Added a new 'cut' directive to charmaps (.chr files) which specifies that

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

diff --git a/index/extract.c b/index/extract.c
index 412bc66..593c9e8 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.149 2004-01-22 15:40:25 heikki Exp $
+/* $Id: extract.c,v 1.157 2004-07-28 09:47:41 adam Exp $
    Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
@@ -23,6 +23,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 
 #include <stdio.h>
 #include <assert.h>
+#include <ctype.h>
 #ifdef WIN32
 #include <io.h>
 #else
@@ -755,6 +756,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     char ext[128];
     char ext_res[128];
     struct file_read_info *fi;
+    const char *original_record_type = 0;
 
     if (!zh->m_group || !*zh->m_group)
         *gprefix = '\0';
@@ -774,6 +776,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
             break;
         }
     /* determine file type - depending on extension */
+    original_record_type = zh->m_record_type;
     if (!zh->m_record_type)
     {
         sprintf (ext_res, "%srecordType.%s", gprefix, ext);
@@ -811,6 +814,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
         {
             logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
+           zh->m_record_type = original_record_type;
             return 0;
         }
     }
@@ -823,6 +827,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
     file_read_stop (fi);
     if (fd != -1)
         close (fd);
+    zh->m_record_type = original_record_type;
     return r;
 }
 
@@ -1027,7 +1032,7 @@ int buffer_extract_record (ZebraHandle zh,
                extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
                rec_rm (&rec);
                logRecord(zh);
-               return 0;
+               return -1;
            }
        }
 
@@ -1283,7 +1288,9 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         if (zh->reg->key_buf_used + 1024 > 
             (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
             extract_flushWriteKeys (zh,0);
+        assert(zh->reg->ptr_i >= 0);
         ++(zh->reg->ptr_i);
+        assert(zh->reg->ptr_i > 0);
         (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
            (char*)zh->reg->key_buf + zh->reg->key_buf_used;
 #if SU_SCHEME
@@ -1353,15 +1360,21 @@ void extract_flushWriteKeys (ZebraHandle zh, int final)
     qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
                sizeof(char*), key_qsort_compare);
 
-    /* Case 1: always use temp files (old way) */
-    /* Case 2: use temp files, if more than one (auto) */
-    /*         = if this is both the last and the first */
-    /* Case 3: never bother with temp files (new) */
-    temp_policy=2;
-    /* FIXME - will come from config file into zh */
+    /* zebra.cfg: tempfiles:  
+       Y: always use temp files (old way) 
+       A: use temp files, if more than one (auto) 
+          = if this is both the last and the first 
+       N: never bother with temp files (new) */
 
-    if (   ( temp_policy ==3 )   ||     /* always from memory */
-         ( ( temp_policy ==2 ) &&       /* automatic */
+    temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
+    if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
+        logf (LOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ", 
+                        temp_policy);
+        temp_policy='A';
+    }
+
+    if (   ( temp_policy =='N' )   ||     /* always from memory */
+         ( ( temp_policy =='A' ) &&       /* automatic */
              (zh->reg->key_file_no == 1) &&  /* this is first time */
              (final) ) )                     /* and last (=only) time */
     { /* go directly from memory */
@@ -1369,7 +1382,7 @@ void extract_flushWriteKeys (ZebraHandle zh, int final)
         zebra_index_merge(zh); 
         zh->reg->ptr_i = 0;
         zh->reg->key_buf_used = 0; 
-        return; /*!*/
+        return; 
     }
 
     /* Not doing directly from memory, write into a temp file */
@@ -1472,6 +1485,10 @@ void extract_add_index_string (RecWord *p, const char *string,
     }
     dst = keys->buf + keys->buf_used;
 
+    /* leader byte is encoded as follows:
+       bit 0 : 1 if attrset is unchanged; 0 if attrset is changed
+       bit 1 : 1 if attruse is unchanged; 0 if attruse is changed
+    */
     attrSet = p->attrSet;
     if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
         lead |= 1;
@@ -1638,6 +1655,7 @@ static void extract_add_complete_field (RecWord *p)
        while (map && *map && **map == *CHR_SPACE)
        {
            remain = p->length - (b - p->string);
+
            if (remain > 0)
                map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
            else
@@ -1652,10 +1670,17 @@ static void extract_add_complete_field (RecWord *p)
        {
            const char *cp = *map;
 
-           if (i >= IT_MAX_WORD)
-               break;
-           while (i < IT_MAX_WORD && *cp)
-               buf[i++] = *(cp++);
+           if (**map == *CHR_CUT)
+           {
+               i = 0;
+           }
+           else
+           {
+               if (i >= IT_MAX_WORD)
+                   break;
+               while (i < IT_MAX_WORD && *cp)
+                   buf[i++] = *(cp++);
+           }
            remain = p->length  - (b - p->string);
            if (remain > 0)
                map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,