X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=efc5c71ba3082cfa07f9fdf05201e85f7aeae08d;hb=85c5e09eacc5c89eda6e1ffa6f039fa1e8dd7221;hp=665640e752bbdcd76d15275d7d58b72dbcef15bf;hpb=8a5853c99ad09bcdb1ea567a594e3b586715fe90;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 665640e..efc5c71 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,5 +1,5 @@
-/* $Id: extract.c,v 1.245 2006-12-11 17:08:03 adam Exp $
-   Copyright (C) 1995-2006
+/* $Id: extract.c,v 1.249 2007-02-06 09:34:56 adam Exp $
+   Copyright (C) 1995-2007
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -59,8 +59,18 @@ static void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
 static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
 static void extract_token_add (RecWord *p);
 
+static void check_log_limit(ZebraHandle zh)
+{
+    if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
+    {
+        yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
+                zh->m_file_verbose_limit);
+    }
+}
+
 static void logRecord (ZebraHandle zh)
 {
+    check_log_limit(zh);
     ++zh->records_processed;
     if (!(zh->records_processed % 1000))
     {
@@ -351,8 +361,11 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     }
     if (!zh->m_record_type)
     {
-	if (zh->records_processed < zh->m_file_verbose_limit)
+        check_log_limit(zh);
+	if (zh->records_processed + zh->records_skipped
+            < zh->m_file_verbose_limit)
             yaz_log (YLOG_LOG, "? %s", fname);
+        zh->records_skipped++;
         return 0;
     }
     /* determine match criteria */
@@ -521,7 +534,8 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
     Record rec;
     off_t start_offset = 0, end_offset = 0;
     const char *pr_fname = fname;  /* filename to print .. */
-    int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+    int show_progress = zh->records_processed + zh->records_skipped 
+        < zh->m_file_verbose_limit ? 1:0;
 
     zebra_init_log_level();
 
@@ -1158,6 +1172,20 @@ static void extract_add_sort_string(RecWord *p, const char *str, int length)
     zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
 }
 
+static void extract_add_staticrank_string(RecWord *p,
+                                          const char *str, int length)
+{
+    char valz[40];
+    struct recExtractCtrl *ctrl = p->extractCtrl;
+
+    if (length > sizeof(valz)-1)
+        length = sizeof(valz)-1;
+
+    memcpy(valz, str, length);
+    valz[length] = '\0';
+    ctrl->staticrank = atozint(valz);
+}
+
 static void extract_add_string(RecWord *p, const char *string, int length)
 {
     ZebraHandle zh = p->extractCtrl->handle;
@@ -1166,9 +1194,7 @@ static void extract_add_string(RecWord *p, const char *string, int length)
     if (!p->index_name)
         return;
 
-    if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
-	extract_add_sort_string(p, string, length);
-    else
+    if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
     {
 	extract_add_index_string(p, zinfo_index_category_index,
                                  string, length);
@@ -1182,6 +1208,14 @@ static void extract_add_string(RecWord *p, const char *string, int length)
                 &word, zinfo_index_category_alwaysmatches, "", 0);
         }
     }
+    else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
+    {
+	extract_add_sort_string(p, string, length);
+    }
+    else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
+    {
+	extract_add_staticrank_string(p, string, length);
+    }
 }
 
 static void extract_add_incomplete_field(RecWord *p)
@@ -1361,24 +1395,30 @@ static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid)
 void extract_flush_sort_keys(ZebraHandle zh, zint sysno,
                              int cmd, zebra_rec_keys_t reckeys)
 {
+#if 0
+    yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT,
+            cmd, sysno);
+    extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG);
+#endif
+
     if (zebra_rec_keys_rewind(reckeys))
     {
-        SortIdx sortIdx = zh->reg->sortIdx;
+        zebra_sort_index_t si = zh->reg->sort_index;
 	size_t slen;
 	const char *str;
 	struct it_key key_in;
 
-        sortIdx_sysno (sortIdx, sysno);
+        zebra_sort_sysno(si, sysno);
 
 	while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
         {
             int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
             
-            sortIdx_type(sortIdx, ord);
+            zebra_sort_type(si, ord);
             if (cmd == 1)
-                sortIdx_add(sortIdx, str, slen);
+                zebra_sort_add(si, str, slen);
             else
-                sortIdx_add(sortIdx, "", 1);
+                zebra_sort_delete(si);
         }
     }
 }