Use Text::Iconv for better error handling.

1 files changed, 44 insertions, 11 deletions

diff --git a/httpd/cgi-bin/check b/httpd/cgi-bin/check
index eca50c5..6ea3f9f 100755
--- a/httpd/cgi-bin/check
+++ b/httpd/cgi-bin/check
@@ -8,7 +8,7 @@
 # This source code is available under the license at:
 #     http://www.w3.org/Consortium/Legal/copyright-software
 #
-# $Id: check,v 1.77 2001-02-20 10:47:53 gerald Exp $
+# $Id: check,v 1.78 2001-02-21 08:56:32 link Exp $
 
 #
 # We need Perl 5.004.
@@ -22,6 +22,7 @@ use URI::Escape;
 use CGI::Carp;
 use CGI qw(:cgi -newstyle_urls -private_tempfiles);
 use Text::Wrap;
+use Text::Iconv;
 
 
 #############################################################################
@@ -61,7 +62,6 @@ my $temp      = "/tmp/validate.$$"; # @@ Use POSIX/IO::File tmpfiles instead!
 #
 # Executables and binaries
 my $sp      = '/usr/local/bin/lq-nsgmls';
-my $iconv   = '/usr/bin/iconv';
 my $weblint = '/usr/bin/weblint';
 
 #
@@ -74,9 +74,9 @@ my $element_ref = 'http://www.htmlhelp.com/reference/html40/';
 
 #
 # Strings
-$VERSION    =  q$Revision: 1.77 $;
+$VERSION    =  q$Revision: 1.78 $;
 $VERSION    =~ s/Revision: ([\d\.]+) /$1/;
-$DATE       =  q$Date: 2001-02-20 10:47:53 $;
+$DATE       =  q$Date: 2001-02-21 08:56:32 $;
 $MAINTAINER =  'gerald@w3.org';
 my $notice  =  ''; # "<p><strong>Note: This service will be ...</strong>";
 
@@ -304,14 +304,10 @@ if ($File->{HTTP_Charset}) {
 
 
 #
-# Setup conversion filter and SP environment for the effective charset.
-my $codeconv = '';
+# Setup SP environment for the charset.
 if ($File->{Charset} ne 'unknown' and $File->{Charset} ne 'us-ascii') {
   $ENV{SP_CHARSET_FIXED} = 'YES';
   $ENV{SP_ENCODING}      = 'utf-8';
-  if ($File->{Charset} ne 'utf-8') {
-    $codeconv = "$iconv -f $File->{Charset} -t utf-8 |";
-  }
 }
 
 #
@@ -377,7 +373,7 @@ if ($File->{Type} eq 'xhtml') {
   $catalog = "-c $sgmlstuff/catalog";
 }
 
-my $command  = "$codeconv $sp -E0 $xmlflags $catalog $decl";
+my $command  = "$sp -E0 $xmlflags $catalog $decl";
 
 if ( $ENV{SERVER_PORT} eq "8000" ) {	# running on the test server?
     print "    <li>nsgmls command line: <code>$command</code>\n";
@@ -387,7 +383,20 @@ open CHECKER, "|$command - >$temp.esis 2>$temp"
   or die "open(|$command - >$temp.esis 2>$temp) returned: $!\n";
 
 print CHECKER $doctype, "\n" if $guessed_doctype == 1;
-for (@{$File->{Content}}) {print CHECKER $_, "\n"}
+if (   $File->{Charset} eq 'utf-8'
+    or $File->{Charset} eq 'us-ascii'
+    or $File->{Charset} eq 'unknown') {
+  for (@{$File->{Content}}) {print CHECKER $_, "\n"};
+} else {
+  eval {my $c = Text::Iconv->new($File->{Charset}, 'utf-8')};
+  if (not $@) {
+    my $c = Text::Iconv->new($File->{Charset}, 'utf-8');
+    for (@{$File->{Content}}) {print CHECKER $c->convert("$_\n")};
+  } else {
+    &print_charset_error($@, $File->{Charset});
+    &clean_up_and_exit;
+  }
+}
 close CHECKER;
 
 open ERRORS, "<$temp"   or die  "open($temp) returned: $!\n";
@@ -1265,6 +1274,30 @@ EOF
 
 }
 
+
+#
+# Complain about strange charsets.
+sub print_charset_error {
+  my $error   = shift;
+  my $charset = shift;
+
+  print <<".EOF.";
+  <p>
+    A fatal error occured when attempting to transliterate the document charset.
+    Either we do not support this charset yet, or you have specified a non-existant
+    character set (typically a misspelling such as "iso8859-1" for "iso-8859-1").
+  </p>
+  <p>The detected charset was "$charset".</p>
+  <p>The error was "$error".</p>
+  <p>
+    If you believe the charset to be valid you can submit a request for that
+    character set (see the <a href="/feedback.html">feedback page</a> for
+    details) and we will look into supporting it in the future.
+  </p>
+.EOF.
+}
+
+
 sub output_css_validator_blurb {
   my $uri = shift;
   $uri = ent($uri);