-/* $Id: logic.c,v 1.22 2007-04-26 11:03:54 marc Exp $
+/* $Id: logic.c,v 1.35 2007-06-02 04:32:28 quinn Exp $
Copyright (c) 2006-2007, Index Data.
This file is part of Pazpar2.
#include "database.h"
#include "client.h"
#include "settings.h"
+#include "normalize7bit.h"
#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)
{
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",
for (m = sdb->map; m; m = m->next){
xmlDoc *new = 0;
-#if 1
{
xmlNodePtr root = 0;
new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
return 0;
}
}
-#endif
-
-#if 0
- // do it another way to detect transformation errors right now
- // but does not seem to work either!
- {
- 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;
}
for (cl = se->clients; cl; cl = client_next_in_session(cl))
- {
client_prep_connection(cl);
- }
return 0;
}
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;
{
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);
+ record_assign_client(record, cl);
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)
{
struct conf_metadata *ser_md = 0;
struct conf_sortkey *ser_sk = 0;
- struct record_metadata **wheretoput, *newm;
+ struct record_metadata **wheretoput = 0;
+ struct record_metadata *rec_md = 0;
int md_field_id = -1;
int sk_field_id = -1;
int first, last;
if (!type || !value)
continue;
- // First, find out what field we're looking at
- for (md_field_id = 0; md_field_id < service->num_metadata;
- md_field_id++)
- if (!strcmp((const char *) type,
- service->metadata[md_field_id].name))
- {
- ser_md = &service->metadata[md_field_id];
- if (ser_md->sortkey_offset >= 0)
- ser_sk = &service->sortkeys[ser_md->sortkey_offset];
- break;
- }
- if (!ser_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
+ 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 = &res->metadata[md_field_id];
+ wheretoput = &record->metadata[md_field_id];
else
wheretoput = &cluster->metadata[md_field_id];
- // Put it there
- newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
- newm->next = 0;
+ // 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 (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 (ser_md->type == Metadata_type_year
- && ser_md->merge != Metadata_merge_range)
- {
- yaz_log(YLOG_WARN, "Only range merging supported for years");
- 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,
- newm->data.text))
+ rec_md->data.text))
break;
if (!mnode)
{
- newm->next = *wheretoput;
- *wheretoput = newm;
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
}
}
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;
+ *wheretoput = rec_md;
if (ser_sk)
{
- char *s = nmem_strdup(se->nmem, newm->data.text);
- if (!cluster->sortkeys[ser_md->sortkey_offset])
- cluster->sortkeys[ser_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,
+ normalize7bit_mergekey(s,
(ser_sk->type == Metadata_sortkey_skiparticle));
- cluster->sortkeys[ser_md->sortkey_offset]->text = s;
+ cluster->sortkeys[sk_field_id]->text = s;
}
}
}
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 (ser_md->merge == Metadata_merge_range)
{
- assert(ser_md->type == Metadata_type_year);
if (!*wheretoput)
{
- *wheretoput = newm;
+ *wheretoput = rec_md;
(*wheretoput)->data.number.min = first;
(*wheretoput)->data.number.max = last;
if (ser_sk)
- cluster->sortkeys[ser_md->sortkey_offset]
- = &newm->data;
+ 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",
- ser_md->name);
+
+ // ranking of _all_ fields enabled ...
if (ser_md->rank)
relevance_countwords(se->relevance, cluster,
(char *) value, ser_md->rank);
+
+ // construct facets ...
if (ser_md->termlist)
{
if (ser_md->type == Metadata_type_year)
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;
}