filter: use IPC::Run and improve lynx error handling

We may occasionally encounter horrid HTML which lynx cannot
handle, so improve error reporting.

diff --git a/INSTALL b/INSTALL
index a69782ad23625e92908f59448f5d7234c6516628..8e512a63a37f9614d4f25adee859f6f42177f58c 100644 (file)
--- a/INSTALL
+++ b/INSTALL
@@ -25,7 +25,6 @@ Requirements (server MDA)
   - Email::MIME::ContentType   libemail-mime-contenttype-perl
   - Encode::MIME::Header       perl
   - File::Path::Expand         libfile-path-expand-perl
-  - IPC::Open2                 perl
   - IPC::Run                   libipc-run-perl
   - Mail::Thread (2.5+)        libmail-thread-perl
   - URI::Escape                liburi-perl

diff --git a/Makefile.PL b/Makefile.PL
index f74d146a677110a5d10074b44378c44eb00207e0..7f2d586df30a4bc2397ea8ba4eab1fd0601bdd60 100644 (file)
--- a/Makefile.PL
+++ b/Makefile.PL
@@ -21,7 +21,6 @@ WriteMakefile(
                'Email::MIME::ContentType' => 0,
                'Encode::MIME::Header' => 0,
                'File::Path::Expand' => 0,
-               'IPC::Open2' => 0,
                'IPC::Run' => 0,
                'Mail::Thread' => '2.5', # 2.5+ needed for Email::Simple compat
                'URI::Escape' => 0,

diff --git a/lib/PublicInbox/Filter.pm b/lib/PublicInbox/Filter.pm
index 4a3482170422c7a16965a198abf89b7dac140d25..e5a8fafe52d66cd07fb06b73018c8b8cf3218d18 100644 (file)
--- a/lib/PublicInbox/Filter.pm
+++ b/lib/PublicInbox/Filter.pm
@@ -10,7 +10,7 @@ use warnings;
 use Email::MIME;
 use Email::MIME::ContentType qw/parse_content_type/;
 use Email::Filter;
-use IPC::Open2;
+use IPC::Run;
 our $VERSION = '0.0.1';
 
 # start with the same defaults as mailman
@@ -41,7 +41,7 @@ sub run {
                # HTML-only, non-multipart
                my $body = $simple->body;
                my $ct_parsed = parse_content_type($content_type);
-               dump_html($body, $ct_parsed->{attributes}->{charset});
+               dump_html(\$body, $ct_parsed->{attributes}->{charset});
                replace_body($simple, $body);
                return 1;
        } elsif ($content_type =~ m!\bmultipart/!i) {
@@ -80,28 +80,29 @@ sub html_part_to_text {
        my ($simple, $part) = @_;
        my $body = $part->body;
        my $ct_parsed = parse_content_type($part->content_type);
-       dump_html($body, $ct_parsed->{attributes}->{charset});
+       dump_html(\$body, $ct_parsed->{attributes}->{charset});
        replace_part($simple, $part, $body, 'text/plain');
 }
 
 # modifies $_[0] in place
 sub dump_html {
-       my $charset = $_[1] || 'US-ASCII';
-       my $cmd = "lynx -stdin -dump";
+       my ($body, $charset) = @_;
+       $charset ||= 'US-ASCII';
+       my @cmd = qw(lynx -stdin -stderr -dump);
+       my $out = "";
+       my $err = "";
 
        # be careful about remote command injection!
        if ($charset =~ /\A[A-Za-z0-9\-]+\z/) {
-               $cmd .= " -assume_charset=$charset";
+               push @cmd, "-assume_charset=$charset";
        }
-
-       my $pid = open2(my $out, my $in, $cmd);
-       print $in $_[0];
-       close $in;
-       {
-               local $/;
-               $_[0] = <$out>;
+       if (IPC::Run::run(\@cmd, $body, \$out, \$err)) {
+               $$body = $out;
+       } else {
+               # give them an ugly version:
+               $$body = "public-inbox HTML conversion failed: $err\n" .
+                        $$body . "\n";
        }
-       waitpid($pid, 0);
 }
 
 # this is to correct user errors and not expected to cover all corner cases