-/* $Id: logic.c,v 1.12 2007-04-20 04:32:33 quinn Exp $
+/* $Id: logic.c,v 1.52 2007-07-16 09:00:22 adam Exp $
Copyright (c) 2006-2007, Index Data.
This file is part of Pazpar2.
02111-1307, USA.
*/
+/** \file logic.c
+ \brief high-level logic; mostly user sessions and settings
+*/
+
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <yaz/nmem.h>
#include <yaz/query-charset.h>
#include <yaz/querytowrbuf.h>
-#if YAZ_VERSIONL >= 0x020163
#include <yaz/oid_db.h>
-#endif
#if HAVE_CONFIG_H
#include "cconfig.h"
#include "relevance.h"
#include "config.h"
#include "database.h"
+#include "client.h"
#include "settings.h"
+#include "normalize7bit.h"
-#define MAX_CHUNK 15
+#define TERMLIST_HIGH_SCORE 25
-static void client_fatal(struct client *cl);
-static void connection_destroy(struct connection *co);
-static int client_prep_connection(struct client *cl);
-static void ingest_records(struct client *cl, Z_Records *r);
-void session_alert_watch(struct session *s, int what);
-
-static struct connection *connection_freelist = 0;
-static struct client *client_freelist = 0;
-
-static char *client_states[] = {
- "Client_Connecting",
- "Client_Connected",
- "Client_Idle",
- "Client_Initializing",
- "Client_Searching",
- "Client_Presenting",
- "Client_Error",
- "Client_Failed",
- "Client_Disconnected",
- "Client_Stopped"
-};
+#define MAX_CHUNK 15
// Note: Some things in this structure will eventually move to configuration
struct parameters global_parameters =
{
"",
"",
- "",
- "",
- 0,
+ "",
0,
+ 0, /* dump_records */
+ 0, /* debug_mode */
30,
"81",
"Index Data PazPar2",
0
};
-static int send_apdu(struct client *c, Z_APDU *a)
-{
- struct connection *co = c->connection;
- char *buf;
- int len, r;
-
- if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
- {
- odr_perror(global_parameters.odr_out, "Encoding APDU");
- abort();
- }
- buf = odr_getbuf(global_parameters.odr_out, &len, 0);
- r = cs_put(co->link, buf, len);
- if (r < 0)
- {
- yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
- return -1;
- }
- else if (r == 1)
- {
- fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
- exit(1);
- }
- odr_reset(global_parameters.odr_out); /* release the APDU structure */
- co->state = Conn_Waiting;
- return 0;
-}
-
-// Set authentication token in init if one is set for the client
-// TODO: Extend this to handle other schemes than open (should be simple)
-static void init_authentication(struct client *cl, Z_InitRequest *req)
-{
- struct session_database *sdb = cl->database;
- char *auth = session_setting_oneval(sdb, PZ_AUTHENTICATION);
-
- if (auth)
- {
- Z_IdAuthentication *idAuth = odr_malloc(global_parameters.odr_out,
- sizeof(*idAuth));
- idAuth->which = Z_IdAuthentication_open;
- idAuth->u.open = auth;
- req->idAuthentication = idAuth;
- }
-}
-
-static void send_init(IOCHAN i)
-{
-
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
-
- a->u.initRequest->implementationId = global_parameters.implementationId;
- a->u.initRequest->implementationName = global_parameters.implementationName;
- a->u.initRequest->implementationVersion =
- global_parameters.implementationVersion;
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
- ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
-
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
- ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
-
- init_authentication(cl, a->u.initRequest);
-
- /* add virtual host if tunneling through Z39.50 proxy */
-
- if (0 < strlen(global_parameters.zproxy_override)
- && 0 < strlen(cl->database->database->url))
- {
-#if YAZ_VERSIONL >= 0x020163
- yaz_oi_set_string_oid(&a->u.initRequest->otherInfo,
- global_parameters.odr_out,
- yaz_oid_userinfo_proxy,
- 1, cl->database->database->url);
-#else
- yaz_oi_set_string_oidval(&a->u.initRequest->otherInfo,
- global_parameters.odr_out, VAL_PROXY,
- 1, cl->database->database->url);
-#endif
- }
-
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Initializing;
- }
- else
- cl->state = Client_Error;
- odr_reset(global_parameters.odr_out);
-}
// Recursively traverse query structure to extract terms.
-static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
+void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
{
char **words;
int numwords;
}
}
-// 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);
- struct client *cl = co->client;
- struct session *se = cl->session;
- struct session_database *sdb = cl->database;
- Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
- int ndb;
- char **databaselist;
- Z_Query *zquery;
- int ssub = 0, lslb = 100000, mspn = 10;
- char *recsyn = 0;
- char *piggyback = 0;
- char *queryenc = 0;
- yaz_iconv_t iconv = 0;
-
- yaz_log(YLOG_DEBUG, "Sending search to %s", cl->database->database->url);
-
- // constructing RPN query
- a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
- sizeof(Z_Query));
- zquery->which = Z_Query_type_1;
- zquery->u.type_1 = p_query_rpn(global_parameters.odr_out, cl->pquery);
-
- // converting to target encoding
- if ((queryenc = session_setting_oneval(sdb, PZ_QUERYENCODING))){
- iconv = yaz_iconv_open(queryenc, "UTF-8");
- if (iconv){
- yaz_query_charset_convert_rpnquery(zquery->u.type_1,
- global_parameters.odr_out,
- iconv);
- yaz_iconv_close(iconv);
- } else
- yaz_log(YLOG_WARN, "Query encoding failed %s %s",
- cl->database->database->url, queryenc);
- }
-
- for (ndb = 0; sdb->database->databases[ndb]; ndb++)
- ;
- databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
- for (ndb = 0; sdb->database->databases[ndb]; ndb++)
- databaselist[ndb] = sdb->database->databases[ndb];
-
- if (!(piggyback = session_setting_oneval(sdb, PZ_PIGGYBACK)) || *piggyback == '1')
- {
- if ((recsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX)))
- {
-#if YAZ_VERSIONL >= 0x020163
- a->u.searchRequest->preferredRecordSyntax =
- yaz_string_to_oid_odr(yaz_oid_std(),
- CLASS_RECSYN, recsyn,
- global_parameters.odr_out);
-#else
- a->u.searchRequest->preferredRecordSyntax =
- yaz_str_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, recsyn);
-#endif
- }
- 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;
-
-
- { //scope for sending and logging queries
- WRBUF wbquery = wrbuf_alloc();
- yaz_query_to_wrbuf(wbquery, zquery);
-
-
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Searching;
- cl->requestid = se->requestid;
- yaz_log(YLOG_LOG, "SearchRequest %s %s %s",
- cl->database->database->url,
- queryenc ? queryenc : "UTF-8",
- wrbuf_cstr(wbquery));
- }
- else {
- cl->state = Client_Error;
- yaz_log(YLOG_WARN, "Failed SearchRequest %s %s %s",
- cl->database->database->url,
- queryenc ? queryenc : "UTF-8",
- wrbuf_cstr(wbquery));
- }
-
- wrbuf_destroy(wbquery);
- }
-
- odr_reset(global_parameters.odr_out);
-}
-
-static void send_present(IOCHAN i)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session_database *sdb = 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)
- toget = global_parameters.toget - cl->records;
- if (toget > cl->hits - cl->records)
- toget = cl->hits - cl->records;
-
- yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
-
- a->u.presentRequest->resultSetStartPoint = &start;
- a->u.presentRequest->numberOfRecordsRequested = &toget;
-
- a->u.presentRequest->resultSetId = "Default";
-
- if ((recsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX)))
- {
-#if YAZ_VERSIONL >= 0x020163
- a->u.presentRequest->preferredRecordSyntax =
- yaz_string_to_oid_odr(yaz_oid_std(),
- CLASS_RECSYN, recsyn,
- global_parameters.odr_out);
-#else
- a->u.presentRequest->preferredRecordSyntax =
- yaz_str_to_z3950oid(global_parameters.odr_out,
- CLASS_RECSYN, recsyn);
-#endif
- }
-
- if (send_apdu(cl, a) >= 0)
- {
- iochan_setflags(i, EVENT_INPUT);
- cl->state = Client_Presenting;
- }
- else
- cl->state = Client_Error;
- odr_reset(global_parameters.odr_out);
-}
-
-static void do_initResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_InitResponse *r = a->u.initResponse;
-
- yaz_log(YLOG_DEBUG, "Init response %s", cl->database->database->url);
-
- if (*r->result)
- {
- cl->state = Client_Idle;
- }
- else
- cl->state = Client_Failed; // FIXME need to do something to the connection
-}
-
-static void do_searchResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session *se = cl->session;
- Z_SearchResponse *r = a->u.searchResponse;
-
- yaz_log(YLOG_DEBUG, "Search response %s (status=%d)",
- cl->database->database->url, *r->searchStatus);
-
- if (*r->searchStatus)
- {
- cl->hits = *r->resultCount;
- se->total_hits += cl->hits;
- if (r->presentStatus && !*r->presentStatus && r->records)
- {
- yaz_log(YLOG_DEBUG, "Records in search response %s",
- cl->database->database->url);
- ingest_records(cl, r->records);
- }
- cl->state = Client_Idle;
- }
- else
- { /*"FAILED"*/
- cl->hits = 0;
- cl->state = Client_Error;
- if (r->records) {
- Z_Records *recs = r->records;
- if (recs->which == Z_Records_NSD)
- {
- yaz_log(YLOG_WARN,
- "Search response: Non-surrogate diagnostic %s",
- cl->database->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->database->url);
-
- cl->state = Client_Failed;
- connection_destroy(co);
-}
-
-
-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;
-}
static void add_facet(struct session *s, const char *type, const char *value)
{
if (i == SESSION_MAX_TERMLISTS)
{
yaz_log(YLOG_FATAL, "Too many termlists");
- exit(1);
+ return;
}
+
s->termlists[i].name = nmem_strdup(s->nmem, type);
- s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
+ s->termlists[i].termlist
+ = termlist_create(s->nmem, s->expected_maxrecs,
+ TERMLIST_HIGH_SCORE);
s->num_termlists = i + 1;
}
termlist_insert(s->termlists[i].termlist, value);
}
-static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
+xmlDoc *record_to_xml(struct session_database *sdb, Z_External *rec)
{
- struct database_retrievalmap *m;
- struct database *db = cl->database->database;
- xmlNode *res;
- xmlDoc *rdoc;
+ struct database *db = sdb->database;
+ xmlDoc *rdoc = 0;
+ const Odr_oid *oid = rec->direct_reference;
- // First normalize to XML
- if (db->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",
- cl->database->database->url);
+ 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(db->yaz_marc, buf, len) < 0)
+ }
+ else if (rec->which == Z_External_OPAC)
+ {
+ if (!sdb->yaz_marc)
{
- yaz_log(YLOG_WARN, "Failed to decode MARC %s",
- cl->database->database->url);
+ yaz_log(YLOG_WARN, "MARC decoding not configured");
return 0;
}
-
- yaz_marc_write_using_libxml2(db->yaz_marc, 1);
- if (yaz_marc_write_xml(db->yaz_marc, &res,
- "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
+ else
{
- yaz_log(YLOG_WARN, "Failed to encode as XML %s",
- cl->database->database->url);
- return 0;
+ /* 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");
+ wrbuf_destroy(wrbuf_opac);
}
- rdoc = xmlNewDoc((xmlChar *) "1.0");
- xmlDocSetRootElement(rdoc, res);
-
}
- else
+ else if (oid && yaz_oid_is_iso2709(oid))
{
- yaz_log(YLOG_FATAL,
- "Unknown native_syntax in normalize_record from %s",
- cl->database->database->url);
- exit(1);
- }
-
- if (global_parameters.dump_records){
- fprintf(stderr,
- "Input Record (normalized) from %s\n----------------\n",
- cl->database->database->url);
-#if LIBXML_VERSION >= 20600
- xmlDocFormatDump(stderr, rdoc, 1);
-#else
- xmlDocDump(stderr, rdoc);
-#endif
- }
-
- for (m = db->map; m; m = m->next){
- xmlDoc *new = 0;
-
-#if 1
- {
- xmlNodePtr root = 0;
- new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
- root= xmlDocGetRootElement(new);
- if (!new || !root || !(root->children))
+ /* ISO2709 gets converted to MARCXML */
+ if (!sdb->yaz_marc)
{
- yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
- cl->database->database->url);
- xmlFreeDoc(new);
- xmlFreeDoc(rdoc);
+ yaz_log(YLOG_WARN, "MARC decoding not configured");
return 0;
}
- }
-#endif
-
-#if 0
- // do it another way to detect transformation errors right now
- // but does not seem to work either!
+ else
{
- 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);
+ 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);
}
-#endif
-
- xmlFreeDoc(rdoc);
- rdoc = new;
}
+ 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,
+ "Unable to handle record of type %s from %s",
+ oid_name, db->url);
+ return 0;
+ }
+
if (global_parameters.dump_records)
{
- fprintf(stderr, "Record from %s\n----------------\n",
- cl->database->database->url);
+ FILE *lf = yaz_log_file();
+ if (lf)
+ {
+ yaz_log(YLOG_LOG, "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
- }
- return rdoc;
-}
-
-// 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;
+ fprintf(lf, "\n");
}
- buf = e;
}
- return *first;
+ return rdoc;
}
-static struct record *ingest_record(struct client *cl, Z_External *rec)
+xmlDoc *normalize_record(struct session_database *sdb, Z_External *rec)
{
- xmlDoc *xdoc = normalize_record(cl, rec);
- xmlNode *root, *n;
- struct record *res;
- struct record_cluster *cluster;
- struct session *se = cl->session;
- xmlChar *mergekey, *mergekey_norm;
- xmlChar *type = 0;
- xmlChar *value = 0;
- struct conf_service *service = global_parameters.server->service;
-
- if (!xdoc)
- return 0;
-
- root = xmlDocGetRootElement(xdoc);
- if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
- {
- yaz_log(YLOG_WARN, "No mergekey found in record");
- xmlFreeDoc(xdoc);
- 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);
-
- mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
- xmlFree(mergekey);
- normalize_mergekey((char *) mergekey_norm, 0);
-
- cluster = reclist_insert(se->reclist,
- global_parameters.server->service,
- 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->database->url, cl->records);
- if (!cluster)
- {
- /* no room for record */
- xmlFreeDoc(xdoc);
- return 0;
- }
- relevance_newrec(se->relevance, cluster);
-
- for (n = root->children; n; n = n->next)
+ struct database_retrievalmap *m;
+ xmlDoc *rdoc = record_to_xml(sdb, rec);
+ if (rdoc)
{
- if (type)
- xmlFree(type);
- if (value)
- xmlFree(value);
- type = value = 0;
-
- if (n->type != XML_ELEMENT_NODE)
- 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;
- int first, last;
-
- type = xmlGetProp(n, (xmlChar *) "type");
- value = xmlNodeListGetString(xdoc, n->children, 0);
-
- if (!type || !value)
- continue;
-
- // First, find out what field we're looking at
- for (imeta = 0; imeta < service->num_metadata; imeta++)
- if (!strcmp((const char *) type, service->metadata[imeta].name))
- {
- md = &service->metadata[imeta];
- if (md->sortkey_offset >= 0)
- sk = &service->sortkeys[md->sortkey_offset];
- break;
- }
- if (!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
- wheretoput = &cluster->metadata[imeta];
+ for (m = sdb->map; m; m = m->next){
+ xmlDoc *new = 0;
- // Put it there
- newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
- newm->next = 0;
- if (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);
-
- }
- else if (md->type == Metadata_type_year)
- {
- if (extract_years((char *) value, &first, &last) < 0)
- continue;
- }
- else
- {
- yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
- continue;
- }
- if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
- {
- yaz_log(YLOG_WARN, "Only range merging supported for years");
- continue;
- }
- if (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))
- break;
- if (!mnode)
- {
- newm->next = *wheretoput;
- *wheretoput = newm;
- }
- }
- else if (md->merge == Metadata_merge_longest)
{
- if (!*wheretoput ||
- strlen(newm->data.text) > strlen((*wheretoput)->data.text))
+ xmlNodePtr root = 0;
+ new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
+ root= xmlDocGetRootElement(new);
+ if (!new || !root || !(root->children))
{
- *wheretoput = newm;
- if (sk)
- {
- char *s = nmem_strdup(se->nmem, newm->data.text);
- if (!cluster->sortkeys[md->sortkey_offset])
- cluster->sortkeys[md->sortkey_offset] =
- nmem_malloc(se->nmem, sizeof(union data_types));
- normalize_mergekey(s,
- (sk->type == Metadata_sortkey_skiparticle));
- cluster->sortkeys[md->sortkey_offset]->text = s;
- }
+ yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
+ sdb->database->url);
+ xmlFreeDoc(new);
+ xmlFreeDoc(rdoc);
+ return 0;
}
}
- else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
- {
- newm->next = *wheretoput;
- *wheretoput = newm;
- }
- else if (md->merge == Metadata_merge_range)
- {
- assert(md->type == Metadata_type_year);
- if (!*wheretoput)
- {
- *wheretoput = newm;
- (*wheretoput)->data.number.min = first;
- (*wheretoput)->data.number.max = last;
- if (sk)
- cluster->sortkeys[md->sortkey_offset] = &newm->data;
- }
- else
- {
- if (first < (*wheretoput)->data.number.min)
- (*wheretoput)->data.number.min = first;
- if (last > (*wheretoput)->data.number.max)
- (*wheretoput)->data.number.max = last;
- }
-#ifdef GAGA
- if (sk)
- {
- union data_types *sdata = cluster->sortkeys[md->sortkey_offset];
- yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max);
- }
+
+ xmlFreeDoc(rdoc);
+ rdoc = new;
+ }
+ if (global_parameters.dump_records)
+ {
+ fprintf(stderr, "Record from %s\n----------------\n",
+ sdb->database->url);
+#if LIBXML_VERSION >= 20600
+ xmlDocFormatDump(stderr, rdoc, 1);
+#else
+ xmlDocDump(stderr, rdoc);
#endif
- }
- else
- yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
-
- if (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, (char *) type, year);
- if (first != last)
- {
- sprintf(year, "%d", first);
- add_facet(se, (char *) type, year);
- }
- }
- else
- add_facet(se, (char *) type, (char *) value);
- }
- xmlFree(type);
- xmlFree(value);
- type = value = 0;
}
- else
- yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
}
- if (type)
- xmlFree(type);
- if (value)
- xmlFree(value);
-
- xmlFreeDoc(xdoc);
-
- relevance_donerecord(se->relevance, cluster);
- se->total_records++;
-
- return res;
+ return rdoc;
}
// Retrieve first defined value for 'name' for given database.
return db->settings[offset]->value;
}
-static void ingest_records(struct client *cl, Z_Records *r)
-{
-#if USE_TIMING
- yaz_timing_t t = yaz_timing_create();
-#endif
- struct record *rec;
- struct session *s = cl->session;
- Z_NamePlusRecordList *rlist;
- int i;
-
- if (r->which != Z_Records_DBOSD)
- return;
- rlist = r->u.databaseOrSurDiagnostics;
- for (i = 0; i < rlist->num_records; i++)
- {
- Z_NamePlusRecord *npr = rlist->records[i];
-
- cl->records++;
- if (npr->which != Z_NamePlusRecord_databaseRecord)
- {
- yaz_log(YLOG_WARN,
- "Unexpected record type, probably diagnostic %s",
- cl->database->database->url);
- continue;
- }
-
- rec = ingest_record(cl, npr->u.databaseRecord);
- if (!rec)
- continue;
- }
- if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
- session_alert_watch(s, SESSION_WATCH_RECORDS);
-
-#if USE_TIMING
- yaz_timing_stop(t);
- yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
- yaz_timing_get_real(t), yaz_timing_get_user(t),
- yaz_timing_get_sys(t));
- yaz_timing_destroy(&t);
-#endif
-}
-
-static void do_presentResponse(IOCHAN i, Z_APDU *a)
-{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- Z_PresentResponse *r = a->u.presentResponse;
-
- if (r->records) {
- Z_Records *recs = r->records;
- if (recs->which == Z_Records_NSD)
- {
- yaz_log(YLOG_WARN, "Non-surrogate diagnostic %s",
- cl->database->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 %s",
- cl->database->database->url);
- ingest_records(cl, r->records);
- cl->state = Client_Idle;
- }
- else if (*r->presentStatus)
- {
- yaz_log(YLOG_WARN, "Bad Present response %s",
- cl->database->database->url);
- cl->state = Client_Error;
- }
-}
-static void handler(IOCHAN i, int event)
+// Initialize YAZ Map structures for MARC-based targets
+static int prepare_yazmarc(struct session_database *sdb)
{
- struct connection *co = iochan_getdata(i);
- struct client *cl = co->client;
- struct session *se = 0;
+ char *s;
- if (cl)
- se = cl->session;
- else
- {
- yaz_log(YLOG_WARN, "Destroying orphan connection");
- connection_destroy(co);
- return;
- }
-
- if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
+ if (!sdb->settings)
{
- int errcode;
- socklen_t errlen = sizeof(errcode);
-
- if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
- &errlen) < 0 || errcode != 0)
- {
- client_fatal(cl);
- return;
- }
- else
- {
- yaz_log(YLOG_DEBUG, "Connect OK");
- co->state = Conn_Open;
- if (cl)
- cl->state = Client_Connected;
- }
+ yaz_log(YLOG_WARN, "No settings for %s", sdb->database->url);
+ return -1;
}
-
- else if (event & EVENT_INPUT)
+ if ((s = session_setting_oneval(sdb, PZ_NATIVESYNTAX))
+ && !strncmp(s, "iso2709", 7))
{
- int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
-
- if (len < 0)
- {
- yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from %s",
- cl->database->database->url);
- connection_destroy(co);
- return;
- }
- else if (len == 0)
- {
- yaz_log(YLOG_WARN, "EOF reading from %s", cl->database->database->url);
- connection_destroy(co);
- return;
- }
- else if (len > 1) // We discard input if we have no connection
- {
- co->state = Conn_Open;
-
- if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
- {
- Z_APDU *a;
-
- odr_reset(global_parameters.odr_in);
- odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
- if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
- {
- client_fatal(cl);
- return;
- }
- switch (a->which)
- {
- case Z_APDU_initResponse:
- do_initResponse(i, a);
- break;
- case Z_APDU_searchResponse:
- do_searchResponse(i, a);
- break;
- case Z_APDU_presentResponse:
- do_presentResponse(i, a);
- break;
- case Z_APDU_close:
- do_closeResponse(i, a);
- break;
- default:
- yaz_log(YLOG_WARN,
- "Unexpected Z39.50 response from %s",
- cl->database->database->url);
- client_fatal(cl);
- return;
- }
- // We aren't expecting staggered output from target
- // if (cs_more(t->link))
- // iochan_setevent(i, EVENT_INPUT);
- }
- else // we throw away response and go to idle mode
- {
- yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
- cl->state = Client_Idle;
- }
- }
- /* if len==1 we do nothing but wait for more input */
- }
+ char *encoding = "marc-8s", *e;
+ yaz_iconv_t cm;
- if (cl->state == Client_Connected) {
- send_init(i);
- }
+ // See if a native encoding is specified
+ if ((e = strchr(s, ';')))
+ encoding = e + 1;
- if (cl->state == Client_Idle)
- {
- if (cl->requestid != se->requestid && cl->pquery) {
- send_search(i);
- }
- else if (cl->hits > 0 && cl->records < global_parameters.toget &&
- cl->records < cl->hits) {
- send_present(i);
+ sdb->yaz_marc = yaz_marc_create();
+ yaz_marc_subfield_str(sdb->yaz_marc, "\t");
+
+ cm = yaz_iconv_open("utf-8", encoding);
+ if (!cm)
+ {
+ yaz_log(YLOG_FATAL,
+ "Unable to map from %s to UTF-8 for target %s",
+ encoding, sdb->database->url);
+ return -1;
}
+ yaz_marc_iconv(sdb->yaz_marc, cm);
}
+ return 0;
}
-// Disassociate connection from client
-static void connection_release(struct connection *co)
-{
- struct client *cl = co->client;
-
- yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
- if (!cl)
- return;
- cl->connection = 0;
- co->client = 0;
-}
-
-// Close connection and recycle structure
-static void connection_destroy(struct connection *co)
+// Prepare XSLT stylesheets for record normalization
+// Structures are allocated on the session_wide nmem to avoid having
+// to recompute this for every search. This would lead
+// to leaking if a single session was to repeatedly change the PZ_XSLT
+// setting. However, this is not a realistic use scenario.
+static int prepare_map(struct session *se, struct session_database *sdb)
{
- struct host *h = co->host;
- cs_close(co->link);
- iochan_destroy(co->iochan);
+ char *s;
- yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
- if (h->connections == co)
- h->connections = co->next;
- else
+ if (!sdb->settings)
{
- struct connection *pco;
- for (pco = h->connections; pco && pco->next != co; pco = pco->next)
- ;
- if (pco)
- pco->next = co->next;
- else
- abort();
+ yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
+ return -1;
}
- if (co->client)
+ if ((s = session_setting_oneval(sdb, PZ_XSLT)))
{
- if (co->client->state != Client_Idle)
- co->client->state = Client_Disconnected;
- co->client->connection = 0;
- }
- co->next = connection_freelist;
- connection_freelist = co;
-}
+ char **stylesheets;
+ struct database_retrievalmap **m = &sdb->map;
+ int num, i;
-// Creates a new connection for client, associated with the host of
-// client's database
-static struct connection *connection_create(struct client *cl)
-{
- struct connection *new;
- COMSTACK link;
- int res;
- void *addr;
-
-
- if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
+ nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
+ for (i = 0; i < num; i++)
{
- yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
- exit(1);
- }
-
- if (0 == strlen(global_parameters.zproxy_override)){
- /* no Z39.50 proxy needed - direct connect */
- yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->database->url);
-
- if (!(addr = cs_straddr(link, cl->database->database->host->ipport)))
- {
- yaz_log(YLOG_WARN|YLOG_ERRNO,
- "Lookup of IP address %s failed",
- cl->database->database->host->ipport);
- return 0;
- }
-
- } else {
- /* Z39.50 proxy connect */
- yaz_log(YLOG_DEBUG, "Connection create %s proxy %s",
- cl->database->database->url, global_parameters.zproxy_override);
-
- if (!(addr = cs_straddr(link, global_parameters.zproxy_override)))
+ (*m) = nmem_malloc(se->session_nmem, sizeof(**m));
+ (*m)->next = 0;
+ if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
{
- yaz_log(YLOG_WARN|YLOG_ERRNO,
- "Lookup of IP address %s failed",
- global_parameters.zproxy_override);
- return 0;
+ yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
+ stylesheets[i]);
+ return -1;
}
- }
-
- res = cs_connect(link, addr);
- if (res < 0)
- {
- yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->database->url);
- return 0;
- }
-
- if ((new = connection_freelist))
- connection_freelist = new->next;
- else
- {
- new = xmalloc(sizeof (struct connection));
- new->ibuf = 0;
- new->ibufsize = 0;
- }
- new->state = Conn_Connecting;
- new->host = cl->database->database->host;
- new->next = new->host->connections;
- new->host->connections = new;
- new->client = cl;
- cl->connection = new;
- new->link = link;
-
- new->iochan = iochan_create(cs_fileno(link), handler, 0);
- iochan_setdata(new->iochan, new);
- pazpar2_add_channel(new->iochan);
- return new;
-}
-
-// Close connection and set state to error
-static void client_fatal(struct client *cl)
-{
- yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->database->url);
- connection_destroy(cl->connection);
- cl->state = Client_Error;
-}
-
-// Ensure that client has a connection associated
-static int client_prep_connection(struct client *cl)
-{
- struct connection *co;
- struct session *se = cl->session;
- struct host *host = cl->database->database->host;
-
- co = cl->connection;
-
- yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->database->url);
-
- if (!co)
- {
- // See if someone else has an idle connection
- // We should look at timestamps here to select the longest-idle connection
- for (co = host->connections; co; co = co->next)
- if (co->state == Conn_Open && (!co->client || co->client->session != se))
- break;
- if (co)
- {
- connection_release(co);
- cl->connection = co;
- co->client = cl;
- }
- else
- co = connection_create(cl);
- }
- if (co)
- {
- if (co->state == Conn_Connecting)
- {
- cl->state = Client_Connecting;
- iochan_setflag(co->iochan, EVENT_OUTPUT);
+ m = &(*m)->next;
}
- else if (co->state == Conn_Open)
- {
- if (cl->state == Client_Error || cl->state == Client_Disconnected)
- cl->state = Client_Idle;
- iochan_setflag(co->iochan, EVENT_OUTPUT);
- }
- return 1;
}
- else
- return 0;
+ if (!sdb->map)
+ yaz_log(YLOG_WARN, "No Normalization stylesheet for target %s",
+ sdb->database->url);
+ return 0;
}
-// Initialize CCL map for a target
-static CCL_bibset prepare_cclmap(struct client *cl)
+// This analyzes settings and recomputes any supporting data structures
+// if necessary.
+static int prepare_session_database(struct session *se,
+ struct session_database *sdb)
{
- struct session_database *sdb = cl->database;
- struct setting *s;
- CCL_bibset res;
-
if (!sdb->settings)
- return 0;
- res = ccl_qual_mk();
- for (s = sdb->settings[PZ_CCLMAP]; s; s = s->next)
{
- char *p = strchr(s->name + 3, ':');
- if (!p)
- {
- yaz_log(YLOG_WARN, "Malformed cclmap name: %s", s->name);
- ccl_qual_rm(&res);
- return 0;
- }
- p++;
- ccl_qual_fitem(res, s->value, p);
- }
- return res;
-}
-
-// Parse the query given the settings specific to this client
-static int client_parse_query(struct client *cl, const char *query)
-{
- struct session *se = cl->session;
- struct ccl_rpn_node *cn;
- int cerror, cpos;
- CCL_bibset ccl_map = prepare_cclmap(cl);
-
- if (!ccl_map)
- return -1;
- cn = ccl_find_str(ccl_map, query, &cerror, &cpos);
- ccl_qual_rm(&ccl_map);
- if (!cn)
- {
- cl->state = Client_Error;
- yaz_log(YLOG_WARN, "Failed to parse query for %s",
- cl->database->database->url);
+ yaz_log(YLOG_WARN,
+ "No settings associated with %s", sdb->database->url);
return -1;
}
- wrbuf_rewind(se->wrbuf);
- ccl_pquery(se->wrbuf, cn);
- wrbuf_putc(se->wrbuf, '\0');
- if (cl->pquery)
- xfree(cl->pquery);
- cl->pquery = xstrdup(wrbuf_buf(se->wrbuf));
-
- if (!se->relevance)
+ if (sdb->settings[PZ_NATIVESYNTAX] && !sdb->yaz_marc)
{
- // 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);
+ if (prepare_yazmarc(sdb) < 0)
+ return -1;
}
-
- ccl_rpn_delete(cn);
- return 0;
-}
-
-static struct client *client_create(void)
-{
- struct client *r;
- if (client_freelist)
+ if (sdb->settings[PZ_XSLT] && !sdb->map)
{
- r = client_freelist;
- client_freelist = client_freelist->next;
+ if (prepare_map(se, sdb) < 0)
+ return -1;
}
- else
- r = xmalloc(sizeof(struct client));
- r->pquery = 0;
- r->database = 0;
- r->connection = 0;
- r->session = 0;
- r->hits = 0;
- r->records = 0;
- r->setno = 0;
- r->requestid = -1;
- r->diagnostic = 0;
- r->state = Client_Disconnected;
- r->next = 0;
- return r;
+ return 0;
}
-void client_destroy(struct client *c)
+// 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 *se = c->session;
- if (c == se->clients)
- se->clients = c->next;
- else
- {
- struct client *cc;
- for (cc = se->clients; cc && cc->next != c; cc = cc->next)
- ;
- if (cc)
- cc->next = c->next;
- }
- if (c->connection)
- connection_release(c->connection);
- c->next = client_freelist;
- client_freelist = 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
{
struct session *se = (struct session*) context;
struct client *cl = client_create();
- cl->database = db;
- cl->session = se;
- cl->next = se->clients;
- se->clients = cl;
+ client_set_database(cl, db);
+ client_set_session(cl, se);
}
// Associates a set of clients with a session;
-// Note: Session-databases represent databases with per-session setting overrides
+// Note: Session-databases represent databases with per-session
+// setting overrides
int select_targets(struct session *se, struct database_criterion *crit)
{
while (se->clients)
struct client *c;
int res = 0;
- for (c = s->clients; c; c = c->next)
- if (c->connection && (c->state == Client_Connecting ||
- c->state == Client_Initializing ||
- c->state == Client_Searching ||
- c->state == Client_Presenting))
+ for (c = s->clients; c; c = client_next_in_session(c))
+ if (client_is_active(c))
res++;
return res;
new->values = 0;
for (subi = 0; subi < subnum; subi++)
{
- struct database_criterion_value *newv = nmem_malloc(m, sizeof(*newv));
+ struct database_criterion_value *newv
+ = nmem_malloc(m, sizeof(*newv));
newv->value = subvalues[subi];
newv->next = new->values;
new->values = newv;
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);
criteria = parse_filter(se->nmem, filter);
se->requestid++;
se->expected_maxrecs = maxrecs;
}
else
- return "NOTARGETS";
+ return PAZPAR2_NO_TARGETS;
se->relevance = 0;
- for (cl = se->clients; cl; cl = cl->next)
- if (client_parse_query(cl, query) < 0) // Query must parse for all targets
- return "QUERY";
- for (cl = se->clients; cl; cl = cl->next)
- client_prep_connection(cl);
- return 0;
-}
-
-// Apply a session override to a database
-void session_apply_setting(struct session *se, char *dbname, char *setting, char *value)
-{
- struct session_database *sdb;
-
- for (sdb = se->databases; sdb; sdb = sdb->next)
- if (!strcmp(dbname, sdb->database->url))
+ for (cl = se->clients; cl; cl = client_next_in_session(cl))
+ {
+ if (prepare_session_database(se, client_get_database(cl)) < 0)
{
- struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
- int offset = settings_offset(setting);
-
- if (offset < 0)
- {
- yaz_log(YLOG_WARN, "Unknown setting %s", setting);
- return;
- }
- new->precedence = 0;
- new->target = dbname;
- new->name = setting;
- new->value = value;
- new->next = sdb->settings[offset];
- sdb->settings[offset] = new;
- break;
+ *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);
}
- if (!sdb)
- yaz_log(YLOG_WARN, "Unknown database in setting override: %s", dbname);
+ }
+
+ // 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;
}
-void session_init_databases_fun(void *context, struct database *db)
+// Creates a new session_database object for a database
+static void session_init_databases_fun(void *context, struct database *db)
{
struct session *se = (struct session *) context;
struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
int i;
new->database = db;
- new->settings = nmem_malloc(se->session_nmem, sizeof(struct settings *) * num);
- for (i = 0; i < num; i++)
- new->settings[i] = db->settings[i];
+ 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);
+ memset(new->settings, 0, sizeof(struct settings*) * num);
+
+ if (db->settings)
+ {
+ for (i = 0; i < num; i++)
+ new->settings[i] = db->settings[i];
+ }
new->next = se->databases;
se->databases = new;
}
+// Doesn't free memory associated with sdb -- nmem takes care of that
+static void session_database_destroy(struct session_database *sdb)
+{
+ struct database_retrievalmap *m;
+
+ for (m = sdb->map; m; m = m->next)
+ 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
// of the database list -- subject to modification by the settings ws command
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
+// called here.
+static struct session_database *load_session_database(struct session *se,
+ char *id)
+{
+ struct database *db = find_database(id, 0);
+
+ session_init_databases_fun((void*) se, db);
+ // New sdb is head of se->databases list
+ return se->databases;
+}
+
+// Find an existing session database. If not found, load it
+static struct session_database *find_session_database(struct session *se,
+ char *id)
+{
+ struct session_database *sdb;
+
+ for (sdb = se->databases; sdb; sdb = sdb->next)
+ if (!strcmp(sdb->database->url, id))
+ return sdb;
+ return load_session_database(se, id);
+}
+
+// Apply a session override to a database
+void session_apply_setting(struct session *se, char *dbname, char *setting,
+ char *value)
+{
+ struct session_database *sdb = find_session_database(se, dbname);
+ struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
+ 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;
+ new->value = value;
+ new->next = sdb->settings[offset];
+ sdb->settings[offset] = new;
+
+ // Force later recompute of settings-driven data structures
+ // (happens when a search starts and client connections are prepared)
+ switch (offset)
+ {
+ case PZ_NATIVESYNTAX:
+ if (sdb->yaz_marc)
+ {
+ yaz_marc_destroy(sdb->yaz_marc);
+ sdb->yaz_marc = 0;
+ }
+ break;
+ case PZ_XSLT:
+ if (sdb->map)
+ {
+ struct database_retrievalmap *m;
+ // We don't worry about the map structure -- it's in nmem
+ for (m = sdb->map; m; m = m->next)
+ xsltFreeStylesheet(m->stylesheet);
+ sdb->map = 0;
+ }
+ break;
+ }
}
void destroy_session(struct session *s)
{
- yaz_log(YLOG_LOG, "Destroying session");
+ struct session_database *sdb;
+
while (s->clients)
client_destroy(s->clients);
+ for (sdb = s->databases; sdb; sdb = sdb->next)
+ session_database_destroy(sdb);
nmem_destroy(s->nmem);
wrbuf_destroy(s->wrbuf);
}
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;
struct client *cl;
*count = 0;
- for (cl = se->clients; cl; cl = cl->next)
+ for (cl = se->clients; cl; cl = client_next_in_session(cl))
{
- char *name = session_setting_oneval(cl->database, PZ_NAME);
+ char *name = session_setting_oneval(client_get_database(cl), PZ_NAME);
- res[*count].id = cl->database->database->url;
+ res[*count].id = client_get_database(cl)->database->url;
res[*count].name = *name ? name : "Unknown";
- res[*count].hits = cl->hits;
- res[*count].records = cl->records;
- res[*count].diagnostic = cl->diagnostic;
- res[*count].state = client_states[cl->state];
- res[*count].connected = cl->connection ? 1 : 0;
+ res[*count].hits = client_get_hits(cl);
+ res[*count].records = client_get_num_records(cl);
+ res[*count].diagnostic = client_get_diagnostic(cl);
+ res[*count].state = client_get_state_str(cl);
+ res[*count].connected = client_get_connection(cl) ? 1 : 0;
(*count)++;
}
return 0;
}
-struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start,
- int *num, int *total, int *sumhits, NMEM nmem_show)
+struct record_cluster **show(struct session *s, struct reclist_sortparms *sp,
+ int start, int *num, int *total, int *sumhits,
+ NMEM nmem_show)
{
struct record_cluster **recs = nmem_malloc(nmem_show, *num
* sizeof(struct record_cluster *));
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);
int count = 0;
memset(stat, 0, sizeof(*stat));
- for (cl = se->clients; cl; cl = cl->next)
+ for (cl = se->clients; cl; cl = client_next_in_session(cl))
{
- if (!cl->connection)
+ if (!client_get_connection(cl))
stat->num_no_connection++;
- switch (cl->state)
+ switch (client_get_state(cl))
{
case Client_Connecting: stat->num_connecting++; break;
case Client_Initializing: stat->num_initializing++; break;
http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
}
-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;
-}
// Master list of connections we're handling events to
static IOCHAN channel_list = 0;
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);
+ xmlNode *root, *n;
+ struct record *record;
+ struct record_cluster *cluster;
+ struct session *se = client_get_session(cl);
+ xmlChar *mergekey, *mergekey_norm;
+ xmlChar *type = 0;
+ xmlChar *value = 0;
+ struct conf_service *service = global_parameters.server->service;
+
+ if (!xdoc)
+ return 0;
+
+ root = xmlDocGetRootElement(xdoc);
+ if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
+ {
+ yaz_log(YLOG_WARN, "No mergekey found in record");
+ xmlFreeDoc(xdoc);
+ return 0;
+ }
+
+ record = record_create(se->nmem,
+ service->num_metadata, service->num_sortkeys, cl,
+ record_no);
+
+ mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
+ xmlFree(mergekey);
+ normalize7bit_mergekey((char *) mergekey_norm, 0);
+
+ cluster = reclist_insert(se->reclist,
+ global_parameters.server->service,
+ record, (char *) mergekey_norm,
+ &se->total_merged);
+ if (global_parameters.dump_records)
+ yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
+ client_get_database(cl)->database->url, record_no);
+ if (!cluster)
+ {
+ /* no room for record */
+ xmlFreeDoc(xdoc);
+ return 0;
+ }
+ 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)
+ xmlFree(type);
+ if (value)
+ xmlFree(value);
+ type = value = 0;
+
+ if (n->type != XML_ELEMENT_NODE)
+ continue;
+ if (!strcmp((const char *) n->name, "metadata"))
+ {
+ 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;
+
+ type = xmlGetProp(n, (xmlChar *) "type");
+ value = xmlNodeListGetString(xdoc, n->children, 1);
+
+ if (!type || !value)
+ continue;
+
+ 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];
+ }
+
+ // non-merged metadata
+ rec_md = record_metadata_init(se->nmem, (char *) value,
+ ser_md->type);
+ if (!rec_md)
+ {
+ 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
+ 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,
+ rec_md->data.text))
+ break;
+ if (!mnode)
+ {
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
+ }
+ }
+ else if (ser_md->merge == Metadata_merge_longest)
+ {
+ if (!*wheretoput
+ || strlen(rec_md->data.text)
+ > strlen((*wheretoput)->data.text))
+ {
+ *wheretoput = rec_md;
+ if (ser_sk)
+ {
+ 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));
+ normalize7bit_mergekey(s,
+ (ser_sk->type == Metadata_sortkey_skiparticle));
+ cluster->sortkeys[sk_field_id]->text = s;
+ }
+ }
+ }
+ else if (ser_md->merge == Metadata_merge_all)
+ {
+ rec_md->next = *wheretoput;
+ *wheretoput = rec_md;
+ }
+ else if (ser_md->merge == Metadata_merge_range)
+ {
+ if (!*wheretoput)
+ {
+ *wheretoput = rec_md;
+ if (ser_sk)
+ cluster->sortkeys[sk_field_id]
+ = &rec_md->data;
+ }
+ else
+ {
+ 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)
+ {
+ union data_types *sdata
+ = cluster->sortkeys[sk_field_id];
+ yaz_log(YLOG_LOG, "SK range: %d-%d",
+ sdata->number.min, sdata->number.max);
+ }
+#endif
+ }
+
+
+ // 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)
+ {
+ char year[64];
+ sprintf(year, "%d", rec_md->data.number.max);
+ add_facet(se, (char *) type, year);
+ if (rec_md->data.number.max != rec_md->data.number.min)
+ {
+ sprintf(year, "%d", rec_md->data.number.min);
+ add_facet(se, (char *) type, year);
+ }
+ }
+ else
+ add_facet(se, (char *) type, (char *) value);
+ }
+
+ // cleaning up
+ xmlFree(type);
+ xmlFree(value);
+ type = value = 0;
+ }
+ else
+ yaz_log(YLOG_WARN,
+ "Unexpected element %s in internal record", n->name);
+ }
+ if (type)
+ xmlFree(type);
+ if (value)
+ xmlFree(value);
+
+ xmlFreeDoc(xdoc);
+
+ relevance_donerecord(se->relevance, cluster);
+ se->total_records++;
+
+ return record;
+}
+
+
+
/*
* Local variables:
* c-basic-offset: 4