X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Freclists.c;h=7912ca21b76fde18a3f181c8acbd7076337ef003;hb=705f49b5a94dcdba10bba40a3f63d312ebbc054c;hp=fc6a578148ca51a6468dd11ce7404ce0018aa412;hpb=aa06e37ed93f09b830953e51f28a0c90dc5138f5;p=pazpar2-moved-to-github.git

diff --git a/src/reclists.c b/src/reclists.c
index fc6a578..7912ca2 100644
--- a/src/reclists.c
+++ b/src/reclists.c
@@ -1,5 +1,5 @@
 /* This file is part of Pazpar2.
-   Copyright (C) 2006-2009 Index Data
+   Copyright (C) 2006-2010 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
@@ -25,30 +25,30 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 #include <yaz/yaz-util.h>
 
-#include "pazpar2.h"
+#include "ppmutex.h"
+#include "session.h"
 #include "reclists.h"
 #include "jenkins_hash.h"
 
 struct reclist
 {
     struct reclist_bucket **hashtable;
-    int hashtable_size;
-    int hashmask;
+    unsigned hash_size;
 
-    struct record_cluster **flatlist;
-    int flatlist_size;
     int num_records;
-    int pointer;
-
+    struct reclist_bucket *sorted_list;
+    struct reclist_bucket *sorted_ptr;
+    struct reclist_bucket **last;
     NMEM nmem;
+    YAZ_MUTEX mutex;
 };
 
-static struct reclist_sortparms *qsort_sortparms = 0; /* thread pr */
-
 struct reclist_bucket
 {
     struct record_cluster *record;
-    struct reclist_bucket *next;
+    struct reclist_bucket *hnext;
+    struct reclist_bucket *snext;
+    struct reclist_sortparms *sort_parms;
 };
 
 struct reclist_sortparms *reclist_parse_sortparms(NMEM nmem, const char *parms,
@@ -121,117 +121,157 @@ struct reclist_sortparms *reclist_parse_sortparms(NMEM nmem, const char *parms,
 
 static int reclist_cmp(const void *p1, const void *p2)
 {
-    struct record_cluster *r1 = (*(struct record_cluster**) p1);
-    struct record_cluster *r2 = (*(struct record_cluster**) p2);
+    struct reclist_sortparms *sortparms = 
+        (*(struct reclist_bucket **) p1)->sort_parms;
+    struct record_cluster *r1 = (*(struct reclist_bucket**) p1)->record;
+    struct record_cluster *r2 = (*(struct reclist_bucket**) p2)->record;
     struct reclist_sortparms *s;
     int res = 0;
 
-    for (s = qsort_sortparms; s && res == 0; s = s->next)
+    for (s = sortparms; s && res == 0; s = s->next)
     {
         union data_types *ut1 = r1->sortkeys[s->offset];
         union data_types *ut2 = r2->sortkeys[s->offset];
+        const char *s1, *s2;
         switch (s->type)
         {
-            const char *s1, *s2;
-            
-            case Metadata_sortkey_relevance:
-                res = r2->relevance - r1->relevance;
-                break;
-            case Metadata_sortkey_string:
-                s1 = ut1 ? ut1->text.sort : "";
-                s2 = ut2 ? ut2->text.sort : "";
-                res = strcmp(s2, s1);
-                if (res)
-                {
-                    if (s->increasing)
-                        res *= -1;
-                }
-                break;
-            case Metadata_sortkey_numeric:
-                if (ut1 && ut2)
-                {
-                    if (s->increasing)
-                        res = ut1->number.min  - ut2->number.min;
-                    else
-                        res = ut2->number.max  - ut1->number.max;
-                }
-                else if (ut1 && !ut2)
-                    res = -1;
-                else if (!ut1 && ut2)
-                    res = 1;
+        case Metadata_sortkey_relevance:
+            res = r2->relevance_score - r1->relevance_score;
+            break;
+        case Metadata_sortkey_string:
+            s1 = ut1 ? ut1->text.sort : "";
+            s2 = ut2 ? ut2->text.sort : "";
+            res = strcmp(s2, s1);
+            if (res)
+            {
+                if (s->increasing)
+                    res *= -1;
+            }
+            break;
+        case Metadata_sortkey_numeric:
+            if (ut1 && ut2)
+            {
+                if (s->increasing)
+                    res = ut1->number.min  - ut2->number.min;
                 else
-                    res = 0;
-                break;
-            default:
-                yaz_log(YLOG_WARN, "Bad sort type: %d", s->type);
+                    res = ut2->number.max  - ut1->number.max;
+            }
+            else if (ut1 && !ut2)
+                res = -1;
+            else if (!ut1 && ut2)
+                res = 1;
+            else
                 res = 0;
+            break;
+        default:
+            yaz_log(YLOG_WARN, "Bad sort type: %d", s->type);
+            res = 0;
         }
     }
+    if (res == 0)
+        res = strcmp(r1->recid, r2->recid);
     return res;
 }
 
 void reclist_sort(struct reclist *l, struct reclist_sortparms *parms)
 {
-    qsort_sortparms = parms;
-    qsort(l->flatlist, l->num_records, 
-          sizeof(struct record_cluster*), reclist_cmp);
-    reclist_rewind(l);
+    struct reclist_bucket **flatlist = xmalloc(sizeof(*flatlist) * l->num_records);
+    struct reclist_bucket *ptr;
+    struct reclist_bucket **prev;
+    int i = 0;
+
+    reclist_enter(l);
+
+    ptr = l->sorted_list;
+    prev = &l->sorted_list;
+    while (ptr)
+    {
+        ptr->sort_parms = parms;
+        flatlist[i] = ptr;
+        ptr = ptr->snext;
+        i++;
+    }
+    assert(i == l->num_records);
+
+    qsort(flatlist, l->num_records, sizeof(*flatlist), reclist_cmp);
+    for (i = 0; i < l->num_records; i++)
+    {
+        *prev = flatlist[i];
+        prev = &flatlist[i]->snext;
+    }
+    *prev = 0;
+    l->last = prev;
+
+    xfree(flatlist);
+
+    reclist_leave(l);
 }
 
 struct record_cluster *reclist_read_record(struct reclist *l)
 {
-    if (l && l->pointer < l->num_records)
-        return l->flatlist[l->pointer++];
+    if (l && l->sorted_ptr)
+    {
+        struct record_cluster *t = l->sorted_ptr->record;
+        l->sorted_ptr = l->sorted_ptr->snext;
+        return t;
+    }
     else
         return 0;
 }
 
-void reclist_rewind(struct reclist *l)
+void reclist_enter(struct reclist *l)
 {
+    yaz_mutex_enter(l->mutex);
     if (l)
-        l->pointer = 0;
+        l->sorted_ptr = l->sorted_list;
 }
 
-struct reclist *reclist_create(NMEM nmem, int numrecs)
+
+void reclist_leave(struct reclist *l)
 {
-    int hashsize = 1;
-    struct reclist *res;
+    yaz_mutex_leave(l->mutex);
+    if (l)
+        l->sorted_ptr = l->sorted_list;
+}
 
-    assert(numrecs);
-    while (hashsize < numrecs)
-        hashsize <<= 1;
-    res = nmem_malloc(nmem, sizeof(struct reclist));
+
+struct reclist *reclist_create(NMEM nmem)
+{
+    struct reclist *res = nmem_malloc(nmem, sizeof(struct reclist));
+    res->hash_size = 399;
     res->hashtable 
-        = nmem_malloc(nmem, hashsize * sizeof(struct reclist_bucket*));
-    memset(res->hashtable, 0, hashsize * sizeof(struct reclist_bucket*));
-    res->hashtable_size = hashsize;
+        = nmem_malloc(nmem, res->hash_size * sizeof(struct reclist_bucket*));
+    memset(res->hashtable, 0, res->hash_size * sizeof(struct reclist_bucket*));
     res->nmem = nmem;
-    res->hashmask = hashsize - 1; // Creates a bitmask
 
-    res->num_records = 0;
-    res->flatlist = nmem_malloc(nmem, numrecs * sizeof(struct record_cluster*));
-    res->flatlist_size = numrecs;
+    res->sorted_ptr = 0;
+    res->sorted_list = 0;
+    res->last = &res->sorted_list;
 
+    res->num_records = 0;
+    res->mutex = 0;
+    pazpar2_mutex_create(&res->mutex, "reclist");
     return res;
 }
 
-int reclist_get_num_records(struct reclist *l)
+void reclist_destroy(struct reclist *l)
 {
     if (l)
-        return l->num_records;
-    return 0;
+        yaz_mutex_destroy(&l->mutex);
 }
 
-struct record_cluster *reclist_get_cluster(struct reclist *l, int i)
+int reclist_get_num_records(struct reclist *l)
 {
-    return l->flatlist[i];
+    if (l)
+        return l->num_records;
+    return 0;
 }
 
 // Insert a record. Return record cluster (newly formed or pre-existing)
-struct record_cluster *reclist_insert( struct reclist *l,
-                                       struct conf_service *service, 
-                                       struct record  *record,
-                                       char *merge_key, int *total)
+struct record_cluster *reclist_insert(struct reclist *l,
+                                      struct conf_service *service, 
+                                      struct record *record,
+                                      const char *merge_key, int *total)
 {
     unsigned int bucket;
     struct reclist_bucket **p;
@@ -243,9 +283,10 @@ struct record_cluster *reclist_insert( struct reclist *l,
     assert(merge_key);
     assert(total);
 
-    bucket = jenkins_hash((unsigned char*) merge_key) & l->hashmask;
+    bucket = jenkins_hash((unsigned char*) merge_key) % l->hash_size;
 
-    for (p = &l->hashtable[bucket]; *p; p = &(*p)->next)
+    yaz_mutex_enter(l->mutex);
+    for (p = &l->hashtable[bucket]; *p; p = &(*p)->hnext)
     {
         // We found a matching record. Merge them
         if (!strcmp(merge_key, (*p)->record->merge_key))
@@ -257,39 +298,47 @@ struct record_cluster *reclist_insert( struct reclist *l,
             break;
         }
     }
-    if (!cluster && l->num_records < l->flatlist_size)
+    if (!cluster)
     {
         struct reclist_bucket *new =
-            nmem_malloc(l->nmem, sizeof(struct reclist_bucket));
-        struct record_cluster *newc =
-            nmem_malloc(l->nmem, sizeof(struct record_cluster));
+            nmem_malloc(l->nmem, sizeof(*new));
+
+        cluster = nmem_malloc(l->nmem, sizeof(*cluster));
         
         record->next = 0;
-        new->record = newc;
-        new->next = 0;
-        newc->records = record;
-        newc->merge_key = merge_key;
-        newc->relevance = 0;
-        newc->term_frequency_vec = 0;
-        newc->recid = merge_key;
+        new->record = cluster;
+        new->hnext = 0;
+        cluster->records = record;
+        cluster->merge_key = nmem_strdup(l->nmem, merge_key);
+        cluster->relevance_score = 0;
+        cluster->term_frequency_vec = 0;
+        cluster->recid = nmem_strdup(l->nmem, merge_key);
         (*total)++;
-        newc->metadata = nmem_malloc(l->nmem,
-                sizeof(struct record_metadata*) * service->num_metadata);
-        memset(newc->metadata, 0, 
+        cluster->metadata =
+            nmem_malloc(l->nmem,
+                        sizeof(struct record_metadata*) * service->num_metadata);
+        memset(cluster->metadata, 0, 
                sizeof(struct record_metadata*) * service->num_metadata);
-        newc->sortkeys = nmem_malloc(l->nmem,
-                sizeof(struct record_metadata*) * service->num_sortkeys);
-        memset(newc->sortkeys, 0,
+        cluster->sortkeys = 
+            nmem_malloc(l->nmem, sizeof(struct record_metadata*) * service->num_sortkeys);
+        memset(cluster->sortkeys, 0,
                sizeof(union data_types*) * service->num_sortkeys);
+ 
+        /* attach to hash list */
+        *p = new; 
 
-        *p = new;
-        l->flatlist[l->num_records++] = newc;
-        cluster = newc;
+        /* append to sorted_list */
+        *l->last = new;
+        l->last = &new->snext;
+        l->sorted_ptr = l->sorted_list;
+        new->snext = 0;
+
+        l->num_records++;
     }
+    yaz_mutex_leave(l->mutex);
     return cluster;
 }
 
-
 /*
  * Local variables:
  * c-basic-offset: 4