_report_facets(cl, resultset); + cl->record_offset = cl->startrecs; cl->hits = ZOOM_resultset_size(resultset); - se->total_hits += cl->hits; + yaz_log(YLOG_DEBUG, "client_search_response: hits " ODR_INT_PRINTF, cl->hits); + if (cl->suggestions) + client_suggestions_destroy(cl); + cl->suggestions = client_suggestions_create(ZOOM_resultset_option_get(resultset, "suggestions")); + } +} + +void client_got_records(struct client *cl) +{ + struct session *se = cl->session; + if (se) + { + if (reclist_get_num_records(se->reclist) > 0) + { + client_unlock(cl); + session_alert_watch(se, SESSION_WATCH_SHOW); + session_alert_watch(se, SESSION_WATCH_BYTARGET); + session_alert_watch(se, SESSION_WATCH_TERMLIST); + session_alert_watch(se, SESSION_WATCH_RECORD); + client_lock(cl); + } } } +static void client_record_ingest(struct client *cl) +{ + const char *msg, *addinfo; + ZOOM_record rec = 0; + ZOOM_resultset resultset = cl->resultset; + int offset = cl->record_offset; + if ((rec = ZOOM_resultset_record_immediate(resultset, offset))) + { + cl->record_offset++; + if (cl->session == 0) { + /* no operation */ + } + else if (ZOOM_record_error(rec, &msg, &addinfo, 0)) + { + yaz_log(YLOG_WARN, "Record error %s (%s): %s (rec #%d)", + msg, addinfo, client_get_id(cl), cl->record_offset); + } + else + { + struct session_database *sdb = client_get_database(cl); + NMEM nmem = nmem_create(); + const char *xmlrec; + char type[80]; + + const char *s = session_setting_oneval(sdb, PZ_NATIVESYNTAX); + if (nativesyntax_to_type(s, type, rec)) + yaz_log(YLOG_WARN, "Failed to determine record type"); + xmlrec = ZOOM_record_get(rec, type, NULL); + if (!xmlrec) + yaz_log(YLOG_WARN, "ZOOM_record_get failed from %s", + client_get_id(cl)); + else + { + /* OK = 0, -1 = failure, -2 = Filtered */ + int rc = ingest_record(cl, xmlrec, cl->record_offset, nmem); + if (rc == -1) + yaz_log(YLOG_WARN, "Failed to ingest from %s", client_get_id(cl)); + if (rc == -2) + cl->filtered += 1; + } + nmem_destroy(nmem); + } + } + else + { + yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d", + offset); + } +} void client_record_response(struct client *cl) { struct connection *co = cl->connection; ZOOM_connection link = connection_get_link(co); - ZOOM_resultset resultset = connection_get_resultset(co); + ZOOM_resultset resultset = cl->resultset; const char *error, *addinfo; - yaz_log(YLOG_LOG, "client_record_response"); if (ZOOM_connection_error(link, &error, &addinfo)) { - cl->state = Client_Error; + client_set_state(cl, Client_Error); yaz_log(YLOG_WARN, "Search error %s (%s): %s", - error, addinfo, client_get_url(cl)); + error, addinfo, client_get_id(cl)); } else { - ZOOM_record rec = 0; - const char *msg, *addinfo; - - yaz_log(YLOG_LOG, "show_raw=%p show_raw->active=%d", - cl->show_raw, cl->show_raw ? cl->show_raw->active : 0); if (cl->show_raw && cl->show_raw->active) { - if ((rec = ZOOM_resultset_record(resultset, - cl->show_raw->position-1))) + ZOOM_record rec = 0; + if ((rec = ZOOM_resultset_record_immediate( + resultset, cl->show_raw->position-1))) { cl->show_raw->active = 0; ingest_raw_record(cl, rec); } + else + { + yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d", + cl->show_raw->position-1); + } } else { - int offset = cl->records; - if ((rec = ZOOM_resultset_record(resultset, offset))) - { - yaz_log(YLOG_LOG, "Record with offset %d", offset); - - cl->records++; - if (ZOOM_record_error(rec, &msg, &addinfo, 0)) - yaz_log(YLOG_WARN, "Record error %s (%s): %s (rec #%d)", - error, addinfo, client_get_url(cl), cl->records); - else - { - struct session_database *sdb = client_get_database(cl); - const char *xmlrec; - char type[80]; - nativesyntax_to_type(sdb, type); - if ((xmlrec = ZOOM_record_get(rec, type, NULL))) - { - if (ingest_record(cl, xmlrec, cl->records)) - { - session_alert_watch(cl->session, SESSION_WATCH_SHOW); - session_alert_watch(cl->session, SESSION_WATCH_RECORD); - } - else - yaz_log(YLOG_WARN, "Failed to ingest"); - } - else - yaz_log(YLOG_WARN, "Failed to extract ZOOM record"); - } + client_record_ingest(cl); + } + } +} - } +void client_reingest(struct client *cl) +{ + int i = cl->startrecs; + int to = cl->record_offset; + cl->filtered = 0; + + cl->record_offset = i; + for (; i < to; i++) + client_record_ingest(cl); +} + +static void client_set_facets_request(struct client *cl, ZOOM_connection link) +{ + struct session_database *sdb = client_get_database(cl); + + WRBUF w = wrbuf_alloc(); + + struct setting *s; + + for (s = sdb->settings[PZ_FACETMAP]; s; s = s->next) + { + const char *p = strchr(s->name + 3, ':'); + if (!p) + { + yaz_log(YLOG_WARN, "Malformed facetmap name: %s", s->name); + } + else if (s->value && s->value[0]) + { + wrbuf_puts(w, "@attr 1="); + yaz_encode_pqf_term(w, s->value, strlen(s->value)); + if (s->next) + wrbuf_puts(w, ","); + } + } + yaz_log(YLOG_DEBUG, "using facets str: %s", wrbuf_cstr(w)); + ZOOM_connection_option_set(link, "facets", + wrbuf_len(w) ? wrbuf_cstr(w) : 0); + wrbuf_destroy(w); +} + +int client_has_facet(struct client *cl, const char *name) +{ + struct session_database *sdb = client_get_database(cl); + struct setting *s; + + for (s = sdb->settings[PZ_FACETMAP]; s; s = s->next) + { + const char *p = strchr(s->name + 3, ':'); + if (p && !strcmp(name, p + 1)) + return 1; + } + return 0; +} + +static const char *get_strategy_plus_sort(struct client *l, const char *field) +{ + struct session_database *sdb = client_get_database(l); + struct setting *s; + + const char *strategy_plus_sort = 0; + + for (s = sdb->settings[PZ_SORTMAP]; s; s = s->next) + { + char *p = strchr(s->name + 3, ':'); + if (!p) + { + yaz_log(YLOG_WARN, "Malformed sortmap name: %s", s->name); + continue; + } + p++; + if (!strcmp(p, field)) + { + strategy_plus_sort = s->value; + break; } - if (!rec) - yaz_log(YLOG_WARN, "Expected record, but got NULL"); } + return strategy_plus_sort; } void client_start_search(struct client *cl) @@ -432,19 +741,39 @@ void client_start_search(struct client *cl) struct session_database *sdb = client_get_database(cl); struct connection *co = client_get_connection(cl); ZOOM_connection link = connection_get_link(co); + struct session *se = client_get_session(cl); ZOOM_resultset rs; - char *databaseName = sdb->database->databases[0]; - const char *opt_piggyback = session_setting_oneval(sdb, PZ_PIGGYBACK); - const char *opt_queryenc = session_setting_oneval(sdb, PZ_QUERYENCODING); - const char *opt_elements = session_setting_oneval(sdb, PZ_ELEMENTS); - const char *opt_requestsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX); - const char *opt_maxrecs = session_setting_oneval(sdb, PZ_MAXRECS); - + const char *opt_piggyback = session_setting_oneval(sdb, PZ_PIGGYBACK); + const char *opt_queryenc = session_setting_oneval(sdb, PZ_QUERYENCODING); + const char *opt_elements = session_setting_oneval(sdb, PZ_ELEMENTS); + const char *opt_requestsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX); + const char *opt_maxrecs = session_setting_oneval(sdb, PZ_MAXRECS); + const char *opt_sru = session_setting_oneval(sdb, PZ_SRU); + const char *opt_sort = session_setting_oneval(sdb, PZ_SORT); + const char *opt_preferred = session_setting_oneval(sdb, PZ_PREFERRED); + const char *extra_args = session_setting_oneval(sdb, PZ_EXTRA_ARGS); + const char *opt_present_chunk = session_setting_oneval(sdb, PZ_PRESENT_CHUNK); + ZOOM_query q; + char maxrecs_str[24], startrecs_str[24], present_chunk_str[24]; + int present_chunk = 20; // Default chunk size + if (opt_present_chunk && strcmp(opt_present_chunk,"")) { + present_chunk = atoi(opt_present_chunk); + yaz_log(YLOG_DEBUG, "Present chunk set to %d", present_chunk); + } assert(link); - cl->hits = -1; - cl->records = 0; cl->diagnostic = 0; + cl->filtered = 0; + + if (extra_args && *extra_args) + ZOOM_connection_option_set(link, "extraArgs", extra_args); + + if (opt_preferred) { + cl->preferred = atoi(opt_preferred); + if (cl->preferred) + yaz_log(YLOG_LOG, "Target %s has preferred status: %d", + client_get_id(cl), cl->preferred); + } if (*opt_piggyback) ZOOM_connection_option_set(link, "piggyback", opt_piggyback); @@ -452,78 +781,192 @@ void client_start_search(struct client *cl) ZOOM_connection_option_set(link, "piggyback", "1"); if (*opt_queryenc) ZOOM_connection_option_set(link, "rpnCharset", opt_queryenc); - if (*opt_elements) + if (*opt_sru && *opt_elements) + ZOOM_connection_option_set(link, "schema", opt_elements); + else if (*opt_elements) ZOOM_connection_option_set(link, "elementSetName", opt_elements); if (*opt_requestsyn) ZOOM_connection_option_set(link, "preferredRecordSyntax", opt_requestsyn); - if (*opt_maxrecs) - ZOOM_connection_option_set(link, "count", opt_maxrecs); - else + + if (opt_maxrecs && *opt_maxrecs) { - char n[128]; - sprintf(n, "%d", global_parameters.toget); - ZOOM_connection_option_set(link, "count", n); + cl->maxrecs = atoi(opt_maxrecs); } - if (!databaseName || !*databaseName) - databaseName = "Default"; - ZOOM_connection_option_set(link, "databaseName", databaseName); - ZOOM_connection_option_set(link, "presentChunk", "20"); + /* convert back to string representation used in ZOOM API */ + sprintf(maxrecs_str, "%d", cl->maxrecs); + ZOOM_connection_option_set(link, "count", maxrecs_str); - rs = ZOOM_connection_search_pqf(link, cl->pquery); - connection_set_resultset(co, rs); - connection_continue(co); -} + /* A present_chunk less than 1 will disable chunking. */ + if (present_chunk > 0 && cl->maxrecs > present_chunk) { + sprintf(present_chunk_str, "%d", present_chunk); + ZOOM_connection_option_set(link, "presentChunk", present_chunk_str); + yaz_log(YLOG_DEBUG, "Present chunk set to %s", present_chunk_str); + } + else { + ZOOM_connection_option_set(link, "presentChunk", maxrecs_str); + yaz_log(YLOG_DEBUG, "Present chunk set to %s (maxrecs)", maxrecs_str); + } + sprintf(startrecs_str, "%d", cl->startrecs); + ZOOM_connection_option_set(link, "start", startrecs_str); -struct client *client_create(void) -{ - struct client *r; - if (client_freelist) + /* TODO Verify does it break something for CQL targets(non-SOLR) ? */ + /* facets definition is in PQF */ + client_set_facets_request(cl, link); + + q = ZOOM_query_create(); + if (cl->cqlquery) { - r = client_freelist; - client_freelist = client_freelist->next; + yaz_log(YLOG_LOG, "Search %s CQL: %s", client_get_id(cl), + cl->cqlquery); + ZOOM_query_cql(q, cl->cqlquery); + if (*opt_sort) + ZOOM_query_sortby(q, opt_sort); } else - r = xmalloc(sizeof(struct client)); - r->pquery = 0; - r->database = 0; - r->connection = 0; - r->session = 0; - r->hits = 0; - r->records = 0; - r->setno = 0; - r->requestid = -1; - r->diagnostic = 0; - r->state = Client_Disconnected; - r->show_raw = 0; - r->next = 0; - return r; + { + yaz_log(YLOG_LOG, "Search %s PQF: %s", client_get_id(cl), cl->pquery); + + ZOOM_query_prefix(q, cl->pquery); + } + if (se->sorted_results) + { /* first entry is current sorting ! */ + const char *sort_strategy_and_spec = + get_strategy_plus_sort(cl, se->sorted_results->field); + int increasing = se->sorted_results->increasing; + if (sort_strategy_and_spec && strlen(sort_strategy_and_spec) < 40) + { + char spec[50], *p; + strcpy(spec, sort_strategy_and_spec); + p = strchr(spec, ':'); + if (p) + { + *p++ = '\0'; /* cut the string in two */ + while (*p == ' ') + p++; + if (increasing) + strcat(p, " <"); + else + strcat(p, " >"); + yaz_log(YLOG_LOG, "applying %s %s", spec, p); + ZOOM_query_sortby2(q, spec, p); + } + } + else + { + /* no native sorting.. If this is not the first search, then + skip it entirely */ + if (se->sorted_results->next) + { + ZOOM_query_destroy(q); + return; + } + } + } + client_set_state(cl, Client_Working); + cl->hits = 0; + cl->record_offset = 0; + rs = ZOOM_connection_search(link, q); + ZOOM_query_destroy(q); + ZOOM_resultset_destroy(cl->resultset); + cl->resultset = rs; + connection_continue(co); } -void client_destroy(struct client *c) +struct client *client_create(const char *id) { - struct session *se = c->session; - if (c == se->clients) - se->clients = c->next; - else + struct client *cl = xmalloc(sizeof(*cl)); + cl->maxrecs = 100; + cl->startrecs = 0; + cl->pquery = 0; + cl->cqlquery = 0; + cl->addinfo = 0; + cl->database = 0; + cl->connection = 0; + cl->session = 0; + cl->hits = 0; + cl->record_offset = 0; + cl->filtered = 0; + cl->diagnostic = 0; + cl->state = Client_Disconnected; + cl->show_raw = 0; + cl->resultset = 0; + cl->suggestions = 0; + cl->mutex = 0; + pazpar2_mutex_create(&cl->mutex, "client"); + cl->preferred = 0; + cl->ref_count = 1; + cl->facet_limits = 0; + assert(id); + cl->id = xstrdup(id); + client_use(1); + + return cl; +} + +void client_lock(struct client *c) +{ + yaz_mutex_enter(c->mutex); +} + +void client_unlock(struct client *c) +{ + yaz_mutex_leave(c->mutex); +} + +void client_incref(struct client *c) +{ + pazpar2_incref(&c->ref_count, c->mutex); + yaz_log(YLOG_DEBUG, "client_incref c=%p %s cnt=%d", + c, client_get_id(c), c->ref_count); +} + +int client_destroy(struct client *c) +{ + if (c) { - struct client *cc; - for (cc = se->clients; cc && cc->next != c; cc = cc->next) - ; - if (cc) - cc->next = c->next; + yaz_log(YLOG_DEBUG, "client_destroy c=%p %s cnt=%d", + c, client_get_id(c), c->ref_count); + if (!pazpar2_decref(&c->ref_count, c->mutex)) + { + xfree(c->pquery); + c->pquery = 0; + xfree(c->cqlquery); + c->cqlquery = 0; + xfree(c->addinfo); + c->addinfo = 0; + xfree(c->id); + assert(!c->connection); + facet_limits_destroy(c->facet_limits); + + if (c->resultset) + { + ZOOM_resultset_destroy(c->resultset); + } + yaz_mutex_destroy(&c->mutex); + xfree(c); + client_use(-1); + return 1; + } } - xfree(c->pquery); - - if (c->connection) - connection_release(c->connection); - c->next = client_freelist; - client_freelist = c; + return 0; } void client_set_connection(struct client *cl, struct connection *con) { - cl->connection = con; + if (cl->resultset) + ZOOM_resultset_release(cl->resultset); + if (con) + { + assert(cl->connection == 0); + cl->connection = con; + client_incref(cl); + } + else + { + cl->connection = con; + client_destroy(cl); + } } void client_disconnect(struct client *cl) @@ -533,14 +976,6 @@ void client_disconnect(struct client *cl) client_set_connection(cl, 0); } -// Extract terms from query into null-terminated termlist -static void extract_terms(NMEM nmem, struct ccl_rpn_node *query, char **termlist) -{ - int num = 0; - - pull_terms(nmem, query, termlist, &num); - termlist[num] = 0; -} // Initialize CCL map for a target static CCL_bibset prepare_cclmap(struct client *cl) @@ -567,104 +1002,462 @@ static CCL_bibset prepare_cclmap(struct client *cl) return res; } +// returns a xmalloced CQL query corresponding to the pquery in client +static char *make_cqlquery(struct client *cl, Z_RPNQuery *zquery) +{ + cql_transform_t cqlt = cql_transform_create(); + char *r = 0; + WRBUF wrb = wrbuf_alloc(); + int status; + + if ((status = cql_transform_rpn2cql_wrbuf(cqlt, wrb, zquery))) + { + yaz_log(YLOG_WARN, "Failed to generate CQL query, code=%d", status); + } + else + { + r = xstrdup(wrbuf_cstr(wrb)); + } + wrbuf_destroy(wrb); + cql_transform_close(cqlt); + return r; +} + +// returns a xmalloced SOLR query corresponding to the pquery in client +// TODO Could prob. be merge with the similar make_cqlquery +static char *make_solrquery(struct client *cl, Z_RPNQuery *zquery) +{ + solr_transform_t sqlt = solr_transform_create(); + char *r = 0; + WRBUF wrb = wrbuf_alloc(); + int status; + + if ((status = solr_transform_rpn2solr_wrbuf(sqlt, wrb, zquery))) + { + yaz_log(YLOG_WARN, "Failed to generate SOLR query, code=%d", status); + } + else + { + r = xstrdup(wrbuf_cstr(wrb)); + } + wrbuf_destroy(wrb); + solr_transform_close(sqlt); + return r; +} + +const char *client_get_facet_limit_local(struct client *cl, + struct session_database *sdb, + int *l, + NMEM nmem, int *num, char ***values) +{ + const char *name = 0; + const char *value = 0; + for (; (name = facet_limits_get(cl->facet_limits, *l, &value)); (*l)++) + { + struct setting *s = 0; + + for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next) + { + const char *p = strchr(s->name + 3, ':'); + if (p && !strcmp(p + 1, name) && s->value && + !strncmp(s->value, "local:", 6)) + { + const char *cp = s->value + 6; + while (*cp == ' ') + cp++; + + nmem_strsplit_escape2(nmem, "|", value, values, + num, 1, '\\', 1); + (*l)++; + return *cp ? cp : name; + } + } + } + return 0; +} + +static int apply_limit(struct session_database *sdb, + facet_limits_t facet_limits, + WRBUF w_pqf, WRBUF w_ccl) +{ + int ret = 0; + int i = 0; + const char *name; + const char *value; + NMEM nmem_tmp = nmem_create(); + for (i = 0; (name = facet_limits_get(facet_limits, i, &value)); i++) + { + struct setting *s = 0; + nmem_reset(nmem_tmp); + for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next) + { + const char *p = strchr(s->name + 3, ':'); + if (p && !strcmp(p + 1, name) && s->value) + { + char **values = 0; + int i, num = 0; + nmem_strsplit_escape2(nmem_tmp, "|", value, &values, + &num, 1, '\\', 1); + + if (!strncmp(s->value, "rpn:", 4)) + { + const char *pqf = s->value + 4; + + wrbuf_puts(w_pqf, "@and "); + wrbuf_puts(w_pqf, pqf); + wrbuf_puts(w_pqf, " "); + for (i = 0; i < num; i++) + { + if (i < num - 1) + wrbuf_puts(w_pqf, "@or "); + yaz_encode_pqf_term(w_pqf, values[i], + strlen(values[i])); + } + } + else if (!strncmp(s->value, "ccl:", 4)) + { + const char *ccl = s->value + 4; + + wrbuf_puts(w_ccl, " and ("); + + for (i = 0; i < num; i++) + { + if (i) + wrbuf_puts(w_ccl, " or "); + wrbuf_puts(w_ccl, ccl); + wrbuf_puts(w_ccl, "=\""); + wrbuf_puts(w_ccl, values[i]); + wrbuf_puts(w_ccl, "\""); + } + wrbuf_puts(w_ccl, ")"); + + } + else if (!strncmp(s->value, "local:", 6)) { + /* no operation */ + } + else + { + yaz_log(YLOG_WARN, "Target %s: Bad limitmap '%s'", + sdb->database->id, s->value); + ret = -1; /* bad limitmap */ + } + break; + } + } + if (!s) + { + yaz_log(YLOG_WARN, "Target %s: limit %s used, but no limitmap defined", + (sdb->database ? sdb->database->id : ""), name); + } + } + nmem_destroy(nmem_tmp); + return ret; +} + // Parse the query given the settings specific to this client -int client_parse_query(struct client *cl, const char *query) +// return 0 if query is OK but different from before +// return 1 if query is OK but same as before +// return -1 on query error +// return -2 on limit error +int client_parse_query(struct client *cl, const char *query, + facet_limits_t facet_limits, + const char *startrecs, const char *maxrecs) { struct session *se = client_get_session(cl); + struct session_database *sdb = client_get_database(cl); struct ccl_rpn_node *cn; int cerror, cpos; + ODR odr_out; CCL_bibset ccl_map = prepare_cclmap(cl); + const char *sru = session_setting_oneval(sdb, PZ_SRU); + const char *pqf_prefix = session_setting_oneval(sdb, PZ_PQF_PREFIX); + const char *pqf_strftime = session_setting_oneval(sdb, PZ_PQF_STRFTIME); + const char *query_syntax = session_setting_oneval(sdb, PZ_QUERY_SYNTAX); + WRBUF w_ccl, w_pqf; + int ret_value = 1; + Z_RPNQuery *zquery; if (!ccl_map) return -1; - cn = ccl_find_str(ccl_map, query, &cerror, &cpos); + if (maxrecs && atoi(maxrecs) != cl->maxrecs) + { + ret_value = 0; + cl->maxrecs = atoi(maxrecs); + } + + if (startrecs && atoi(startrecs) != cl->startrecs) + { + ret_value = 0; + cl->startrecs = atoi(startrecs); + } + + w_ccl = wrbuf_alloc(); + wrbuf_puts(w_ccl, query); + + w_pqf = wrbuf_alloc(); + if (*pqf_prefix) + { + wrbuf_puts(w_pqf, pqf_prefix); + wrbuf_puts(w_pqf, " "); + } + + if (apply_limit(sdb, facet_limits, w_pqf, w_ccl)) + return -2; + + facet_limits_destroy(cl->facet_limits); + cl->facet_limits = facet_limits_dup(facet_limits); + + yaz_log(YLOG_LOG, "CCL query: %s", wrbuf_cstr(w_ccl)); + cn = ccl_find_str(ccl_map, wrbuf_cstr(w_ccl), &cerror, &cpos); ccl_qual_rm(&ccl_map); if (!cn) { - cl->state = Client_Error; - yaz_log(YLOG_WARN, "Failed to parse query for %s", - client_get_database(cl)->database->url); + client_set_state(cl, Client_Error); + session_log(se, YLOG_WARN, "Failed to parse CCL query '%s' for %s", + wrbuf_cstr(w_ccl), + client_get_id(cl)); + wrbuf_destroy(w_ccl); + wrbuf_destroy(w_pqf); return -1; } - wrbuf_rewind(se->wrbuf); - ccl_pquery(se->wrbuf, cn); - xfree(cl->pquery); - cl->pquery = xstrdup(wrbuf_cstr(se->wrbuf)); + wrbuf_destroy(w_ccl); + + if (!pqf_strftime || !*pqf_strftime) + ccl_pquery(w_pqf, cn); + else + { + time_t cur_time = time(0); + struct tm *tm = localtime(&cur_time); + char tmp_str[300]; + const char *cp = tmp_str; + + /* see man strftime(3) for things .. In particular %% gets converted + to %.. And That's our original query .. */ + strftime(tmp_str, sizeof(tmp_str)-1, pqf_strftime, tm); + for (; *cp; cp++) + { + if (cp[0] == '%') + ccl_pquery(w_pqf, cn); + else + wrbuf_putc(w_pqf, cp[0]); + } + } + + if (!cl->pquery || strcmp(cl->pquery, wrbuf_cstr(w_pqf))) + { + xfree(cl->pquery); + cl->pquery = xstrdup(wrbuf_cstr(w_pqf)); + ret_value = 0; + } + wrbuf_destroy(w_pqf); + + xfree(cl->cqlquery); + cl->cqlquery = 0; + + odr_out = odr_createmem(ODR_ENCODE); + zquery = p_query_rpn(odr_out, cl->pquery); + if (!zquery) + { + + session_log(se, YLOG_WARN, "Invalid PQF query for %s: %s", + client_get_id(cl), cl->pquery); + ret_value = -1; + } + else + { + session_log(se, YLOG_LOG, "PQF for %s: %s", + client_get_id(cl), cl->pquery); + + /* Support for PQF on SRU targets. */ + if (strcmp(query_syntax, "pqf") != 0 && *sru) + { + if (!strcmp(sru, "solr")) + cl->cqlquery = make_solrquery(cl, zquery); + else + cl->cqlquery = make_cqlquery(cl, zquery); + if (!cl->cqlquery) + ret_value = -1; + } + } + odr_destroy(odr_out); + /* TODO FIX Not thread safe */ if (!se->relevance) { // Initialize relevance structure with query terms - char *p[512]; - extract_terms(se->nmem, cn, p); - se->relevance = relevance_create( - global_parameters.server->relevance_pct, - se->nmem, (const char **) p, - se->expected_maxrecs); + se->relevance = relevance_create_ccl(se->service->charsets, cn); } - ccl_rpn_delete(cn); - return 0; + return ret_value; } void client_set_session(struct client *cl, struct session *se) { cl->session = se; - cl->next = se->clients; - se->clients = cl; } int client_is_active(struct client *cl) { - if (cl->connection && (cl->state == Client_Continue || - cl->state == Client_Connecting || + if (cl->connection && (cl->state == Client_Connecting || cl->state == Client_Working)) return 1; return 0; } -struct client *client_next_in_session(struct client *cl) +int client_is_active_preferred(struct client *cl) { - if (cl) - return cl->next; + /* only count if this is a preferred target. */ + if (!cl->preferred) + return 0; + /* TODO No sure this the condition that Seb wants */ + if (cl->connection && (cl->state == Client_Connecting || + cl->state == Client_Working)) + return 1; return 0; +} +Odr_int client_get_hits(struct client *cl) +{ + return cl->hits; } -int client_get_hits(struct client *cl) +Odr_int client_get_approximation(struct client *cl) { + if (cl->record_offset > 0) { + Odr_int approx = ((10 * cl->hits * (cl->record_offset - cl->filtered)) / cl->record_offset + 5) /10; + yaz_log(YLOG_DEBUG, "%s: Approx: %lld * %d / %d = %lld ", client_get_id(cl), cl->hits, cl->record_offset - cl->filtered, cl->record_offset, approx); + return approx; + } return cl->hits; } int client_get_num_records(struct client *cl) { - return cl->records; + return cl->record_offset; +} + +int client_get_num_records_filtered(struct client *cl) +{ + return cl->filtered; +} + +void client_set_diagnostic(struct client *cl, int diagnostic, + const char *addinfo) +{ + cl->diagnostic = diagnostic; + xfree(cl->addinfo); + cl->addinfo = 0; + if (addinfo) + cl->addinfo = xstrdup(addinfo); } -int client_get_diagnostic(struct client *cl) +int client_get_diagnostic(struct client *cl, const char **addinfo) { + if (addinfo) + *addinfo = cl->addinfo; return cl->diagnostic; } +const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf) +{ + /* int idx; */ + struct suggestions *suggestions = cl->suggestions; + + if (!suggestions) { + //yaz_log(YLOG_DEBUG, "No suggestions found"); + return ""; + } + if (suggestions->passthrough) { + yaz_log(YLOG_DEBUG, "Passthrough Suggestions: \n%s\n", suggestions->passthrough); + return suggestions->passthrough; + } + if (suggestions->num == 0) { + return ""; + } + /* + for (idx = 0; idx < suggestions->num; idx++) { + wrbuf_printf(wrbuf, "suggest[idx]); + if (suggestions->misspelled[idx] && suggestions->misspelled[idx]) { + wrbuf_puts(wrbuf, suggestions->misspelled[idx]); + wrbuf_puts(wrbuf, "\n"); + } + else + wrbuf_puts(wrbuf, "/>\n"); + } + */ + return wrbuf_cstr(wrbuf); +} + + void client_set_database(struct client *cl, struct session_database *db) { cl->database = db; } -struct host *client_get_host(struct client *cl) +const char *client_get_id(struct client *cl) +{ + return cl->id; +} + +int client_get_maxrecs(struct client *cl) { - return client_get_database(cl)->database->host; + return cl->maxrecs; } -const char *client_get_url(struct client *cl) +void client_set_preferred(struct client *cl, int v) { - return client_get_database(cl)->database->url; + cl->preferred = v; +} + + +struct suggestions* client_suggestions_create(const char* suggestions_string) +{ + int i; + NMEM nmem; + struct suggestions *suggestions; + if (suggestions_string == 0) + return 0; + nmem = nmem_create(); + suggestions = nmem_malloc(nmem, sizeof(*suggestions)); + yaz_log(YLOG_DEBUG, "client target suggestions: %s", suggestions_string); + + suggestions->nmem = nmem; + suggestions->num = 0; + suggestions->misspelled = 0; + suggestions->suggest = 0; + suggestions->passthrough = nmem_strdup_null(nmem, suggestions_string); + + if (suggestions_string) + nmem_strsplit_escape2(suggestions->nmem, "\n", suggestions_string, &suggestions->suggest, + &suggestions->num, 1, '\\', 0); + /* Set up misspelled array */ + suggestions->misspelled = (char **) nmem_malloc(nmem, suggestions->num * sizeof(**suggestions->misspelled)); + /* replace = with \0 .. for each item */ + for (i = 0; i < suggestions->num; i++) + { + char *cp = strchr(suggestions->suggest[i], '='); + if (cp) { + *cp = '\0'; + suggestions->misspelled[i] = cp+1; + } + } + return suggestions; +} + +static void client_suggestions_destroy(struct client *cl) +{ + NMEM nmem = cl->suggestions->nmem; + cl->suggestions = 0; + nmem_destroy(nmem); } /* * Local variables: * c-basic-offset: 4 + * c-file-style: "Stroustrup" * indent-tabs-mode: nil * End: * vim: shiftwidth=4 tabstop=8 expandtab */ +