/* This file is part of Pazpar2.
- Copyright (C) 2006-2010 Index Data
+ Copyright (C) 2006-2011 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
#include <config.h>
#endif
+#include <time.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#if HAVE_UNISTD_H
#include <unistd.h>
#endif
+#ifdef WIN32
+#include <windows.h>
+#endif
#include <signal.h>
#include <ctype.h>
#include <assert.h>
+#include <math.h>
#include <yaz/marcdisp.h>
#include <yaz/comstack.h>
#include <yaz/querytowrbuf.h>
#include <yaz/oid_db.h>
#include <yaz/snprintf.h>
+#include <yaz/gettimeofday.h>
#define USE_TIMING 0
#if USE_TIMING
#define MAX_CHUNK 15
+#define MAX(a,b) ((a)>(b)?(a):(b))
+
// Note: Some things in this structure will eventually move to configuration
struct parameters global_parameters =
{
0, // dump_records
- 0 // debug_mode
+ 0, // debug_mode
+ 0, // predictable sessions
};
struct client_list {
struct client_list *next;
};
+struct session_sorted_results {
+ const char *field;
+ int increasing;
+ struct session_sorted_results *next;
+};
+
+/* session counting (1) , disable client counting (0) */
+static YAZ_MUTEX g_session_mutex = 0;
+static int no_sessions = 0;
+static int no_session_total = 0;
+
+static int session_use(int delta)
+{
+ int sessions;
+ if (!g_session_mutex)
+ yaz_mutex_create(&g_session_mutex);
+ yaz_mutex_enter(g_session_mutex);
+ no_sessions += delta;
+ if (delta > 0)
+ no_session_total += delta;
+ sessions = no_sessions;
+ yaz_mutex_leave(g_session_mutex);
+ yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
+ return sessions;
+}
+
+int sessions_count(void)
+{
+ return session_use(0);
+}
+
+int session_count_total(void)
+{
+ int total = 0;
+ if (!g_session_mutex)
+ return 0;
+ yaz_mutex_enter(g_session_mutex);
+ total = no_session_total;
+ yaz_mutex_leave(g_session_mutex);
+ return total;
+}
+
static void log_xml_doc(xmlDoc *doc)
{
FILE *lf = yaz_log_file();
yaz_mutex_leave(s->session_mutex);
}
-// Recursively traverse query structure to extract terms.
-void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
+void add_facet(struct session *s, const char *type, const char *value, int count)
{
- char **words;
- int numwords;
+ struct conf_service *service = s->service;
+ pp2_charset_token_t prt;
+ const char *facet_component;
+ WRBUF facet_wrbuf = wrbuf_alloc();
+ WRBUF display_wrbuf = wrbuf_alloc();
int i;
+ const char *icu_chain_id = 0;
- switch (n->kind)
+ for (i = 0; i < service->num_metadata; i++)
+ if (!strcmp((service->metadata + i)->name, type))
+ icu_chain_id = (service->metadata + i)->facetrule;
+ if (!icu_chain_id)
+ icu_chain_id = "facet";
+ prt = pp2_charset_token_create(service->charsets, icu_chain_id);
+ if (!prt)
{
- case CCL_RPN_AND:
- case CCL_RPN_OR:
- case CCL_RPN_NOT:
- case CCL_RPN_PROX:
- pull_terms(nmem, n->u.p[0], termlist, num);
- pull_terms(nmem, n->u.p[1], termlist, num);
- break;
- case CCL_RPN_TERM:
- nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords);
- for (i = 0; i < numwords; i++)
- termlist[(*num)++] = words[i];
- break;
- default: // NOOP
- break;
- }
-}
-
-
-static void add_facet(struct session *s, const char *type, const char *value)
-{
- int i;
-
- if (!*value)
+ yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
+ icu_chain_id, type);
+ wrbuf_destroy(facet_wrbuf);
+ wrbuf_destroy(display_wrbuf);
return;
- for (i = 0; i < s->num_termlists; i++)
- if (!strcmp(s->termlists[i].name, type))
- break;
- if (i == s->num_termlists)
+ }
+ pp2_charset_token_first(prt, value, 0);
+ while ((facet_component = pp2_charset_token_next(prt)))
{
- if (i == SESSION_MAX_TERMLISTS)
+ const char *display_component;
+ if (*facet_component)
{
- yaz_log(YLOG_FATAL, "Too many termlists");
- return;
+ if (wrbuf_len(facet_wrbuf))
+ wrbuf_puts(facet_wrbuf, " ");
+ wrbuf_puts(facet_wrbuf, facet_component);
+ }
+ display_component = pp2_get_display(prt);
+ if (display_component)
+ {
+ if (wrbuf_len(display_wrbuf))
+ wrbuf_puts(display_wrbuf, " ");
+ wrbuf_puts(display_wrbuf, display_component);
}
-
- s->termlists[i].name = nmem_strdup(s->nmem, type);
- s->termlists[i].termlist
- = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
- s->num_termlists = i + 1;
}
- termlist_insert(s->termlists[i].termlist, value);
+ pp2_charset_token_destroy(prt);
+
+ if (wrbuf_len(facet_wrbuf))
+ {
+ int i;
+ for (i = 0; i < s->num_termlists; i++)
+ if (!strcmp(s->termlists[i].name, type))
+ break;
+ if (i == s->num_termlists)
+ {
+ if (i == SESSION_MAX_TERMLISTS)
+ {
+ session_log(s, YLOG_FATAL, "Too many termlists");
+ wrbuf_destroy(facet_wrbuf);
+ wrbuf_destroy(display_wrbuf);
+ return;
+ }
+
+ s->termlists[i].name = nmem_strdup(s->nmem, type);
+ s->termlists[i].termlist
+ = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
+ s->num_termlists = i + 1;
+ }
+
+#if 0
+ session_log(s, YLOG_DEBUG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
+#endif
+ termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
+ wrbuf_cstr(facet_wrbuf), count);
+ }
+ wrbuf_destroy(facet_wrbuf);
+ wrbuf_destroy(display_wrbuf);
}
-static xmlDoc *record_to_xml(struct session_database *sdb, const char *rec)
+static xmlDoc *record_to_xml(struct session *se,
+ struct session_database *sdb, const char *rec)
{
struct database *db = sdb->database;
xmlDoc *rdoc = 0;
if (!rdoc)
{
- yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
- db->url);
+ session_log(se, YLOG_FATAL, "Non-wellformed XML received from %s",
+ db->id);
return 0;
}
if (global_parameters.dump_records)
{
- yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url);
+ session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
log_xml_doc(rdoc);
}
}
}
-static xmlDoc *normalize_record(struct session_database *sdb,
+static xmlDoc *normalize_record(struct session *se,
+ struct session_database *sdb,
struct conf_service *service,
const char *rec, NMEM nmem)
{
- xmlDoc *rdoc = record_to_xml(sdb, rec);
+ xmlDoc *rdoc = record_to_xml(se, sdb, rec);
if (rdoc)
{
if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
{
- yaz_log(YLOG_WARN, "Normalize failed from %s", sdb->database->url);
+ session_log(se, YLOG_WARN, "Normalize failed from %s",
+ sdb->database->id);
}
else
{
if (global_parameters.dump_records)
{
- yaz_log(YLOG_LOG, "Normalized record from %s",
- sdb->database->url);
+ session_log(se, YLOG_LOG, "Normalized record from %s",
+ sdb->database->id);
log_xml_doc(rdoc);
}
}
{
const char *s;
- if (!sdb->settings)
- {
- yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
- return -1;
- }
- if ((s = session_setting_oneval(sdb, PZ_XSLT)))
+ if (sdb->settings && sdb->settings[PZ_XSLT] && !sdb->map &&
+ (s = session_setting_oneval(sdb, PZ_XSLT)))
{
char auto_stylesheet[256];
}
else
{
- yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet");
+ session_log(se, YLOG_WARN,
+ "No pz:requestsyntax for auto stylesheet");
}
}
sdb->map = normalize_cache_get(se->normalize_cache,
return 0;
}
-// This analyzes settings and recomputes any supporting data structures
-// if necessary.
-static int prepare_session_database(struct session *se,
- struct session_database *sdb)
-{
- if (!sdb->settings)
- {
- yaz_log(YLOG_WARN,
- "No settings associated with %s", sdb->database->url);
- return -1;
- }
- if (sdb->settings[PZ_XSLT] && !sdb->map)
- {
- if (prepare_map(se, sdb) < 0)
- return -1;
- }
- return 0;
-}
-
// called if watch should be removed because http_channel is to be destroyed
static void session_watch_cancel(void *data, struct http_channel *c,
void *data2)
ret = 0;
}
session_leave(s);
- return 0;
+ return ret;
}
void session_alert_watch(struct session *s, int what)
session_watchfun fun;
http_remove_observer(s->watchlist[what].obs);
- fun = s->watchlist[what].fun;
+ fun = s->watchlist[what].fun;
data = s->watchlist[what].data;
/* reset watch before fun is invoked - in case fun wants to set
s->watchlist[what].obs = 0;
session_leave(s);
+ session_log(s, YLOG_DEBUG,
+ "Alert Watch: %d calling function: %p", what, fun);
fun(data);
}
else
}
//callback for grep_databases
-static void select_targets_callback(void *context, struct session_database *db)
+static void select_targets_callback(struct session *se,
+ struct session_database *db)
{
- struct session *se = (struct session*) context;
- struct client *cl = client_create();
+ struct client *cl;
struct client_list *l;
- client_set_database(cl, db);
- client_set_session(cl, se);
+ for (l = se->clients_cached; l; l = l->next)
+ if (client_get_database(l->client) == db)
+ break;
+
+ if (l)
+ cl = l->client;
+ else
+ {
+ cl = client_create(db->database->id);
+ client_set_database(cl, db);
+ client_set_session(cl, se);
+
+ l = xmalloc(sizeof(*l));
+ l->client = cl;
+ l->next = se->clients_cached;
+ se->clients_cached = l;
+ }
+
l = xmalloc(sizeof(*l));
l->client = cl;
- l->next = se->clients;
- se->clients = l;
+ l->next = se->clients_active;
+ se->clients_active = l;
}
-static void session_remove_clients(struct session *se)
+static void session_reset_active_clients(struct session *se,
+ struct client_list *new_list)
{
struct client_list *l;
session_enter(se);
- l = se->clients;
- se->clients = 0;
+ l = se->clients_active;
+ se->clients_active = new_list;
+ session_leave(se);
+
+ while (l)
+ {
+ struct client_list *l_next = l->next;
+ xfree(l);
+ l = l_next;
+ }
+}
+
+static void session_remove_cached_clients(struct session *se)
+{
+ struct client_list *l;
+
+ session_reset_active_clients(se, 0);
+
+ session_enter(se);
+ l = se->clients_cached;
+ se->clients_cached = 0;
session_leave(se);
while (l)
struct client_list *l;
int res = 0;
- for (l = s->clients; l; l = l->next)
+ for (l = s->clients_active; l; l = l->next)
if (client_is_active(l->client))
res++;
return res;
}
+int session_is_preferred_clients_ready(struct session *s)
+{
+ struct client_list *l;
+ int res = 0;
+
+ for (l = s->clients_active; l; l = l->next)
+ if (client_is_active_preferred(l->client))
+ res++;
+ session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
+ return res == 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;
+ }
+ }
+ return strategy_plus_sort;
+}
+
+void session_sort(struct session *se, const char *field, int increasing)
+{
+ struct session_sorted_results *sr;
+ struct client_list *l;
+
+ session_enter(se);
+
+ /* see if we already have sorted for this critieria */
+ for (sr = se->sorted_results; sr; sr = sr->next)
+ {
+ if (!strcmp(field, sr->field) && increasing == sr->increasing)
+ break;
+ }
+ if (sr)
+ {
+ yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched",
+ field, increasing);
+ session_leave(se);
+ return;
+ }
+ yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch",
+ field, increasing);
+ sr = nmem_malloc(se->nmem, sizeof(*sr));
+ sr->field = nmem_strdup(se->nmem, field);
+ sr->increasing = increasing;
+ sr->next = se->sorted_results;
+ se->sorted_results = sr;
+
+ for (l = se->clients_active; l; l = l->next)
+ {
+ struct client *cl = l->client;
+ const char *strategy_plus_sort = get_strategy_plus_sort(cl, field);
+ if (strategy_plus_sort)
+ {
+ struct timeval tval;
+ if (client_prep_connection(cl, se->service->z3950_operation_timeout,
+ se->service->z3950_session_timeout,
+ se->service->server->iochan_man,
+ &tval))
+ client_start_search(cl, strategy_plus_sort, increasing);
+ }
+ }
+ session_leave(se);
+}
-enum pazpar2_error_code search(struct session *se,
- const char *query,
- const char *startrecs, const char *maxrecs,
- const char *filter,
- const char **addinfo)
+enum pazpar2_error_code session_search(struct session *se,
+ const char *query,
+ const char *startrecs,
+ const char *maxrecs,
+ const char *filter,
+ const char *limit,
+ const char **addinfo,
+ const char *sort_field, int increasing)
{
int live_channels = 0;
int no_working = 0;
int no_failed = 0;
- struct client_list *l;
+ struct client_list *l, *l0;
+ struct timeval tval;
+ facet_limits_t facet_limits;
- yaz_log(YLOG_DEBUG, "Search");
+ session_log(se, YLOG_DEBUG, "Search");
*addinfo = 0;
- session_remove_clients(se);
+ if (se->settings_modified)
+ session_remove_cached_clients(se);
+ else
+ session_reset_active_clients(se, 0);
session_enter(se);
reclist_destroy(se->reclist);
se->reclist = 0;
+ se->settings_modified = 0;
relevance_destroy(&se->relevance);
nmem_reset(se->nmem);
- se->total_records = se->total_hits = se->total_merged = 0;
+ se->total_records = se->total_merged = 0;
se->num_termlists = 0;
+
+ /* reset list of sorted results and clear to relevance search */
+ se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
+ se->sorted_results->field = nmem_strdup(se->nmem, sort_field);
+ se->sorted_results->increasing = increasing;
+ se->sorted_results->next = 0;
+
live_channels = select_targets(se, filter);
if (!live_channels)
{
}
se->reclist = reclist_create(se->nmem);
- for (l = se->clients; l; l = l->next)
+ yaz_gettimeofday(&tval);
+
+ tval.tv_sec += 5;
+
+ facet_limits = facet_limits_create(limit);
+ if (!facet_limits)
+ {
+ *addinfo = "limit";
+ session_leave(se);
+ return PAZPAR2_MALFORMED_PARAMETER_VALUE;
+ }
+
+ l0 = se->clients_active;
+ se->clients_active = 0;
+ session_leave(se);
+
+ for (l = l0; l; l = l->next)
{
+ int parse_ret;
struct client *cl = l->client;
+ const char *strategy_plus_sort = get_strategy_plus_sort(cl, sort_field);
+
+ if (prepare_map(se, client_get_database(cl)) < 0)
+ continue;
- if (maxrecs)
- client_set_maxrecs(cl, atoi(maxrecs));
- if (startrecs)
- client_set_startrecs(cl, atoi(startrecs));
- if (prepare_session_database(se, client_get_database(cl)) < 0)
- ;
- else if (client_parse_query(cl, query) < 0)
+ parse_ret = client_parse_query(cl, query, facet_limits, startrecs,
+ maxrecs);
+ if (parse_ret < 0)
no_failed++;
+ else if (parse_ret == 0)
+ {
+ session_log(se, YLOG_LOG, "client NEW %s", client_get_id(cl));
+ no_working++;
+ if (client_prep_connection(cl, se->service->z3950_operation_timeout,
+ se->service->z3950_session_timeout,
+ se->service->server->iochan_man,
+ &tval))
+ client_start_search(cl, strategy_plus_sort, increasing);
+ }
else
{
+ session_log(se, YLOG_LOG, "client REUSE %s", client_get_id(cl));
no_working++;
if (client_prep_connection(cl, se->service->z3950_operation_timeout,
se->service->z3950_session_timeout,
- se->service->server->iochan_man))
- client_start_search(cl);
+ se->service->server->iochan_man,
+ &tval))
+ {
+ client_reingest(cl);
+ }
}
}
- session_leave(se);
+ facet_limits_destroy(facet_limits);
+ session_reset_active_clients(se, l0);
+
if (no_working == 0)
{
if (no_failed > 0)
else
return PAZPAR2_NO_TARGETS;
}
+ yaz_log(YLOG_LOG, "session_start_search done");
return PAZPAR2_NO_ERROR;
}
{
struct database *db = new_database(id, se->session_nmem);
- resolve_database(se->service, db);
-
session_init_databases_fun((void*) se, db);
// New sdb is head of se->databases list
struct session_database *sdb;
for (sdb = se->databases; sdb; sdb = sdb->next)
- if (!strcmp(sdb->database->url, id))
+ if (!strcmp(sdb->database->id, id))
return sdb;
return load_session_database(se, id);
}
new->next = sdb->settings[offset];
sdb->settings[offset] = new;
+ se->settings_modified = 1;
+
// Force later recompute of settings-driven data structures
// (happens when a search starts and client connections are prepared)
switch (offset)
}
}
-void destroy_session(struct session *s)
+void session_destroy(struct session *se)
{
struct session_database *sdb;
+ session_log(se, YLOG_DEBUG, "Destroying");
+ session_use(-1);
+ session_remove_cached_clients(se);
- session_remove_clients(s);
-
- for (sdb = s->databases; sdb; sdb = sdb->next)
+ for (sdb = se->databases; sdb; sdb = sdb->next)
session_database_destroy(sdb);
- normalize_cache_destroy(s->normalize_cache);
- relevance_destroy(&s->relevance);
- reclist_destroy(s->reclist);
- nmem_destroy(s->nmem);
- service_destroy(s->service);
- yaz_mutex_destroy(&s->session_mutex);
- wrbuf_destroy(s->wrbuf);
+ normalize_cache_destroy(se->normalize_cache);
+ relevance_destroy(&se->relevance);
+ reclist_destroy(se->reclist);
+ nmem_destroy(se->nmem);
+ service_destroy(se->service);
+ yaz_mutex_destroy(&se->session_mutex);
}
+size_t session_get_memory_status(struct session *session) {
+ size_t session_nmem;
+ if (session == 0)
+ return 0;
+ session_enter(session);
+ session_nmem = nmem_total(session->nmem);
+ session_leave(session);
+ return session_nmem;
+}
+
+
struct session *new_session(NMEM nmem, struct conf_service *service,
- const char *name)
+ unsigned session_id)
{
int i;
struct session *session = nmem_malloc(nmem, sizeof(*session));
- yaz_log(YLOG_DEBUG, "New Pazpar2 session");
+ char tmp_str[50];
+ sprintf(tmp_str, "session#%u", session_id);
+
+ session->session_id = session_id;
+ session_log(session, YLOG_DEBUG, "New");
session->service = service;
session->relevance = 0;
- session->total_hits = 0;
session->total_records = 0;
session->number_of_warnings_unknown_elements = 0;
session->number_of_warnings_unknown_metadata = 0;
session->num_termlists = 0;
session->reclist = 0;
- session->clients = 0;
+ session->clients_active = 0;
+ session->clients_cached = 0;
+ session->settings_modified = 0;
session->session_nmem = nmem;
session->nmem = nmem_create();
- session->wrbuf = wrbuf_alloc();
session->databases = 0;
for (i = 0; i <= SESSION_WATCH_MAX; i++)
{
}
session->normalize_cache = normalize_cache_create();
session->session_mutex = 0;
- pazpar2_mutex_create(&session->session_mutex, name);
-
+ pazpar2_mutex_create(&session->session_mutex, tmp_str);
+ session_use(1);
return session;
}
-struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem)
+const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
+
+static struct hitsbytarget *hitsbytarget_nb(struct session *se,
+ int *count, NMEM nmem)
{
struct hitsbytarget *res = 0;
struct client_list *l;
size_t sz = 0;
- session_enter(se);
- for (l = se->clients; l; l = l->next)
+ for (l = se->clients_active; l; l = l->next)
sz++;
res = nmem_malloc(nmem, sizeof(*res) * sz);
*count = 0;
- for (l = se->clients; l; l = l->next)
+ for (l = se->clients_active; l; l = l->next)
{
struct client *cl = l->client;
WRBUF w = wrbuf_alloc();
const char *name = session_setting_oneval(client_get_database(cl),
PZ_NAME);
- res[*count].id = client_get_database(cl)->database->url;
+ res[*count].id = client_get_id(cl);
res[*count].name = *name ? name : "Unknown";
res[*count].hits = client_get_hits(cl);
res[*count].records = client_get_num_records(cl);
res[*count].state = client_get_state_str(cl);
res[*count].connected = client_get_connection(cl) ? 1 : 0;
session_settings_dump(se, client_get_database(cl), w);
- res[*count].settings_xml = w;
+ res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
+ wrbuf_rewind(w);
+ wrbuf_puts(w, "");
+ res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
+ wrbuf_destroy(w);
(*count)++;
}
- session_leave(se);
return res;
}
-struct termlist_score **termlist(struct session *s, const char *name, int *num)
+struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
+{
+ struct hitsbytarget *p;
+ session_enter(se);
+ p = hitsbytarget_nb(se, count, nmem);
+ session_leave(se);
+ return p;
+}
+
+struct termlist_score **get_termlist_score(struct session *se,
+ const char *name, int *num)
{
int i;
struct termlist_score **tl = 0;
- session_enter(s);
- for (i = 0; i < s->num_termlists; i++)
- if (!strcmp((const char *) s->termlists[i].name, name))
+ session_enter(se);
+ for (i = 0; i < se->num_termlists; i++)
+ if (!strcmp((const char *) se->termlists[i].name, name))
{
- tl = termlist_highscore(s->termlists[i].termlist, num);
+ tl = termlist_highscore(se->termlists[i].termlist, num);
break;
}
- session_leave(s);
+ session_leave(se);
return tl;
}
+// Compares two hitsbytarget nodes by hitcount
+static int cmp_ht(const void *p1, const void *p2)
+{
+ const struct hitsbytarget *h1 = p1;
+ const struct hitsbytarget *h2 = p2;
+ return h2->hits - h1->hits;
+}
+
+static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
+ NMEM nmem)
+{
+ struct hitsbytarget *ht;
+ int count, i;
+
+ ht = hitsbytarget_nb(se, &count, nmem);
+ qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
+ for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
+ {
+
+ // do only print terms which have display names
+
+ wrbuf_puts(wrbuf, "<term>\n");
+
+ wrbuf_puts(wrbuf, "<id>");
+ wrbuf_xmlputs(wrbuf, ht[i].id);
+ wrbuf_puts(wrbuf, "</id>\n");
+
+ wrbuf_puts(wrbuf, "<name>");
+ if (!ht[i].name || !ht[i].name[0])
+ wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
+ else
+ wrbuf_xmlputs(wrbuf, ht[i].name);
+ wrbuf_puts(wrbuf, "</name>\n");
+
+ wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
+ ht[i].hits);
+
+ wrbuf_puts(wrbuf, "<state>");
+ wrbuf_xmlputs(wrbuf, ht[i].state);
+ wrbuf_puts(wrbuf, "</state>\n");
+
+ wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
+ ht[i].diagnostic);
+ wrbuf_puts(wrbuf, "</term>\n");
+ }
+ return count;
+}
+
+void perform_termlist(struct http_channel *c, struct session *se,
+ const char *name, int num)
+{
+ int i, j;
+ NMEM nmem_tmp = nmem_create();
+ char **names;
+ int num_names = 0;
+
+ if (name)
+ nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
+
+ session_enter(se);
+
+ for (j = 0; j < num_names; j++)
+ {
+ const char *tname;
+
+ wrbuf_puts(c->wrbuf, "<list name=\"");
+ wrbuf_xmlputs(c->wrbuf, names[j]);
+ wrbuf_puts(c->wrbuf, "\">\n");
+
+ for (i = 0; i < se->num_termlists; i++)
+ {
+ tname = se->termlists[i].name;
+ if (num_names > 0 && !strcmp(names[j], tname))
+ {
+ struct termlist_score **p = 0;
+ int len;
+ p = termlist_highscore(se->termlists[i].termlist, &len);
+ if (p)
+ {
+ int i;
+ for (i = 0; i < len && i < num; i++)
+ {
+ // prevent sending empty term elements
+ if (!p[i]->display_term || !p[i]->display_term[0])
+ continue;
+
+ wrbuf_puts(c->wrbuf, "<term>");
+ wrbuf_puts(c->wrbuf, "<name>");
+ wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
+ wrbuf_puts(c->wrbuf, "</name>");
+
+ wrbuf_printf(c->wrbuf,
+ "<frequency>%d</frequency>",
+ p[i]->frequency);
+ wrbuf_puts(c->wrbuf, "</term>\n");
+ }
+ }
+ }
+ }
+ tname = "xtargets";
+ if (num_names > 0 && !strcmp(names[j], tname))
+ {
+ targets_termlist_nb(c->wrbuf, se, num, c->nmem);
+ }
+ wrbuf_puts(c->wrbuf, "</list>\n");
+ }
+ session_leave(se);
+ nmem_destroy(nmem_tmp);
+}
+
#ifdef MISSING_HEADERS
void report_nmem_stats(void)
{
}
#endif
-struct record_cluster *show_single_start(struct session *s, const char *id,
+struct record_cluster *show_single_start(struct session *se, const char *id,
struct record_cluster **prev_r,
struct record_cluster **next_r)
{
- struct record_cluster *r;
+ struct record_cluster *r = 0;
- session_enter(s);
- reclist_enter(s->reclist);
+ session_enter(se);
*prev_r = 0;
*next_r = 0;
- while ((r = reclist_read_record(s->reclist)))
+ if (se->reclist)
{
- if (!strcmp(r->recid, id))
+ reclist_enter(se->reclist);
+ while ((r = reclist_read_record(se->reclist)))
{
- *next_r = reclist_read_record(s->reclist);
- break;
+ if (!strcmp(r->recid, id))
+ {
+ *next_r = reclist_read_record(se->reclist);
+ break;
+ }
+ *prev_r = r;
}
- *prev_r = r;
+ reclist_leave(se->reclist);
}
- reclist_leave(s->reclist);
if (!r)
- session_leave(s);
+ session_leave(se);
return r;
}
-void show_single_stop(struct session *s, struct record_cluster *rec)
+void show_single_stop(struct session *se, struct record_cluster *rec)
{
- session_leave(s);
+ session_leave(se);
}
-struct record_cluster **show_range_start(struct session *s,
+struct record_cluster **show_range_start(struct session *se,
struct reclist_sortparms *sp,
int start, int *num, int *total, Odr_int *sumhits)
{
- struct record_cluster **recs = nmem_malloc(s->nmem, *num
- * sizeof(struct record_cluster *));
+ struct record_cluster **recs;
struct reclist_sortparms *spp;
int i;
#if USE_TIMING
yaz_timing_t t = yaz_timing_create();
#endif
-
- session_enter(s);
- if (!s->relevance)
+ session_enter(se);
+ recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
+ if (!se->relevance)
{
*num = 0;
*total = 0;
}
else
{
+ struct client_list *l;
+
for (spp = sp; spp; spp = spp->next)
if (spp->type == Metadata_sortkey_relevance)
{
- relevance_prepare_read(s->relevance, s->reclist);
+ relevance_prepare_read(se->relevance, se->reclist);
break;
}
- reclist_sort(s->reclist, sp);
+ reclist_sort(se->reclist, sp);
- reclist_enter(s->reclist);
- *total = reclist_get_num_records(s->reclist);
- *sumhits = s->total_hits;
+ reclist_enter(se->reclist);
+ *total = reclist_get_num_records(se->reclist);
+
+ *sumhits = 0;
+ for (l = se->clients_active; l; l = l->next)
+ *sumhits += client_get_hits(l->client);
for (i = 0; i < start; i++)
- if (!reclist_read_record(s->reclist))
+ if (!reclist_read_record(se->reclist))
{
*num = 0;
recs = 0;
for (i = 0; i < *num; i++)
{
- struct record_cluster *r = reclist_read_record(s->reclist);
+ struct record_cluster *r = reclist_read_record(se->reclist);
if (!r)
{
*num = i;
}
recs[i] = r;
}
- reclist_leave(s->reclist);
+ reclist_leave(se->reclist);
}
#if USE_TIMING
yaz_timing_stop(t);
return recs;
}
-void show_range_stop(struct session *s, struct record_cluster **recs)
+void show_range_stop(struct session *se, struct record_cluster **recs)
{
- session_leave(s);
+ session_leave(se);
}
void statistics(struct session *se, struct statistics *stat)
int count = 0;
memset(stat, 0, sizeof(*stat));
- for (l = se->clients; l; l = l->next)
+ stat->num_hits = 0;
+ for (l = se->clients_active; l; l = l->next)
{
struct client *cl = l->client;
if (!client_get_connection(cl))
stat->num_no_connection++;
+ stat->num_hits += client_get_hits(cl);
switch (client_get_state(cl))
{
case Client_Connecting: stat->num_connecting++; break;
}
count++;
}
- stat->num_hits = se->total_hits;
stat->num_records = se->total_records;
stat->num_clients = count;
if (!strcmp((const char *) n->name, "metadata"))
{
xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
- if (!strcmp(name, (const char *) type))
+ if (type == NULL) {
+ yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
+ }
+ else 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);
+ pp2_charset_token_t prt =
+ pp2_charset_token_create(service->charsets, "mergekey");
- pp2_relevance_first(prt, (const char *) value, 0);
+ pp2_charset_token_first(prt, (const char *) value, 0);
if (wrbuf_len(norm_wr) > 0)
wrbuf_puts(norm_wr, " ");
wrbuf_puts(norm_wr, name);
while ((norm_str =
- pp2_relevance_token_next(prt)))
+ pp2_charset_token_next(prt)))
{
if (*norm_str)
{
}
}
xmlFree(value);
- pp2_relevance_token_destroy(prt);
+ pp2_charset_token_destroy(prt);
no_found++;
}
}
if (mergekey)
{
const char *norm_str;
- pp2_relevance_token_t prt =
- pp2_relevance_tokenize(service->mergekey_pct);
+ pp2_charset_token_t prt =
+ pp2_charset_token_create(service->charsets, "mergekey");
- pp2_relevance_first(prt, (const char *) mergekey, 0);
- while ((norm_str = pp2_relevance_token_next(prt)))
+ pp2_charset_token_first(prt, (const char *) mergekey, 0);
+ while ((norm_str = pp2_charset_token_next(prt)))
{
if (*norm_str)
{
wrbuf_puts(norm_wr, norm_str);
}
}
- pp2_relevance_token_destroy(prt);
+ pp2_charset_token_destroy(prt);
xmlFree(mergekey);
}
else
/* generate unique key if none is not generated already or is empty */
if (wrbuf_len(norm_wr) == 0)
{
- wrbuf_printf(norm_wr, "%s-%d",
- client_get_database(cl)->database->url, record_no);
+ wrbuf_printf(norm_wr, "position: %s-%d",
+ client_get_id(cl), record_no);
+ }
+ else
+ {
+ const char *lead = "content: ";
+ wrbuf_insert(norm_wr, 0, lead, strlen(lead));
}
if (wrbuf_len(norm_wr) > 0)
mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
if (type)
{
size_t len;
- const char *eq = strchr(s, '~');
- if (eq)
- len = eq - s;
+ int substring;
+ const char *eq;
+
+ if ((eq = strchr(s, '=')))
+ substring = 0;
+ else if ((eq = strchr(s, '~')))
+ substring = 1;
+ if (eq)
+ len = eq - s;
else
len = strlen(s);
if (len == strlen((const char *)type) &&
xmlChar *value = xmlNodeGetContent(n);
if (value && *value)
{
- if (!eq || strstr((const char *) value, eq+1))
+ if (!eq ||
+ (substring && strstr((const char *) value, eq+1)) ||
+ (!substring && !strcmp((const char *) value, eq + 1)))
match = 1;
}
xmlFree(value);
\param cl client holds the result set for record
\param rec record buffer (0 terminated)
\param record_no record position (1, 2, ..)
+ \param nmem working NMEM
\retval 0 OK
\retval -1 failure
+ \retval -2 Filtered
*/
int ingest_record(struct client *cl, const char *rec,
int record_no, NMEM nmem)
int ret = 0;
struct session_database *sdb = client_get_database(cl);
struct conf_service *service = se->service;
- xmlDoc *xdoc = normalize_record(sdb, service, rec, nmem);
+ xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
xmlNode *root;
const char *mergekey_norm;
if (!check_record_filter(root, sdb))
{
- yaz_log(YLOG_WARN, "Filtered out record no %d from %s", record_no,
- sdb->database->url);
+ session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
+ record_no, sdb->database->id);
xmlFreeDoc(xdoc);
- return -1;
+ return -2;
}
mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
if (!mergekey_norm)
{
- yaz_log(YLOG_WARN, "Got no mergekey");
+ session_log(se, YLOG_WARN, "Got no mergekey");
xmlFreeDoc(xdoc);
return -1;
}
- client_unlock(cl);
session_enter(se);
- client_lock(cl);
if (client_get_session(cl) == se)
- ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
- client_unlock(cl);
+ ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
session_leave(se);
- client_lock(cl);
xmlFreeDoc(xdoc);
return ret;
xmlNode *n;
xmlChar *type = 0;
xmlChar *value = 0;
- struct session_database *sdb = client_get_database(cl);
struct session *se = client_get_session(cl);
struct conf_service *service = se->service;
+ int term_factor = 1;
+ struct record_cluster *cluster;
+ struct session_database *sdb = client_get_database(cl);
struct record *record = record_create(se->nmem,
service->num_metadata,
service->num_sortkeys, cl,
record_no);
- struct record_cluster *cluster = reclist_insert(se->reclist,
- service,
- record,
- mergekey_norm,
- &se->total_merged);
- if (!cluster)
- return -1;
- 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
+
for (n = root->children; n; n = n->next)
{
- pp2_relevance_token_t prt;
if (type)
xmlFree(type);
if (value)
if (!strcmp((const char *) n->name, "metadata"))
{
struct conf_metadata *ser_md = 0;
- struct conf_sortkey *ser_sk = 0;
struct record_metadata **wheretoput = 0;
struct record_metadata *rec_md = 0;
int md_field_id = -1;
- int sk_field_id = -1;
type = xmlGetProp(n, (xmlChar *) "type");
value = xmlNodeListGetString(xdoc, n->children, 1);
{
if (se->number_of_warnings_unknown_metadata == 0)
{
- yaz_log(YLOG_WARN,
+ session_log(se, YLOG_WARN,
"Ignoring unknown metadata element: %s", type);
}
se->number_of_warnings_unknown_metadata++;
continue;
}
-
+
ser_md = &service->metadata[md_field_id];
-
- if (ser_md->sortkey_offset >= 0){
- sk_field_id = ser_md->sortkey_offset;
- ser_sk = &service->sortkeys[sk_field_id];
- }
// non-merged metadata
rec_md = record_metadata_init(se->nmem, (const char *) value,
ser_md->type, n->properties);
if (!rec_md)
{
- yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'",
- value, type);
+ session_log(se, YLOG_WARN, "bad metadata data '%s' "
+ "for element '%s'", value, type);
continue;
}
wheretoput = &record->metadata[md_field_id];
while (*wheretoput)
wheretoput = &(*wheretoput)->next;
*wheretoput = rec_md;
+ }
+ }
+
+ cluster = reclist_insert(se->reclist, service, record,
+ mergekey_norm, &se->total_merged);
+ if (!cluster)
+ return -1;
+
+ {
+ const char *use_term_factor_str =
+ session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
+ if (use_term_factor_str && use_term_factor_str[0] == '1')
+ {
+ int maxrecs = client_get_maxrecs(cl);
+ int hits = (int) client_get_hits(cl);
+ term_factor = MAX(hits, maxrecs) / MAX(1, maxrecs);
+ assert(term_factor >= 1);
+ yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
+ }
+ }
+
+ if (global_parameters.dump_records)
+ session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
+ sdb->database->id, record_no);
+
+
+ relevance_newrec(se->relevance, cluster);
+
+ // now parsing XML record and adding data to cluster or record metadata
+ for (n = root->children; n; n = n->next)
+ {
+ pp2_charset_token_t prt;
+ if (type)
+ xmlFree(type);
+ if (value)
+ xmlFree(value);
+ type = value = 0;
+
+ if (n->type != XML_ELEMENT_NODE)
+ continue;
+ if (!strcmp((const char *) n->name, "metadata"))
+ {
+ struct conf_metadata *ser_md = 0;
+ struct conf_sortkey *ser_sk = 0;
+ struct record_metadata **wheretoput = 0;
+ struct record_metadata *rec_md = 0;
+ int md_field_id = -1;
+ int sk_field_id = -1;
+
+ type = xmlGetProp(n, (xmlChar *) "type");
+ value = xmlNodeListGetString(xdoc, n->children, 1);
+
+ if (!type || !value || !*value)
+ continue;
+
+ md_field_id
+ = conf_service_metadata_field_id(service, (const char *) type);
+ if (md_field_id < 0)
+ continue;
+
+ ser_md = &service->metadata[md_field_id];
+
+ if (ser_md->sortkey_offset >= 0)
+ {
+ sk_field_id = ser_md->sortkey_offset;
+ ser_sk = &service->sortkeys[sk_field_id];
+ }
// merged metadata
rec_md = record_metadata_init(se->nmem, (const char *) value,
ser_md->type, 0);
+ if (!rec_md)
+ continue;
+
wheretoput = &cluster->metadata[md_field_id];
// and polulate with data:
// assign cluster or record based on merge action
if (ser_md->merge == Metadata_merge_unique)
{
- struct record_metadata *mnode;
- for (mnode = *wheretoput; mnode; mnode = mnode->next)
- if (!strcmp((const char *) mnode->data.text.disp,
+ while (*wheretoput)
+ {
+ if (!strcmp((const char *) (*wheretoput)->data.text.disp,
rec_md->data.text.disp))
break;
- if (!mnode)
- {
- rec_md->next = *wheretoput;
- *wheretoput = rec_md;
+ wheretoput = &(*wheretoput)->next;
}
+ if (!*wheretoput)
+ *wheretoput = rec_md;
}
else if (ser_md->merge == Metadata_merge_longest)
{
nmem_malloc(se->nmem,
sizeof(union data_types));
- prt = pp2_relevance_tokenize(service->sort_pct);
+ prt =
+ pp2_charset_token_create(service->charsets, "sort");
- pp2_relevance_first(prt, rec_md->data.text.disp,
- skip_article);
+ pp2_charset_token_first(prt, rec_md->data.text.disp,
+ skip_article);
- pp2_relevance_token_next(prt);
+ pp2_charset_token_next(prt);
sort_str = pp2_get_sort(prt);
if (!sort_str)
{
sort_str = rec_md->data.text.disp;
- yaz_log(YLOG_WARN,
+ session_log(se, YLOG_WARN,
"Could not make sortkey. Bug #1858");
}
cluster->sortkeys[sk_field_id]->text.sort =
nmem_strdup(se->nmem, sort_str);
-#if 0
- yaz_log(YLOG_LOG, "text disp=%s",
- cluster->sortkeys[sk_field_id]->text.disp);
- yaz_log(YLOG_LOG, "text sort=%s",
- cluster->sortkeys[sk_field_id]->text.sort);
-#endif
- pp2_relevance_token_destroy(prt);
+ pp2_charset_token_destroy(prt);
}
}
}
else if (ser_md->merge == Metadata_merge_all)
{
- rec_md->next = *wheretoput;
+ while (*wheretoput)
+ wheretoput = &(*wheretoput)->next;
*wheretoput = rec_md;
}
else if (ser_md->merge == Metadata_merge_range)
(char *) value, ser_md->rank,
ser_md->name);
- // construct facets ...
- if (ser_md->termlist)
+ // construct facets ... unless the client already has reported them
+ if (ser_md->termlist && !client_has_facet(cl, (char *) type))
{
if (ser_md->type == Metadata_type_year)
{
char year[64];
sprintf(year, "%d", rec_md->data.number.max);
- add_facet(se, (char *) type, year);
+
+ add_facet(se, (char *) type, year, term_factor);
if (rec_md->data.number.max != rec_md->data.number.min)
{
sprintf(year, "%d", rec_md->data.number.min);
- add_facet(se, (char *) type, year);
+ add_facet(se, (char *) type, year, term_factor);
}
}
else
- add_facet(se, (char *) type, (char *) value);
+ add_facet(se, (char *) type, (char *) value, term_factor);
}
// cleaning up
else
{
if (se->number_of_warnings_unknown_elements == 0)
- yaz_log(YLOG_WARN,
+ session_log(se, YLOG_WARN,
"Unexpected element in internal record: %s", n->name);
se->number_of_warnings_unknown_elements++;
}
return 0;
}
+void session_log(struct session *s, int level, const char *fmt, ...)
+{
+ char buf[1024];
+ va_list ap;
+ va_start(ap, fmt);
+
+ yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
+ yaz_log(level, "Session (%u): %s", s->session_id, buf);
+
+ va_end(ap);
+}
+
/*
* Local variables:
* c-basic-offset: 4
projects / pazpar2-moved-to-github.git / blobdiff