/* 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
#include <yaz/timing.h>
#endif
+#include "ppmutex.h"
#include "parameters.h"
#include "session.h"
#include "eventl.h"
#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 *client;
+ struct client_list *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 sesions=%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();
static void session_enter(struct session *s)
{
- yaz_mutex_enter(s->mutex);
+ yaz_mutex_enter(s->session_mutex);
}
static void session_leave(struct session *s)
{
- yaz_mutex_leave(s->mutex);
+ yaz_mutex_leave(s->session_mutex);
}
// Recursively traverse query structure to extract terms.
}
-static void add_facet(struct session *s, const char *type, const char *value)
+void add_facet(struct session *s, const char *type, const char *value, int count)
{
+ 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;
- if (!*value)
+ 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)
+ {
+ 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->url);
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->url);
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->url);
}
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->url);
log_xml_doc(rdoc);
}
}
if (!sdb->settings)
{
- yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
+ session_log(se, YLOG_WARN, "No settings on %s", sdb->database->url);
return -1;
}
if ((s = session_setting_oneval(sdb, PZ_XSLT)))
}
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,
- se->service, s);
+ se->service->server->config, s);
if (!sdb->map)
return -1;
}
{
if (!sdb->settings)
{
- yaz_log(YLOG_WARN,
+ session_log(se, YLOG_WARN,
"No settings associated with %s", sdb->database->url);
return -1;
}
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
{
struct session *se = (struct session*) context;
struct client *cl = client_create();
+ struct client_list *l;
client_set_database(cl, db);
+
client_set_session(cl, se);
+
+ l = xmalloc(sizeof(*l));
+ l->client = cl;
+ l->next = se->clients;
+ se->clients = l;
}
static void session_remove_clients(struct session *se)
{
- struct client *cl = se->clients;
- while (cl)
+ struct client_list *l;
+
+ session_enter(se);
+ l = se->clients;
+ se->clients = 0;
+ session_leave(se);
+
+ while (l)
{
- struct client *cl_next = client_next_in_session(cl);
- client_remove_from_session(cl);
- client_destroy(cl);
- cl = cl_next;
+ struct client_list *l_next = l->next;
+ client_lock(l->client);
+ client_set_session(l->client, 0);
+ client_set_database(l->client, 0);
+ client_unlock(l->client);
+ client_destroy(l->client);
+ xfree(l);
+ l = l_next;
}
- se->clients = 0;
}
// Associates a set of clients with a session;
// setting overrides
static int select_targets(struct session *se, const char *filter)
{
- session_remove_clients(se);
return session_grep_databases(se, filter, select_targets_callback);
}
int session_active_clients(struct session *s)
{
- struct client *c;
+ struct client_list *l;
int res = 0;
- for (c = s->clients; c; c = client_next_in_session(c))
- if (client_is_active(c))
+ for (l = s->clients; 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; 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;
+}
enum pazpar2_error_code search(struct session *se,
const char *query,
const char *startrecs, const char *maxrecs,
const char *filter,
+ const char *limit,
const char **addinfo)
{
int live_channels = 0;
int no_working = 0;
int no_failed = 0;
- struct client *cl;
+ struct client_list *l;
+ 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);
session_enter(se);
reclist_destroy(se->reclist);
se->reclist = 0;
+ relevance_destroy(&se->relevance);
nmem_reset(se->nmem);
- se->relevance = 0;
se->total_records = se->total_hits = se->total_merged = 0;
se->num_termlists = 0;
live_channels = select_targets(se, filter);
}
se->reclist = reclist_create(se->nmem);
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ 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;
+ }
+ for (l = se->clients; l; l = l->next)
{
+ struct client *cl = l->client;
+
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)
- continue;
- // Parse query for target
- if (client_parse_query(cl, query) < 0)
+ ;
+ else if (client_parse_query(cl, query, facet_limits) < 0)
no_failed++;
else
{
no_working++;
if (client_prep_connection(cl, se->service->z3950_operation_timeout,
se->service->z3950_session_timeout,
- se->service->server->iochan_man))
+ se->service->server->iochan_man,
+ &tval))
client_start_search(cl);
}
}
+ facet_limits_destroy(facet_limits);
session_leave(se);
if (no_working == 0)
{
}
}
-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_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);
- reclist_destroy(s->reclist);
- nmem_destroy(s->nmem);
- service_destroy(s->service);
- yaz_mutex_destroy(&s->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);
}
-struct session *new_session(NMEM nmem, struct conf_service *service)
+/* Depreciated: use session_destroy */
+void destroy_session(struct session *se)
+{
+ session_destroy(se);
+}
+
+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,
+ 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->clients = 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->watchlist[i].fun = 0;
}
session->normalize_cache = normalize_cache_create();
- session->mutex = 0;
- yaz_mutex_create(&session->mutex);
-
+ session->session_mutex = 0;
+ pazpar2_mutex_create(&session->session_mutex, tmp_str);
+ session_use(1);
return session;
}
struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem)
{
struct hitsbytarget *res = 0;
- struct client *cl;
+ struct client_list *l;
size_t sz = 0;
session_enter(se);
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ for (l = se->clients; l; l = l->next)
sz++;
res = nmem_malloc(nmem, sizeof(*res) * sz);
*count = 0;
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ for (l = se->clients; 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].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_destroy(w);
(*count)++;
}
session_leave(se);
return res;
}
-struct termlist_score **termlist(struct session *s, const char *name, int *num)
+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;
}
}
#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;
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 = se->total_hits;
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)
{
- struct client *cl;
+ struct client_list *l;
int count = 0;
memset(stat, 0, sizeof(*stat));
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ for (l = se->clients; l; l = l->next)
{
+ struct client *cl = l->client;
if (!client_get_connection(cl))
stat->num_no_connection++;
switch (client_get_state(cl))
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,
- (const char *) value, 0);
+ pp2_charset_token_t prt =
+ pp2_charset_token_create(service->charsets, "mergekey");
+ 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,
- (const char *) mergekey, 0);
-
- while ((norm_str = pp2_relevance_token_next(prt)))
+ pp2_charset_token_t prt =
+ pp2_charset_token_create(service->charsets, "mergekey");
+
+ 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
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)
{
struct session *se = client_get_session(cl);
- int ret;
+ 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->url);
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;
}
session_enter(se);
- ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
+ if (client_get_session(cl) == se)
+ ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
session_leave(se);
xmlFreeDoc(xdoc);
record,
mergekey_norm,
&se->total_merged);
+
+ const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
+ int use_term_factor = 0;
+ int term_factor = 1;
+ if (use_term_factor_str && use_term_factor_str[0] != 0)
+ use_term_factor = atoi(use_term_factor_str);
+ if (use_term_factor) {
+ 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 (!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);
+ session_log(se, 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;
+ pp2_charset_token_t prt;
if (type)
xmlFree(type);
if (value)
{
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++;
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];
// 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,
- rec_md->data.text.disp, skip_article);
+ prt =
+ pp2_charset_token_create(service->charsets, "sort");
+
+ 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