X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Flogic.c;h=697da4d50ef95859c8a50121d653ed5a2f5df9fe;hb=8cea11ee29fe71f11f61d1d05295b4e4290f6a98;hp=96a13ed4d4c6d852ffb247e5fb9885da58533ecb;hpb=b485fddaff33e8aa6ee0e8f8c8673a1ef8f786b3;p=pazpar2-moved-to-github.git

diff --git a/src/logic.c b/src/logic.c
index 96a13ed..697da4d 100644
--- a/src/logic.c
+++ b/src/logic.c
@@ -1,5 +1,5 @@
 /* This file is part of Pazpar2.
-   Copyright (C) 2006-2008 Index Data
+   Copyright (C) 2006-2009 Index Data
 
 Pazpar2 is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free
@@ -57,7 +57,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <yaz/timing.h>
 #endif
 
-
+#include "parameters.h"
 #include "pazpar2.h"
 #include "eventl.h"
 #include "http.h"
@@ -68,6 +68,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include "client.h"
 #include "settings.h"
 #include "normalize7bit.h"
+#include "marcmap.h"
 
 #define TERMLIST_HIGH_SCORE 25
 
@@ -76,23 +77,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 // Note: Some things in this structure will eventually move to configuration
 struct parameters global_parameters = 
 {
-    "",
-    "",
-    "", 
-    0,
     0,   // dump_records
     0,   // debug_mode
-    30,  // operations timeout 
-    "81",
-    "Index Data PazPar2",
-    VERSION,
-    60,   // session timeout 
     100,
-    MAX_CHUNK,
-    0,
-    0,
-    180, // Z39.50 session timeout
-    15   // Connect timeout
 };
 
 static void log_xml_doc(xmlDoc *doc)
@@ -196,7 +183,7 @@ xmlDoc *record_to_xml(struct session_database *sdb, const char *rec)
 static void insert_settings_parameters(struct session_database *sdb,
                                        struct session *se, char **parms)
 {
-    struct conf_service *service = global_parameters.server->service;
+    struct conf_service *service = se->service;
     int i;
     int nparms = 0;
     int offset = 0;
@@ -207,7 +194,7 @@ static void insert_settings_parameters(struct session_database *sdb,
         int setting;
 
         if (md->setting == Metadata_setting_parameter &&
-            (setting = settings_offset(md->name)) > 0)
+            (setting = settings_offset(service, md->name)) > 0)
         {
             const char *val = session_setting_oneval(sdb, setting);
             if (val && nparms < MAX_XSLT_ARGS)
@@ -229,9 +216,9 @@ static void insert_settings_parameters(struct session_database *sdb,
 }
 
 // Add static values from session database settings if applicable
-static void insert_settings_values(struct session_database *sdb, xmlDoc *doc)
+static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
+    struct conf_service *service)
 {
-    struct conf_service *service = global_parameters.server->service;
     int i;
 
     for (i = 0; i < service->num_metadata; i++)
@@ -240,7 +227,7 @@ static void insert_settings_values(struct session_database *sdb, xmlDoc *doc)
         int offset;
 
         if (md->setting == Metadata_setting_postproc &&
-            (offset = settings_offset(md->name)) > 0)
+            (offset = settings_offset(service, md->name)) > 0)
         {
             const char *val = session_setting_oneval(sdb, offset);
             if (val)
@@ -271,8 +258,17 @@ xmlDoc *normalize_record(struct session_database *sdb, struct session *se,
 
                 insert_settings_parameters(sdb, se, parms);
 
-                new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms);
-                root= xmlDocGetRootElement(new);
+                if (m->stylesheet)
+                {
+                    new = xsltApplyStylesheet(m->stylesheet, rdoc, (const char **) parms);
+                }
+                else if (m->marcmap)
+                {
+                    new = marcmap_apply(m->marcmap, rdoc);
+                }
+
+                root = xmlDocGetRootElement(new);
+
                 if (!new || !root || !(root->children))
                 {
                     yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
@@ -287,7 +283,7 @@ xmlDoc *normalize_record(struct session_database *sdb, struct session *se,
             rdoc = new;
         }
 
-        insert_settings_values(sdb, rdoc);
+        insert_settings_values(sdb, rdoc, se->service);
 
         if (global_parameters.dump_records)
         {
@@ -354,14 +350,37 @@ static int prepare_map(struct session *se, struct session_database *sdb)
         nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
         for (i = 0; i < num; i++)
         {
+            WRBUF fname = conf_get_fname(se->service, stylesheets[i]);
+            
             (*m) = nmem_malloc(se->session_nmem, sizeof(**m));
             (*m)->next = 0;
-            if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
+            
+            // XSLT
+            if (!strcmp(&stylesheets[i][strlen(stylesheets[i])-4], ".xsl")) 
+            {    
+                (*m)->marcmap = NULL;
+                if (!((*m)->stylesheet =
+                      xsltParseStylesheetFile((xmlChar *) wrbuf_cstr(fname))))
+                {
+                    yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
+                            stylesheets[i]);
+                    wrbuf_destroy(fname);
+                    return -1;
+                }
+            }
+            // marcmap
+            else if (!strcmp(&stylesheets[i][strlen(stylesheets[i])-5], ".mmap"))
             {
-                yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load stylesheet: %s",
-                        stylesheets[i]);
-                return -1;
+                (*m)->stylesheet = NULL;
+                if (!((*m)->marcmap = marcmap_load(wrbuf_cstr(fname), se->session_nmem)))
+                {
+                    yaz_log(YLOG_FATAL|YLOG_ERRNO, "Unable to load marcmap: %s",
+                            stylesheets[i]);
+                    wrbuf_destroy(fname);
+                    return -1;
+                }
             }
+            wrbuf_destroy(fname);
             m = &(*m)->next;
         }
     }
@@ -486,10 +505,14 @@ static struct database_criterion *parse_filter(NMEM m, const char *buf)
         int subnum;
         int subi;
         struct database_criterion *new = nmem_malloc(m, sizeof(*new));
-        char *eq = strchr(values[i], '=');
+        char *eq;
+        if (eq = strchr(values[i], '='))
+            new->type = PAZPAR2_STRING_MATCH;
+        if (eq = strchr(values[i], '~'))
+            new->type = PAZPAR2_SUBSTRING_MATCH;
         if (!eq)
         {
-            yaz_log(YLOG_WARN, "Missing equal-sign in filter");
+            yaz_log(YLOG_WARN, "Missing equal-signi/tilde in filter");
             return 0;
         }
         *(eq++) = '\0';
@@ -511,7 +534,7 @@ static struct database_criterion *parse_filter(NMEM m, const char *buf)
 }
 
 enum pazpar2_error_code search(struct session *se,
-                               char *query, char *filter,
+                               const char *query, const char *filter,
                                const char **addinfo)
 {
     int live_channels = 0;
@@ -532,7 +555,7 @@ enum pazpar2_error_code search(struct session *se,
     live_channels = select_targets(se, criteria);
     if (live_channels)
     {
-        int maxrecs = live_channels * global_parameters.toget;
+        int maxrecs = live_channels * global_parameters.toget; // This is buggy!!!
         se->reclist = reclist_create(se->nmem, maxrecs);
         se->expected_maxrecs = maxrecs;
     }
@@ -552,7 +575,8 @@ enum pazpar2_error_code search(struct session *se,
         else
         {
             no_working++;
-            if (client_prep_connection(cl))
+            if (client_prep_connection(cl, se->service->z3950_operation_timeout,
+                    se->service->z3950_session_timeout))
                 client_start_search(cl);
         }
     }
@@ -570,8 +594,9 @@ enum pazpar2_error_code search(struct session *se,
 static void session_init_databases_fun(void *context, struct database *db)
 {
     struct session *se = (struct session *) context;
+    struct conf_service *service = se->service;
     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
-    int num = settings_num();
+    int num = settings_num(service);
     int i;
 
     new->database = db;
@@ -604,7 +629,7 @@ static void session_database_destroy(struct session_database *sdb)
 void session_init_databases(struct session *se)
 {
     se->databases = 0;
-    predef_grep_databases(se, 0, session_init_databases_fun);
+    predef_grep_databases(se, se->service, 0, session_init_databases_fun);
 }
 
 // Probably session_init_databases_fun should be refactored instead of
@@ -612,7 +637,9 @@ void session_init_databases(struct session *se)
 static struct session_database *load_session_database(struct session *se, 
                                                       char *id)
 {
-    struct database *db = find_database(id, 0);
+    struct database *db = find_database(id, 0, se->service);
+
+    resolve_database(db);
 
     session_init_databases_fun((void*) se, db);
     // New sdb is head of se->databases list
@@ -636,8 +663,9 @@ void session_apply_setting(struct session *se, char *dbname, char *setting,
                            char *value)
 {
     struct session_database *sdb = find_session_database(se, dbname);
+    struct conf_service *service = se->service;
     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
-    int offset = settings_offset_cprefix(setting);
+    int offset = settings_offset_cprefix(service, setting);
 
     if (offset < 0)
     {
@@ -683,16 +711,18 @@ void destroy_session(struct session *s)
     for (sdb = s->databases; sdb; sdb = sdb->next)
         session_database_destroy(sdb);
     nmem_destroy(s->nmem);
+    service_destroy(s->service);
     wrbuf_destroy(s->wrbuf);
 }
 
-struct session *new_session(NMEM nmem) 
+struct session *new_session(NMEM nmem, struct conf_service *service) 
 {
     int i;
     struct session *session = nmem_malloc(nmem, sizeof(*session));
 
     yaz_log(YLOG_DEBUG, "New Pazpar2 session");
-    
+
+    session->service = service;
     session->relevance = 0;
     session->total_hits = 0;
     session->total_records = 0;
@@ -765,14 +795,24 @@ void report_nmem_stats(void)
 }
 #endif
 
-struct record_cluster *show_single(struct session *s, const char *id)
+struct record_cluster *show_single(struct session *s, const char *id,
+                                   struct record_cluster **prev_r,
+                                   struct record_cluster **next_r)
 {
     struct record_cluster *r;
 
     reclist_rewind(s->reclist);
+    *prev_r = 0;
+    *next_r = 0;
     while ((r = reclist_read_record(s->reclist)))
+    {
         if (!strcmp(r->recid, id))
+        {
+            *next_r = reclist_read_record(s->reclist);
             return r;
+        }
+        *prev_r = r;
+    }
     return 0;
 }
 
@@ -864,52 +904,10 @@ void statistics(struct session *se, struct statistics *stat)
     stat->num_clients = count;
 }
 
-int 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);
-        }
-    }
-    return http_init(hp);
-}
-
-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 : "");
-}
-
 
 // Master list of connections we're handling events to
-static IOCHAN channel_list = 0; 
+static IOCHAN channel_list = 0;  /* thread pr */
+
 void pazpar2_add_channel(IOCHAN chan)
 {
     chan->next = channel_list;
@@ -951,68 +949,206 @@ static struct record_metadata *record_metadata_init(
     return rec_md;
 }
 
+static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
+                                struct conf_service *service, NMEM nmem)
+{
+    char *mergekey_norm = 0;
+    xmlNode *root = xmlDocGetRootElement(doc);
+    WRBUF norm_wr = wrbuf_alloc();
+    xmlNode *n;
+
+    /* consider mergekey from XSL first */
+    xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
+    if (mergekey)
+    {
+        const char *norm_str;
+        pp2_relevance_token_t prt =
+            pp2_relevance_tokenize(
+                service->mergekey_pct,
+                (const char *) mergekey);
+        
+        while ((norm_str = pp2_relevance_token_next(prt)))
+        {
+            if (*norm_str)
+            {
+                if (wrbuf_len(norm_wr))
+                    wrbuf_puts(norm_wr, " ");
+                wrbuf_puts(norm_wr, norm_str);
+            }
+        }
+        pp2_relevance_token_destroy(prt);
+        xmlFree(mergekey);
+    }
+    else
+    {
+        /* no mergekey defined in XSL. Look for mergekey metadata instead */
+        for (n = root->children; n; n = n->next)
+        {
+            if (n->type != XML_ELEMENT_NODE)
+                continue;
+            if (!strcmp((const char *) n->name, "metadata"))
+            {
+                struct conf_metadata *ser_md = 0;
+                int md_field_id = -1;
+                
+                xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
+                
+                if (!type)
+                    continue;
+                
+                md_field_id 
+                    = conf_service_metadata_field_id(service, 
+                                                     (const char *) type);
+                if (md_field_id >= 0)
+                {
+                    ser_md = &service->metadata[md_field_id];
+                    if (ser_md->mergekey == Metadata_mergekey_yes)
+                    {
+                        xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
+                        if (value)
+                        {
+                            const char *norm_str;
+                            pp2_relevance_token_t prt =
+                                pp2_relevance_tokenize(
+                                    service->mergekey_pct,
+                                    (const char *) value);
+                            
+                            while ((norm_str = pp2_relevance_token_next(prt)))
+                            {
+                                if (*norm_str)
+                                {
+                                    if (wrbuf_len(norm_wr))
+                                        wrbuf_puts(norm_wr, " ");
+                                    wrbuf_puts(norm_wr, norm_str);
+                                }
+                            }
+                            xmlFree(value);
+                            pp2_relevance_token_destroy(prt);
+                        }
+                    }
+                }
+                xmlFree(type);
+            }
+        }
+    }
+
+    /* generate unique key if none is not generated already or is empty */
+    if (wrbuf_len(norm_wr) == 0)
+    {
+        wrbuf_printf(norm_wr, "%s-%d",
+                     client_get_database(cl)->database->url, record_no);
+    }
+    if (wrbuf_len(norm_wr) > 0)
+        mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
+    wrbuf_destroy(norm_wr);
+    return mergekey_norm;
+}
+
+/** \brief see if metadata for pz:recordfilter exists 
+    \param root xml root element of normalized record
+    \param sdb session database for client
+    \retval 0 if there is no metadata for pz:recordfilter
+    \retval 1 if there is metadata for pz:recordfilter
+
+    If there is no pz:recordfilter defined, this function returns 1
+    as well.
+*/
+    
+static int check_record_filter(xmlNode *root, struct session_database *sdb)
+{
+    int match = 0;
+    xmlNode *n;
+    const char *s;
+    s = session_setting_oneval(sdb, PZ_RECORDFILTER);
+
+    if (!s || !*s)
+        return 1;
+
+    for (n = root->children; n; n = n->next)
+    {
+        if (n->type != XML_ELEMENT_NODE)
+            continue;
+        if (!strcmp((const char *) n->name, "metadata"))
+        {
+            xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
+            if (type)
+            {
+                size_t len;
+                const char *eq = strchr(s, '=');
+                if (eq)
+                    len = eq - s;
+                else
+                    len = strlen(s);
+                if (len == strlen((const char *)type) &&
+                    !memcmp((const char *) type, s, len))
+                {
+                    xmlChar *value = xmlNodeGetContent(n);
+                    if (value && *value)
+                    {
+                        if (!eq || strstr((const char *) value, eq+1))
+                            match = 1;
+                    }
+                    xmlFree(value);
+                }
+                xmlFree(type);
+            }
+        }
+    }
+    return match;
+}
+
+
+/** \brief ingest XML record
+    \param cl client holds the result set for record
+    \param rec record buffer (0 terminated)
+    \param record_no record position (1, 2, ..)
+    \returns resulting record or NULL on failure
+*/
 struct record *ingest_record(struct client *cl, const char *rec,
                              int record_no)
 {
-    xmlDoc *xdoc = normalize_record(client_get_database(cl),
-                                    client_get_session(cl), rec);
+    struct session_database *sdb = client_get_database(cl);
+    struct session *se = client_get_session(cl);
+    xmlDoc *xdoc = normalize_record(sdb, se, rec);
     xmlNode *root, *n;
     struct record *record;
     struct record_cluster *cluster;
-    struct session *se = client_get_session(cl);
-    xmlChar *mergekey, *mergekey_norm;
+    const char *mergekey_norm;
     xmlChar *type = 0;
     xmlChar *value = 0;
-    struct conf_service *service = global_parameters.server->service;
-    const char *norm_str = 0;
-    pp2_relevance_token_t prt = 0;
-    WRBUF norm_wr = 0;
+    struct conf_service *service = se->service;
 
     if (!xdoc)
         return 0;
 
     root = xmlDocGetRootElement(xdoc);
-    if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
+
+    if (!check_record_filter(root, sdb))
     {
-        yaz_log(YLOG_WARN, "No mergekey found in record");
+        yaz_log(YLOG_WARN, "Filtered out record no %d from %s", record_no,
+            sdb->database->url);
         xmlFreeDoc(xdoc);
         return 0;
     }
-    
-    record = record_create(se->nmem, 
-                           service->num_metadata, service->num_sortkeys, cl,
-                           record_no);
-
-    prt = pp2_relevance_tokenize(
-        global_parameters.server->mergekey_pct, (const char *) mergekey);
-
 
-    norm_wr = wrbuf_alloc();
-    
-    while ((norm_str = pp2_relevance_token_next(prt)))
+    mergekey_norm = get_mergekey(xdoc, cl, record_no, service, se->nmem);
+    if (!mergekey_norm)
     {
-        if (*norm_str)
-        {
-            if (wrbuf_len(norm_wr))
-                wrbuf_puts(norm_wr, " ");
-            wrbuf_puts(norm_wr, norm_str);
-        }
+        yaz_log(YLOG_WARN, "Got no mergekey");
+        xmlFreeDoc(xdoc);
+        return 0;
     }
-        
-    mergekey_norm = (xmlChar *)nmem_strdup(se->nmem, wrbuf_cstr(norm_wr));
-    wrbuf_destroy(norm_wr);
-
-    pp2_relevance_token_destroy(prt);
+    record = record_create(se->nmem, 
+                           service->num_metadata, service->num_sortkeys, cl,
+                           record_no);
 
-    xmlFree(mergekey);
-    
     cluster = reclist_insert(se->reclist, 
-                             global_parameters.server->service, 
+                             service, 
                              record, (char *) mergekey_norm, 
                              &se->total_merged);
     if (global_parameters.dump_records)
         yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
-                client_get_database(cl)->database->url, record_no);
+                sdb->database->url, record_no);
     if (!cluster)
     {
         /* no room for record */
@@ -1021,10 +1157,10 @@ struct record *ingest_record(struct client *cl, const char *rec,
     }
     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)
     {
+        pp2_relevance_token_t prt;
         if (type)
             xmlFree(type);
         if (value)
@@ -1121,7 +1257,7 @@ struct record *ingest_record(struct client *cl, const char *rec,
                                             sizeof(union data_types));
                          
                         prt = pp2_relevance_tokenize(
-                            global_parameters.server->sort_pct,
+                            service->sort_pct,
                             rec_md->data.text.disp);
 
                         pp2_relevance_token_next(prt);
@@ -1228,7 +1364,9 @@ struct record *ingest_record(struct client *cl, const char *rec,
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab
  */
+