-/* $Id: logic.c,v 1.43 2007-06-13 21:29:04 adam Exp $
- Copyright (c) 2006-2007, Index Data.
-
-This file is part of Pazpar2.
+/* This file is part of Pazpar2.
+ Copyright (C) 2006-2008 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
for more details.
You should have received a copy of the GNU General Public License
-along with Pazpar2; see the file LICENSE. If not, write to the
-Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
-02111-1307, USA.
- */
+along with this program; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+
+*/
/** \file logic.c
\brief high-level logic; mostly user sessions and settings
*/
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
+
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
+#if HAVE_SYS_TIME_H
#include <sys/time.h>
+#endif
+#if HAVE_UNISTD_H
#include <unistd.h>
+#endif
+#if HAVE_SYS_SOCKET_H
#include <sys/socket.h>
+#endif
+#if HAVE_NETDB_H
#include <netdb.h>
+#endif
#include <signal.h>
#include <ctype.h>
#include <assert.h>
#include <yaz/nmem.h>
#include <yaz/query-charset.h>
#include <yaz/querytowrbuf.h>
-#if YAZ_VERSIONL >= 0x020163
#include <yaz/oid_db.h>
-#endif
-
-#if HAVE_CONFIG_H
-#include "cconfig.h"
-#endif
+#include <yaz/snprintf.h>
#define USE_TIMING 0
#if USE_TIMING
#include <yaz/timing.h>
#endif
+#if HAVE_NETINET_IN_H
#include <netinet/in.h>
+#endif
#include "pazpar2.h"
#include "eventl.h"
#include "termlists.h"
#include "reclists.h"
#include "relevance.h"
-#include "config.h"
#include "database.h"
#include "client.h"
#include "settings.h"
"",
"",
0,
- 0, /* dump_records */
- 0, /* debug_mode */
- 30,
+ 0, // dump_records
+ 0, // debug_mode
+ 30, // operations timeout
"81",
"Index Data PazPar2",
VERSION,
- 600, // 10 minutes
- 60,
+ 60, // session timeout
100,
MAX_CHUNK,
0,
- 0
+ 0,
+ 180, // Z39.50 session timeout
+ 15 // Connect timeout
};
-
// Recursively traverse query structure to extract terms.
void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
{
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;
+ 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;
}
}
termlist_insert(s->termlists[i].termlist, value);
}
-xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec)
+xmlDoc *record_to_xml(struct session_database *sdb, const char *rec)
{
- struct database_retrievalmap *m;
struct database *db = sdb->database;
xmlDoc *rdoc = 0;
- const Odr_oid *oid = rec->direct_reference;
- /* convert response record to XML somehow */
- if (rec->which == Z_External_octet && oid
- && !oid_oidcmp(oid, yaz_oid_recsyn_xml))
+ rdoc = xmlParseMemory(rec, strlen(rec));
+
+ if (!rdoc)
{
- /* xml already */
- rdoc = xmlParseMemory((char*) rec->u.octet_aligned->buf,
- rec->u.octet_aligned->len);
- if (!rdoc)
- {
- yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
- db->url);
- return 0;
- }
+ yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
+ db->url);
+ return 0;
}
- else if (oid && yaz_oid_is_iso2709(oid))
+
+ if (global_parameters.dump_records)
{
- /* ISO2709 gets converted to MARCXML */
- if (!sdb->yaz_marc)
+ FILE *lf = yaz_log_file();
+ if (lf)
{
- yaz_log(YLOG_FATAL, "Unable to handle ISO2709 record");
- return 0;
+ yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url);
+#if LIBXML_VERSION >= 20600
+ xmlDocFormatDump(lf, rdoc, 1);
+#else
+ xmlDocDump(lf, rdoc);
+#endif
+ fprintf(lf, "\n");
}
- else
+ }
+
+ return rdoc;
+}
+
+#define MAX_XSLT_ARGS 16
+
+// Add static values from session database settings if applicable
+static void insert_settings_parameters(struct session_database *sdb,
+ struct session *se, char **parms)
+{
+ struct conf_service *service = global_parameters.server->service;
+ int i;
+ int nparms = 0;
+ int offset = 0;
+
+ for (i = 0; i < service->num_metadata; i++)
+ {
+ struct conf_metadata *md = &service->metadata[i];
+ int setting;
+
+ if (md->setting == Metadata_setting_parameter &&
+ (setting = settings_offset(md->name)) > 0)
{
- xmlNode *res;
- char *buf;
- int len;
-
- if (rec->which != Z_External_octet)
+ const char *val = session_setting_oneval(sdb, setting);
+ if (val && nparms < MAX_XSLT_ARGS)
{
- yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
- db->url);
- return 0;
+ char *buf;
+ int len = strlen(val);
+ buf = nmem_malloc(se->nmem, len + 3);
+ buf[0] = '\'';
+ strcpy(buf + 1, val);
+ buf[len+1] = '\'';
+ buf[len+2] = '\0';
+ parms[offset++] = md->name;
+ parms[offset++] = buf;
+ nparms++;
}
- buf = (char*) rec->u.octet_aligned->buf;
- len = rec->u.octet_aligned->len;
- if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0)
- {
- yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url);
- return 0;
- }
-
- yaz_marc_write_using_libxml2(sdb->yaz_marc, 1);
- if (yaz_marc_write_xml(sdb->yaz_marc, &res,
- "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
+ }
+ }
+ parms[offset] = 0;
+}
+
+// Add static values from session database settings if applicable
+static void insert_settings_values(struct session_database *sdb, xmlDoc *doc)
+{
+ struct conf_service *service = global_parameters.server->service;
+ int i;
+
+ for (i = 0; i < service->num_metadata; i++)
+ {
+ struct conf_metadata *md = &service->metadata[i];
+ int offset;
+
+ if (md->setting == Metadata_setting_postproc &&
+ (offset = settings_offset(md->name)) > 0)
+ {
+ const char *val = session_setting_oneval(sdb, offset);
+ if (val)
{
- yaz_log(YLOG_WARN, "Failed to encode as XML %s",
- db->url);
- return 0;
+ xmlNode *r = xmlDocGetRootElement(doc);
+ xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
+ (xmlChar *) val);
+ xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
}
- rdoc = xmlNewDoc((xmlChar *) "1.0");
- xmlDocSetRootElement(rdoc, res);
}
}
- else
+}
+
+xmlDoc *normalize_record(struct session_database *sdb, struct session *se,
+ const char *rec)
+{
+ struct database_retrievalmap *m;
+ xmlDoc *rdoc = record_to_xml(sdb, rec);
+ if (rdoc)
{
- char oid_name_buf[OID_STR_MAX];
- const char *oid_name = yaz_oid_to_string_buf(oid, 0, oid_name_buf);
- yaz_log(YLOG_FATAL,
- "Unable to handle record of type %s from %s",
- oid_name, db->url);
- return 0;
- }
-
- if (global_parameters.dump_records){
- fprintf(stderr,
- "Input Record (normalized) from %s\n----------------\n",
- db->url);
-#if LIBXML_VERSION >= 20600
- xmlDocFormatDump(stderr, rdoc, 1);
-#else
- xmlDocDump(stderr, rdoc);
-#endif
- }
+ for (m = sdb->map; m; m = m->next)
+ {
+ xmlDoc *new = 0;
+
+ {
+ xmlNodePtr root = 0;
+ char *parms[MAX_XSLT_ARGS*2+1];
- for (m = sdb->map; m; m = m->next){
- xmlDoc *new = 0;
+ insert_settings_parameters(sdb, se, parms);
- {
- xmlNodePtr root = 0;
- new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
- root= xmlDocGetRootElement(new);
- if (!new || !root || !(root->children))
- {
- yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
- sdb->database->url);
- xmlFreeDoc(new);
+ new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms);
+ root= xmlDocGetRootElement(new);
+ if (!new || !root || !(root->children))
+ {
+ yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
+ sdb->database->url);
+ xmlFreeDoc(new);
+ xmlFreeDoc(rdoc);
+ return 0;
+ }
+ }
+
xmlFreeDoc(rdoc);
- return 0;
- }
+ rdoc = new;
}
-
- xmlFreeDoc(rdoc);
- rdoc = new;
- }
- if (global_parameters.dump_records)
- {
- fprintf(stderr, "Record from %s\n----------------\n",
- sdb->database->url);
+
+ insert_settings_values(sdb, rdoc);
+
+ if (global_parameters.dump_records)
+ {
+ FILE *lf = yaz_log_file();
+
+ if (lf)
+ {
+ yaz_log(YLOG_LOG, "Normalized record from %s",
+ sdb->database->url);
#if LIBXML_VERSION >= 20600
- xmlDocFormatDump(stderr, rdoc, 1);
+ xmlDocFormatDump(lf, rdoc, 1);
#else
- xmlDocDump(stderr, rdoc);
+ xmlDocDump(lf, rdoc);
#endif
+ fprintf(lf, "\n");
+ }
+ }
}
return rdoc;
}
// Retrieve first defined value for 'name' for given database.
// Will be extended to take into account user associated with session
-char *session_setting_oneval(struct session_database *db, int offset)
+const char *session_setting_oneval(struct session_database *db, int offset)
{
if (!db->settings[offset])
return "";
return db->settings[offset]->value;
}
-
-
-// Initialize YAZ Map structures for MARC-based targets
-static int prepare_yazmarc(struct session_database *sdb)
-{
- char *s;
-
- if (!sdb->settings)
- {
- yaz_log(YLOG_WARN, "No settings for %s", sdb->database->url);
- return -1;
- }
- if ((s = session_setting_oneval(sdb, PZ_NATIVESYNTAX))
- && !strncmp(s, "iso2709", 7))
- {
- char *encoding = "marc-8s", *e;
- yaz_iconv_t cm;
-
- // See if a native encoding is specified
- if ((e = strchr(s, ';')))
- encoding = e + 1;
-
- sdb->yaz_marc = yaz_marc_create();
- yaz_marc_subfield_str(sdb->yaz_marc, "\t");
-
- cm = yaz_iconv_open("utf-8", encoding);
- if (!cm)
- {
- yaz_log(YLOG_FATAL,
- "Unable to map from %s to UTF-8 for target %s",
- encoding, sdb->database->url);
- return -1;
- }
- yaz_marc_iconv(sdb->yaz_marc, cm);
- }
- return 0;
-}
-
// Prepare XSLT stylesheets for record normalization
// Structures are allocated on the session_wide nmem to avoid having
// to recompute this for every search. This would lead
// setting. However, this is not a realistic use scenario.
static int prepare_map(struct session *se, struct session_database *sdb)
{
- char *s;
+ const char *s;
if (!sdb->settings)
{
char **stylesheets;
struct database_retrievalmap **m = &sdb->map;
int num, i;
+ char auto_stylesheet[256];
+ if (!strcmp(s, "auto"))
+ {
+ const char *request_syntax = session_setting_oneval(
+ sdb, PZ_REQUESTSYNTAX);
+ if (request_syntax)
+ {
+ char *cp;
+ yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
+ "%s.xsl", request_syntax);
+ for (cp = auto_stylesheet; *cp; cp++)
+ {
+ /* deliberately only consider ASCII */
+ if (*cp > 32 && *cp < 127)
+ *cp = tolower(*cp);
+ }
+ s = auto_stylesheet;
+ }
+ else
+ {
+ yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet");
+ }
+ }
nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
for (i = 0; i < num; i++)
{
(*m)->next = 0;
if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
{
- yaz_log(YLOG_FATAL, "Unable to load stylesheet: %s",
+ yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
stylesheets[i]);
return -1;
}
"No settings associated with %s", sdb->database->url);
return -1;
}
- if (sdb->settings[PZ_NATIVESYNTAX] && !sdb->yaz_marc)
- {
- if (prepare_yazmarc(sdb) < 0)
- return -1;
- }
if (sdb->settings[PZ_XSLT] && !sdb->map)
{
if (prepare_map(se, sdb) < 0)
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)
+{
+ struct session_watchentry *ent = data;
-void session_set_watch(struct session *s, int what,
- session_watchfun fun, void *data)
+ ent->fun = 0;
+ ent->data = 0;
+ ent->obs = 0;
+}
+
+// set watch. Returns 0=OK, -1 if watch is already set
+int session_set_watch(struct session *s, int what,
+ session_watchfun fun, void *data,
+ struct http_channel *chan)
{
+ if (s->watchlist[what].fun)
+ return -1;
s->watchlist[what].fun = fun;
s->watchlist[what].data = data;
+ s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
+ session_watch_cancel);
+ return 0;
}
void session_alert_watch(struct session *s, int what)
{
- if (!s->watchlist[what].fun)
- return;
- (*s->watchlist[what].fun)(s->watchlist[what].data);
- s->watchlist[what].fun = 0;
- s->watchlist[what].data = 0;
+ if (s->watchlist[what].fun)
+ {
+ /* our watch is no longer associated with http_channel */
+ void *data;
+ session_watchfun fun;
+
+ http_remove_observer(s->watchlist[what].obs);
+ fun = s->watchlist[what].fun;
+ data = s->watchlist[what].data;
+
+ /* reset watch before fun is invoked - in case fun wants to set
+ it again */
+ s->watchlist[what].fun = 0;
+ s->watchlist[what].data = 0;
+ s->watchlist[what].obs = 0;
+
+ fun(data);
+ }
}
//callback for grep_databases
const char **addinfo)
{
int live_channels = 0;
+ int no_working = 0;
+ int no_failed = 0;
struct client *cl;
struct database_criterion *criteria;
*addinfo = 0;
nmem_reset(se->nmem);
+ se->relevance = 0;
+ se->total_records = se->total_hits = se->total_merged = 0;
+ se->reclist = 0;
+ se->num_termlists = 0;
criteria = parse_filter(se->nmem, filter);
se->requestid++;
live_channels = select_targets(se, criteria);
if (live_channels)
{
int maxrecs = live_channels * global_parameters.toget;
- se->num_termlists = 0;
se->reclist = reclist_create(se->nmem, maxrecs);
- // This will be initialized in send_search()
- se->total_records = se->total_hits = se->total_merged = 0;
se->expected_maxrecs = maxrecs;
}
else
return PAZPAR2_NO_TARGETS;
- se->relevance = 0;
-
for (cl = se->clients; cl; cl = client_next_in_session(cl))
{
if (prepare_session_database(se, client_get_database(cl)) < 0)
*addinfo = client_get_database(cl)->database->url;
return PAZPAR2_CONFIG_TARGET;
}
- // Query must parse for all targets
+ // Parse query for target
if (client_parse_query(cl, query) < 0)
+ no_failed++;
+ else
{
- *addinfo = "query";
- return PAZPAR2_MALFORMED_PARAMETER_VALUE;
+ no_working++;
+ if (client_prep_connection(cl))
+ client_start_search(cl);
}
}
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
- client_prep_connection(cl);
-
+ // If no queries could be mapped, we signal an error
+ if (no_working == 0)
+ {
+ *addinfo = "query";
+ return PAZPAR2_MALFORMED_PARAMETER_VALUE;
+ }
return PAZPAR2_NO_ERROR;
}
int i;
new->database = db;
- new->yaz_marc = 0;
-#ifdef HAVE_ICU
- if (global_parameters.server && global_parameters.server->icu_chn)
- new->pct = pp2_charset_create(global_parameters.server->icu_chn);
- else
- new->pct = pp2_charset_create(0);
-#else // HAVE_ICU
- new->pct = pp2_charset_create(0);
-#endif // HAVE_ICU
-
new->map = 0;
new->settings
= nmem_malloc(se->session_nmem, sizeof(struct settings *) * num);
for (m = sdb->map; m; m = m->next)
xsltFreeStylesheet(m->stylesheet);
- if (sdb->yaz_marc)
- yaz_marc_destroy(sdb->yaz_marc);
- if (sdb->pct)
- pp2_charset_destroy(sdb->pct);
}
// Initialize session_database list -- this represents this session's view
void session_init_databases(struct session *se)
{
se->databases = 0;
- grep_databases(se, 0, session_init_databases_fun);
+ predef_grep_databases(se, 0, session_init_databases_fun);
}
// Probably session_init_databases_fun should be refactored instead of
}
// Jakub: This breaks the filter setting.
/*if (offset == PZ_ID)
- {
- yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring");
- return;
- }*/
+ {
+ yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring");
+ return;
+ }*/
new->precedence = 0;
new->target = dbname;
new->name = setting;
// (happens when a search starts and client connections are prepared)
switch (offset)
{
- case PZ_NATIVESYNTAX:
- if (sdb->yaz_marc)
- {
- yaz_marc_destroy(sdb->yaz_marc);
- sdb->yaz_marc = 0;
- }
- break;
- case PZ_XSLT:
- if (sdb->map)
- {
- struct database_retrievalmap *m;
- // We don't worry about the map structure -- it's in nmem
- for (m = sdb->map; m; m = m->next)
- xsltFreeStylesheet(m->stylesheet);
- sdb->map = 0;
- }
- break;
+ case PZ_XSLT:
+ if (sdb->map)
+ {
+ struct database_retrievalmap *m;
+ // We don't worry about the map structure -- it's in nmem
+ for (m = sdb->map; m; m = m->next)
+ xsltFreeStylesheet(m->stylesheet);
+ sdb->map = 0;
+ }
+ break;
}
}
session->session_nmem = nmem;
session->nmem = nmem_create();
session->wrbuf = wrbuf_alloc();
- session_init_databases(session);
+ session->databases = 0;
for (i = 0; i <= SESSION_WATCH_MAX; i++)
{
session->watchlist[i].data = 0;
session->watchlist[i].fun = 0;
}
-
return session;
}
-struct hitsbytarget *hitsbytarget(struct session *se, int *count)
+struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem)
{
- static struct hitsbytarget res[1000]; // FIXME MM
+ struct hitsbytarget *res = 0;
struct client *cl;
+ size_t sz = 0;
+
+ for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ sz++;
+ res = nmem_malloc(nmem, sizeof(*res) * sz);
*count = 0;
for (cl = se->clients; cl; cl = client_next_in_session(cl))
{
- char *name = session_setting_oneval(client_get_database(cl), PZ_NAME);
+ const char *name = session_setting_oneval(client_get_database(cl),
+ PZ_NAME);
res[*count].id = client_get_database(cl)->database->url;
res[*count].name = *name ? name : "Unknown";
res[*count].connected = client_get_connection(cl) ? 1 : 0;
(*count)++;
}
-
return res;
}
}
#endif
-struct record_cluster *show_single(struct session *s, int id)
+struct record_cluster *show_single(struct session *s, const char *id)
{
struct record_cluster *r;
reclist_rewind(s->reclist);
while ((r = reclist_read_record(s->reclist)))
- if (r->recid == id)
+ if (!strcmp(r->recid, id))
return r;
return 0;
}
NMEM nmem_show)
{
struct record_cluster **recs = nmem_malloc(nmem_show, *num
- * sizeof(struct record_cluster *));
+ * sizeof(struct record_cluster *));
struct reclist_sortparms *spp;
int i;
#if USE_TIMING
stat->num_no_connection++;
switch (client_get_state(cl))
{
- case Client_Connecting: stat->num_connecting++; break;
- case Client_Initializing: stat->num_initializing++; break;
- case Client_Searching: stat->num_searching++; break;
- case Client_Presenting: stat->num_presenting++; break;
- case Client_Idle: stat->num_idle++; break;
- case Client_Failed: stat->num_failed++; break;
- case Client_Error: stat->num_error++; break;
- default: break;
+ case Client_Connecting: stat->num_connecting++; break;
+ case Client_Working: stat->num_working++; break;
+ case Client_Idle: stat->num_idle++; break;
+ case Client_Failed: stat->num_failed++; break;
+ case Client_Error: stat->num_error++; break;
+ default: break;
}
count++;
}
stat->num_clients = count;
}
-void start_http_listener(void)
+int start_http_listener(void)
{
char hp[128] = "";
struct conf_server *ser = global_parameters.server;
sprintf(hp + strlen(hp), "%d", ser->port);
}
}
- http_init(hp);
+ return http_init(hp);
}
void start_proxy(void)
event_loop(&channel_list);
}
-struct record *ingest_record(struct client *cl, Z_External *rec,
+static struct record_metadata *record_metadata_init(
+ NMEM nmem, char *value, enum conf_metadata_type type)
+{
+ struct record_metadata *rec_md = record_metadata_create(nmem);
+ if (type == Metadata_type_generic)
+ {
+ char * p = value;
+ p = normalize7bit_generic(p, " ,/.:([");
+
+ rec_md->data.text.disp = nmem_strdup(nmem, p);
+ rec_md->data.text.sort = 0;
+ }
+ else if (type == Metadata_type_year || type == Metadata_type_date)
+ {
+ int first, last;
+ int longdate = 0;
+
+ if (type == Metadata_type_date)
+ longdate = 1;
+ if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
+ return 0;
+
+ rec_md->data.number.min = first;
+ rec_md->data.number.max = last;
+ }
+ else
+ return 0;
+ return rec_md;
+}
+
+struct record *ingest_record(struct client *cl, const char *rec,
int record_no)
{
- xmlDoc *xdoc = normalize_record(client_get_database(cl), rec);
+ xmlDoc *xdoc = normalize_record(client_get_database(cl),
+ client_get_session(cl), rec);
xmlNode *root, *n;
struct record *record;
struct record_cluster *cluster;
xmlChar *type = 0;
xmlChar *value = 0;
struct conf_service *service = global_parameters.server->service;
+ const char *norm_str = 0;
+ pp2_relevance_token_t prt = 0;
+ WRBUF norm_wr = 0;
if (!xdoc)
return 0;
xmlFreeDoc(xdoc);
return 0;
}
-
+
record = record_create(se->nmem,
service->num_metadata, service->num_sortkeys, cl,
record_no);
- mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
- xmlFree(mergekey);
- normalize7bit_mergekey((char *) mergekey_norm, 0);
+ prt = pp2_relevance_tokenize(
+ global_parameters.server->mergekey_pct, (const char *) mergekey);
+
+
+ norm_wr = wrbuf_alloc();
+
+ while ((norm_str = pp2_relevance_token_next(prt)))
+ {
+ if (*norm_str)
+ {
+ if (wrbuf_len(norm_wr))
+ wrbuf_puts(norm_wr, " ");
+ wrbuf_puts(norm_wr, norm_str);
+ }
+ }
+
+ mergekey_norm = (xmlChar *)nmem_strdup(se->nmem, wrbuf_cstr(norm_wr));
+ wrbuf_destroy(norm_wr);
+
+ pp2_relevance_token_destroy(prt);
+ xmlFree(mergekey);
+
cluster = reclist_insert(se->reclist,
global_parameters.server->service,
record, (char *) mergekey_norm,
&se->total_merged);
if (global_parameters.dump_records)
- yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
+ yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
client_get_database(cl)->database->url, record_no);
if (!cluster)
{
return 0;
}
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)
- {
- 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;
- int first, last;
-
- type = xmlGetProp(n, (xmlChar *) "type");
- value = xmlNodeListGetString(xdoc, n->children, 1);
-
- if (!type || !value)
- continue;
-
- md_field_id
- = conf_service_metadata_field_id(service, (const char *) type);
- if (md_field_id < 0)
- {
- yaz_log(YLOG_WARN,
- "Ignoring unknown metadata element: %s", type);
- 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];
- }
-
- // Find out where we are putting it - based on merge or not
- if (ser_md->merge == Metadata_merge_no)
- wheretoput = &record->metadata[md_field_id];
- else
- wheretoput = &cluster->metadata[md_field_id];
-
- // create new record_metadata
- rec_md = record_metadata_create(se->nmem);
-
- // and polulate with data:
- // type based charmapping decisions follow here
- if (ser_md->type == Metadata_type_generic)
- {
-
- char * p = (char *) value;
- p = normalize7bit_generic(p, " ,/.:([");
-
- rec_md->data.text = nmem_strdup(se->nmem, p);
-
- }
- else if (ser_md->type == Metadata_type_year)
- {
- if (extract7bit_years((char *) value, &first, &last) < 0)
- continue;
- }
- else
- {
- yaz_log(YLOG_WARN,
- "Unknown type in metadata element %s", type);
- continue;
- }
-
- // 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,
- rec_md->data.text))
- break;
- if (!mnode)
- {
- rec_md->next = *wheretoput;
- *wheretoput = rec_md;
- }
- }
- else if (ser_md->merge == Metadata_merge_longest)
- {
- if (!*wheretoput
- || strlen(rec_md->data.text)
- > strlen((*wheretoput)->data.text))
- {
- *wheretoput = rec_md;
- if (ser_sk)
- {
- char *s = nmem_strdup(se->nmem, rec_md->data.text);
- if (!cluster->sortkeys[sk_field_id])
- cluster->sortkeys[sk_field_id] =
- nmem_malloc(se->nmem,
- sizeof(union data_types));
- normalize7bit_mergekey(s,
- (ser_sk->type == Metadata_sortkey_skiparticle));
- cluster->sortkeys[sk_field_id]->text = s;
+
+
+ // now parsing XML record and adding data to cluster or record metadata
+ for (n = root->children; n; n = n->next)
+ {
+ 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)
+ {
+ yaz_log(YLOG_WARN,
+ "Ignoring unknown metadata element: %s", type);
+ 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, (char *) value,
+ ser_md->type);
+ if (!rec_md)
+ {
+ yaz_log(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;
+
+ // merged metadata
+ rec_md = record_metadata_init(se->nmem, (char *) value,
+ ser_md->type);
+ wheretoput = &cluster->metadata[md_field_id];
+
+ // and polulate with data:
+ // assign cluster or record based on merge action
+ if (ser_md->merge == Metadata_merge_no)
+ {
+ while (*wheretoput)
+ wheretoput = &(*wheretoput)->next;
+ *wheretoput = rec_md;
+ }
+ 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,
+ rec_md->data.text.disp))
+ break;
+ if (!mnode)
+ {
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
+ }
+ }
+ else if (ser_md->merge == Metadata_merge_longest)
+ {
+ if (!*wheretoput
+ || strlen(rec_md->data.text.disp)
+ > strlen((*wheretoput)->data.text.disp))
+ {
+ *wheretoput = rec_md;
+ if (ser_sk)
+ {
+ const char *sort_str = 0;
+ int skip_article =
+ ser_sk->type == Metadata_sortkey_skiparticle;
+
+ if (!cluster->sortkeys[sk_field_id])
+ cluster->sortkeys[sk_field_id] =
+ nmem_malloc(se->nmem,
+ sizeof(union data_types));
+
+ prt = pp2_relevance_tokenize(
+ global_parameters.server->sort_pct,
+ rec_md->data.text.disp);
+
+ pp2_relevance_token_next(prt);
+
+ sort_str = pp2_get_sort(prt, skip_article);
+
+ cluster->sortkeys[sk_field_id]->text.disp =
+ rec_md->data.text.disp;
+ if (!sort_str)
+ {
+ sort_str = rec_md->data.text.disp;
+ yaz_log(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);
}
}
}
- else if (ser_md->merge == Metadata_merge_all
- || ser_md->merge == Metadata_merge_no)
+ else if (ser_md->merge == Metadata_merge_all)
{
rec_md->next = *wheretoput;
*wheretoput = rec_md;
if (!*wheretoput)
{
*wheretoput = rec_md;
- (*wheretoput)->data.number.min = first;
- (*wheretoput)->data.number.max = last;
if (ser_sk)
cluster->sortkeys[sk_field_id]
= &rec_md->data;
}
else
{
- if (first < (*wheretoput)->data.number.min)
- (*wheretoput)->data.number.min = first;
- if (last > (*wheretoput)->data.number.max)
- (*wheretoput)->data.number.max = last;
+ int this_min = rec_md->data.number.min;
+ int this_max = rec_md->data.number.max;
+ if (this_min < (*wheretoput)->data.number.min)
+ (*wheretoput)->data.number.min = this_min;
+ if (this_max > (*wheretoput)->data.number.max)
+ (*wheretoput)->data.number.max = this_max;
}
-#ifdef GAGA
- if (ser_sk)
- {
- union data_types *sdata
- = cluster->sortkeys[sk_field_id];
- yaz_log(YLOG_LOG, "SK range: %d-%d",
- sdata->number.min, sdata->number.max);
- }
-#endif
}
if (ser_md->type == Metadata_type_year)
{
char year[64];
- sprintf(year, "%d", last);
+ sprintf(year, "%d", rec_md->data.number.max);
add_facet(se, (char *) type, year);
- if (first != last)
+ if (rec_md->data.number.max != rec_md->data.number.min)
{
- sprintf(year, "%d", first);
+ sprintf(year, "%d", rec_md->data.number.min);
add_facet(se, (char *) type, year);
}
}