handle <tbody> more smartly

remove several tags for fashion

1 files changed, 108 insertions, 59 deletions

diff --git a/doc/bin/html2sgml.pl b/doc/bin/html2sgml.pl
index c5e1c28d..aed33dbb 100755
--- a/doc/bin/html2sgml.pl
+++ b/doc/bin/html2sgml.pl
@@ -5,87 +5,136 @@
 #   All rights reserved. This program is free software; you can
 #   redistribute it and/or modify it under the same terms as Perl itself. 
 #
-# $FML: html2sgml.pl,v 1.1 2001/04/23 14:37:42 fukachan Exp $
+# $FML: html2sgml.pl,v 1.2 2001/04/24 03:41:52 fukachan Exp $
 #
 
 use strict;
 use Carp;
 use vars qw($TR);
+use FileHandle;
 
 my $para = '';
 my $hr   = 0;
 my $tr   = 0;
 my $listitem = 0;
 
-while (<>) {
-    # <P>
-    if (m@<P>@i) { 
-	print "<para>\n";
-	$para = 1;
-	next;
-    }
+for my $file (@ARGV) {
+    my $rh;
 
-    if (/^\s*$/ && $para) {
-	print "\t" if $para eq 'tab';
-	print "</para>\n\n";
-	$para = 0;
-    }
+    my $tmpf = "/tmp/html2sgml.$$";
+    my $wh = new FileHandle "> $tmpf";
+    convert($file, $rh, $wh);
+    $wh->close;
 
-    # HR
-    if (m@<HR>@i) {
-	print "</sect1>\n" if $hr;
-	s@<HR>@<sect1>@gi;
-	$hr = 1;
-    }
+    my $rh = new FileHandle $tmpf;
+    fix_output($rh);
 
-    # A
-    s@<A\s*HREF=@<ulink url=@gi;
-    s@</A>@</ulink>@gi;
+    unlink $tmpf;
+}
 
-    # PRE
-    s@<PRE>@<programlisting>@gi;
-    s@</PRE>@</programlisting>@gi;
+exit 0;
 
-    # UL
-    s@<UL>@<itemizedlist>@gi;
-    s@</UL>@</itemizedlist>@gi;
 
-    if (m@<LI>@i) {
-	print "\t" if $para eq 'tab' && $listitem;
-	print "\t</listitem>\n\n" if $listitem;
-	s@<LI>@<listitem>\n\t<para>\n\t@gi;
-	$para = "tab";
-	$listitem = 1;
-    }
+sub convert
+{
+    my ($file, $rh, $wh) = @_;
 
-    if (m@</itemizedlist>@) {
-	print "\t" if $para eq 'tab' && $listitem;;
-	print "\t</listitem>\n\n" if $listitem;
-	$listitem = 0;
-    }
+    my $fh = new FileHandle $file;
 
+    while (<$fh>) {
+	# <P>
+	if (m@<P>@i) { 
+	    print $wh  "<para>\n";
+	    $para = 1;
+	    next;
+	}
 
-    # BR
-    s@<BR>@@gi;
+	if (/^\s*$/ && $para) {
+	    print $wh  "\t" if $para eq 'tab';
+	    print $wh  "</para>\n\n";
+	    $para = 0;
+	}
 
-    # TABLE
-    s@<TABLE.*>@<table>@gi;
-    if (m@<TR>@) {
-	$TR = $TR ? 'tbody' : 'thead';
-	$tr++;
-	s@<TR>@<${TR}>\n\t<row>@gi;
-    }
-    if (m@<TD>@) {
-	s@<TD>@<entry>@;
-	s@$@</entry>@;
+	# HR
+	if (m@<HR>@i) {
+	    print $wh  "</sect1>\n" if $hr;
+	    s@<HR>@<sect1>@gi;
+	    $hr = 1;
+	}
+
+	# A
+	s@<A\s*HREF=@\n<ulink url=@gi;
+	s@</A>@\n</ulink>\n@gi;
+
+	# PRE
+	s@<PRE>@<programlisting>@gi;
+	s@</PRE>@</programlisting>@gi;
+
+	# UL
+	s@<UL>@<itemizedlist>@gi;
+	s@</UL>@</itemizedlist>@gi;
+
+	if (m@<LI>@i) {
+	    print $wh  "\t" if $para eq 'tab' && $listitem;
+	    print $wh  "\t</listitem>\n\n" if $listitem;
+	    s@<LI>@<listitem>\n\t<para>\n\t@gi;
+	    $para = "tab";
+	    $listitem = 1;
+	}
+
+	if (m@</itemizedlist>@) {
+	    print $wh  "\t" if $para eq 'tab' && $listitem;;
+	    print $wh  "\t</listitem>\n\n" if $listitem;
+	    $listitem = 0;
+	}
+
+
+	# BR
+	s@<HR>@@gi;
+	s@<BR>@@gi;
+	s@<EM>@@gi;
+	s@</EM>@@gi;
+	s@<B>@@gi;
+	s@</B>@@gi;
+	s@<I>@@gi;
+	s@</I>@@gi;
+
+
+	# TABLE
+	s@<TABLE.*>@<table>\n <tgroup cols=2>\n@gi;
+	s@</TABLE>@\n </tgroup>\n</table>@gi;
+	if (m@<TR>@) {
+	    $TR = $TR ? 'tbody' : 'thead';
+	    $tr++;
+	    s@<TR>@<$TR>\n\t<row>@gi;
+	}
+
+	if (m@<TD>@) {
+	    s@<TD>@<entry>@;
+	    s@$@\n\t</entry>@;
+	}
+
+	if ($tr && /^\s*$/) {
+	    print $wh  "\t</row>\n";
+	    print $wh  "   </${TR}>\n";
+	    $tr = 0;
+	}
+
+	print $wh $_;
     }
-    if ($tr && /^\s*$/) {
-	print "\t</row>\n";
-	print "   </${TR}>\n";
-	$tr = 0;
+}
+
+
+sub fix_output
+{
+    my ($rh) = @_;
+    my $buf;
+
+    while (<$rh>) {
+	$buf .= $_;
     }
 
-    print;
-}
+    $buf =~ s@</tbody>[\s\n]*<tbody>@\n@g;
 
-exit 0;
+    print $buf;
+}