/* This file is part of Pazpar2.
- Copyright (C) 2006-2013 Index Data
+ Copyright (C) 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
// Add static values from session database settings if applicable
static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
- struct conf_service *service)
+ xmlNode *root,
+ struct conf_service *service)
{
int i;
const char *val = session_setting_oneval(sdb, offset);
if (val)
{
- xmlNode *r = xmlDocGetRootElement(doc);
- xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
+ xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
(xmlChar *) val);
xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
}
{
session_log(se, YLOG_WARN, "Normalize failed");
}
- else
- {
- insert_settings_values(sdb, rdoc, service);
-
- if (global_parameters.dump_records)
- {
- session_log(se, YLOG_LOG, "Normalized record from %s",
- sdb->database->id);
- log_xml_doc(rdoc);
- }
- }
}
return rdoc;
}
void session_destroy(struct session *se)
{
struct session_database *sdb;
- session_log(se, YLOG_DEBUG, "Destroying");
+ session_log(se, YLOG_LOG, "destroy");
session_use(-1);
session_remove_cached_clients(se);
session->normalize_cache = normalize_cache_create();
session->session_mutex = 0;
pazpar2_mutex_create(&session->session_mutex, tmp_str);
+ session_log(session, YLOG_LOG, "create");
+
session_use(1);
return session;
}
if (se->relevance)
{
for (spp = sp; spp; spp = spp->next)
- if (spp->type == Metadata_sortkey_relevance)
+ if (spp->type == Metadata_type_relevance)
{
relevance_prepare_read(se->relevance, se->reclist);
break;
}
*attrp = 0;
- if (type == Metadata_type_generic)
+ switch (type)
{
- char *p = nmem_strdup(nmem, value);
-
- p = normalize7bit_generic(p, " ,/.:([");
-
- rec_md->data.text.disp = p;
+ case Metadata_type_generic:
+ case Metadata_type_skiparticle:
+ rec_md->data.text.disp =
+ normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
rec_md->data.text.sort = 0;
rec_md->data.text.snippet = 0;
- }
- else if (type == Metadata_type_year || type == Metadata_type_date)
+ break;
+ case Metadata_type_year:
+ case Metadata_type_date:
{
int first, last;
int longdate = 0;
rec_md->data.number.min = first;
rec_md->data.number.max = last;
}
- else
+ break;
+ case Metadata_type_float:
+ rec_md->data.fnumber = atof(value);
+ break;
+ case Metadata_type_relevance:
+ case Metadata_type_position:
return 0;
+ }
return rec_md;
}
return no_found;
}
-static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
+static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
+ struct client *cl, int record_no,
struct conf_service *service, NMEM nmem,
const char *session_mergekey)
{
char *mergekey_norm = 0;
- xmlNode *root = xmlDocGetRootElement(doc);
WRBUF norm_wr = wrbuf_alloc();
xmlChar *mergekey;
xmlDoc *xdoc,
xmlNode *root,
int record_no,
- const char *mergekey_norm);
+ struct record_metadata_attr *mergekey);
static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
int record_no, NMEM nmem,
- struct session_database *sdb)
+ struct session_database *sdb,
+ struct record_metadata_attr *mergekeys)
{
int ret = 0;
- const char *mergekey_norm;
struct session *se = client_get_session(cl);
struct conf_service *service = se->service;
- if (!check_record_filter(root, sdb))
- {
- session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
- xmlFreeDoc(xdoc);
- return -2;
- }
+ insert_settings_values(sdb, xdoc, root, service);
- mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem,
- se->mergekey);
- if (!mergekey_norm)
+ if (!check_record_filter(root, sdb))
{
- session_log(se, YLOG_WARN, "Got no mergekey");
- xmlFreeDoc(xdoc);
- return -1;
+ session_log(se, YLOG_LOG,
+ "Filtered out record no %d from %s",
+ record_no, sdb->database->id);
+ return 0;
}
session_enter(se, "ingest_sub_record");
if (client_get_session(cl) == se && se->relevance)
- ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
+ ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
session_leave(se, "ingest_sub_record");
- xmlFreeDoc(xdoc);
return ret;
}
struct session_database *sdb = client_get_database(cl);
struct conf_service *service = se->service;
xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
+ int r = 0;
xmlNode *root;
if (!xdoc)
return -1;
+ if (global_parameters.dump_records)
+ {
+ session_log(se, YLOG_LOG, "Normalized record from %s",
+ sdb->database->id);
+ log_xml_doc(xdoc);
+ }
+
root = xmlDocGetRootElement(xdoc);
if (!strcmp((const char *) root->name, "cluster"))
{
- root = root->children;
- return ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb);
+ int no_merge_keys = 0;
+ int no_merge_dups = 0;
+ xmlNode *sroot;
+ struct record_metadata_attr *mk = 0;
+
+ for (sroot = root->children; sroot; sroot = sroot->next)
+ if (sroot->type == XML_ELEMENT_NODE &&
+ !strcmp((const char *) sroot->name, "record"))
+ {
+ struct record_metadata_attr **mkp;
+ const char *mergekey_norm =
+ get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
+ se->mergekey);
+ if (!mergekey_norm)
+ {
+ r = -1;
+ break;
+ }
+ for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
+ if (!strcmp((*mkp)->value, mergekey_norm))
+ break;
+ if (!*mkp)
+ {
+ *mkp = (struct record_metadata_attr*)
+ nmem_malloc(nmem, sizeof(**mkp));
+ (*mkp)->name = 0;
+ (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
+ (*mkp)->next = 0;
+ no_merge_keys++;
+ }
+ else
+ no_merge_dups++;
+ }
+ if (no_merge_keys > 1 || no_merge_dups > 0)
+ {
+ yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
+ no_merge_keys, no_merge_dups, record_no);
+ }
+ for (sroot = root->children; !r && sroot; sroot = sroot->next)
+ if (sroot->type == XML_ELEMENT_NODE &&
+ !strcmp((const char *) sroot->name, "record"))
+ {
+ r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
+ mk);
+ }
}
else if (!strcmp((const char *) root->name, "record"))
{
- return ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb);
+ const char *mergekey_norm =
+ get_mergekey(xdoc, root, cl, record_no, service, nmem,
+ se->mergekey);
+ if (mergekey_norm)
+ {
+ struct record_metadata_attr *mk = (struct record_metadata_attr*)
+ nmem_malloc(nmem, sizeof(*mk));
+ mk->name = 0;
+ mk->value = nmem_strdup(nmem, mergekey_norm);
+ mk->next = 0;
+
+ r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
+ }
}
else
{
session_log(se, YLOG_WARN, "Bad pz root element: %s",
(const char *) root->name);
- return -1;
+ r = -1;
}
+ xmlFreeDoc(xdoc);
+ return r;
}
xmlDoc *xdoc,
xmlNode *root,
int record_no,
- const char *mergekey_norm)
+ struct record_metadata_attr *merge_keys)
{
xmlNode *n;
xmlChar *type = 0;
xmlFree(value);
return -2;
}
- cluster = reclist_insert(se->reclist, service, record,
- mergekey_norm, &se->total_merged);
+ cluster = reclist_insert(se->reclist, se->relevance, service, record,
+ merge_keys, &se->total_merged);
if (!cluster)
+ {
+ if (type)
+ xmlFree(type);
+ if (value)
+ xmlFree(value);
return 0; // complete match with existing record
+ }
{
const char *use_term_factor_str =
session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
sdb->database->id, record_no);
-
- relevance_newrec(se->relevance, cluster);
-
// original metadata, to check if first existence of a field
metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
memcpy(metadata0, cluster->metadata,
{
const char *sort_str = 0;
int skip_article =
- ser_sk->type == Metadata_sortkey_skiparticle;
+ ser_sk->type == Metadata_type_skiparticle;
if (!cluster->sortkeys[sk_field_id])
cluster->sortkeys[sk_field_id] =
projects / pazpar2-moved-to-github.git / blobdiff