New feature: databases. Implemented as prefix to words in dictionary.

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

diff --git a/index/extract.c b/index/extract.c
index 718ef0c..b557a7f 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,16 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.19  1995-10-04 12:55:16  adam
+ * Revision 1.22  1995-10-17 18:02:07  adam
+ * New feature: databases. Implemented as prefix to words in dictionary.
+ *
+ * Revision 1.21  1995/10/10  12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.20  1995/10/06  13:52:05  adam
+ * Bug fixes. Handler may abort further scanning.
+ *
+ * Revision 1.19  1995/10/04  12:55:16  adam
  * Bug fix in ranked search. Use=Any keys inserted.
  *
  * Revision 1.18  1995/10/04  09:37:08  quinn
@@ -82,6 +91,7 @@ static int sys_idx_fd = -1;
 
 static int key_cmd;
 static int key_sysno;
+static char *key_databaseName;
 static char **key_buf;
 static size_t ptr_top;
 static size_t ptr_i;
@@ -116,12 +126,73 @@ void key_open (int mem)
         exit (1);
     }
 }
-    
+
+struct encode_info {
+    int  sysno;
+    int  seqno;
+    char buf[512];
+};
+
+void encode_key_init (struct encode_info *i)
+{
+    i->sysno = 0;
+    i->seqno = 0;
+}
+
+char *encode_key_int (int d, char *bp)
+{
+    if (d <= 63)
+        *bp++ = d;
+    else if (d <= 16383)
+    {
+        *bp++ = 64 + (d>>8);
+        *bp++ = d  & 255;
+    }
+    else if (d <= 4194303)
+    {
+        *bp++ = 128 + (d>>16);
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    else
+    {
+        *bp++ = 192 + (d>>24);
+        *bp++ = (d>>16) & 255;
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    return bp;
+}
+
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+    struct it_key key;
+    char *bp = i->buf;
+
+    while ((*bp++ = *k++))
+        ;
+    memcpy (&key, k+1, sizeof(struct it_key));
+    bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
+    if (i->sysno != key.sysno)
+    {
+        i->sysno = key.sysno;
+        i->seqno = 0;
+    }
+    bp = encode_key_int (key.seqno - i->seqno, bp);
+    i->seqno = key.seqno;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
+    }
+}
+
 void key_flush (void)
 {
     FILE *outf;
     char out_fname[200];
     char *prevcp, *cp;
+    struct encode_info encode_info;
     
     if (ptr_i <= 0)
         return;
@@ -139,33 +210,19 @@ void key_flush (void)
     logf (LOG_LOG, "writing section %d", key_file_no);
     prevcp = cp = key_buf[ptr_top-ptr_i];
     
-    if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-        exit (1);
-    }
+    encode_key_init (&encode_info);
+    encode_key_write (cp, &encode_info, outf);
     while (--ptr_i > 0)
     {
         cp = key_buf[ptr_top-ptr_i];
         if (strcmp (cp, prevcp))
         {
-            if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
-                        outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
+            encode_key_init (&encode_info);
+            encode_key_write (cp, &encode_info, outf);
             prevcp = cp;
         }
         else
-        {
-            cp = strlen (cp) + cp;
-            if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
-        }
+            encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
     if (fclose (outf))
     {
@@ -204,7 +261,8 @@ static void wordAdd (const RecWord *p)
     ++ptr_i;
     key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
     kused += index_word_prefix ((char*)key_buf + kused,
-                                p->attrSet, p->attrUse);
+                                p->attrSet, p->attrUse,
+                                1, &key_databaseName);
     switch (p->which)
     {
     case Word_String:
@@ -224,7 +282,7 @@ static void wordAdd (const RecWord *p)
 
 static void wordAddAny (const RecWord *p)
 {
-    if (p->attrSet != 1 && p->attrUse != 1016)
+    if (p->attrSet != 1 || p->attrUse != 1016)
     {
         RecWord w;
 
@@ -305,7 +363,8 @@ static int file_read (int fd, char *buf, size_t count)
     return read (fd, buf, count);
 }
 #endif
-void file_extract (int cmd, const char *fname, const char *kname)
+void file_extract (int cmd, const char *fname, const char *kname,
+                   char *databaseName)
 {
     int i, r;
     char ext[128];
@@ -316,6 +375,7 @@ void file_extract (int cmd, const char *fname, const char *kname)
     struct recExtractCtrl extractCtrl;
     RecType rt;
 
+    key_databaseName = databaseName;
     for (i = strlen(fname); --i >= 0; )
         if (fname[i] == '/')
         {