huge refactor of encoding handling

[public-inbox.git] / lib / PublicInbox / View.pm

diff --git a/lib/PublicInbox/View.pm b/lib/PublicInbox/View.pm
index 1a912a6edf6694d44860dedbc1153796c9ca5bb6..0a8ab549c97b878c8f2fa215b6a1230acec1850d 100644 (file)
--- a/lib/PublicInbox/View.pm
+++ b/lib/PublicInbox/View.pm
@@ -3,22 +3,23 @@
 package PublicInbox::View;
 use strict;
 use warnings;
+use PublicInbox::Hval;
 use URI::Escape qw/uri_escape/;
-use CGI qw/escapeHTML/;
 use Encode qw/find_encoding/;
 use Encode::MIME::Header;
 use Email::MIME::ContentType qw/parse_content_type/;
 use constant MAX_INLINE_QUOTED => 5;
+use constant MAX_TRUNC_LEN => 72;
+*ascii_html = *PublicInbox::Hval::ascii_html;
 
-my $enc_utf8 = find_encoding('utf8');
-my $enc_ascii = find_encoding('us-ascii');
+my $enc_utf8 = find_encoding('UTF-8');
 my $enc_mime = find_encoding('MIME-Header');
 
 # public functions:
 sub as_html {
        my ($class, $mime, $full_pfx) = @_;
 
-       headers_to_html_header($mime) .
+       headers_to_html_header($mime, $full_pfx) .
                multipart_text_as_html($mime, $full_pfx) .
                '</pre></body></html>';
 }
@@ -95,14 +96,20 @@ sub add_text_body_short {
                if (@lines > MAX_INLINE_QUOTED) {
                        # show a short snippet of quoted text
                        $cur = join(' ', @lines);
-                       $cur =~ s/&gt; ?//g;
+                       $cur =~ s/^&gt;\s*//;
 
                        my @sum = split(/\s+/, $cur);
                        $cur = '';
                        do {
-                               $cur .= shift(@sum) . ' ';
-                       } while (@sum && length($cur) < 64);
-                       $cur=~ s/ \z/ .../;
+                               my $tmp = shift(@sum);
+                               my $len = length($tmp) + length($cur);
+                               if ($len > MAX_TRUNC_LEN) {
+                                       @sum = ();
+                               } else {
+                                       $cur .= $tmp . ' ';
+                               }
+                       } while (@sum && length($cur) < MAX_TRUNC_LEN);
+                       $cur =~ s/ \z/ .../;
                        "&gt; &lt;<a href=\"${full_pfx}#q${part_nr}_" . $n++ .
                                "\">$cur<\/a>&gt;\n";
                } else {
@@ -128,58 +135,44 @@ sub add_text_body_full {
        $s;
 }
 
-sub trim_message_id {
-       my ($mid) = @_;
-       $mid = $enc_mime->decode($mid);
-       $mid =~ s/\A\s*<//;
-       $mid =~ s/>\s*\z//;
-       my $html = ascii_html($mid);
-       my $href = ascii_html(uri_escape($mid));
-
-       ($html, $href);
-}
-
-sub ascii_html {
-       $enc_ascii->encode(escapeHTML($_[0]), Encode::HTMLCREF);
-}
-
 sub headers_to_html_header {
-       my ($simple) = @_;
+       my ($mime, $full_pfx) = @_;
 
        my $rv = "";
        my @title;
        foreach my $h (qw(From To Cc Subject Date)) {
-               my $v = $simple->header($h);
-               defined $v or next;
-               $v =~ tr/\n/ /s;
-               $v =~ tr/\r//d;
-               my $raw = $enc_mime->decode($v);
-               $v = ascii_html($raw);
-               $rv .= "$h: $v\n";
+               my $v = $mime->header($h);
+               defined($v) && length($v) or next;
+               $v = PublicInbox::Hval->new_oneline($v);
+               $rv .= "$h: " . $v->as_html . "\n";
 
                if ($h eq 'From') {
-                       my @from = Email::Address->parse($raw);
-                       $raw = $from[0]->name;
-                       unless (defined($raw) && length($raw)) {
-                               $raw = '<' . $from[0]->address . '>';
+                       my @from = Email::Address->parse($v->raw);
+                       $v = $from[0]->name;
+                       unless (defined($v) && length($v)) {
+                               $v = '<' . $from[0]->address . '>';
                        }
-                       $title[1] = ascii_html($raw);
-
+                       $title[1] = ascii_html($v);
                } elsif ($h eq 'Subject') {
-                       $title[0] = $v;
+                       $title[0] = $v->as_html;
                }
        }
 
-       my $mid = $simple->header('Message-ID');
+       my $header_obj = $mime->header_obj;
+       my $mid = $header_obj->header_raw('Message-ID');
        if (defined $mid) {
-               my ($html, $href) = trim_message_id($mid);
-               $rv .= "Message-ID: &lt;$html&gt; ";
+               $mid = PublicInbox::Hval->new_msgid($mid);
+               $rv .= 'Message-ID: &lt;' . $mid->as_html . '&gt; ';
+               my $href = $mid->as_href;
+               $href = "../m/$href" unless $full_pfx;
                $rv .= "(<a href=\"$href.txt\">original</a>)\n";
        }
 
-       my $irp = $simple->header('In-Reply-To');
+       my $irp = $header_obj->header_raw('In-Reply-To');
        if (defined $irp) {
-               my ($html, $href) = trim_message_id($irp);
+               $irp = PublicInbox::Hval->new_msgid($irp);
+               my $html = $irp->as_html;
+               my $href = $irp->as_href;
                $rv .= "In-Reply-To: &lt;";
                $rv .= "<a href=\"$href.html\">$html</a>&gt;\n";
        }