Basic full record display logic.
[pazpar2-moved-to-github.git] / src / pazpar2.c
index 8ee4ea9..1f23d6c 100644 (file)
@@ -1,4 +1,4 @@
-/* $Id: pazpar2.c,v 1.21 2007-01-09 18:06:28 quinn Exp $ */;
+/* $Id: pazpar2.c,v 1.35 2007-01-16 03:38:49 quinn Exp $ */
 
 #include <stdlib.h>
 #include <stdio.h>
@@ -29,6 +29,8 @@
 #include <yaz/timing.h>
 #endif
 
+#include <netinet/in.h>
+
 #include "pazpar2.h"
 #include "eventl.h"
 #include "http.h"
@@ -70,6 +72,8 @@ static char *client_states[] = {
 // Note: Some things in this structure will eventually move to configuration
 struct parameters global_parameters = 
 {
+    "",
+    "",
     0,
     0,
     30,
@@ -282,10 +286,27 @@ 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 +319,10 @@ char *normalize_mergekey(char *buf)
             p++;
     }
     if (buf != pout)
-        *pout = '\0';
+        do {
+            *(pout--) = '\0';
+        }
+        while (pout > buf && *pout == ' ');
 
     return buf;
 }
@@ -343,6 +367,8 @@ 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;
@@ -419,7 +445,11 @@ 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;
 }
@@ -478,18 +508,15 @@ 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->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 (!cluster)
     {
         /* no room for record */
@@ -511,18 +538,21 @@ 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);
             // 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)
@@ -542,7 +572,12 @@ static struct record *ingest_record(struct client *cl, Z_External *rec)
             newm->next = 0;
             if (md->type == Metadata_type_generic)
             {
+                char *p;
                 newm->data.text = nmem_strdup(se->nmem, value);
+                for (p = newm->data.text + strlen(newm->data.text) - 1;
+                        p > newm->data.text && strchr(" ,/.", *p); p--)
+                    *p = '\0';
+
             }
             else if (md->type == Metadata_type_year)
             {
@@ -563,7 +598,7 @@ 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)
                 {
@@ -575,7 +610,19 @@ 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)
             {
@@ -588,16 +635,31 @@ static struct record *ingest_record(struct client *cl, Z_External *rec)
                 if (!*wheretoput)
                 {
                     *wheretoput = newm;
-                    (*wheretoput)->data.year.year1 = first;
-                    (*wheretoput)->data.year.year2 = last;
+                    (*wheretoput)->data.number.min = first;
+                    (*wheretoput)->data.number.max = last;
+                    if (sk)
+                        cluster->sortkeys[md->sortkey_offset] = &newm->data;
                 }
                 else
                 {
-                    if (first < (*wheretoput)->data.year.year1)
-                        (*wheretoput)->data.year.year1 = first;
-                    if (last > (*wheretoput)->data.year.year2)
-                        (*wheretoput)->data.year.year2 = last;
+                    if (first < (*wheretoput)->data.number.min)
+                        (*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);
@@ -862,7 +924,8 @@ 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;
     }
 
@@ -1018,7 +1081,7 @@ 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);
@@ -1215,7 +1278,7 @@ 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
@@ -1304,16 +1367,35 @@ 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;
@@ -1351,7 +1433,7 @@ 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)
@@ -1386,13 +1468,55 @@ 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);
+}
+
 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);
@@ -1405,14 +1529,13 @@ 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);
@@ -1438,12 +1561,8 @@ int main(int argc, char **argv)
     }
     global_parameters.server = config->servers;
 
-    if (!setport)
-    {
-        fprintf(stderr, "Set command port with -h\n");
-        exit(1);
-    }
-
+    start_http_listener();
+    start_proxy();
     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");