-/* $Id: logic.c,v 1.21 2007-04-25 08:55:01 marc Exp $
+/* $Id: logic.c,v 1.45 2007-06-15 19:35:17 adam Exp $
Copyright (c) 2006-2007, Index Data.
This file is part of Pazpar2.
#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"
#include "database.h"
#include "client.h"
#include "settings.h"
+#include "normalize7bit.h"
+
+#define TERMLIST_HIGH_SCORE 25
#define MAX_CHUNK 15
{
"",
"",
- "",
- "",
- 0,
+ "",
0,
+ 0, /* dump_records */
+ 0, /* debug_mode */
30,
"81",
"Index Data PazPar2",
}
}
-char *normalize_mergekey(char *buf, int skiparticle)
-{
- char *p = buf, *pout = buf;
-
- if (skiparticle)
- {
- char firstword[64];
- char articles[] = "the den der die des an a "; // must end in space
-
- while (*p && !isalnum(*p))
- p++;
- pout = firstword;
- while (*p && *p != ' ' && pout - firstword < 62)
- *(pout++) = tolower(*(p++));
- *(pout++) = ' ';
- *(pout++) = '\0';
- if (!strstr(articles, firstword))
- p = buf;
- pout = buf;
- }
-
- while (*p)
- {
- while (*p && !isalnum(*p))
- p++;
- while (isalnum(*p))
- *(pout++) = tolower(*(p++));
- if (*p)
- *(pout++) = ' ';
- while (*p && !isalnum(*p))
- p++;
- }
- if (buf != pout)
- do {
- *(pout--) = '\0';
- }
- while (pout > buf && *pout == ' ');
-
- return buf;
-}
-
-// Extract what appears to be years from buf, storing highest and
-// lowest values.
-static int extract_years(const char *buf, int *first, int *last)
-{
- *first = -1;
- *last = -1;
- while (*buf)
- {
- const char *e;
- int len;
-
- while (*buf && !isdigit(*buf))
- buf++;
- len = 0;
- for (e = buf; *e && isdigit(*e); e++)
- len++;
- if (len == 4)
- {
- int value = atoi(buf);
- if (*first < 0 || value < *first)
- *first = value;
- if (*last < 0 || value > *last)
- *last = value;
- }
- buf = e;
- }
- return *first;
-}
static void add_facet(struct session *s, const char *type, const char *value)
s->termlists[i].name = nmem_strdup(s->nmem, type);
s->termlists[i].termlist
- = termlist_create(s->nmem, s->expected_maxrecs, 15);
+ = termlist_create(s->nmem, s->expected_maxrecs,
+ TERMLIST_HIGH_SCORE);
s->num_termlists = i + 1;
}
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, Z_External *rec)
{
- struct database_retrievalmap *m;
struct database *db = sdb->database;
- xmlNode *res;
- xmlDoc *rdoc;
+ xmlDoc *rdoc = 0;
+ const Odr_oid *oid = rec->direct_reference;
- // First normalize to XML
- if (sdb->yaz_marc)
+ /* convert response record to XML somehow */
+ if (rec->which == Z_External_octet && oid
+ && !oid_oidcmp(oid, yaz_oid_recsyn_xml))
{
- char *buf;
- int len;
- if (rec->which != Z_External_octet)
+ /* xml already */
+ rdoc = xmlParseMemory((char*) rec->u.octet_aligned->buf,
+ rec->u.octet_aligned->len);
+ if (!rdoc)
{
- yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
+ yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
db->url);
return 0;
}
- buf = (char*) rec->u.octet_aligned->buf;
- len = rec->u.octet_aligned->len;
- if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0)
+ }
+ else if (oid && yaz_oid_is_iso2709(oid))
+ {
+ /* ISO2709 gets converted to MARCXML */
+ if (!sdb->yaz_marc)
{
- yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url);
+ yaz_log(YLOG_FATAL, "Unable to handle ISO2709 record");
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)
+ else
{
- yaz_log(YLOG_WARN, "Failed to encode as XML %s",
- db->url);
- return 0;
+ xmlNode *res;
+ char *buf;
+ int len;
+
+ if (rec->which != Z_External_octet)
+ {
+ yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
+ db->url);
+ return 0;
+ }
+ 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)
+ {
+ yaz_log(YLOG_WARN, "Failed to encode as XML %s",
+ db->url);
+ return 0;
+ }
+ rdoc = xmlNewDoc((xmlChar *) "1.0");
+ xmlDocSetRootElement(rdoc, res);
}
- rdoc = xmlNewDoc((xmlChar *) "1.0");
- xmlDocSetRootElement(rdoc, res);
-
}
else
{
+ 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,
- "Unknown native_syntax in normalize_record from %s",
- db->url);
+ "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",
xmlDocDump(stderr, rdoc);
#endif
}
+ return rdoc;
+}
- for (m = sdb->map; m; m = m->next){
- xmlDoc *new = 0;
-
-#if 1
- {
- 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);
+xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec)
+{
+ struct database_retrievalmap *m;
+ xmlDoc *rdoc = record_to_xml(sdb, rec);
+ if (rdoc)
+ {
+ for (m = sdb->map; m; m = m->next){
+ xmlDoc *new = 0;
+
+ {
+ 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);
+ xmlFreeDoc(rdoc);
+ return 0;
+ }
+ }
+
xmlFreeDoc(rdoc);
- return 0;
- }
+ rdoc = new;
}
-#endif
-
-#if 0
- // do it another way to detect transformation errors right now
- // but does not seem to work either!
+ if (global_parameters.dump_records)
{
- xsltTransformContextPtr ctxt;
- ctxt = xsltNewTransformContext(m->stylesheet, rdoc);
- new = xsltApplyStylesheetUser(m->stylesheet, rdoc, 0, 0, 0, ctxt);
- if ((ctxt->state == XSLT_STATE_ERROR) ||
- (ctxt->state == XSLT_STATE_STOPPED)){
- yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
- cl->database->database->url);
- xmlFreeDoc(new);
- xmlFreeDoc(rdoc);
- return 0;
- }
- }
-#endif
-
- xmlFreeDoc(rdoc);
- rdoc = new;
- }
- if (global_parameters.dump_records)
- {
- fprintf(stderr, "Record from %s\n----------------\n",
- sdb->database->url);
+ fprintf(stderr, "Record from %s\n----------------\n",
+ sdb->database->url);
#if LIBXML_VERSION >= 20600
- xmlDocFormatDump(stderr, rdoc, 1);
+ xmlDocFormatDump(stderr, rdoc, 1);
#else
- xmlDocDump(stderr, rdoc);
+ xmlDocDump(stderr, rdoc);
#endif
+ }
}
return rdoc;
}
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)
+{
+ 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;
+ http_remove_observer(s->watchlist[what].obs);
(*s->watchlist[what].fun)(s->watchlist[what].data);
s->watchlist[what].fun = 0;
s->watchlist[what].data = 0;
+ s->watchlist[what].obs = 0;
}
//callback for grep_databases
return res;
}
-char *search(struct session *se, char *query, char *filter)
+enum pazpar2_error_code search(struct session *se,
+ char *query, char *filter,
+ const char **addinfo)
{
int live_channels = 0;
struct client *cl;
yaz_log(YLOG_DEBUG, "Search");
+ *addinfo = 0;
nmem_reset(se->nmem);
criteria = parse_filter(se->nmem, filter);
se->requestid++;
se->expected_maxrecs = maxrecs;
}
else
- return "NOTARGETS";
+ 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)
- return "CONFIG_ERROR";
+ {
+ *addinfo = client_get_database(cl)->database->url;
+ return PAZPAR2_CONFIG_TARGET;
+ }
// Query must parse for all targets
- if (client_parse_query(cl, query) < 0)
- return "QUERY";
+ if (client_parse_query(cl, query) < 0)
+ {
+ *addinfo = "query";
+ return PAZPAR2_MALFORMED_PARAMETER_VALUE;
+ }
}
for (cl = se->clients; cl; cl = client_next_in_session(cl))
- {
client_prep_connection(cl);
- }
- return 0;
+ return PAZPAR2_NO_ERROR;
}
// Creates a new session_database object for a database
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);
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
{
struct session_database *sdb = find_session_database(se, dbname);
struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
- int offset = settings_offset(setting);
+ int offset = settings_offset_cprefix(setting);
if (offset < 0)
{
yaz_log(YLOG_WARN, "Unknown setting %s", setting);
return;
}
+ // Jakub: This breaks the filter setting.
+ /*if (offset == PZ_ID)
+ {
+ yaz_log(YLOG_WARN, "No need to set pz:id setting. Ignoring");
+ return;
+ }*/
new->precedence = 0;
new->target = dbname;
new->name = setting;
{
struct session_database *sdb;
- yaz_log(YLOG_LOG, "Destroying session");
while (s->clients)
client_destroy(s->clients);
for (sdb = s->databases; sdb; sdb = sdb->next)
yaz_log(YLOG_DEBUG, "New Pazpar2 session");
+ session->relevance = 0;
session->total_hits = 0;
session->total_records = 0;
session->num_termlists = 0;
yaz_timing_t t = yaz_timing_create();
#endif
- for (spp = sp; spp; spp = spp->next)
- if (spp->type == Metadata_sortkey_relevance)
- {
- relevance_prepare_read(s->relevance, s->reclist);
- break;
- }
- reclist_sort(s->reclist, sp);
-
- *total = s->reclist->num_records;
- *sumhits = s->total_hits;
-
- for (i = 0; i < start; i++)
- if (!reclist_read_record(s->reclist))
- {
- *num = 0;
- recs = 0;
- break;
- }
-
- for (i = 0; i < *num; i++)
+ if (!s->relevance)
{
- struct record_cluster *r = reclist_read_record(s->reclist);
- if (!r)
+ *num = 0;
+ *total = 0;
+ *sumhits = 0;
+ recs = 0;
+ }
+ else
+ {
+ for (spp = sp; spp; spp = spp->next)
+ if (spp->type == Metadata_sortkey_relevance)
+ {
+ relevance_prepare_read(s->relevance, s->reclist);
+ break;
+ }
+ reclist_sort(s->reclist, sp);
+
+ *total = s->reclist->num_records;
+ *sumhits = s->total_hits;
+
+ for (i = 0; i < start; i++)
+ if (!reclist_read_record(s->reclist))
+ {
+ *num = 0;
+ recs = 0;
+ break;
+ }
+
+ for (i = 0; i < *num; i++)
{
- *num = i;
- break;
+ struct record_cluster *r = reclist_read_record(s->reclist);
+ if (!r)
+ {
+ *num = i;
+ break;
+ }
+ recs[i] = r;
}
- recs[i] = r;
}
#if USE_TIMING
yaz_timing_stop(t);
http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
}
-void start_zproxy(void)
-{
- struct conf_server *ser = global_parameters.server;
-
- if (*global_parameters.zproxy_override){
- yaz_log(YLOG_LOG, "Z39.50 proxy %s",
- global_parameters.zproxy_override);
- return;
- }
-
- else if (ser->zproxy_host || ser->zproxy_port)
- {
- char hp[128] = "";
-
- strcpy(hp, ser->zproxy_host ? ser->zproxy_host : "");
- if (ser->zproxy_port)
- {
- if (*hp)
- strcat(hp, ":");
- else
- strcat(hp, "@:");
-
- sprintf(hp + strlen(hp), "%d", ser->zproxy_port);
- }
- strcpy(global_parameters.zproxy_override, hp);
- yaz_log(YLOG_LOG, "Z39.50 proxy %s",
- global_parameters.zproxy_override);
-
- }
- else
- return;
-}
// Master list of connections we're handling events to
static IOCHAN channel_list = 0;
{
xmlDoc *xdoc = normalize_record(client_get_database(cl), rec);
xmlNode *root, *n;
- struct record *res;
+ struct record *record;
struct record_cluster *cluster;
struct session *se = client_get_session(cl);
xmlChar *mergekey, *mergekey_norm;
return 0;
}
- res = nmem_malloc(se->nmem, sizeof(struct record));
- res->next = 0;
- res->client = cl;
- res->metadata = nmem_malloc(se->nmem,
- sizeof(struct record_metadata*) * service->num_metadata);
- memset(res->metadata, 0,
- sizeof(struct record_metadata*) * service->num_metadata);
+ 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);
- normalize_mergekey((char *) mergekey_norm, 0);
+ normalize7bit_mergekey((char *) mergekey_norm, 0);
cluster = reclist_insert(se->reclist,
global_parameters.server->service,
- res, (char *) mergekey_norm,
+ record, (char *) mergekey_norm,
&se->total_merged);
if (global_parameters.dump_records)
yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
}
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)
continue;
if (!strcmp((const char *) n->name, "metadata"))
{
- struct conf_metadata *md = 0;
- struct conf_sortkey *sk = 0;
- struct record_metadata **wheretoput, *newm;
- int imeta;
+ 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, 0);
+ value = xmlNodeListGetString(xdoc, n->children, 1);
if (!type || !value)
continue;
- // First, find out what field we're looking at
- for (imeta = 0; imeta < service->num_metadata; imeta++)
- if (!strcmp((const char *) type,
- service->metadata[imeta].name))
- {
- md = &service->metadata[imeta];
- if (md->sortkey_offset >= 0)
- sk = &service->sortkeys[md->sortkey_offset];
- break;
- }
- if (!md)
+ 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;
}
- // Find out where we are putting it
- if (md->merge == Metadata_merge_no)
- wheretoput = &res->metadata[imeta];
+ 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[imeta];
+ wheretoput = &cluster->metadata[md_field_id];
- // Put it there
- newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
- newm->next = 0;
- if (md->type == Metadata_type_generic)
+ // 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, *pe;
- for (p = (char *) value; *p && isspace(*p); p++)
- ;
- for (pe = p + strlen(p) - 1;
- pe > p && strchr(" ,/.:([", *pe); pe--)
- *pe = '\0';
- newm->data.text = nmem_strdup(se->nmem, p);
+
+ char * p = (char *) value;
+ p = normalize7bit_generic(p, " ,/.:([");
+
+ rec_md->data.text = nmem_strdup(se->nmem, p);
}
- else if (md->type == Metadata_type_year)
+ else if (ser_md->type == Metadata_type_year)
{
- if (extract_years((char *) value, &first, &last) < 0)
+ if (extract7bit_years((char *) value, &first, &last) < 0)
continue;
}
else
"Unknown type in metadata element %s", type);
continue;
}
- if (md->type == Metadata_type_year
- && md->merge != Metadata_merge_range)
- {
- yaz_log(YLOG_WARN, "Only range merging supported for years");
- continue;
- }
- if (md->merge == Metadata_merge_unique)
+
+ // 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,
- newm->data.text))
+ rec_md->data.text))
break;
if (!mnode)
{
- newm->next = *wheretoput;
- *wheretoput = newm;
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
}
}
- else if (md->merge == Metadata_merge_longest)
+ else if (ser_md->merge == Metadata_merge_longest)
{
if (!*wheretoput
- || strlen(newm->data.text)
+ || strlen(rec_md->data.text)
> strlen((*wheretoput)->data.text))
{
- *wheretoput = newm;
- if (sk)
+ *wheretoput = rec_md;
+ if (ser_sk)
{
- char *s = nmem_strdup(se->nmem, newm->data.text);
- if (!cluster->sortkeys[md->sortkey_offset])
- cluster->sortkeys[md->sortkey_offset] =
+ 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));
- normalize_mergekey(s,
- (sk->type == Metadata_sortkey_skiparticle));
- cluster->sortkeys[md->sortkey_offset]->text = s;
+ normalize7bit_mergekey(s,
+ (ser_sk->type == Metadata_sortkey_skiparticle));
+ cluster->sortkeys[sk_field_id]->text = s;
}
}
}
- else if (md->merge == Metadata_merge_all
- || md->merge == Metadata_merge_no)
+ else if (ser_md->merge == Metadata_merge_all
+ || ser_md->merge == Metadata_merge_no)
{
- newm->next = *wheretoput;
- *wheretoput = newm;
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
}
- else if (md->merge == Metadata_merge_range)
+ else if (ser_md->merge == Metadata_merge_range)
{
- assert(md->type == Metadata_type_year);
if (!*wheretoput)
{
- *wheretoput = newm;
+ *wheretoput = rec_md;
(*wheretoput)->data.number.min = first;
(*wheretoput)->data.number.max = last;
- if (sk)
- cluster->sortkeys[md->sortkey_offset] = &newm->data;
+ if (ser_sk)
+ cluster->sortkeys[sk_field_id]
+ = &rec_md->data;
}
else
{
(*wheretoput)->data.number.max = last;
}
#ifdef GAGA
- if (sk)
+ if (ser_sk)
{
union data_types *sdata
- = cluster->sortkeys[md->sortkey_offset];
+ = cluster->sortkeys[sk_field_id];
yaz_log(YLOG_LOG, "SK range: %d-%d",
sdata->number.min, sdata->number.max);
}
#endif
}
- else
- yaz_log(YLOG_WARN,
- "Don't know how to merge on element name %s",
- md->name);
- if (md->rank)
+
+ // ranking of _all_ fields enabled ...
+ if (ser_md->rank)
relevance_countwords(se->relevance, cluster,
- (char *) value, md->rank);
- if (md->termlist)
+ (char *) value, ser_md->rank);
+
+ // construct facets ...
+ if (ser_md->termlist)
{
- if (md->type == Metadata_type_year)
+ if (ser_md->type == Metadata_type_year)
{
char year[64];
sprintf(year, "%d", last);
else
add_facet(se, (char *) type, (char *) value);
}
+
+ // cleaning up
xmlFree(type);
xmlFree(value);
type = value = 0;
relevance_donerecord(se->relevance, cluster);
se->total_records++;
- return res;
+ return record;
}