X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fclient.c;h=e5d3e4c101e9020eaeebfbb7f48b6251132f5082;hb=37482a306afb1c7952bdf304530498b6d068db92;hp=59fb027290d8c077ec120c248d3ef22ac4705e11;hpb=a6d3bfef7037ade6ce0abd1ba732e5f6c6854170;p=pazpar2-moved-to-github.git diff --git a/src/client.c b/src/client.c index 59fb027..e5d3e4c 100644 --- a/src/client.c +++ b/src/client.c @@ -1,5 +1,5 @@ /* This file is part of Pazpar2. - Copyright (C) 2006-2012 Index Data + Copyright (C) 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 @@ -18,7 +18,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ /** \file client.c - \brief Z39.50 client + \brief Z39.50 client */ #if HAVE_CONFIG_H @@ -33,6 +33,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #if HAVE_UNISTD_H #include #endif +#ifdef WIN32 +#include +#endif #include #include @@ -52,6 +55,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include +#include #define USE_TIMING 0 #if USE_TIMING @@ -67,9 +71,10 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include "relevance.h" #include "incref.h" +#define XDOC_CACHE_SIZE 100 + static YAZ_MUTEX g_mutex = 0; static int no_clients = 0; -static int no_clients_total = 0; static int client_use(int delta) { @@ -78,29 +83,18 @@ static int client_use(int delta) 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); + yaz_log(YLOG_DEBUG, "%s clients=%d", + delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), clients); return clients; } -int clients_count(void) { +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 { struct session_database *database; @@ -111,10 +105,14 @@ struct client { 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 show_stat_no; + int filtered; /* number of records ignored for local filtering */ + int ingest_failures; /* number of records where XSLT/other failed */ + int record_failures; /* number of records where ZOOM reported error */ int maxrecs; int startrecs; int diagnostic; + char *message; int preferred; struct suggestions *suggestions; enum client_state state; @@ -124,6 +122,10 @@ struct client { int ref_count; char *id; facet_limits_t facet_limits; + int same_search; + char *sort_strategy; + char *sort_criteria; + xmlDoc **xdoc; }; struct suggestions { @@ -195,6 +197,50 @@ void client_set_state(struct client *cl, enum client_state st) } } +static void client_init_xdoc(struct client *cl) +{ + int i; + + cl->xdoc = xmalloc(sizeof(*cl->xdoc) * XDOC_CACHE_SIZE); + for (i = 0; i < XDOC_CACHE_SIZE; i++) + cl->xdoc[i] = 0; +} + +static void client_destroy_xdoc(struct client *cl) +{ + int i; + + assert(cl->xdoc); + for (i = 0; i < XDOC_CACHE_SIZE; i++) + if (cl->xdoc[i]) + xmlFreeDoc(cl->xdoc[i]); + xfree(cl->xdoc); +} + +xmlDoc *client_get_xdoc(struct client *cl, int record_no) +{ + assert(cl->xdoc); + if (record_no >= 0 && record_no < XDOC_CACHE_SIZE) + return cl->xdoc[record_no]; + return 0; +} + +void client_store_xdoc(struct client *cl, int record_no, xmlDoc *xdoc) +{ + assert(cl->xdoc); + if (record_no >= 0 && record_no < XDOC_CACHE_SIZE) + { + if (cl->xdoc[record_no]) + xmlFreeDoc(cl->xdoc[record_no]); + cl->xdoc[record_no] = xdoc; + } + else + { + xmlFreeDoc(xdoc); + } +} + + static void client_show_raw_error(struct client *cl, const char *addinfo); struct connection *client_get_connection(struct client *cl) @@ -212,11 +258,6 @@ struct session *client_get_session(struct client *cl) return cl->session; } -const char *client_get_pquery(struct client *cl) -{ - return cl->pquery; -} - static void client_send_raw_present(struct client *cl); static int nativesyntax_to_type(const char *s, char *type, ZOOM_record rec); @@ -286,7 +327,7 @@ int client_show_raw_begin(struct client *cl, int position, if (!cl->connection) return -1; - + rr = xmalloc(sizeof(*rr)); rr->position = position; @@ -306,13 +347,13 @@ int client_show_raw_begin(struct client *cl, int position, 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"); @@ -350,7 +391,7 @@ void client_show_raw_remove(struct client *cl, void *data) } } -void client_show_raw_dequeue(struct client *cl) +static void client_show_raw_dequeue(struct client *cl) { struct show_raw *rr = cl->show_raw; @@ -412,16 +453,12 @@ static int nativesyntax_to_type(const char *s, 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)) - { - strcpy(type, "xml"); - } else if (!strncmp(s, "txml", 4)) { const char *cp = strchr(s, ';'); yaz_snprintf(type, 80, "txml; charset=%s", cp ? cp+1 : "marc-8s"); } - else + else /* pass verbatim to ZOOM - including "xml" */ strcpy(type, s); return 0; } @@ -450,7 +487,7 @@ static int nativesyntax_to_type(const char *s, char *type, * TODO Consider thread safety!!! * */ -void client_report_facets(struct client *cl, ZOOM_resultset rs) +static 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); @@ -506,27 +543,6 @@ static void ingest_raw_record(struct client *cl, ZOOM_record rec) 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); @@ -535,25 +551,29 @@ void client_search_response(struct client *cl) struct connection *co = cl->connection; ZOOM_connection link = connection_get_link(co); ZOOM_resultset resultset = cl->resultset; + struct session *se = client_get_session(cl); const char *error, *addinfo = 0; - + if (ZOOM_connection_error(link, &error, &addinfo)) { cl->hits = 0; + session_log(se, YLOG_WARN, "%s: Error %s (%s)", + client_get_id(cl), error, addinfo); client_set_state(cl, Client_Error); - yaz_log(YLOG_WARN, "Search error %s (%s): %s", - error, addinfo, client_get_id(cl)); } else { client_report_facets(cl, resultset); cl->record_offset = cl->startrecs; cl->hits = ZOOM_resultset_size(resultset); - yaz_log(YLOG_DEBUG, "client_search_response: hits " ODR_INT_PRINTF, cl->hits); + session_log(se, YLOG_LOG, "%s: hits: " ODR_INT_PRINTF, + client_get_id(cl), cl->hits); if (cl->suggestions) client_suggestions_destroy(cl); - cl->suggestions = client_suggestions_create(ZOOM_resultset_option_get(resultset, "suggestions")); + cl->suggestions = + client_suggestions_create(ZOOM_resultset_option_get( + resultset, "suggestions")); } } @@ -562,6 +582,11 @@ void client_got_records(struct client *cl) struct session *se = cl->session; 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); + client_lock(cl); if (reclist_get_num_records(se->reclist) > 0) { client_unlock(cl); @@ -579,17 +604,39 @@ 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))) + struct session *se = client_get_session(cl); + xmlDoc *xdoc; + int offset = cl->record_offset + 1; /* 0 versus 1 numbered offsets */ + + xdoc = client_get_xdoc(cl, offset); + if (xdoc) { - cl->record_offset++; - if (cl->session == 0) { - /* no operation */ + if (cl->session) + { + NMEM nmem = nmem_create(); + int rc = ingest_xml_record(cl, xdoc, offset, nmem, 1); + if (rc == -1) + { + session_log(se, YLOG_WARN, + "%s: #%d: failed to ingest xdoc", + client_get_id(cl), offset); + cl->ingest_failures++; + } + else if (rc == -2) + cl->filtered++; + nmem_destroy(nmem); } + } + else if ((rec = ZOOM_resultset_record_immediate(resultset, + 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); + session_log(se, YLOG_WARN, "Record error %s (%s): %s #%d", + msg, addinfo, client_get_id(cl), offset); + cl->record_failures++; } else { @@ -597,34 +644,53 @@ static void client_record_ingest(struct client *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"); + session_log(se, 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)); + { + const char *rec_syn = ZOOM_record_get(rec, "syntax", NULL); + session_log(se, YLOG_WARN, "%s: #%d: ZOOM_record_get failed", + client_get_id(cl), offset); + session_log(se, YLOG_LOG, "pz:nativesyntax=%s . " + "ZOOM record type=%s . Actual record syntax=%s", + s ? s : "null", type, + rec_syn ? rec_syn : "null"); + cl->ingest_failures++; + } else { /* OK = 0, -1 = failure, -2 = Filtered */ - int rc = ingest_record(cl, xmlrec, cl->record_offset, nmem); + int rc = ingest_record(cl, xmlrec, offset, nmem); if (rc == -1) - yaz_log(YLOG_WARN, "Failed to ingest from %s", client_get_id(cl)); - if (rc == -2) - cl->filtered += 1; + { + const char *rec_syn = ZOOM_record_get(rec, "syntax", NULL); + session_log(se, YLOG_WARN, + "%s: #%d: failed to ingest record", + client_get_id(cl), offset); + session_log(se, YLOG_LOG, "pz:nativesyntax=%s . " + "ZOOM record type=%s . Actual record syntax=%s", + s ? s : "null", type, + rec_syn ? rec_syn : "null"); + cl->ingest_failures++; + } + else if (rc == -2) + cl->filtered++; } nmem_destroy(nmem); } } else { - yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d", - offset); + session_log(se, YLOG_WARN, "Got NULL record from %s #%d", + client_get_id(cl), offset); } + cl->record_offset++; } -void client_record_response(struct client *cl) +void client_record_response(struct client *cl, int *got_records) { struct connection *co = cl->connection; ZOOM_connection link = connection_get_link(co); @@ -633,9 +699,10 @@ void client_record_response(struct client *cl) if (ZOOM_connection_error(link, &error, &addinfo)) { + struct session *se = client_get_session(cl); + session_log(se, YLOG_WARN, "%s: Error %s (%s)", + client_get_id(cl), error, addinfo); client_set_state(cl, Client_Error); - yaz_log(YLOG_WARN, "Search error %s (%s): %s", - error, addinfo, client_get_id(cl)); } else { @@ -657,19 +724,21 @@ void client_record_response(struct client *cl) else { client_record_ingest(cl); + *got_records = 1; } } } -void client_reingest(struct client *cl) +int client_reingest(struct client *cl) { int i = cl->startrecs; int to = cl->record_offset; - cl->filtered = 0; + cl->record_failures = cl->ingest_failures = cl->filtered = 0; cl->record_offset = i; for (; i < to; i++) client_record_ingest(cl); + return 0; } static void client_set_facets_request(struct client *cl, ZOOM_connection link) @@ -677,7 +746,7 @@ 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) @@ -695,7 +764,7 @@ static void client_set_facets_request(struct client *cl, ZOOM_connection link) wrbuf_puts(w, ","); } } - yaz_log(YLOG_LOG, "using facets str: %s", wrbuf_cstr(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); @@ -721,7 +790,7 @@ static const char *get_strategy_plus_sort(struct client *l, const char *field) 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, ':'); @@ -740,11 +809,90 @@ static const char *get_strategy_plus_sort(struct client *l, const char *field) return strategy_plus_sort; } -void client_start_search(struct client *cl) +void client_update_show_stat(struct client *cl, int cmd) +{ + if (cmd == 0) + cl->show_stat_no = 0; + else if (cmd == 1) + cl->show_stat_no++; +} + +int client_fetch_more(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); + const char *str; + int extend_recs = 0; + int number = cl->hits - cl->record_offset; + + str = session_setting_oneval(sdb, PZ_EXTENDRECS); + if (!str || !*str) + return 0; + + extend_recs = atoi(str); + + yaz_log(YLOG_LOG, "cl=%s show_stat_no=%d got=%d", + client_get_id(cl), cl->show_stat_no, cl->record_offset); + if (cl->show_stat_no < cl->record_offset) + return 0; + yaz_log(YLOG_LOG, "cl=%s Trying to fetch more", client_get_id(cl)); + + if (number > extend_recs) + number = extend_recs; + if (number > 0) + { + ZOOM_resultset set = cl->resultset; + struct connection *co = client_get_connection(cl); + + str = session_setting_oneval(sdb, PZ_REQUESTSYNTAX); + ZOOM_resultset_option_set(set, "preferredRecordSyntax", str); + str = session_setting_oneval(sdb, PZ_ELEMENTS); + if (str && *str) + ZOOM_resultset_option_set(set, "elementSetName", str); + + ZOOM_resultset_records(set, 0, cl->record_offset, number); + client_set_state(cl, Client_Working); + connection_continue(co); + return 1; + } + else + { + yaz_log(YLOG_LOG, "cl=%s. OK no more in total set", client_get_id(cl)); + } + return 0; +} + +int client_parse_init(struct client *cl, int same_search) +{ + cl->same_search = same_search; + return 0; +} + +/* + * TODO consider how to extend the range + * */ +int client_parse_range(struct client *cl, const char *startrecs, + const char *maxrecs) +{ + if (maxrecs && atoi(maxrecs) != cl->maxrecs) + { + cl->same_search = 0; + cl->maxrecs = atoi(maxrecs); + } + + if (startrecs && atoi(startrecs) != cl->startrecs) + { + cl->same_search = 0; + cl->startrecs = atoi(startrecs); + } + + return 0; +} + +int client_start_search(struct client *cl) +{ + struct session_database *sdb = client_get_database(cl); + struct connection *co = 0; + ZOOM_connection link = 0; struct session *se = client_get_session(cl); ZOOM_resultset rs; const char *opt_piggyback = session_setting_oneval(sdb, PZ_PIGGYBACK); @@ -757,16 +905,55 @@ void client_start_search(struct client *cl) 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; + const char *opt_timeout = session_setting_oneval(sdb, PZ_TIMEOUT); + ZOOM_query query; char maxrecs_str[24], startrecs_str[24], present_chunk_str[24]; + struct timeval tval; int present_chunk = 20; // Default chunk size + int rc_prep_connection; + int operation_timeout = se->service->z3950_operation_timeout; + + cl->diagnostic = 0; + cl->record_failures = cl->ingest_failures = cl->filtered = 0; + + yaz_gettimeofday(&tval); + tval.tv_sec += 5; + + if (opt_timeout && *opt_timeout) + operation_timeout = atoi(opt_timeout); + 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); } - + rc_prep_connection = + client_prep_connection(cl, operation_timeout, + se->service->z3950_session_timeout, + se->service->server->iochan_man, + &tval); + /* Nothing has changed and we already have a result */ + if (cl->same_search == 1 && rc_prep_connection == 2) + { + session_log(se, YLOG_LOG, "%s: reuse result", client_get_id(cl)); + client_report_facets(cl, cl->resultset); + return client_reingest(cl); + } + else if (!rc_prep_connection) + { + session_log(se, YLOG_LOG, "%s: postponing search: No connection", + client_get_id(cl)); + client_set_state_nb(cl, Client_Working); + return -1; + } + co = client_get_connection(cl); + assert(cl); + link = connection_get_link(co); assert(link); - cl->diagnostic = 0; + session_log(se, YLOG_LOG, "%s: new search", client_get_id(cl)); + + client_destroy_xdoc(cl); + client_init_xdoc(cl); if (extra_args && *extra_args) ZOOM_connection_option_set(link, "extraArgs", extra_args); @@ -803,11 +990,13 @@ void client_start_search(struct client *cl) /* 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", opt_present_chunk); + ZOOM_connection_option_set(link, "presentChunk", present_chunk_str); + yaz_log(YLOG_DEBUG, "Present chunk set to %s", present_chunk_str); } - else + 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); @@ -815,63 +1004,38 @@ void client_start_search(struct client *cl) /* facets definition is in PQF */ client_set_facets_request(cl, link); - q = ZOOM_query_create(); + query = ZOOM_query_create(); if (cl->cqlquery) { - yaz_log(YLOG_LOG, "Search %s CQL: %s", client_get_id(cl), - cl->cqlquery); - ZOOM_query_cql(q, cl->cqlquery); + session_log(se, YLOG_LOG, "%s: Search CQL: %s", client_get_id(cl), + cl->cqlquery); + ZOOM_query_cql(query, cl->cqlquery); if (*opt_sort) - ZOOM_query_sortby(q, opt_sort); + ZOOM_query_sortby(query, opt_sort); } else { - yaz_log(YLOG_LOG, "Search %s PQF: %s", client_get_id(cl), cl->pquery); - - ZOOM_query_prefix(q, cl->pquery); + session_log(se, YLOG_LOG, "%s: Search PQF: %s", client_get_id(cl), + cl->pquery); + ZOOM_query_prefix(query, 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; - } - } + if (cl->sort_strategy && cl->sort_criteria) { + yaz_log(YLOG_LOG, "Client %s: " + "Set ZOOM sort strategy and criteria: %s %s", + client_get_id(cl), cl->sort_strategy, cl->sort_criteria); + ZOOM_query_sortby2(query, cl->sort_strategy, cl->sort_criteria); } + + yaz_log(YLOG_DEBUG,"Client %s: Starting search", client_get_id(cl)); client_set_state(cl, Client_Working); cl->hits = 0; cl->record_offset = 0; - rs = ZOOM_connection_search(link, q); - ZOOM_query_destroy(q); + rs = ZOOM_connection_search(link, query); + ZOOM_query_destroy(query); ZOOM_resultset_destroy(cl->resultset); cl->resultset = rs; connection_continue(co); + return 0; } struct client *client_create(const char *id) @@ -882,6 +1046,7 @@ struct client *client_create(const char *id) cl->pquery = 0; cl->cqlquery = 0; cl->addinfo = 0; + cl->message = 0; cl->database = 0; cl->connection = 0; cl->session = 0; @@ -898,10 +1063,14 @@ struct client *client_create(const char *id) cl->preferred = 0; cl->ref_count = 1; cl->facet_limits = 0; + cl->sort_strategy = 0; + cl->sort_criteria = 0; assert(id); cl->id = xstrdup(id); + client_init_xdoc(cl); client_use(1); - + + yaz_log(YLOG_DEBUG, "client_create c=%p %s", cl, id); return cl; } @@ -936,10 +1105,15 @@ int client_destroy(struct client *c) c->cqlquery = 0; xfree(c->addinfo); c->addinfo = 0; + xfree(c->message); + c->message = 0; xfree(c->id); + xfree(c->sort_strategy); + xfree(c->sort_criteria); assert(!c->connection); facet_limits_destroy(c->facet_limits); + client_destroy_xdoc(c); if (c->resultset) { ZOOM_resultset_destroy(c->resultset); @@ -965,7 +1139,9 @@ void client_set_connection(struct client *cl, struct connection *con) } else { + client_lock(cl); cl->connection = con; + client_unlock(cl); client_destroy(cl); } } @@ -977,9 +1153,28 @@ void client_disconnect(struct client *cl) client_set_connection(cl, 0); } +void client_stop(struct client *cl) +{ + client_lock(cl); + if (cl->state == Client_Working || cl->state == Client_Connecting) + { + yaz_log(YLOG_LOG, "client_stop: %s release", client_get_id(cl)); + if (cl->connection) + { + connection_release2(cl->connection); + assert(cl->ref_count > 1); + cl->ref_count--; + cl->connection = 0; + } + cl->state = Client_Disconnected; + } + else + yaz_log(YLOG_LOG, "client_stop: %s ignore", client_get_id(cl)); + client_unlock(cl); +} // Initialize CCL map for a target -static CCL_bibset prepare_cclmap(struct client *cl) +static CCL_bibset prepare_cclmap(struct client *cl, CCL_bibset base_bibset) { struct session_database *sdb = client_get_database(cl); struct setting *s; @@ -987,18 +1182,43 @@ static CCL_bibset prepare_cclmap(struct client *cl) if (!sdb->settings) return 0; - res = ccl_qual_mk(); + if (base_bibset) + res = ccl_qual_dup(base_bibset); + else + res = ccl_qual_mk(); for (s = sdb->settings[PZ_CCLMAP]; s; s = s->next) { + const char *addinfo = 0; char *p = strchr(s->name + 3, ':'); if (!p) { - yaz_log(YLOG_WARN, "Malformed cclmap name: %s", s->name); + WRBUF w = wrbuf_alloc(); + wrbuf_printf(w, "Malformed cclmap. name=%s", s->name); + yaz_log(YLOG_WARN, "%s: %s", client_get_id(cl), wrbuf_cstr(w)); + client_set_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG, + ZOOM_diag_str(ZOOM_ERROR_CCL_CONFIG), + wrbuf_cstr(w)); + client_set_state_nb(cl, Client_Error); ccl_qual_rm(&res); + wrbuf_destroy(w); return 0; } p++; - ccl_qual_fitem(res, s->value, p); + if (ccl_qual_fitem2(res, s->value, p, &addinfo)) + { + WRBUF w = wrbuf_alloc(); + + wrbuf_printf(w, "Malformed cclmap. name=%s: value=%s (%s)", + s->name, p, addinfo); + yaz_log(YLOG_WARN, "%s: %s", client_get_id(cl), wrbuf_cstr(w)); + client_set_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG, + ZOOM_diag_str(ZOOM_ERROR_CCL_CONFIG), + wrbuf_cstr(w)); + client_set_state_nb(cl, Client_Error); + ccl_qual_rm(&res); + wrbuf_destroy(w); + return 0; + } } return res; } @@ -1018,7 +1238,7 @@ static char *make_cqlquery(struct client *cl, Z_RPNQuery *zquery) else { r = xstrdup(wrbuf_cstr(wrb)); - } + } wrbuf_destroy(wrb); cql_transform_close(cqlt); return r; @@ -1032,7 +1252,7 @@ static char *make_solrquery(struct client *cl, Z_RPNQuery *zquery) 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); @@ -1056,40 +1276,79 @@ const char *client_get_facet_limit_local(struct client *cl, 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)) + if (p && !strcmp(p + 1, name) && s->value) { - const char *cp = s->value + 6; - while (*cp == ' ') - cp++; - - nmem_strsplit_escape2(nmem, "|", value, values, - num, 1, '\\', 1); - (*l)++; - return *cp ? cp : name; + int j, cnum; + char **cvalues; + nmem_strsplit_escape2(nmem, ",", s->value, &cvalues, + &cnum, 1, '\\', 1); + for (j = 0; j < cnum; j++) + { + const char *cvalue = cvalues[j]; + while (*cvalue == ' ') + cvalue++; + if (!strncmp(cvalue, "local:", 6)) + { + const char *cp = cvalue + 6; + while (*cp == ' ') + cp++; + nmem_strsplit_escape2(nmem, "|", value, values, + num, 1, '\\', 1); + (*l)++; + return *cp ? cp : name; + } + } } } } return 0; } +static void ccl_quote_map_term(CCL_bibset ccl_map, WRBUF w, + const char *term) +{ + int quote_it = 0; + const char *cp; + for (cp = term; *cp; cp++) + if ((*cp >= '0' && *cp <= '9') || strchr(" +-", *cp)) + ; + else + quote_it = 1; + if (!quote_it) + wrbuf_puts(w, term); + else + { + wrbuf_putc(w, '\"'); + for (cp = term; *cp; cp++) + { + if (strchr( "\\\"", *cp)) + wrbuf_putc(w, '\\'); + wrbuf_putc(w, *cp); + } + wrbuf_putc(w, '\"'); + } +} + static int apply_limit(struct session_database *sdb, facet_limits_t facet_limits, - WRBUF w_pqf, WRBUF w_ccl) + WRBUF w_pqf, CCL_bibset ccl_map, + struct conf_service *service) { 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); + /* name="pz:limitmap:author" value="rpn:@attr 1=4|local:other" */ for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next) { const char *p = strchr(s->name + 3, ':'); @@ -1097,80 +1356,116 @@ static int apply_limit(struct session_database *sdb, { char **values = 0; int i, num = 0; + char **cvalues = 0; + int j, cnum = 0; nmem_strsplit_escape2(nmem_tmp, "|", value, &values, &num, 1, '\\', 1); - if (!strncmp(s->value, "rpn:", 4)) - { - const char *pqf = s->value + 4; + nmem_strsplit_escape2(nmem_tmp, ",", s->value, &cvalues, + &cnum, 1, '\\', 1); - wrbuf_puts(w_pqf, "@and "); - wrbuf_puts(w_pqf, pqf); - wrbuf_puts(w_pqf, " "); - for (i = 0; i < num; i++) + for (j = 0; ret == 0 && j < cnum; j++) + { + const char *cvalue = cvalues[j]; + while (*cvalue == ' ') + cvalue++; + if (!strncmp(cvalue, "rpn:", 4)) { - if (i < num - 1) - wrbuf_puts(w_pqf, "@or "); - yaz_encode_pqf_term(w_pqf, values[i], - strlen(values[i])); + const char *pqf = cvalue + 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++) + else if (!strncmp(cvalue, "ccl:", 4)) + { + const char *ccl = cvalue + 4; + WRBUF ccl_w = wrbuf_alloc(); + for (i = 0; i < num; i++) + { + int cerror, cpos; + struct ccl_rpn_node *cn; + wrbuf_rewind(ccl_w); + wrbuf_puts(ccl_w, ccl); + wrbuf_putc(ccl_w, '='); + ccl_quote_map_term(ccl_map, ccl_w, values[i]); + cn = ccl_find_str(ccl_map, wrbuf_cstr(ccl_w), + &cerror, &cpos); + if (cn) + { + if (i == 0) + wrbuf_printf(w_pqf, "@and "); + + /* or multiple values.. could be bad if last + CCL parse fails, but this is unlikely to + happen */ + if (i < num - 1) + wrbuf_printf(w_pqf, "@or "); + ccl_pquery(w_pqf, cn); + ccl_rpn_delete(cn); + } + } + wrbuf_destroy(ccl_w); + } + else if (!strncmp(cvalue, "local:", 6)) { + /* no operation */ + } + else { - 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, "\""); + yaz_log(YLOG_WARN, "Target %s: Bad limitmap '%s'", + sdb->database->id, cvalue); + ret = -1; /* bad limitmap */ } - 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; } } - nmem_reset(nmem_tmp); if (!s) { - yaz_log(YLOG_WARN, "Target %s: limit %s used, but no limitmap defined", - (sdb->database ? sdb->database->id : ""), name); + int i; + for (i = 0; i < service->num_metadata; i++) + { + struct conf_metadata *md = service->metadata + i; + if (!strcmp(md->name, name) && md->limitcluster) + { + yaz_log(YLOG_LOG, "limitcluster in use for %s", + md->name); + break; + } + } + if (i == service->num_metadata) + { + 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 -// return 0 if query is OK but different from before -// return 1 if query is OK but same as before +// client variable same_search is set as below as well as returned: +// 0 if query is OK but different from before +// 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) + facet_limits_t facet_limits, const char **error_msg) { struct session *se = client_get_session(cl); + struct conf_service *service = se->service; 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); + CCL_bibset ccl_map = prepare_cclmap(cl, service->ccl_bibset); 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); @@ -1180,19 +1475,7 @@ int client_parse_query(struct client *cl, const char *query, Z_RPNQuery *zquery; if (!ccl_map) - return -1; - - 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); - } + return -3; w_ccl = wrbuf_alloc(); wrbuf_puts(w_ccl, query); @@ -1204,21 +1487,27 @@ int client_parse_query(struct client *cl, const char *query, wrbuf_puts(w_pqf, " "); } - if (apply_limit(sdb, facet_limits, w_pqf, w_ccl)) + if (apply_limit(sdb, facet_limits, w_pqf, ccl_map, service)) + { + ccl_qual_rm(&ccl_map); 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)); + yaz_log(YLOG_LOG, "Client %s: CCL query: %s limit: %s", + client_get_id(cl), wrbuf_cstr(w_ccl), wrbuf_cstr(w_pqf)); cn = ccl_find_str(ccl_map, wrbuf_cstr(w_ccl), &cerror, &cpos); ccl_qual_rm(&ccl_map); if (!cn) { + if (error_msg) + *error_msg = ccl_err_msg(cerror); 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)); + session_log(se, YLOG_WARN, "Client %s: Failed to parse CCL query '%s'", + client_get_id(cl), + wrbuf_cstr(w_ccl)); wrbuf_destroy(w_ccl); wrbuf_destroy(w_pqf); return -1; @@ -1246,31 +1535,40 @@ int client_parse_query(struct client *cl, const char *query, } } + /* Compares query and limit with old one. If different we need to research */ if (!cl->pquery || strcmp(cl->pquery, wrbuf_cstr(w_pqf))) { + if (cl->pquery) + session_log(se, YLOG_LOG, "Client %s: " + "Re-search due query/limit change: %s to %s", + client_get_id(cl), cl->pquery, wrbuf_cstr(w_pqf)); xfree(cl->pquery); cl->pquery = xstrdup(wrbuf_cstr(w_pqf)); + // return value is no longer used. ret_value = 0; + // Need to (re)search + cl->same_search= 0; } wrbuf_destroy(w_pqf); - + xfree(cl->cqlquery); cl->cqlquery = 0; - odr_out = odr_createmem(ODR_ENCODE); + 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", + session_log(se, YLOG_WARN, "Invalid PQF query for Client %s: %s", client_get_id(cl), cl->pquery); ret_value = -1; + *error_msg = "Invalid PQF after CCL to PQF conversion"; } else { - session_log(se, YLOG_LOG, "PQF for %s: %s", + session_log(se, YLOG_LOG, "PQF for Client %s: %s", client_get_id(cl), cl->pquery); - + /* Support for PQF on SRU targets. */ if (strcmp(query_syntax, "pqf") != 0 && *sru) { @@ -1279,7 +1577,13 @@ int client_parse_query(struct client *cl, const char *query, else cl->cqlquery = make_cqlquery(cl, zquery); if (!cl->cqlquery) + { + *error_msg = "Cannot convert PQF to Solr/CQL"; ret_value = -1; + } + else + session_log(se, YLOG_LOG, "Client %s native query: %s (%s)", + client_get_id(cl), cl->cqlquery, sru); } } odr_destroy(odr_out); @@ -1288,13 +1592,78 @@ int client_parse_query(struct client *cl, const char *query, if (!se->relevance) { // Initialize relevance structure with query terms - se->relevance = relevance_create_ccl( - se->service->charsets, se->nmem, cn); + se->relevance = relevance_create_ccl(se->service->charsets, cn, + se->service->rank_cluster, + se->service->rank_follow, + se->service->rank_lead, + se->service->rank_length); } ccl_rpn_delete(cn); return ret_value; } +int client_parse_sort(struct client *cl, struct reclist_sortparms *sp) +{ + if (sp) + { + const char *sort_strategy_and_spec = + get_strategy_plus_sort(cl, sp->name); + int increasing = sp->increasing; + if (!strcmp(sp->name, "relevance")) + increasing = 1; + if (sort_strategy_and_spec && strlen(sort_strategy_and_spec) < 40) + { + char strategy[50], *p; + strcpy(strategy, sort_strategy_and_spec); + p = strchr(strategy, ':'); + if (p) + { + // Split the string in two + *p++ = 0; + while (*p == ' ') + p++; + if (increasing) + strcat(p, " <"); + else + strcat(p, " >"); + yaz_log(YLOG_LOG, "Client %s: " + "applying sorting %s %s", client_get_id(cl), + strategy, p); + if (!cl->sort_strategy || strcmp(cl->sort_strategy, strategy)) + cl->same_search = 0; + if (!cl->sort_criteria || strcmp(cl->sort_criteria, p)) + cl->same_search = 0; + if (cl->same_search == 0) { + xfree(cl->sort_strategy); + cl->sort_strategy = xstrdup(strategy); + xfree(cl->sort_criteria); + cl->sort_criteria = xstrdup(p); + } + } + else { + yaz_log(YLOG_LOG, "Client %s: " + "Invalid sort strategy and spec found %s", + client_get_id(cl), sort_strategy_and_spec); + xfree(cl->sort_strategy); + cl->sort_strategy = 0; + xfree(cl->sort_criteria); + cl->sort_criteria = 0; + } + } + else + { + yaz_log(YLOG_DEBUG, "Client %s: " + "No sort strategy and spec found.", client_get_id(cl)); + xfree(cl->sort_strategy); + cl->sort_strategy = 0; + xfree(cl->sort_criteria); + cl->sort_criteria = 0; + } + + } + return !cl->same_search; +} + void client_set_session(struct client *cl, struct session *se) { cl->session = se; @@ -1327,37 +1696,47 @@ Odr_int client_get_hits(struct client *cl) Odr_int client_get_approximation(struct client *cl) { - int records = cl->record_offset + cl->filtered; - if (records > 0) { - Odr_int approx = (cl->hits * cl->record_offset) / records; - yaz_log(YLOG_LOG, "%s: Approx: %lld * %d / %d = %lld ", client_get_id(cl), cl->hits, cl->record_offset, records, approx); + 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) +int client_get_num_records(struct client *cl, int *filtered, int *ingest, + int *failed) { + if (filtered) + *filtered = cl->filtered; + if (ingest) + *ingest = cl->ingest_failures; + if (failed) + *failed = cl->record_failures; 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) + const char *message, const char *addinfo) { cl->diagnostic = diagnostic; + xfree(cl->message); + cl->message = xstrdup(message); xfree(cl->addinfo); cl->addinfo = 0; if (addinfo) cl->addinfo = xstrdup(addinfo); } -int client_get_diagnostic(struct client *cl, const char **addinfo) +int client_get_diagnostic(struct client *cl, const char **message, + const char **addinfo) { + if (message) + *message = cl->message; if (addinfo) *addinfo = cl->addinfo; return cl->diagnostic; @@ -1368,17 +1747,16 @@ 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"); + if (!suggestions) return ""; - } - if (suggestions->passthrough) { - yaz_log(YLOG_DEBUG, "Passthrough Suggestions: \n%s\n", suggestions->passthrough); + if (suggestions->passthrough) + { + yaz_log(YLOG_DEBUG, "Passthrough Suggestions: \n%s\n", + suggestions->passthrough); return suggestions->passthrough; } - if (suggestions->num == 0) { + if (suggestions->num == 0) return ""; - } /* for (idx = 0; idx < suggestions->num; idx++) { wrbuf_printf(wrbuf, "suggest[idx]); @@ -1420,11 +1798,11 @@ struct suggestions* client_suggestions_create(const char* suggestions_string) int i; NMEM nmem; struct suggestions *suggestions; - if (suggestions_string == 0) + if (suggestions_string == 0 || suggestions_string[0] == 0 ) return 0; nmem = nmem_create(); suggestions = nmem_malloc(nmem, sizeof(*suggestions)); - yaz_log(YLOG_DEBUG, "client target suggestions: %s", suggestions_string); + yaz_log(YLOG_DEBUG, "client target suggestions: %s.", suggestions_string); suggestions->nmem = nmem; suggestions->num = 0; @@ -1436,7 +1814,8 @@ struct suggestions* client_suggestions_create(const char* 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)); + suggestions->misspelled = (char **) + nmem_malloc(nmem, suggestions->num * sizeof(*suggestions->misspelled)); /* replace = with \0 .. for each item */ for (i = 0; i < suggestions->num; i++) {