#! /usr/bin/perl
-# $Cambridge: exim/doc/doc-docbook/Tidytxt,v 1.6 2008/02/04 17:18:18 fanf2 Exp $
-
# Script to tidy up the output of w3m when it makes a text file. First we
# convert sequences of blank lines into a single blank line, to get everything
# uniform. Then we go through and insert blank lines before chapter and
$line =~ s/\x{c2}\x{a0}/ /g;
# (2) bullet -> asterisk
$line =~ s/\x{e2}\x{97}\x{8f}/*/g;
+ $line =~ s/\x{e2}\x{80}\x{a2}/*/g; # OpenSUSE
+ $line =~ s/\x{e2}\x{96}\x{a1}/*/g; # OpenSUSE
# (3a) horizontal box drawing -> hyphen
$line =~ s/\x{e2}\x{94}[\x{80}\x{81}\x{84}\x{85}\x{88}\x{89}]/-/g;
$line =~ s/\x{e2}\x{95}[\x{8c}\x{8d}\x{90}]/-/g;
$line =~ s/\x{e2}\x{95}\x{b2}/\\/g;
$line =~ s/\x{e2}\x{95}\x{b3}/X/g;
+ # w3m rendering issue apparently only seen by pdp
+ # affects section numbers after the ToC, some info on spool-file -lines, etc
+ # always appears to be a spurious extra character, safely just dropped.
+ $line =~ s/\x{c2}//g;
+
if ($line =~ /^\s*$/)
{
$line = "" if $lastwasblank;