X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fpazpar2.c;h=acc494be48f526c34965b3712495f209de01e80b;hb=9a660575bb51d190608e5cdec416bb8a9acf3389;hp=22ec100118f8cdf777627433abc878c4e85ab762;hpb=77b4d51d81340f3374ae8a97aea0177a0fbcd336;p=pazpar2-moved-to-github.git diff --git a/src/pazpar2.c b/src/pazpar2.c index 22ec100..acc494b 100644 --- a/src/pazpar2.c +++ b/src/pazpar2.c @@ -1,4 +1,4 @@ -/* $Id: pazpar2.c,v 1.19 2007-01-08 18:32:35 quinn Exp $ */; +/* $Id: pazpar2.c,v 1.49 2007-03-15 16:50:56 quinn Exp $ */ #include #include @@ -29,6 +29,8 @@ #include #endif +#include + #include "pazpar2.h" #include "eventl.h" #include "http.h" @@ -36,6 +38,7 @@ #include "reclists.h" #include "relevance.h" #include "config.h" +#include "database.h" #define MAX_CHUNK 15 @@ -51,9 +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", @@ -70,6 +70,8 @@ static char *client_states[] = { // Note: Some things in this structure will eventually move to configuration struct parameters global_parameters = { + "", + "", 0, 0, 30, @@ -172,7 +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; @@ -261,7 +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; @@ -282,10 +283,27 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) } } -char *normalize_mergekey(char *buf) +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)) @@ -298,7 +316,10 @@ char *normalize_mergekey(char *buf) p++; } if (buf != pout) - *pout = '\0'; + do { + *(pout--) = '\0'; + } + while (pout > buf && *pout == ' '); return buf; } @@ -343,6 +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; @@ -360,8 +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; @@ -400,6 +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; @@ -419,11 +451,44 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) if (global_parameters.dump_records) { fprintf(stderr, "Record:\n----------------\n"); +#if LIBXML_VERSION >= 20600 xmlDocFormatDump(stderr, rdoc, 1); +#else + xmlDocDump(stderr, 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; + } + buf = e; + } + return *first; +} + static struct record *ingest_record(struct client *cl, Z_External *rec) { xmlDoc *xdoc = normalize_record(cl, rec); @@ -432,8 +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) @@ -449,18 +514,19 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) res = nmem_malloc(se->nmem, sizeof(struct record)); res->next = 0; - res->target_offset = -1; - res->term_frequency_vec = 0; + res->client = cl; res->metadata = nmem_malloc(se->nmem, sizeof(struct record_metadata*) * service->num_metadata); - bzero(res->metadata, sizeof(struct record_metadata*) * service->num_metadata); - res->relevance = 0; + memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata); mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey); xmlFree(mergekey); - normalize_mergekey(mergekey_norm); + normalize_mergekey(mergekey_norm, 0); - cluster = reclist_insert(se->reclist, res, mergekey_norm); + 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 */ @@ -469,7 +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) @@ -482,17 +547,25 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) continue; if (!strcmp(n->name, "metadata")) { - type = xmlGetProp(n, "type"); - value = xmlNodeListGetString(xdoc, n->children, 0); struct conf_metadata *md = 0; + struct conf_sortkey *sk = 0; struct record_metadata **wheretoput, *newm; int imeta; + int first, last; + + 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)) { md = &service->metadata[imeta]; + if (md->sortkey_offset >= 0) + sk = &service->sortkeys[md->sortkey_offset]; break; } if (!md) @@ -512,18 +585,35 @@ 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) + { + if (extract_years(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(mnode->data.text, mnode->data.text)) + if (!strcmp(mnode->data.text, newm->data.text)) break; if (!mnode) { @@ -535,19 +625,72 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) { if (!*wheretoput || strlen(newm->data.text) > strlen((*wheretoput)->data.text)) - *wheretoput = newm; + { + *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; + } + } } 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); + } +#endif + } else yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name); - relevance_countwords(se->relevance, cluster, value, 4); + 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; @@ -555,6 +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); @@ -581,6 +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"); @@ -622,7 +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; } @@ -804,7 +951,8 @@ static struct connection *connection_create(struct client *cl) if (!(addr = cs_straddr(link, cl->database->host->ipport))) { - yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?"); + yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address %s failed?", + cl->database->host->ipport); return 0; } @@ -892,6 +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) @@ -908,19 +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; @@ -960,7 +1115,7 @@ void load_simpletargets(const char *fn) } assert(addrinfo->ai_family == PF_INET); memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4); - sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s", + sprintf(ipport, "%u.%u.%u.%u:%s", addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port); host->ipport = xstrdup(ipport); freeaddrinfo(addrinfo); @@ -973,6 +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); @@ -987,6 +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) @@ -1079,40 +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) { - 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, select_targets_callback); } int session_active_clients(struct session *s) @@ -1157,7 +1303,7 @@ char *search(struct session *se, char *query) se->reclist = reclist_create(se->nmem, maxrecs); extract_terms(se->nmem, query, p); se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs); - se->total_records = se->total_hits = 0; + se->total_records = se->total_hits = se->total_merged = 0; se->expected_maxrecs = maxrecs; } else @@ -1211,7 +1357,8 @@ 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; @@ -1246,16 +1393,35 @@ void report_nmem_stats(void) } #endif -struct record_cluster **show(struct session *s, int start, int *num, int *total, - int *sumhits, NMEM nmem_show) +struct record_cluster *show_single(struct session *s, int id) +{ + struct record_cluster *r; + + reclist_rewind(s->reclist); + while ((r = reclist_read_record(s->reclist))) + if (r->recid == id) + return r; + 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 **recs = nmem_malloc(nmem_show, *num * sizeof(struct record_cluster *)); + struct reclist_sortparms *spp; int i; #if USE_TIMING yaz_timing_t t = yaz_timing_create(); #endif - relevance_prepare_read(s->relevance, s->reclist); + + 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; @@ -1293,7 +1459,7 @@ void statistics(struct session *se, struct statistics *stat) struct client *cl; int count = 0; - bzero(stat, sizeof(*stat)); + memset(stat, 0, sizeof(*stat)); for (cl = se->clients; cl; cl = cl->next) { if (!cl->connection) @@ -1328,13 +1494,55 @@ static CCL_bibset load_cclfile(const char *fn) return res; } +static void start_http_listener(void) +{ + char hp[128] = ""; + struct conf_server *ser = global_parameters.server; + + if (*global_parameters.listener_override) + strcpy(hp, global_parameters.listener_override); + else + { + strcpy(hp, ser->host ? ser->host : ""); + if (ser->port) + { + if (*hp) + strcat(hp, ":"); + sprintf(hp + strlen(hp), "%d", ser->port); + } + } + http_init(hp); +} + +static void start_proxy(void) +{ + char hp[128] = ""; + struct conf_server *ser = global_parameters.server; + + if (*global_parameters.proxy_override) + strcpy(hp, global_parameters.proxy_override); + else if (ser->proxy_host || ser->proxy_port) + { + strcpy(hp, ser->proxy_host ? ser->proxy_host : ""); + if (ser->proxy_port) + { + if (*hp) + strcat(hp, ":"); + sprintf(hp + strlen(hp), "%d", ser->proxy_port); + } + } + else + return; + + http_set_proxyaddr(hp, ser->myurl ? ser->myurl : ""); +} + int main(int argc, char **argv) { int ret; char *arg; - int setport = 0; - if (signal(SIGPIPE, SIG_IGN) < 0) + if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) yaz_log(YLOG_WARN|YLOG_ERRNO, "signal"); yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0); @@ -1347,14 +1555,13 @@ int main(int argc, char **argv) exit(1); break; case 'h': - http_init(arg); - setport++; + strcpy(global_parameters.listener_override, arg); break; case 'C': global_parameters.ccl_filter = load_cclfile(arg); break; case 'p': - http_set_proxyaddr(arg); + strcpy(global_parameters.proxy_override, arg); break; case 's': load_simpletargets(arg); @@ -1368,7 +1575,8 @@ 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); } } @@ -1380,13 +1588,10 @@ int main(int argc, char **argv) } global_parameters.server = config->servers; - if (!setport) - { - fprintf(stderr, "Set command port with -h\n"); - exit(1); - } - - global_parameters.ccl_filter = load_cclfile("../etc/default.bib"); + 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"); global_parameters.odr_in = odr_createmem(ODR_DECODE);