-/* $Id: pazpar2.c,v 1.53 2007-03-26 14:00:21 marc Exp $ */
+/* $Id: pazpar2.c,v 1.64 2007-04-08 20:52:09 quinn Exp $ */
#include <stdlib.h>
#include <stdio.h>
#include <yaz/proto.h>
#include <yaz/readconf.h>
#include <yaz/pquery.h>
+#include <yaz/otherinfo.h>
#include <yaz/yaz-util.h>
#include <yaz/nmem.h>
#include "relevance.h"
#include "config.h"
#include "database.h"
+#include "settings.h"
#define MAX_CHUNK 15
static void ingest_records(struct client *cl, Z_Records *r);
//static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
void session_alert_watch(struct session *s, int what);
+char *session_setting_oneval(struct session *s, struct database *db, int offset);
IOCHAN channel_list = 0; // Master list of connections we're handling events to
"",
"",
"",
+ "",
0,
0,
30,
100,
MAX_CHUNK,
0,
- 0,
- 0,
0
};
odr_reset(global_parameters.odr_out);
}
+static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
+{
+ switch (n->kind)
+ {
+ case CCL_RPN_AND:
+ case CCL_RPN_OR:
+ case CCL_RPN_NOT:
+ case CCL_RPN_PROX:
+ pull_terms(nmem, n->u.p[0], termlist, num);
+ pull_terms(nmem, n->u.p[1], termlist, num);
+ break;
+ case CCL_RPN_TERM:
+ termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
+ break;
+ default: // NOOP
+ break;
+ }
+}
+
+// Extract terms from query into null-terminated termlist
+static void extract_terms(NMEM nmem, struct ccl_rpn_node *query, char **termlist)
+{
+ int num = 0;
+
+ pull_terms(nmem, query, termlist, &num);
+ termlist[num] = 0;
+}
+
static void send_search(IOCHAN i)
{
struct connection *co = iochan_getdata(i);
Z_Query *zquery;
struct ccl_rpn_node *cn;
int ssub = 0, lslb = 100000, mspn = 10;
+ char *recsyn;
+ char *piggyback;
yaz_log(YLOG_DEBUG, "Sending search");
- cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
+ cn = ccl_find_str(db->ccl_map, se->query, &cerror, &cpos);
if (!cn)
return;
+
+ if (!se->relevance)
+ {
+ // Initialize relevance structure with query terms
+ char *p[512];
+ extract_terms(se->nmem, cn, p);
+ se->relevance = relevance_create(se->nmem, (const char **) p,
+ se->expected_maxrecs);
+ }
+
a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
sizeof(Z_Query));
zquery->which = Z_Query_type_1;
for (ndb = 0; db->databases[ndb]; ndb++)
databaselist[ndb] = db->databases[ndb];
- a->u.searchRequest->preferredRecordSyntax =
- yaz_oidval_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, VAL_USMARC);
- a->u.searchRequest->smallSetUpperBound = &ssub;
- a->u.searchRequest->largeSetLowerBound = &lslb;
- a->u.searchRequest->mediumSetPresentNumber = &mspn;
+ if (!(piggyback = session_setting_oneval(se, db, PZ_PIGGYBACK)) || *piggyback == '1')
+ {
+ if ((recsyn = session_setting_oneval(se, db, PZ_NATIVESYNTAX)))
+ a->u.searchRequest->preferredRecordSyntax =
+ yaz_str_to_z3950oid(global_parameters.odr_out,
+ CLASS_RECSYN, recsyn);
+ a->u.searchRequest->smallSetUpperBound = &ssub;
+ a->u.searchRequest->largeSetLowerBound = &lslb;
+ a->u.searchRequest->mediumSetPresentNumber = &mspn;
+ }
a->u.searchRequest->resultSetName = "Default";
a->u.searchRequest->databaseNames = databaselist;
a->u.searchRequest->num_databaseNames = ndb;
{
struct connection *co = iochan_getdata(i);
struct client *cl = co->client;
+ struct session *se = cl->session;
+ struct database *db = cl->database;
Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
int toget;
int start = cl->records + 1;
+ char *recsyn;
toget = global_parameters.chunk;
if (toget > global_parameters.toget - cl->records)
a->u.presentRequest->resultSetId = "Default";
- a->u.presentRequest->preferredRecordSyntax =
- yaz_oidval_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, VAL_USMARC);
+ if ((recsyn = session_setting_oneval(se, db, PZ_NATIVESYNTAX)))
+ a->u.presentRequest->preferredRecordSyntax =
+ yaz_str_to_z3950oid(global_parameters.odr_out,
+ CLASS_RECSYN, recsyn);
if (send_apdu(cl, a) >= 0)
{
struct client *cl = co->client;
Z_InitResponse *r = a->u.initResponse;
- yaz_log(YLOG_DEBUG, "Received init response");
+ yaz_log(YLOG_DEBUG, "Init response %s", cl->database->url);
if (*r->result)
{
struct session *se = cl->session;
Z_SearchResponse *r = a->u.searchResponse;
- yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
+ yaz_log(YLOG_DEBUG, "Search response %s (status=%d)",
+ cl->database->url, *r->searchStatus);
if (*r->searchStatus)
{
se->total_hits += cl->hits;
if (r->presentStatus && !*r->presentStatus && r->records)
{
- yaz_log(YLOG_DEBUG, "Records in search response");
+ yaz_log(YLOG_DEBUG, "Records in search response %s",
+ cl->database->url);
ingest_records(cl, r->records);
}
cl->state = Client_Idle;
Z_Records *recs = r->records;
if (recs->which == Z_Records_NSD)
{
- yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
+ yaz_log(YLOG_WARN,
+ "Search response: Non-surrogate diagnostic %s",
+ cl->database->url);
cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
cl->state = Client_Error;
}
}
}
+static void do_closeResponse(IOCHAN i, Z_APDU *a)
+{
+ struct connection *co = iochan_getdata(i);
+ struct client *cl = co->client;
+ /* Z_Close *r = a->u.close; */
+
+ yaz_log(YLOG_WARN, "Close response %s", cl->database->url);
+
+ cl->state = Client_Failed;
+ connection_destroy(co);
+}
+
+
char *normalize_mergekey(char *buf, int skiparticle)
{
char *p = buf, *pout = buf;
return buf;
}
-
-#ifdef GAGA
-// FIXME needs to be generalized. Should flexibly generate X lists per search
-static void extract_subject(struct session *s, const char *rec)
-{
- const char *field, *subfield;
-
- while ((field = find_field(rec, "650")))
- {
- rec = field;
- if ((subfield = find_subfield(field, 'a')))
- {
- char *e, *ef;
- char buf[1024];
- int len;
-
- ef = index(subfield, '\n');
- if (!ef)
- return;
- if ((e = index(subfield, '\t')) && e < ef)
- ef = e;
- while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
- ef--;
- len = ef - subfield;
- assert(len < 1023);
- memcpy(buf, subfield, len);
- buf[len] = '\0';
-#ifdef FIXME
- if (*buf)
- termlist_insert(s->termlist, buf);
-#endif
- }
- }
-}
-#endif
-
static void add_facet(struct session *s, const char *type, const char *value)
{
int i;
static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
{
- struct conf_retrievalprofile *rprofile = cl->database->rprofile;
- struct conf_retrievalmap *m;
+ struct database_retrievalmap *m;
+ struct database *db = cl->database;
xmlNode *res;
xmlDoc *rdoc;
// First normalize to XML
- if (rprofile->native_syntax == Nativesyn_iso2709)
+ if (db->yaz_marc)
{
char *buf;
int len;
if (rec->which != Z_External_octet)
{
- yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
+ yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
+ cl->database->url);
return 0;
}
buf = (char*) rec->u.octet_aligned->buf;
len = rec->u.octet_aligned->len;
- if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
+ if (yaz_marc_read_iso2709(db->yaz_marc, buf, len) < 0)
{
- yaz_log(YLOG_WARN, "Failed to decode MARC");
+ yaz_log(YLOG_WARN, "Failed to decode MARC %s",
+ cl->database->url);
return 0;
}
- if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
+ if (yaz_marc_write_xml(db->yaz_marc, &res,
"http://www.loc.gov/MARC21/slim", 0, 0) < 0)
{
- yaz_log(YLOG_WARN, "Failed to encode as XML");
+ yaz_log(YLOG_WARN, "Failed to encode as XML %s",
+ cl->database->url);
return 0;
}
- rdoc = xmlNewDoc("1.0");
+ rdoc = xmlNewDoc((xmlChar *) "1.0");
xmlDocSetRootElement(rdoc, res);
}
else
#endif
}
- for (m = rprofile->maplist; m; m = m->next)
+ for (m = db->map; m; m = m->next)
{
xmlDoc *new;
- if (m->type != Map_xslt)
- {
- yaz_log(YLOG_WARN, "Unknown map type");
- return 0;
- }
if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
{
yaz_log(YLOG_WARN, "XSLT transformation failed");
return 0;
root = xmlDocGetRootElement(xdoc);
- if (!(mergekey = xmlGetProp(root, "mergekey")))
+ if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
{
yaz_log(YLOG_WARN, "No mergekey found in record");
xmlFreeDoc(xdoc);
sizeof(struct record_metadata*) * service->num_metadata);
memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata);
- mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
+ mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
xmlFree(mergekey);
- normalize_mergekey(mergekey_norm, 0);
+ normalize_mergekey((char *) mergekey_norm, 0);
- cluster = reclist_insert(se->reclist, res, mergekey_norm, &se->total_merged);
+ cluster = reclist_insert(se->reclist, res, (char *) mergekey_norm,
+ &se->total_merged);
if (global_parameters.dump_records)
yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
cl->database->url, cl->records);
if (n->type != XML_ELEMENT_NODE)
continue;
- if (!strcmp(n->name, "metadata"))
+ if (!strcmp((const char *) n->name, "metadata"))
{
struct conf_metadata *md = 0;
struct conf_sortkey *sk = 0;
int imeta;
int first, last;
- type = xmlGetProp(n, "type");
+ type = xmlGetProp(n, (xmlChar *) "type");
value = xmlNodeListGetString(xdoc, n->children, 0);
if (!type || !value)
// First, find out what field we're looking at
for (imeta = 0; imeta < service->num_metadata; imeta++)
- if (!strcmp(type, service->metadata[imeta].name))
+ if (!strcmp((const char *) type, service->metadata[imeta].name))
{
md = &service->metadata[imeta];
if (md->sortkey_offset >= 0)
if (md->type == Metadata_type_generic)
{
char *p, *pe;
- for (p = value; *p && isspace(*p); p++)
+ for (p = (char *) value; *p && isspace(*p); p++)
;
for (pe = p + strlen(p) - 1;
pe > p && strchr(" ,/.:([", *pe); pe--)
}
else if (md->type == Metadata_type_year)
{
- if (extract_years(value, &first, &last) < 0)
+ if (extract_years((char *) value, &first, &last) < 0)
continue;
}
else
{
struct record_metadata *mnode;
for (mnode = *wheretoput; mnode; mnode = mnode->next)
- if (!strcmp(mnode->data.text, newm->data.text))
+ if (!strcmp((const char *) mnode->data.text, newm->data.text))
break;
if (!mnode)
{
yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
if (md->rank)
- relevance_countwords(se->relevance, cluster, value, md->rank);
+ relevance_countwords(se->relevance, cluster,
+ (char *) value, md->rank);
if (md->termlist)
{
if (md->type == Metadata_type_year)
{
char year[64];
sprintf(year, "%d", last);
- add_facet(se, type, year);
+ add_facet(se, (char *) type, year);
if (first != last)
{
sprintf(year, "%d", first);
- add_facet(se, type, year);
+ add_facet(se, (char *) type, year);
}
}
else
- add_facet(se, type, value);
+ add_facet(se, (char *) type, (char *) value);
}
xmlFree(type);
xmlFree(value);
return res;
}
+// 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 *s, struct database *db, int offset)
+{
+ if (!db->settings[offset])
+ return 0;
+ return db->settings[offset]->value;
+}
+
static void ingest_records(struct client *cl, Z_Records *r)
{
#if USE_TIMING
cl->records++;
if (npr->which != Z_NamePlusRecord_databaseRecord)
{
- yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
+ yaz_log(YLOG_WARN,
+ "Unexpected record type, probably diagnostic %s",
+ cl->database->url);
continue;
}
Z_Records *recs = r->records;
if (recs->which == Z_Records_NSD)
{
- yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
+ yaz_log(YLOG_WARN, "Non-surrogate diagnostic %s",
+ cl->database->url);
cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
cl->state = Client_Error;
}
if (!*r->presentStatus && cl->state != Client_Error)
{
- yaz_log(YLOG_DEBUG, "Good Present response");
+ yaz_log(YLOG_DEBUG, "Good Present response %s",
+ cl->database->url);
ingest_records(cl, r->records);
cl->state = Client_Idle;
}
else if (*r->presentStatus)
{
- yaz_log(YLOG_WARN, "Bad Present response");
+ yaz_log(YLOG_WARN, "Bad Present response %s",
+ cl->database->url);
cl->state = Client_Error;
}
}
if (len < 0)
{
- yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
+ yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from %s",
+ cl->database->url);
connection_destroy(co);
return;
}
else if (len == 0)
{
- yaz_log(YLOG_WARN, "EOF reading from Z server");
+ yaz_log(YLOG_WARN, "EOF reading from %s", cl->database->url);
connection_destroy(co);
return;
}
case Z_APDU_presentResponse:
do_presentResponse(i, a);
break;
+ case Z_APDU_close:
+ do_closeResponse(i, a);
+ break;
default:
- yaz_log(YLOG_WARN, "Unexpected result from server");
+ yaz_log(YLOG_WARN,
+ "Unexpected Z39.50 response from %s",
+ cl->database->url);
client_fatal(cl);
return;
}
yaz_log(YLOG_DEBUG, "Connection create %s proxy %s",
cl->database->url, global_parameters.zproxy_override);
- yaz_log(YLOG_LOG, "Connection cs_create_host %s proxy %s",
- cl->database->url, global_parameters.zproxy_override);
-
if (!(addr = cs_straddr(link, global_parameters.zproxy_override)))
{
yaz_log(YLOG_WARN|YLOG_ERRNO,
cl->connection = new;
new->link = link;
- new->iochan = iochan_create(cs_fileno(link), handler, 0);
+ new->iochan = iochan_create(cs_fileno(link), 0, handler, 0);
iochan_setdata(new->iochan, new);
new->iochan->next = channel_list;
channel_list = new->iochan;
yaz_log(YLOG_LOG, "Target: %s, '%s'", url, db);
for (host = hosts; host; host = host->next)
- if (!strcmp(url, host->hostport))
+ if (!strcmp((const char *) url, host->hostport))
break;
if (!host)
{
#endif
-static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
-{
- switch (n->kind)
- {
- case CCL_RPN_AND:
- case CCL_RPN_OR:
- case CCL_RPN_NOT:
- case CCL_RPN_PROX:
- pull_terms(nmem, n->u.p[0], termlist, num);
- pull_terms(nmem, n->u.p[1], termlist, num);
- break;
- case CCL_RPN_TERM:
- termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
- break;
- default: // NOOP
- break;
- }
-}
-
-// Extract terms from query into null-terminated termlist
-static int extract_terms(NMEM nmem, char *query, char **termlist)
-{
- int error, pos;
- struct ccl_rpn_node *n;
- int num = 0;
-
- n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
- if (!n)
- return -1;
- pull_terms(nmem, n, termlist, &num);
- termlist[num] = 0;
- ccl_rpn_delete(n);
- return 0;
-}
-
static struct client *client_create(void)
{
struct client *r;
}
if (live_channels)
{
- char *p[512];
int maxrecs = live_channels * global_parameters.toget;
se->num_termlists = 0;
se->reclist = reclist_create(se->nmem, maxrecs);
- extract_terms(se->nmem, query, p);
- se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
+ // This will be initialized in send_search()
+ se->relevance = 0;
se->total_records = se->total_hits = se->total_merged = 0;
se->expected_maxrecs = maxrecs;
}
int i;
for (i = 0; i < s->num_termlists; i++)
- if (!strcmp(s->termlists[i].name, name))
+ if (!strcmp((const char *) s->termlists[i].name, name))
return termlist_highscore(s->termlists[i].termlist, num);
return 0;
}
stat->num_clients = count;
}
-static CCL_bibset load_cclfile(const char *fn)
-{
- CCL_bibset res = ccl_qual_mk();
- if (ccl_qual_fname(res, fn) < 0)
- {
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
- exit(1);
- }
- return res;
-}
-
static void start_http_listener(void)
{
char hp[128] = "";
http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
}
+static 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;
+}
+
+
+
int main(int argc, char **argv)
{
int ret;
yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
- while ((ret = options("f:x:h:p:z:C:s:d", argv, argc, &arg)) != -2)
+ while ((ret = options("t:f:x:h:p:z:s:d", argv, argc, &arg)) != -2)
{
switch (ret) {
case 'f':
case 'h':
strcpy(global_parameters.listener_override, arg);
break;
- case 'C':
- global_parameters.ccl_filter = load_cclfile(arg);
- break;
case 'p':
strcpy(global_parameters.proxy_override, arg);
break;
case 'z':
strcpy(global_parameters.zproxy_override, arg);
break;
+ case 't':
+ strcpy(global_parameters.settings_path_override, arg);
+ break;
case 's':
load_simpletargets(arg);
break;
start_http_listener();
start_proxy();
- if (!global_parameters.ccl_filter)
- global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
- global_parameters.yaz_marc = yaz_marc_create();
- yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
+ start_zproxy();
+
+ if (*global_parameters.settings_path_override)
+ settings_read(global_parameters.settings_path_override);
+ else if (global_parameters.server->settings)
+ settings_read(global_parameters.server->settings);
+ else
+ yaz_log(YLOG_WARN, "No settings-directory specified. Problems may well ensue!");
+ prepare_databases();
global_parameters.odr_in = odr_createmem(ODR_DECODE);
global_parameters.odr_out = odr_createmem(ODR_ENCODE);
projects / pazpar2-moved-to-github.git / blobdiff