For estimated hit counts, Zebra returns resultSetStatus=estimtate as

[idzebra-moved-to-github.git] / index / zebraapi.c

diff --git a/index/zebraapi.c b/index/zebraapi.c
index 7788602..a8e8039 100644 (file)
--- a/index/zebraapi.c
+++ b/index/zebraapi.c
@@ -1,5 +1,5 @@
-/* $Id: zebraapi.c,v 1.232 2006-11-17 13:47:22 marc Exp $
-   Copyright (C) 1995-2006
+/* $Id: zebraapi.c,v 1.243 2007-01-16 15:01:15 adam Exp $
+   Copyright (C) 1995-2007
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -84,9 +84,10 @@ static ZEBRA_RES zebra_flush_reg (ZebraHandle zh)
     ZEBRA_CHECK_HANDLE(zh);
     yaz_log(log_level, "zebra_flush_reg");
     zebraExplain_flush (zh->reg->zei, zh);
-    
-    extract_flushWriteKeys (zh, 1 /* final */);
-    zebra_index_merge (zh );
+
+    key_block_flush(zh->reg->key_block, 1);
+
+    zebra_index_merge(zh);
     return ZEBRA_OK;
 }
 
@@ -173,12 +174,12 @@ ZebraHandle zebra_open(ZebraService zs, Res res)
     return zh;
 }
 
-ZebraService zebra_start (const char *configName)
+ZebraService zebra_start(const char *configName)
 {
     return zebra_start_res(configName, 0, 0);
 }
 
-ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res)
+ZebraService zebra_start_res(const char *configName, Res def_res, Res over_res)
 {
     Res res;
 
@@ -208,7 +209,18 @@ ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res)
                res_close(res);
                return 0;
            }
+            if (zebra_check_res(res))
+            {
+                yaz_log(YLOG_FATAL, "Configuration error(s) for %s",
+                        configName);
+                return 0;
+            }
        }
+        else
+        {
+            zebra_check_res(res);
+        }
+
        zh = xmalloc(sizeof(*zh));
         zh->global_res = res;
         zh->sessions = 0;
@@ -246,6 +258,7 @@ ZebraService zebra_start_res (const char *configName, Res def_res, Res over_res)
            }
        }
 
+        zh->timing = yaz_timing_create();
         zh->path_root = res_get (zh->global_res, "root");
        zh->nmem = nmem_create();
        zh->record_classes = recTypeClass_create (zh->global_res, zh->nmem);
@@ -302,6 +315,7 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name,
     const char *recordCompression = 0;
     const char *profilePath;
     char cwd[1024];
+    int sort_type = ZEBRA_SORT_TYPE_FLAT;
     ZEBRA_RES ret = ZEBRA_OK;
 
     ASSERTZS;
@@ -347,8 +361,7 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name,
     }
 
     getcwd(cwd, sizeof(cwd)-1);
-    profilePath = res_get_def(res, "profilePath", DEFAULT_PROFILE_PATH);
-    yaz_log(YLOG_DEBUG, "profilePath=%s cwd=%s", profilePath, cwd);
+    profilePath = res_get_def(res, "profilePath", 0);
 
     data1_set_tabpath (reg->dh, profilePath);
     data1_set_tabroot (reg->dh, reg_path);
@@ -367,22 +380,19 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name,
     }
     reg->rank_classes = NULL;
 
-    reg->key_buf = 0;
-
+    reg->key_block = 0;
     reg->keys = zebra_rec_keys_open();
 
     reg->sortKeys = zebra_rec_keys_open();
 
     reg->records = 0;
     reg->dict = 0;
-    reg->sortIdx = 0;
+    reg->sort_index = 0;
     reg->isams = 0;
     reg->matchDict = 0;
     reg->isamc = 0;
     reg->isamb = 0;
     reg->zei = 0;
-    reg->key_file_no = 0;
-    reg->ptr_i = 0;
     
     /* installing rank classes */
     zebraRankInstall (reg, rank_1_class);
@@ -419,9 +429,22 @@ struct zebra_register *zebra_register_open(ZebraService zs, const char *name,
        yaz_log (YLOG_WARN, "dict_open failed");
        ret = ZEBRA_FAIL;
     }
-    if (!(reg->sortIdx = sortIdx_open (reg->bfs, rw)))
+
+    
+    if (res_get_match (res, "sortindex", "f", "f"))
+        sort_type = ZEBRA_SORT_TYPE_FLAT;
+    else if (res_get_match (res, "sortindex", "i", "f"))
+        sort_type = ZEBRA_SORT_TYPE_ISAMB;
+    else
     {
-       yaz_log (YLOG_WARN, "sortIdx_open failed");
+       yaz_log (YLOG_WARN, "bad_value for 'sortindex'");
+       ret = ZEBRA_FAIL;
+    }
+
+
+    if (!(reg->sort_index = zebra_sort_open(reg->bfs, rw, sort_type)))
+    {
+       yaz_log (YLOG_WARN, "zebra_sort_open failed");
        ret = ZEBRA_FAIL;
     }
     if (res_get_match (res, "isam", "s", ISAM_DEFAULT))
@@ -532,7 +555,7 @@ static void zebra_register_close(ZebraService zs, struct zebra_register *reg)
     dict_close (reg->dict);
     if (reg->matchDict)
        dict_close (reg->matchDict);
-    sortIdx_close (reg->sortIdx);
+    zebra_sort_close(reg->sort_index);
     if (reg->isams)
        isams_close (reg->isams);
     if (reg->isamc)
@@ -550,7 +573,7 @@ static void zebra_register_close(ZebraService zs, struct zebra_register *reg)
     zebra_rec_keys_close(reg->keys);
     zebra_rec_keys_close(reg->sortKeys);
 
-    xfree(reg->key_buf);
+    key_block_destroy(&reg->key_block);
     xfree(reg->name);
     xfree(reg);
 }
@@ -559,8 +582,6 @@ ZEBRA_RES zebra_stop(ZebraService zs)
 {
     if (!zs)
         return ZEBRA_OK;
-    yaz_log (log_level, "zebra_stop");
-
     while (zs->sessions)
     {
         zebra_close (zs->sessions);
@@ -574,6 +595,15 @@ ZEBRA_RES zebra_stop(ZebraService zs)
     recTypeClass_destroy(zs->record_classes);
     nmem_destroy(zs->nmem);
     res_close (zs->global_res);
+
+    yaz_timing_stop(zs->timing);
+    yaz_log (YLOG_LOG, "zebra_stop: %4.2f %4.2f %4.2f",
+             yaz_timing_get_real(zs->timing),
+             yaz_timing_get_user(zs->timing),
+             yaz_timing_get_sys(zs->timing));
+    
+
+    yaz_timing_destroy(&zs->timing);
     xfree(zs);
     return ZEBRA_OK;
 }
@@ -985,8 +1015,10 @@ ZEBRA_RES zebra_set_approx_limit(ZebraHandle zh, zint approx_limit)
     return ZEBRA_OK;
 }
 
-ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query,
-                          const char *setname, zint *hits)
+ZEBRA_RES zebra_search_RPN_x(ZebraHandle zh, ODR o, Z_RPNQuery *query,
+                             const char *setname, zint *hits,
+                             int *estimated_hit_count,
+                             int *partial_resultset)
 {
     ZEBRA_RES r;
     
@@ -997,19 +1029,27 @@ ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query,
     assert(hits);
     assert(setname);
     yaz_log(log_level, "zebra_search_rpn");
-    zh->hits = 0;
-    *hits = 0;
 
     if (zebra_begin_read(zh) == ZEBRA_FAIL)
        return ZEBRA_FAIL;
 
     r = resultSetAddRPN(zh, odr_extract_mem(o), query, 
-                       zh->num_basenames, zh->basenames, setname);
+                       zh->num_basenames, zh->basenames, setname,
+                        hits, estimated_hit_count, partial_resultset);
     zebra_end_read(zh);
-    *hits = zh->hits;
     return r;
 }
 
+ZEBRA_RES zebra_search_RPN(ZebraHandle zh, ODR o, Z_RPNQuery *query,
+                           const char *setname, zint *hits)
+{
+    int estimated_hit_count;
+    int partial_resultset;
+    return zebra_search_RPN_x(zh, o, query, setname, hits,
+                              &estimated_hit_count,
+                              &partial_resultset);
+}
+
 ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream,
                                 const char *setname,
                                 Z_RecordComposition *comp,
@@ -1068,9 +1108,12 @@ ZEBRA_RES zebra_records_retrieve(ZebraHandle zh, ODR stream,
                int len = 0;
                zebra_snippets *hit_snippet = zebra_snippets_create();
 
+                /* we disable hit snippets for now. It does not work well
+                 and it slows retrieval down a lot */
+#if 0
                zebra_snippets_hit_vector(zh, setname, poset[i].sysno, 
                                          hit_snippet);
-
+#endif
                recs[i].errCode =
                    zebra_record_fetch(zh, poset[i].sysno, poset[i].score,
                                       hit_snippet,
@@ -1318,7 +1361,7 @@ ZEBRA_RES zebra_admin_import_end (ZebraHandle zh)
 ZEBRA_RES zebra_admin_import_segment (ZebraHandle zh, Z_Segment *segment)
 {
     ZEBRA_RES res = ZEBRA_OK;
-    SYSNO sysno;
+    zint sysno;
     int i;
     ZEBRA_CHECK_HANDLE(zh);
     yaz_log(log_level, "zebra_admin_import_segment");
@@ -1360,7 +1403,7 @@ ZEBRA_RES zebra_admin_exchange_record(ZebraHandle zh,
     /* 4 = update. Insert/replace */
 {
     ZEBRA_RES res;
-    SYSNO sysno = 0;
+    zint sysno = 0;
     char *rinfo = 0;
     char recid_z[256];
     int db_ord;
@@ -2238,7 +2281,7 @@ ZEBRA_RES zebra_add_record(ZebraHandle zh,
 
 ZEBRA_RES zebra_insert_record(ZebraHandle zh, 
                              const char *recordType,
-                             SYSNO *sysno, const char *match,
+                             zint *sysno, const char *match,
                              const char *fname,
                              const char *buf, int buf_size, int force_update)
 {
@@ -2269,12 +2312,12 @@ ZEBRA_RES zebra_insert_record(ZebraHandle zh,
     return res; 
 }
 
-ZEBRA_RES zebra_update_record (ZebraHandle zh, 
-                              const char *recordType,
-                              SYSNO* sysno, const char *match,
-                              const char *fname,
-                              const char *buf, int buf_size,
-                              int force_update)
+ZEBRA_RES zebra_update_record(ZebraHandle zh, 
+                              const char *recordType,
+                              zint *sysno, const char *match,
+                              const char *fname,
+                              const char *buf, int buf_size,
+                              int force_update)
 {
     ZEBRA_RES res;
 
@@ -2306,12 +2349,12 @@ ZEBRA_RES zebra_update_record (ZebraHandle zh,
     return res; 
 }
 
-ZEBRA_RES zebra_delete_record (ZebraHandle zh, 
-                              const char *recordType,
-                              SYSNO *sysno, const char *match,
-                              const char *fname,
-                              const char *buf, int buf_size,
-                              int force_update) 
+ZEBRA_RES zebra_delete_record(ZebraHandle zh, 
+                              const char *recordType,
+                              zint *sysno, const char *match,
+                              const char *fname,
+                              const char *buf, int buf_size,
+                              int force_update) 
 {
     ZEBRA_RES res;
 
@@ -2354,6 +2397,7 @@ ZEBRA_RES zebra_search_PQF(ZebraHandle zh, const char *pqf_query,
     Z_RPNQuery *query;
     ODR odr;
 
+
     ZEBRA_CHECK_HANDLE(zh);
 
     odr = odr_createmem(ODR_ENCODE);
@@ -2468,11 +2512,11 @@ void zebra_setError_zint(ZebraHandle zh, int code, zint i)
     zh->errString = nmem_strdup(zh->nmem_error, vstr);
 }
 
-void zebra_lock_prefix (Res res, char *path)
+void zebra_lock_prefix(Res res, char *path)
 {
     const char *lock_dir = res_get_def (res, "lockDir", "");
-
-    strcpy (path, lock_dir);
+    
+    strcpy(path, lock_dir);
     if (*path && path[strlen(path)-1] != '/')
         strcat (path, "/");
 }