-/* $Id: logic.c,v 1.21 2007-04-25 08:55:01 marc Exp $
+/* $Id: logic.c,v 1.26 2007-04-27 12:17:04 marc 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
}
}
-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)
return 0;
}
}
-#endif
-
-#if 0
+#else
// do it another way to detect transformation errors right now
// but does not seem to work either!
{
{
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,
+#if 0
+ record = nmem_malloc(se->nmem, sizeof(struct record));
+ record->next = 0;
+ record->client = cl;
+ record->metadata = nmem_malloc(se->nmem,
sizeof(struct record_metadata*) * service->num_metadata);
- memset(res->metadata, 0,
+ memset(record->metadata, 0,
sizeof(struct record_metadata*) * service->num_metadata);
+#else
+ record = record_create(se->nmem,
+ service->num_metadata, service->num_sortkeys);
+ record_assign_client(record, cl);
+#endif
+
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");
if (!type || !value)
continue;
+#if 0
// First, find out what field we're looking at
- for (imeta = 0; imeta < service->num_metadata; imeta++)
+ for (md_field_id = 0; md_field_id < service->num_metadata;
+ md_field_id++)
if (!strcmp((const char *) type,
- service->metadata[imeta].name))
+ service->metadata[md_field_id].name))
{
- md = &service->metadata[imeta];
- if (md->sortkey_offset >= 0)
- sk = &service->sortkeys[md->sortkey_offset];
+ 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];
+ }
break;
}
- if (!md)
+
+ if (!ser_md)
{
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];
+#else
+ 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];
+ }
+#endif
+
+ // 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
+#if 0
+ rec_md = nmem_malloc(se->nmem, sizeof(struct record_metadata));
+ rec_md->next = 0;
+#else
+ rec_md = record_metadata_create(se->nmem);
+#endif
+
+ // and polulate with data:
+ // type based charmapping decisions follow here
+ if (ser_md->type == Metadata_type_generic)
{
+
+#if 0
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);
+#else
+ char * p = (char *) value;
+ p = normalize7bit_generic(p, " ,/.:([");
+#endif
+
+ 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)
+
+#if 0 // this test does not belong here.
+ // the condition is enforced in the constructor
+ // inside conf_metadata_assign()
+ // and will _never_ occur
+ if (ser_md->type == Metadata_type_year
+ && ser_md->merge != Metadata_merge_range)
{
yaz_log(YLOG_WARN, "Only range merging supported for years");
continue;
}
- if (md->merge == Metadata_merge_unique)
+# endif
+
+
+ // 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 0 // this assert does not belong here.
+ // the condition is enforced in the constructor
+ // inside conf_metadata_assign()
+ // and will _never_ occur
+ assert(ser_md->type == Metadata_type_year);
+#endif
+
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
}
+
+#if 0 // this else is only entered when Metadata_merge_no
+ // but I believe then we should _not_ pollute with log messages
else
+
yaz_log(YLOG_WARN,
"Don't know how to merge on element name %s",
- md->name);
+ ser_md->name);
+#endif
+
- 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;
}