X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=src%2Fpazpar2.c;h=f0bd9bee11a2844bd30190ced4325d4a940d5614;hb=8acdfb7a90bd20104f67a2b1dfc9023c54473422;hp=3805d88ee6ddb6400c0edf1d92c66ad3c318b62e;hpb=8ca1269eac32c1ddc19d16dc4f74e9a1e3e0b8f9;p=pazpar2-moved-to-github.git diff --git a/src/pazpar2.c b/src/pazpar2.c index 3805d88..f0bd9be 100644 --- a/src/pazpar2.c +++ b/src/pazpar2.c @@ -1,4 +1,4 @@ -/* $Id: pazpar2.c,v 1.30 2007-01-15 04:34:28 quinn Exp $ */ +/* $Id: pazpar2.c,v 1.51 2007-03-20 07:27:51 adam Exp $ */ #include #include @@ -38,6 +38,7 @@ #include "reclists.h" #include "relevance.h" #include "config.h" +#include "database.h" #define MAX_CHUNK 15 @@ -45,7 +46,7 @@ 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); -static struct conf_retrievalprofile *database_retrieval_profile(struct database *db); +//static struct conf_retrievalprofile *database_retrieval_profile(struct database *db); void session_alert_watch(struct session *s, int what); IOCHAN channel_list = 0; // Master list of connections we're handling events to @@ -53,9 +54,6 @@ IOCHAN channel_list = 0; // Master list of connections we're handling events to static struct connection *connection_freelist = 0; static struct client *client_freelist = 0; -static struct host *hosts = 0; // The hosts we know about -static struct database *databases = 0; // The databases we know about - static char *client_states[] = { "Client_Connecting", "Client_Connected", @@ -176,7 +174,7 @@ static void send_search(IOCHAN i) for (ndb = 0; db->databases[ndb]; ndb++) databaselist[ndb] = db->databases[ndb]; - a->u.presentRequest->preferredRecordSyntax = + a->u.searchRequest->preferredRecordSyntax = yaz_oidval_to_z3950oid(global_parameters.odr_out, CLASS_RECSYN, VAL_USMARC); a->u.searchRequest->smallSetUpperBound = &ssub; @@ -265,7 +263,6 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) if (r->presentStatus && !*r->presentStatus && r->records) { yaz_log(YLOG_DEBUG, "Records in search response"); - cl->records += *r->numberOfRecordsReturned; ingest_records(cl, r->records); } cl->state = Client_Idle; @@ -319,7 +316,10 @@ char *normalize_mergekey(char *buf, int skiparticle) p++; } if (buf != pout) - *pout = '\0'; + do { + *(pout--) = '\0'; + } + while (pout > buf && *pout == ' '); return buf; } @@ -364,6 +364,8 @@ static void add_facet(struct session *s, const char *type, const char *value) { int i; + if (!*value) + return; for (i = 0; i < s->num_termlists; i++) if (!strcmp(s->termlists[i].name, type)) break; @@ -381,8 +383,6 @@ static void add_facet(struct session *s, const char *type, const char *value) termlist_insert(s->termlists[i].termlist, value); } -int yaz_marc_write_xml(); - static xmlDoc *normalize_record(struct client *cl, Z_External *rec) { struct conf_retrievalprofile *rprofile = cl->database->rprofile; @@ -421,6 +421,17 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record"); exit(1); } + + if (global_parameters.dump_records) + { + fprintf(stderr, "Input Record (normalized):\n----------------\n"); +#if LIBXML_VERSION >= 20600 + xmlDocFormatDump(stderr, rdoc, 1); +#else + xmlDocDump(stderr, rdoc); +#endif + } + for (m = rprofile->maplist; m; m = m->next) { xmlDoc *new; @@ -486,8 +497,8 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) struct record_cluster *cluster; struct session *se = cl->session; xmlChar *mergekey, *mergekey_norm; - xmlChar *type; - xmlChar *value; + xmlChar *type = 0; + xmlChar *value = 0; struct conf_service *service = global_parameters.server->service; if (!xdoc) @@ -503,6 +514,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) 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); @@ -512,6 +524,9 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) normalize_mergekey(mergekey_norm, 0); cluster = reclist_insert(se->reclist, res, 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 (!cluster) { /* no room for record */ @@ -520,7 +535,6 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) } relevance_newrec(se->relevance, cluster); - type = value = 0; for (n = root->children; n; n = n->next) { if (type) @@ -541,6 +555,10 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) type = xmlGetProp(n, "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(type, service->metadata[imeta].name)) @@ -567,7 +585,14 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) newm->next = 0; if (md->type == Metadata_type_generic) { - newm->data.text = nmem_strdup(se->nmem, value); + char *p, *pe; + for (p = 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) { @@ -588,7 +613,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) { struct record_metadata *mnode; for (mnode = *wheretoput; mnode; mnode = mnode->next) - if (!strcmp(mnode->data.text, mnode->data.text)) + if (!strcmp(mnode->data.text, newm->data.text)) break; if (!mnode) { @@ -611,7 +636,6 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) normalize_mergekey(s, (sk->type == Metadata_sortkey_skiparticle)); cluster->sortkeys[md->sortkey_offset]->text = s; - yaz_log(YLOG_LOG, "SK Longest: %s", s); } } } @@ -637,18 +661,14 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) (*wheretoput)->data.number.min = first; if (last > (*wheretoput)->data.number.max) (*wheretoput)->data.number.max = last; - if (sk) - { - union data_types *sdata = cluster->sortkeys[md->sortkey_offset]; - sdata->number.min = first; - sdata->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); } +#endif } else yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name); @@ -656,7 +676,21 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) if (md->rank) relevance_countwords(se->relevance, cluster, value, md->rank); if (md->termlist) - add_facet(se, type, value); + { + if (md->type == Metadata_type_year) + { + char year[64]; + sprintf(year, "%d", last); + add_facet(se, type, year); + if (first != last) + { + sprintf(year, "%d", first); + add_facet(se, type, year); + } + } + else + add_facet(se, type, value); + } xmlFree(type); xmlFree(value); type = value = 0; @@ -664,6 +698,10 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) else yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name); } + if (type) + xmlFree(type); + if (value) + xmlFree(value); xmlFreeDoc(xdoc); @@ -690,6 +728,7 @@ static void ingest_records(struct client *cl, Z_Records *r) { Z_NamePlusRecord *npr = rlist->records[i]; + cl->records++; if (npr->which != Z_NamePlusRecord_databaseRecord) { yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic"); @@ -731,7 +770,6 @@ static void do_presentResponse(IOCHAN i, Z_APDU *a) if (!*r->presentStatus && cl->state != Client_Error) { yaz_log(YLOG_DEBUG, "Good Present response"); - cl->records += *r->numberOfRecordsReturned; ingest_records(cl, r->records); cl->state = Client_Idle; } @@ -1002,6 +1040,8 @@ static int client_prep_connection(struct client *cl) return 0; } +#ifdef GAGA // Moved to database.c + // This function will most likely vanish when a proper target profile mechanism is // introduced. void load_simpletargets(const char *fn) @@ -1018,19 +1058,24 @@ void load_simpletargets(const char *fn) while (fgets(line, 255, f)) { char *url, *db; + char *name; struct host *host; struct database *database; if (strncmp(line, "target ", 7)) continue; + line[strlen(line) - 1] = '\0'; + + if ((name = strchr(line, ';'))) + *(name++) = '\0'; + url = line + 7; - url[strlen(url) - 1] = '\0'; - yaz_log(YLOG_DEBUG, "Target: %s", url); if ((db = strchr(url, '/'))) *(db++) = '\0'; else db = "Default"; + yaz_log(YLOG_LOG, "Target: %s, '%s'", url, db); for (host = hosts; host; host = host->next) if (!strcmp(url, host->hostport)) break; @@ -1083,6 +1128,10 @@ void load_simpletargets(const char *fn) strcpy(database->url, url); strcat(database->url, "/"); strcat(database->url, db); + if (name) + database->name = xstrdup(name); + else + database->name = 0; database->databases = xmalloc(2 * sizeof(char *)); database->databases[0] = xstrdup(db); @@ -1097,6 +1146,8 @@ void load_simpletargets(const char *fn) fclose(f); } +#endif + static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) { switch (n->kind) @@ -1189,40 +1240,25 @@ void session_alert_watch(struct session *s, int what) s->watchlist[what].data = 0; } -// This needs to be extended with selection criteria -static struct conf_retrievalprofile *database_retrieval_profile(struct database *db) +//callback for grep_databases +static void select_targets_callback(void *context, struct database *db) { - if (!config) - { - yaz_log(YLOG_FATAL, "Must load configuration (-f)"); - exit(1); - } - if (!config->retrievalprofiles) - { - yaz_log(YLOG_FATAL, "No retrieval profiles defined"); - } - return config->retrievalprofiles; + struct session *se = (struct session*) context; + struct client *cl = client_create(); + cl->database = db; + cl->session = se; + cl->next = se->clients; + se->clients = cl; } // This should be extended with parameters to control selection criteria // Associates a set of clients with a session; -int select_targets(struct session *se) +int select_targets(struct session *se, struct database_criterion *crit) { - struct database *db; - int c = 0; - while (se->clients) client_destroy(se->clients); - for (db = databases; db; db = db->next) - { - struct client *cl = client_create(); - cl->database = db; - cl->session = se; - cl->next = se->clients; - se->clients = cl; - c++; - } - return c; + + return grep_databases(se, crit, select_targets_callback); } int session_active_clients(struct session *s) @@ -1240,22 +1276,51 @@ int session_active_clients(struct session *s) return res; } -char *search(struct session *se, char *query) +// parses crit1=val1,crit2=val2,... +static struct database_criterion *parse_filter(NMEM m, const char *buf) +{ + struct database_criterion *res = 0; + char **values; + int num; + int i; + + if (!buf || !*buf) + return 0; + nmem_strsplit(m, ",", buf, &values, &num); + for (i = 0; i < num; i++) + { + struct database_criterion *new = nmem_malloc(m, sizeof(*new)); + char *eq = strchr(values[i], '='); + if (!eq) + { + yaz_log(YLOG_WARN, "Missing equal-sign in filter"); + return 0; + } + *(eq++) = '\0'; + new->name = values[i]; + new->value = eq; + new->next = res; + res = new; + } + return res; +} + +char *search(struct session *se, char *query, char *filter) { int live_channels = 0; struct client *cl; + struct database_criterion *criteria; yaz_log(YLOG_DEBUG, "Search"); + nmem_reset(se->nmem); + criteria = parse_filter(se->nmem, filter); strcpy(se->query, query); se->requestid++; - nmem_reset(se->nmem); + // Release any existing clients + select_targets(se, criteria); for (cl = se->clients; cl; cl = cl->next) { - cl->hits = -1; - cl->records = 0; - cl->diagnostic = 0; - if (client_prep_connection(cl)) live_channels++; } @@ -1282,7 +1347,7 @@ void destroy_session(struct session *s) while (s->clients) client_destroy(s->clients); nmem_destroy(s->nmem); - wrbuf_free(s->wrbuf, 1); + wrbuf_destroy(s->wrbuf); } struct session *new_session() @@ -1308,8 +1373,6 @@ struct session *new_session() session->watchlist[i].fun = 0; } - select_targets(session); - return session; } @@ -1321,7 +1384,8 @@ struct hitsbytarget *hitsbytarget(struct session *se, int *count) *count = 0; for (cl = se->clients; cl; cl = cl->next) { - strcpy(res[*count].id, cl->database->host->hostport); + res[*count].id = cl->database->url; + res[*count].name = cl->database->name; res[*count].hits = cl->hits; res[*count].records = cl->records; res[*count].diagnostic = cl->diagnostic; @@ -1497,7 +1561,7 @@ static void start_proxy(void) else return; - http_set_proxyaddr(hp); + http_set_proxyaddr(hp, ser->myurl ? ser->myurl : ""); } int main(int argc, char **argv) @@ -1538,7 +1602,8 @@ int main(int argc, char **argv) " -h [host:]port (REST protocol listener)\n" " -C cclconfig\n" " -s simpletargetfile\n" - " -p hostname[:portno] (HTTP proxy)\n"); + " -p hostname[:portno] (HTTP proxy)\n" + " -d (show internal records)\n"); exit(1); } } @@ -1552,7 +1617,8 @@ int main(int argc, char **argv) start_http_listener(); start_proxy(); - global_parameters.ccl_filter = load_cclfile("../etc/default.bib"); + 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"); global_parameters.odr_in = odr_createmem(ODR_DECODE);