X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=ea31d07d2208e3db4e217fa1badfc378709506dc;hb=9af45a7f129664e5f802ff1cdfce08fbce8b1adb;hp=8464bb48ca44ef34942a064a35a5a6fc8f1a67c6;hpb=5bdf6c8ddbafa8515a82079f631ef172d98e055b;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 8464bb4..ea31d07 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,22 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.15  1995-10-02 15:42:53  adam
+ * 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
+ * Fixed bug.
+ *
+ * Revision 1.17  1995/10/03  14:28:57  adam
+ * Buffered read in extract works.
+ *
+ * Revision 1.16  1995/10/03  14:28:45  adam
+ * Work on more effecient read handler in extract.
+ *
+ * Revision 1.15  1995/10/02  15:42:53  adam
  * Extract uses file descriptors instead of FILE pointers.
  *
  * Revision 1.14  1995/10/02  15:29:13  adam
@@ -210,11 +225,89 @@ static void wordAdd (const RecWord *p)
     kused += sizeof(key);
 }
 
+static void wordAddAny (const RecWord *p)
+{
+    if (p->attrSet != 1 || p->attrUse != 1016)
+    {
+        RecWord w;
+
+        memcpy (&w, p, sizeof(w));
+        w.attrSet = 1;
+        w.attrUse = 1016;
+        wordAdd (&w);
+    }
+    wordAdd (p);
+}
+
+
+#define FILE_READ_BUF 1
+#if FILE_READ_BUF
+static char *file_buf;
+static int file_offset;
+static int file_bufsize;
+
+static void file_read_start (int fd)
+{
+    file_offset = 0;
+    file_buf = xmalloc (4096);
+    file_bufsize = read (fd, file_buf, 4096);
+}
+
+static void file_read_stop (int fd)
+{
+    xfree (file_buf);
+}
+
+static int file_read (int fd, char *buf, size_t count)
+{
+    int l = file_bufsize - file_offset;
+
+    if (count > l)
+    {
+        int r;
+        if (l > 0)
+            memcpy (buf, file_buf + file_offset, l);
+        count = count-l;
+        if (count > file_bufsize)
+        {
+            if ((r = read (fd, buf + l, count)) == -1)
+            {
+                logf (LOG_FATAL|LOG_ERRNO, "read");
+                exit (1);
+            }
+            file_bufsize = 0;
+            file_offset = 0;
+            return r;
+        }
+        file_bufsize = r = read (fd, file_buf, 4096);
+        if (r == -1)
+        {
+            logf (LOG_FATAL|LOG_ERRNO, "read");
+            exit (1);
+        }
+        else if (r <= count)
+        {
+            file_offset = r;
+            memcpy (buf + l, file_buf, r);
+            return l + r;
+        }
+        else
+        {
+            file_offset = count;
+            memcpy (buf + l, file_buf, count - l);
+            return count;
+        }
+    }
+    memcpy (buf, file_buf + file_offset, count);
+    file_offset += count;
+    return count;
+}
+#else
 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)
 {
     int i, r;
@@ -262,11 +355,17 @@ void file_extract (int cmd, const char *fname, const char *kname)
     }
     extractCtrl.subType = "";
     extractCtrl.init = wordInit;
-    extractCtrl.add = wordAdd;
+    extractCtrl.add = wordAddAny;
+#if FILE_READ_BUF
+    file_read_start (extractCtrl.fd);
+#endif
     extractCtrl.readf = file_read;
     key_sysno = sysno;
     key_cmd = cmd;
     r = (*rt->extract)(&extractCtrl);
+#if FILE_READ_BUF
+    file_read_stop (extractCtrl.fd);
+#endif
     close (extractCtrl.fd);
     if (r)
         logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);