fix to output <br> when needed + other misc pbs

src/filters/rcldoc

@@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rcldoc,v 1.4 2005-10-20 15:42:29 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rcldoc,v 1.5 2006-01-27 13:37:31 dockes Exp $  (C) 2004 J.F.Dockes
 # Parts taken from Estraier:
 #================================================================
 # Estraier: a personal full-text search system
@@ -20,7 +20,7 @@
 LANG=C ; export LANG
 LC_ALL=C ; export LC_ALL
 progname="rcldoc"
-decoder="antiword -i -1 -m UTF-8"
+decoder="antiword -t -i 1 -m UTF-8"
 # Not ready to use this for now (it outputs html, so the code below has to
 # be simplified.)
 #decoder="wvWare -1 -c UTF-8"
@@ -72,28 +72,36 @@ fi
 $decoder "$infile" |
 awk '
 BEGIN {
-  printf("<html><head><title></title>\n")
-  printf("<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">\n")
-  printf("</head>\n<body><p>");
-  esc = 1
+  print "<html><head><title></title>"
+  print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
+  print "</head>\n<body>\n<p>"
+  cont = ""
 }
 {
-  if ($0 ~ /-$/) {
-    sub(/-$/, "", $0)
-    printf("%s", $0);
-  } else if($0 == "\f") {
-    printf("</p>\n<hr>\n<p>")
-  } else {
-    if(esc > 0) {
-      gsub(/&/, "\\&amp;", $0)
-      gsub(/</, "\\&lt;", $0)
-      gsub(/>/, "\\&gt;", $0)
-    }
-    print $0
+  $0 = cont $0
+  cont = ""
+
+  if ($0 ~ /[­-]$/) {
+    # Note : soft-hyphen is iso8859 0xad
+    # Break at last whitespace
+    match($0, "[ \t][^ \t]+$")
+    line = substr($0, 0, RSTART)
+    cont = substr($0, RSTART, RLENGTH-1)
+    $0 = line
   }
+
+  if($0 == "\f") {
+    print "</p><hr><p>"
+    next
+  } 
+  gsub(/&/, "\\&amp;", $0)
+  gsub(/</, "\\&lt;", $0)
+  gsub(/>/, "\\&gt;", $0)
+
+  print $0 "<br>"
 }
 END {
-    printf("</p></body></html>\n");
+    print "</p></body></html>"
 }' | iconv -f UTF-8 -t UTF-8 -c -s 
 
 # exit normally