/* 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 <signal.h>
#include <ctype.h>
#include <assert.h>
+#include <math.h>
#include <yaz/marcdisp.h>
#include <yaz/comstack.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_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 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)
-{
- char **words;
- int numwords;
- int i;
-
- switch (n->kind)
- {
- 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;
- }
-}
-
-
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)
{
- session_log(s, 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;
}
- session_log(s, YLOG_DEBUG, "Session: facets for %s: %s (%d)",
- type, value, count);
-
- termlist_insert(s->termlists[i].termlist, value, count);
+ 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 *se,
ret = 0;
}
session_leave(s);
- return 0;
+ return ret;
}
void session_alert_watch(struct session *s, int what)
session_leave(s);
session_log(s, YLOG_DEBUG,
- "session_alert_watch: %d calling function: %p", what, fun);
+ "Alert Watch: %d calling function: %p", what, fun);
fun(data);
}
else
client_set_database(cl, db);
client_set_session(cl, se);
+
l = xmalloc(sizeof(*l));
l->client = cl;
l->next = se->clients;
for (l = s->clients; l; l = l->next)
if (client_is_active_preferred(l->client))
res++;
- session_log(s, YLOG_DEBUG, "Session has %d active preferred clients.", 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_failed = 0;
struct client_list *l;
struct timeval tval;
+ facet_limits_t facet_limits;
session_log(se, YLOG_DEBUG, "Search");
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;
client_set_startrecs(cl, atoi(startrecs));
if (prepare_session_database(se, client_get_database(cl)) < 0)
;
- else if (client_parse_query(cl, query) < 0)
+ else if (client_parse_query(cl, query, facet_limits) < 0)
no_failed++;
else
{
client_start_search(cl);
}
}
+ facet_limits_destroy(facet_limits);
session_leave(se);
if (no_working == 0)
{
}
}
-void destroy_session(struct session *se)
-{
+void session_destroy(struct session *se) {
struct session_database *sdb;
-
- session_log(se, YLOG_DEBUG, "Pazpar2 session destroy");
+ session_log(se, YLOG_DEBUG, "Destroying");
+ session_use(-1);
session_remove_clients(se);
for (sdb = se->databases; sdb; sdb = sdb->next)
nmem_destroy(se->nmem);
service_destroy(se->service);
yaz_mutex_destroy(&se->session_mutex);
- wrbuf_destroy(se->wrbuf);
}
+/* 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)
{
sprintf(tmp_str, "session#%u", session_id);
session->session_id = session_id;
- session_log(session, YLOG_DEBUG, "New session");
+ 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->normalize_cache = normalize_cache_create();
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)
+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)
sz++;
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 *se, 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;
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;
+ 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;
+ wrbuf_puts(c->wrbuf, "<list name=\"");
+ wrbuf_xmlputs(c->wrbuf, tname);
+ wrbuf_puts(c->wrbuf, "\">\n");
+ 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");
+ }
+ wrbuf_puts(c->wrbuf, "</list>\n");
+ }
+ }
+ }
+ tname = "xtargets";
+ if (num_names > 0 && !strcmp(names[j], tname))
+ {
+ wrbuf_puts(c->wrbuf, "<list name=\"");
+ wrbuf_xmlputs(c->wrbuf, tname);
+ wrbuf_puts(c->wrbuf, "\">\n");
+ 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)
{
struct record_cluster **prev_r,
struct record_cluster **next_r)
{
- struct record_cluster *r;
+ struct record_cluster *r = 0;
session_enter(se);
- reclist_enter(se->reclist);
*prev_r = 0;
*next_r = 0;
- while ((r = reclist_read_record(se->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(se->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(se->reclist);
if (!r)
session_leave(se);
return r;
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
\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)
if (!check_record_filter(root, sdb))
{
- session_log(se, YLOG_WARN, "Filtered out record no %d from %s",
+ 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);
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)
// 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)
// 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);
}
cluster->sortkeys[sk_field_id]->text.sort =
nmem_strdup(se->nmem, sort_str);
- 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)
// 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, 1);
+
+ 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, 1);
+ add_facet(se, (char *) type, year, term_factor);
}
}
else
- add_facet(se, (char *) type, (char *) value, 1);
+ add_facet(se, (char *) type, (char *) value, term_factor);
}
// cleaning up
va_start(ap, fmt);
yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
- yaz_log(level, "%u %s", s->session_id, buf);
+ yaz_log(level, "Session (%u): %s", s->session_id, buf);
va_end(ap);
}
projects / pazpar2-moved-to-github.git / blobdiff