Improve rcldoc filter and switch back to using it for indexing instead of direct antiword exec. This is slightly slower but it does catch a number of .doc files which would not be indexed otherwise

src/filters/rcldoc

@@ -102,15 +102,18 @@ checkcmds awk antiword iconv
 $decoder "$infile" |
 awk 'BEGIN'\
 ' {
-  print "<html><head><title></title>"
-  print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
-  print "</head>\n<body>\n<p>"
   cont = ""
   gotdata = 0
 }
 {
-  if (!($0 ~ /^[ 	]*$/))
+  if (!($0 ~ /^[ 	]*$/)) {
     gotdata = 1
+    if (gotdata == 0) {
+      print "<html><head><title></title>"
+      print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
+      print "</head>\n<body>\n<p>"
+    }
+  }
   $0 = cont $0
   cont = ""
 
@@ -139,12 +142,32 @@ END {
     print "</p></body></html>"
 }' 
 
-# wvWare is much slower and we don't use it by default, but it handles
-# some files that antiword won't, so use it as a last resort. Yes the
-# html document will have a second header section, but this doesn't
-# seem to be an issue with our brave html input handler...
+
+# Antiword rarely fails, we try to catch the most common reasons:
 if test $? -eq 1 ; then
-    if iscmd wvWare ; then
-	wvWare --nographics --charset=utf-8 "$infile"
+    # Check actual document type 
+    mtype=`file -b -i "$infile" | awk '{sub(";", "", $1);print $1}'`
+
+    if test X"$mtype" = Xtext/rtf; then
+      # RTF document disguising as msword either because it has a .doc
+      # extension or because it's an attachment with a wrong mime.
+      exec `dirname $0`/rclrtf $infile
     fi
+    
+    if test X"$mtype" = Xtext/plain; then
+      # Someone gave a .doc ext to their texts. Happens...
+      exec `dirname $0`/rcltext $infile
+    fi
+
+    if test X"$mtype" = Xapplication/msword; then
+      # Actually application/msword: try wvWare, which is much
+      # slower and we don't use it by default, but it handles some
+      # files that antiword won't, so use it as a last resort.
+      if iscmd wvWare ; then
+        exec wvWare --nographics --charset=utf-8 "$infile"
+      fi
+    fi
+
+    # else let the error be...
+    exit 1
 fi