-/* $Id: logic.c,v 1.40 2007-06-11 12:08:23 adam Exp $
+/* $Id: logic.c,v 1.62 2007-08-17 12:39:11 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
+#include <yaz/snprintf.h>
#if HAVE_CONFIG_H
#include "cconfig.h"
100,
MAX_CHUNK,
0,
- 0
+ 0,
+ 180,
+ 30
};
-
// Recursively traverse query structure to extract terms.
void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
{
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;
xmlDoc *rdoc = 0;
const Odr_oid *oid = rec->direct_reference;
return 0;
}
}
+ else if (rec->which == Z_External_OPAC)
+ {
+ if (!sdb->yaz_marc)
+ {
+ yaz_log(YLOG_WARN, "MARC decoding not configured");
+ return 0;
+ }
+ else
+ {
+ /* OPAC gets converted to XML too */
+ WRBUF wrbuf_opac = wrbuf_alloc();
+ /* MARCXML inside the OPAC XML. Charset is in effect because we
+ use the yaz_marc handle */
+ yaz_marc_xml(sdb->yaz_marc, YAZ_MARC_MARCXML);
+ yaz_opac_decode_wrbuf(sdb->yaz_marc, rec->u.opac, wrbuf_opac);
+
+ rdoc = xmlParseMemory((char*) wrbuf_buf(wrbuf_opac),
+ wrbuf_len(wrbuf_opac));
+ if (!rdoc)
+ {
+ yaz_log(YLOG_WARN, "Unable to parse OPAC XML");
+ /* Was used to debug bug #1348 */
+#if 0
+ FILE *f = fopen("/tmp/opac.xml.txt", "wb");
+ if (f)
+ {
+ fwrite(wrbuf_buf(wrbuf_opac), 1, wrbuf_len(wrbuf_opac), f);
+ fclose(f);
+ }
+#endif
+ }
+ wrbuf_destroy(wrbuf_opac);
+ }
+ }
else if (oid && yaz_oid_is_iso2709(oid))
{
/* ISO2709 gets converted to MARCXML */
if (!sdb->yaz_marc)
{
- yaz_log(YLOG_FATAL, "Unable to handle ISO2709 record");
+ yaz_log(YLOG_WARN, "MARC decoding not configured");
return 0;
}
else
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)
{
return 0;
}
- if (global_parameters.dump_records){
- fprintf(stderr,
- "Input Record (normalized) from %s\n----------------\n",
- db->url);
+ if (global_parameters.dump_records)
+ {
+ FILE *lf = yaz_log_file();
+ if (lf)
+ {
+ yaz_log(YLOG_LOG, "Un-normalized record from %s", db->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;
+}
- for (m = sdb->map; m; m = m->next){
- xmlDoc *new = 0;
+#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)
{
- 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;
+ const char *val = session_setting_oneval(sdb, setting);
+ if (val && nparms < MAX_XSLT_ARGS)
+ {
+ 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++;
+ }
}
+ }
+ 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)
+ {
+ xmlNode *r = xmlDocGetRootElement(doc);
+ xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
+ (xmlChar *) val);
+ xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
+ }
}
-
- xmlFreeDoc(rdoc);
- rdoc = new;
}
- if (global_parameters.dump_records)
+}
+
+xmlDoc *normalize_record(struct session_database *sdb, struct session *se,
+ Z_External *rec)
+{
+ struct database_retrievalmap *m;
+ xmlDoc *rdoc = record_to_xml(sdb, rec);
+ if (rdoc)
{
- fprintf(stderr, "Record from %s\n----------------\n",
- sdb->database->url);
+ for (m = sdb->map; m; m = m->next)
+ {
+ xmlDoc *new = 0;
+
+ {
+ xmlNodePtr root = 0;
+ char *parms[MAX_XSLT_ARGS*2+1];
+
+ insert_settings_parameters(sdb, se, parms);
+
+ 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);
+ rdoc = new;
+ }
+
+ 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 "";
// Initialize YAZ Map structures for MARC-based targets
static int prepare_yazmarc(struct session_database *sdb)
{
- char *s;
+ const char *s;
if (!sdb->settings)
{
// 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;
}
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;
+
+ ent->fun = 0;
+ ent->data = 0;
+ ent->obs = 0;
+}
-void session_set_watch(struct session *s, int what,
- session_watchfun fun, void *data)
+// 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;
+ int no_working = 0;
+ int no_failed = 0;
struct client *cl;
struct database_criterion *criteria;
yaz_log(YLOG_DEBUG, "Search");
+ *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 "NOTARGETS";
-
- se->relevance = 0;
+ return PAZPAR2_NO_TARGETS;
for (cl = se->clients; cl; cl = client_next_in_session(cl))
{
if (prepare_session_database(se, client_get_database(cl)) < 0)
- return "CONFIG_ERROR";
- // Query must parse for all targets
- if (client_parse_query(cl, query) < 0)
- return "QUERY";
+ {
+ *addinfo = client_get_database(cl)->database->url;
+ return PAZPAR2_CONFIG_TARGET;
+ }
+ // Parse query for target
+ if (client_parse_query(cl, query) < 0)
+ no_failed++;
+ else
+ {
+ no_working++;
+ client_prep_connection(cl);
+ }
}
- for (cl = se->clients; cl; cl = client_next_in_session(cl))
- client_prep_connection(cl);
-
- return 0;
+ // 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;
}
// Creates a new session_database object for a database
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
{
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;
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;
*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";
}
#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;
}
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)
+ {
+ *num = 0;
+ *total = 0;
+ *sumhits = 0;
+ recs = 0;
+ }
+ else
{
- struct record_cluster *r = reclist_read_record(s->reclist);
- if (!r)
+ 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);
event_loop(&channel_list);
}
+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 = nmem_strdup(nmem, p);
+ }
+ else if (type == Metadata_type_year)
+ {
+ int first, last;
+ if (extract7bit_years((char *) value, &first, &last) < 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, Z_External *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;
}
record = record_create(se->nmem,
- service->num_metadata, service->num_sortkeys, cl);
+ service->num_metadata, service->num_sortkeys, cl,
+ record_no);
mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
xmlFree(mergekey);
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)
{
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)
+ if (!type || !value || !*value)
continue;
md_field_id
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)
+ // non-merged metadata
+ rec_md = record_metadata_init(se->nmem, (char *) value,
+ ser_md->type);
+ if (!rec_md)
{
- if (extract7bit_years((char *) value, &first, &last) < 0)
- continue;
- }
- else
- {
- yaz_log(YLOG_WARN,
- "Unknown type in metadata element %s", type);
+ yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'",
+ value, type);
continue;
}
+ rec_md->next = record->metadata[md_field_id];
+ record->metadata[md_field_id] = 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
}
}
}
- 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)
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);
}
}
projects / pazpar2-moved-to-github.git / blobdiff