]> Sergey Matveev's repositories - public-inbox.git/commitdiff
EmlContentFoo: Email::MIME::ContentType replacement
authorEric Wong <e@yhbt.net>
Thu, 7 May 2020 21:05:51 +0000 (21:05 +0000)
committerEric Wong <e@yhbt.net>
Sat, 9 May 2020 08:59:10 +0000 (08:59 +0000)
Since we're getting rid of Email::MIME, get rid of
Email::MIME::ContentType, too; since we may introduce
speedups down the line specific to our codebase.

MANIFEST
lib/PublicInbox/Eml.pm
lib/PublicInbox/EmlContentFoo.pm [new file with mode: 0644]
lib/PublicInbox/WwwAttach.pm
t/eml_content_disposition.t [new file with mode: 0644]
t/eml_content_type.t [new file with mode: 0644]

index 0906448ef41d809283e540eb3a228dfb5d7ff8f9..055c8c9a2165f45e105f87e209355e53810d9f63 100644 (file)
--- a/MANIFEST
+++ b/MANIFEST
@@ -106,6 +106,7 @@ lib/PublicInbox/DSPoll.pm
 lib/PublicInbox/Daemon.pm
 lib/PublicInbox/Emergency.pm
 lib/PublicInbox/Eml.pm
+lib/PublicInbox/EmlContentFoo.pm
 lib/PublicInbox/ExtMsg.pm
 lib/PublicInbox/Feed.pm
 lib/PublicInbox/Filter/Base.pm
@@ -231,6 +232,8 @@ t/ds-poll.t
 t/edit.t
 t/emergency.t
 t/eml.t
+t/eml_content_disposition.t
+t/eml_content_type.t
 t/epoll.t
 t/fail-bin/spamc
 t/feed.t
index 0c23bed03e5e3413a11fb27b9bc388119199d66d..1988bdb3dd039633d81ce0e6baaeffedc0f258b2 100644 (file)
@@ -33,10 +33,9 @@ use Text::Wrap qw(wrap); # stdlib, we need Perl 5.6+ for $huge
 
 my $MIME_Header = find_encoding('MIME-Header');
 
-# TODO remove these dependencies
-use Email::MIME::ContentType;
+use PublicInbox::EmlContentFoo qw(parse_content_type parse_content_disposition);
 use Email::MIME::Encodings;
-$Email::MIME::ContentType::STRICT_PARAMS = 0;
+$PublicInbox::EmlContentFoo::STRICT_PARAMS = 0;
 
 our $MAXPARTS = 1000; # same as SpamAssassin
 our $MAXDEPTH = 20; # seems enough, Perl sucks, here
@@ -108,7 +107,7 @@ sub header_raw {
 # pick the first Content-Type header to match Email::MIME behavior.
 # It's usually the right one based on historical archives.
 sub ct ($) {
-       # Email::MIME::ContentType::content_type:
+       # PublicInbox::EmlContentFoo::content_type:
        $_[0]->{ct} //= parse_content_type(header($_[0], 'Content-Type'));
 }
 
diff --git a/lib/PublicInbox/EmlContentFoo.pm b/lib/PublicInbox/EmlContentFoo.pm
new file mode 100644 (file)
index 0000000..f507d54
--- /dev/null
@@ -0,0 +1,294 @@
+# Copyright (C) 2020 all contributors <meta@public-inbox.org>
+# Copyright (C) 2004- Simon Cozens, Casey West, Ricardo SIGNES
+# This library is free software; you can redistribute it and/or modify
+# it under the same terms as Perl itself.
+#
+# License: GPL-1.0+ or Artistic-1.0-Perl
+#  <https://www.gnu.org/licenses/gpl-1.0.txt>
+#  <https://dev.perl.org/licenses/artistic.html>
+#
+# This license differs from the rest of public-inbox
+#
+# This is a fork of the Email::MIME::ContentType 1.022 with
+# minor improvements and incompatibilities; namely changes to
+# quiet warnings with legacy data.
+package PublicInbox::EmlContentFoo;
+use strict;
+use parent qw(Exporter);
+# ABSTRACT: Parse a MIME Content-Type or Content-Disposition Header
+
+use Encode 2.87 qw(find_mime_encoding);
+our @EXPORT_OK = qw(parse_content_type parse_content_disposition);
+
+our $STRICT_PARAMS = 1;
+
+my $ct_default = 'text/plain; charset=us-ascii';
+
+my $re_token = # US-ASCII except SPACE, CTLs and tspecials ()<>@,;:\\"/[]?=
+       qr/[\x21\x23-\x27\x2A\x2B\x2D\x2E\x30-\x39\x41-\x5A\x5E-\x7E]+/;
+
+my $re_token_non_strict = # allow CTLs and above ASCII
+       qr/([\x00-\x08\x0B\x0C\x0E-\x1F\x7E-\xFF]+|$re_token)/;
+
+my $re_qtext = # US-ASCII except CR, LF, white space, backslash and quote
+       qr/[\x01-\x08\x0B\x0C\x0E-\x1F\x21\x23-\x5B\x5D-\x7E\x7F]/;
+my $re_quoted_pair = qr/\\[\x00-\x7F]/;
+my $re_quoted_string = qr/"((?:[ \t]*(?:$re_qtext|$re_quoted_pair))*[ \t]*)"/;
+
+my $re_qtext_non_strict = qr/[\x80-\xFF]|$re_qtext/;
+my $re_quoted_pair_non_strict = qr/\\[\x00-\xFF]/;
+my $re_quoted_string_non_strict =
+qr/"((?:[ \t]*(?:$re_qtext_non_strict|$re_quoted_pair_non_strict))*[ \t]*)"/;
+
+my $re_charset = qr/[!"#\$%&'+\-0-9A-Z\\\^_`a-z\{\|\}~]+/;
+my $re_language = qr/[A-Za-z]{1,8}(?:-[0-9A-Za-z]{1,8})*/;
+my $re_exvalue = qr/($re_charset)?'(?:$re_language)?'(.*)/;
+
+sub parse_content_type {
+       my ($ct) = @_;
+
+       # If the header isn't there or is empty, give default answer.
+       $ct = $ct_default unless defined($ct) && length($ct);
+
+       _unfold_lines($ct);
+       _clean_comments($ct);
+
+       # It is also recommend (sic.) that this default be assumed when a
+       # syntactically invalid Content-Type header field is encountered.
+       unless ($ct =~ s/^($re_token)\/($re_token)//) {
+               unless ($STRICT_PARAMS && $ct =~ s/^($re_token_non_strict)\/
+                                               ($re_token_non_strict)//x) {
+                       #carp "Invalid Content-Type '$ct'";
+                       return parse_content_type($ct_default);
+               }
+       }
+
+       my ($type, $subtype) = (lc $1, lc $2);
+
+       _clean_comments($ct);
+       $ct =~ s/\s+$//;
+
+       my $attributes = {};
+       if ($STRICT_PARAMS && length($ct) && $ct !~ /^;/) {
+               # carp "Missing ';' before first Content-Type parameter '$ct'";
+       } else {
+               $attributes = _process_rfc2231(_parse_attributes($ct));
+       }
+
+       {
+               type       => $type,
+               subtype => $subtype,
+               attributes => $attributes,
+
+               # This is dumb.  Really really dumb.  For backcompat. -- rjbs,
+               # 2013-08-10
+               discrete   => $type,
+               composite  => $subtype,
+       };
+}
+
+my $cd_default = 'attachment';
+
+sub parse_content_disposition {
+       my ($cd) = @_;
+
+       $cd = $cd_default unless defined($cd) && length($cd);
+
+       _unfold_lines($cd);
+       _clean_comments($cd);
+
+       unless ($cd =~ s/^($re_token)//) {
+               unless ($STRICT_PARAMS and $cd =~ s/^($re_token_non_strict)//) {
+                       #carp "Invalid Content-Disposition '$cd'";
+                       return parse_content_disposition($cd_default);
+               }
+       }
+
+       my $type = lc $1;
+
+       _clean_comments($cd);
+       $cd =~ s/\s+$//;
+
+       my $attributes = {};
+       if ($STRICT_PARAMS && length($cd) && $cd !~ /^;/) {
+# carp "Missing ';' before first Content-Disposition parameter '$cd'";
+       } else {
+               $attributes = _process_rfc2231(_parse_attributes($cd));
+       }
+
+       {
+               type       => $type,
+               attributes => $attributes,
+       };
+}
+
+sub _unfold_lines {
+       $_[0] =~ s/(?:\r\n|[\r\n])(?=[ \t])//g;
+}
+
+sub _clean_comments {
+       my $ret = ($_[0] =~ s/^\s+//);
+       while (length $_[0]) {
+               last unless $_[0] =~ s/^\(//;
+               my $level = 1;
+               while (length $_[0]) {
+                       my $ch = substr $_[0], 0, 1, '';
+                       if ($ch eq '(') {
+                               $level++;
+                       } elsif ($ch eq ')') {
+                               $level--;
+                               last if $level == 0;
+                       } elsif ($ch eq '\\') {
+                               substr $_[0], 0, 1, '';
+                       }
+               }
+               # carp "Unbalanced comment" if $level != 0 and $STRICT_PARAMS;
+               $ret |= ($_[0] =~ s/^\s+//);
+       }
+       $ret;
+}
+
+sub _process_rfc2231 {
+       my ($attribs) = @_;
+       my %cont;
+       my %encoded;
+       foreach (keys %{$attribs}) {
+               next unless $_ =~ m/^(.*)\*([0-9])\*?$/;
+               my ($attr, $sec) = ($1, $2);
+               $cont{$attr}->[$sec] = $attribs->{$_};
+               $encoded{$attr}->[$sec] = 1 if $_ =~ m/\*$/;
+               delete $attribs->{$_};
+       }
+       foreach (keys %cont) {
+               my $key = $_;
+               $key .= '*' if $encoded{$_};
+               $attribs->{$key} = join '', @{$cont{$_}};
+       }
+       foreach (keys %{$attribs}) {
+               next unless $_ =~ m/^(.*)\*$/;
+               my $key = $1;
+               next unless $attribs->{$_} =~ m/^$re_exvalue$/;
+               my ($charset, $value) = ($1, $2);
+               $value =~ s/%([0-9A-Fa-f]{2})/pack('C', hex($1))/eg;
+               if (length $charset) {
+                       my $enc = find_mime_encoding($charset);
+                       if (defined $enc) {
+                               $value = $enc->decode($value);
+                       # } else {
+                               #carp "Unknown charset '$charset' in
+                               #attribute '$key' value";
+                       }
+               }
+               $attribs->{$key} = $value;
+               delete $attribs->{$_};
+       }
+       $attribs;
+}
+
+sub _parse_attributes {
+       local $_ = shift;
+       substr($_, 0, 0, '; ') if length $_ and $_ !~ /^;/;
+       my $attribs = {};
+       while (length $_) {
+               s/^;// or $STRICT_PARAMS and do {
+                       #carp "Missing semicolon before parameter '$_'";
+                       return $attribs;
+               };
+               _clean_comments($_);
+               unless (length $_) {
+                       # Some mail software generates a Content-Type like this:
+                       # "Content-Type: text/plain;"
+                       # RFC 1521 section 3 says a parameter must exist if
+                       # there is a semicolon.
+                       #carp "Extra semicolon after last parameter" if
+                       #$STRICT_PARAMS;
+                       return $attribs;
+               }
+               my $attribute;
+               if (s/^($re_token)=//) {
+                       $attribute = lc $1;
+               } else {
+                       if ($STRICT_PARAMS) {
+                               # carp "Illegal parameter '$_'";
+                               return $attribs;
+                       }
+                       if (s/^($re_token_non_strict)=//) {
+                               $attribute = lc $1;
+                       } else {
+                               unless (s/^([^;=\s]+)\s*=//) {
+                                       #carp "Cannot parse parameter '$_'";
+                                       return $attribs;
+                               }
+                               $attribute = lc $1;
+                       }
+               }
+               _clean_comments($_);
+               my $value = _extract_attribute_value();
+               $attribs->{$attribute} = $value;
+               _clean_comments($_);
+       }
+       $attribs;
+}
+
+sub _extract_attribute_value { # EXPECTS AND MODIFIES $_
+       my $value;
+       while (length $_) {
+               if (s/^($re_token)//) {
+                       $value .= $1;
+               } elsif (s/^$re_quoted_string//) {
+                       my $sub = $1;
+                       $sub =~ s/\\(.)/$1/g;
+                       $value .= $sub;
+               } elsif ($STRICT_PARAMS) {
+                       #my $char = substr $_, 0, 1;
+                       #carp "Unquoted '$char' not allowed";
+                       return;
+               } elsif (s/^($re_token_non_strict)//) {
+                       $value .= $1;
+               } elsif (s/^$re_quoted_string_non_strict//) {
+                       my $sub = $1;
+                       $sub =~ s/\\(.)/$1/g;
+                       $value .= $sub;
+               }
+               my $erased = _clean_comments($_);
+               last if !length $_ or /^;/;
+               if ($STRICT_PARAMS) {
+                       #my $char = substr $_, 0, 1;
+                       #carp "Extra '$char' found after parameter";
+                       return;
+               }
+               if ($erased) {
+                       # Sometimes semicolon is missing, so check for = char
+                       last if m/^$re_token_non_strict=/;
+                       $value .= ' ';
+               }
+               $value .= substr $_, 0, 1, '';
+       }
+       $value;
+}
+
+1;
+__END__
+=func parse_content_type
+
+This routine is exported by default.
+
+This routine parses email content type headers according to section 5.1 of RFC
+2045 and also RFC 2231 (Character Set and Parameter Continuations).  It returns
+a hash as above, with entries for the C<type>, the C<subtype>, and a hash of
+C<attributes>.
+
+For backward compatibility with a really unfortunate misunderstanding of RFC
+2045 by the early implementors of this module, C<discrete> and C<composite> are
+also present in the returned hashref, with the values of C<type> and C<subtype>
+respectively.
+
+=func parse_content_disposition
+
+This routine is exported by default.
+
+This routine parses email Content-Disposition headers according to RFC 2183 and
+RFC 2231.  It returns a hash as above, with entries for the C<type>, and a hash
+of C<attributes>.
+
+=cut
index 5b2914b393fdcc28ea3d220eeac3137004ae3b8d..754da13ffded9b0bb00360397140f133b8b7ec39 100644 (file)
@@ -6,7 +6,7 @@ package PublicInbox::WwwAttach; # internal package
 use strict;
 use warnings;
 use bytes (); # only for bytes::length
-use Email::MIME::ContentType qw(parse_content_type);
+use PublicInbox::EmlContentFoo qw(parse_content_type);
 use PublicInbox::Eml;
 
 sub get_attach_i { # ->each_part callback
diff --git a/t/eml_content_disposition.t b/t/eml_content_disposition.t
new file mode 100644 (file)
index 0000000..9bdacc0
--- /dev/null
@@ -0,0 +1,102 @@
+#!perl -w
+# Copyright (C) 2020 all contributors <meta@public-inbox.org>
+# Copyright (C) 2004- Simon Cozens, Casey West, Ricardo SIGNES
+# This library is free software; you can redistribute it and/or modify
+# it under the same terms as Perl itself.
+#
+# License: GPL-1.0+ or Artistic-1.0-Perl
+#  <https://www.gnu.org/licenses/gpl-1.0.txt>
+#  <https://dev.perl.org/licenses/artistic.html>
+use strict;
+use Test::More;
+use PublicInbox::EmlContentFoo qw(parse_content_disposition);
+
+my %cd_tests = (
+       '' => { type => 'attachment', attributes => {} },
+       'inline' => { type => 'inline', attributes => {} },
+       'attachment' => { type => 'attachment', attributes => {} },
+
+       'attachment; filename=genome.jpeg;' .
+       ' modification-date="Wed, 12 Feb 1997 16:29:51 -0500"' => {
+               type => 'attachment',
+               attributes => {
+                       filename => 'genome.jpeg',
+                       'modification-date' => 'Wed, 12 Feb 1997 16:29:51 -0500'
+               }
+       },
+
+       q(attachment; filename*=UTF-8''genome.jpeg;) .
+       q( modification-date="Wed, 12 Feb 1997 16:29:51 -0500") => {
+               type => 'attachment',
+               attributes => {
+                       filename => 'genome.jpeg',
+                       'modification-date' => 'Wed, 12 Feb 1997 16:29:51 -0500'
+               }
+       },
+
+       q(attachment; filename*0*=us-ascii'en'This%20is%20even%20more%20;) .
+       q( filename*1*=%2A%2A%2Afun%2A%2A%2A%20; filename*2="isn't it!") => {
+               type => 'attachment',
+               attributes => {
+                       filename => "This is even more ***fun*** isn't it!"
+               }
+       },
+
+       q(attachment; filename*0*='en'This%20is%20even%20more%20;) .
+       q( filename*1*=%2A%2A%2Afun%2A%2A%2A%20; filename*2="isn't it!") => {
+               type => 'attachment',
+               attributes => {
+                       filename => "This is even more ***fun*** isn't it!"
+               }
+       },
+
+       q(attachment; filename*0*=''This%20is%20even%20more%20;) .
+       q( filename*1*=%2A%2A%2Afun%2A%2A%2A%20; filename*2="isn't it!") => {
+               type => 'attachment',
+               attributes => {
+                       filename => "This is even more ***fun*** isn't it!"
+               }
+       },
+
+       q(attachment; filename*0*=us-ascii''This%20is%20even%20more%20;).
+       q( filename*1*=%2A%2A%2Afun%2A%2A%2A%20; filename*2="isn't it!") => {
+               type => 'attachment',
+               attributes => {
+                       filename => "This is even more ***fun*** isn't it!"
+               }
+       },
+);
+
+my %non_strict_cd_tests = (
+       'attachment; filename=genome.jpeg;' .
+       ' modification-date="Wed, 12 Feb 1997 16:29:51 -0500";' => {
+               type => 'attachment',
+               attributes => {
+                       filename => 'genome.jpeg',
+                       'modification-date' =>
+                               'Wed, 12 Feb 1997 16:29:51 -0500'
+               }
+       },
+);
+
+sub test {
+       my ($string, $expect, $info) = @_;
+       local $_;
+       $info =~ s/\r/\\r/g;
+       $info =~ s/\n/\\n/g;
+       is_deeply(parse_content_disposition($string), $expect, $info);
+}
+
+for (sort keys %cd_tests) {
+       test($_, $cd_tests{$_}, "Can parse C-D <$_>");
+}
+
+local $PublicInbox::EmlContentFoo::STRICT_PARAMS = 0;
+for (sort keys %cd_tests) {
+       test($_, $cd_tests{$_}, "Can parse non-strict C-D <$_>");
+}
+for (sort keys %non_strict_cd_tests) {
+       test($_, $non_strict_cd_tests{$_}, "Can parse non-strict C-D <$_>");
+}
+
+done_testing;
diff --git a/t/eml_content_type.t b/t/eml_content_type.t
new file mode 100644 (file)
index 0000000..5fd7d1d
--- /dev/null
@@ -0,0 +1,289 @@
+#!perl -w
+# Copyright (C) 2020 all contributors <meta@public-inbox.org>
+# Copyright (C) 2004- Simon Cozens, Casey West, Ricardo SIGNES
+# This library is free software; you can redistribute it and/or modify
+# it under the same terms as Perl itself.
+#
+# License: GPL-1.0+ or Artistic-1.0-Perl
+#  <https://www.gnu.org/licenses/gpl-1.0.txt>
+#  <https://dev.perl.org/licenses/artistic.html>
+use strict;
+use Test::More;
+use PublicInbox::EmlContentFoo qw(parse_content_type);
+
+my %ct_tests = (
+       '' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "us-ascii" }
+       },
+
+       "text/plain" => {
+               type => "text",
+               subtype => "plain",
+               attributes => {}
+       },
+       'text/plain; charset=us-ascii' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "us-ascii" }
+       },
+       'text/plain; charset="us-ascii"' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "us-ascii" }
+       },
+       "text/plain; charset=us-ascii (Plain text)" => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "us-ascii" }
+       },
+
+       'text/plain; charset=ISO-8859-1' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "ISO-8859-1" }
+       },
+       'text/plain; charset="ISO-8859-1"' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "ISO-8859-1" }
+       },
+       'text/plain; charset="ISO-8859-1" (comment)' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "ISO-8859-1" }
+       },
+
+       '(c) text/plain (c); (c) charset=ISO-8859-1 (c)' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "ISO-8859-1" }
+       },
+       '(c \( \\\\) (c) text/plain (c) (c) ; (c) (c) charset=utf-8 (c)' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "utf-8" }
+       },
+       'text/plain; (c (nested ()c)another c)() charset=ISO-8859-1' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "ISO-8859-1" }
+       },
+       'text/plain (c \(!nested ()c\)\)(nested\(c())); charset=utf-8' => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { charset => "utf-8" }
+       },
+
+       "application/foo" => {
+               type       => "application",
+               subtype    => "foo",
+               attributes => {}
+       },
+       "multipart/mixed; boundary=unique-boundary-1" => {
+               type       => "multipart",
+               subtype    => "mixed",
+               attributes => { boundary => "unique-boundary-1" }
+       },
+       'message/external-body; access-type=local-file; name="/u/n/m.jpg"' => {
+               type       => "message",
+               subtype    => "external-body",
+               attributes => {
+                       "access-type" => "local-file",
+                       "name"        => "/u/n/m.jpg"
+               }
+       },
+       'multipart/mixed; boundary="----------=_1026452699-10321-0" ' => {
+               'type'       => 'multipart',
+               'subtype'    => 'mixed',
+               'attributes' => {
+                       'boundary' => '----------=_1026452699-10321-0'
+               }
+       },
+       'multipart/report; boundary= "=_0=73e476c3-cd5a-5ba3-b910-2="' => {
+               'type'       => 'multipart',
+               'subtype'    => 'report',
+               'attributes' => {
+                       'boundary' => '=_0=73e476c3-cd5a-5ba3-b910-2='
+               }
+       },
+       'multipart/report; boundary=' . " \t" . '"=_0=7-c-5-b-2="' => {
+               'type'       => 'multipart',
+               'subtype'    => 'report',
+               'attributes' => {
+                       'boundary' => '=_0=7-c-5-b-2='
+               }
+       },
+
+       'message/external-body; access-type=URL;' .
+       ' URL*0="ftp://";' .
+       ' URL*1="example.com/"' => {
+               'type'       => 'message',
+               'subtype'    => 'external-body',
+               'attributes' => {
+                       'access-type' => 'URL',
+                       'url' => 'ftp://example.com/'
+               }
+       },
+       'message/external-body; access-type=URL; URL="ftp://example.com/"' => {
+               'type'       => 'message',
+               'subtype'    => 'external-body',
+               'attributes' => {
+                       'access-type' => 'URL',
+                       'url' => 'ftp://example.com/',
+               }
+       },
+
+       "application/x-stuff; title*=us-ascii'en-us'This%20is%20f%2Ad" => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => 'This is f*d'
+               }
+       },
+       "application/x-stuff; title*=us-ascii''This%20is%20f%2Ad" => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => 'This is f*d'
+               }
+       },
+       "application/x-stuff; title*=''This%20is%20f%2Ad" => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => 'This is f*d'
+               }
+       },
+       "application/x-stuff; title*='en-us'This%20is%20f%2Ad" => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => 'This is f*d'
+               }
+       },
+       q(application/x-stuff;) .
+       q( title*0*=us-ascii'en'This%20is%20even%20more%20;) .
+       q(title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => "This is even more ***fun*** isn't it!"
+               }
+       },
+       q(application/x-stuff;) .
+       q( title*0*='en'This%20is%20even%20more%20;) .
+       q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => "This is even more ***fun*** isn't it!"
+               }
+       },
+       q(application/x-stuff;) .
+       q( title*0*=''This%20is%20even%20more%20;) .
+       q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!") => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => "This is even more ***fun*** isn't it!"
+               }
+       },
+       q(application/x-stuff;).
+       q( title*0*=us-ascii''This%20is%20even%20more%20;).
+       q( title*1*=%2A%2A%2Afun%2A%2A%2A%20; title*2="isn't it!")
+         => {
+               'type'       => 'application',
+               'subtype'    => 'x-stuff',
+               'attributes' => {
+                       'title' => "This is even more ***fun*** isn't it!"
+               }
+       },
+
+       'text/plain; attribute="v\"v\\\\v\(v\>\<\)\@\,\;\:\/\]\[\?\=v v";' .
+       ' charset=us-ascii' => {
+               'type'       => 'text',
+               'subtype'    => 'plain',
+               'attributes' => {
+                       'attribute' => 'v"v\\v(v><)@,;:/][?=v v',
+                       'charset' => 'us-ascii',
+               },
+       },
+
+       qq(text/plain;\r
+        charset=us-ascii;\r
+        attribute="\r value1 \r value2\r\n value3\r\n value4\r\n "\r\n ) => {
+               'type'       => 'text',
+               'subtype'    => 'plain',
+               'attributes' => {
+                       'attribute' => ' value1  value2 value3 value4 ',
+                       'charset'   => 'us-ascii',
+               },
+       },
+);
+
+my %non_strict_ct_tests = (
+       "text/plain;" => { type => "text", subtype => "plain", attributes => {} },
+       "text/plain; " =>
+         { type => "text", subtype => "plain", attributes => {} },
+       'image/jpeg;' .
+       ' x-mac-type="3F3F3F3F";'.
+       ' x-mac-creator="3F3F3F3F" name="file name.jpg";' => {
+               type       => "image",
+               subtype    => "jpeg",
+               attributes => {
+                       'x-mac-type'    => "3F3F3F3F",
+                       'x-mac-creator' => "3F3F3F3F",
+                       'name'          => "file name.jpg"
+               }
+       },
+       "text/plain; key=very long value" => {
+               type       => "text",
+               subtype    => "plain",
+               attributes => { key => "very long value" }
+       },
+       "text/plain; key=very long value key2=value2" => {
+               type    => "text",
+               subtype => "plain",
+               attributes => { key => "very long value", key2 => "value2" }
+       },
+       'multipart/mixed; boundary = "--=_Next_Part_24_Nov_2016_08.09.21"' => {
+               type    => "multipart",
+               subtype => "mixed",
+               attributes => {
+                       boundary => "--=_Next_Part_24_Nov_2016_08.09.21"
+               }
+       },
+);
+
+sub test {
+       my ($string, $expect, $info) = @_;
+
+       # So stupid. -- rjbs, 2013-08-10
+       $expect->{discrete}  = $expect->{type};
+       $expect->{composite} = $expect->{subtype};
+
+       local $_;
+       $info =~ s/\r/\\r/g;
+       $info =~ s/\n/\\n/g;
+       is_deeply(parse_content_type($string), $expect, $info);
+}
+
+for (sort keys %ct_tests) {
+       test($_, $ct_tests{$_}, "Can parse C-T <$_>");
+}
+
+local $PublicInbox::EmlContentFoo::STRICT_PARAMS = 0;
+for (sort keys %ct_tests) {
+       test($_, $ct_tests{$_}, "Can parse non-strict C-T <$_>");
+}
+for (sort keys %non_strict_ct_tests) {
+       test(
+               $_,
+               $non_strict_ct_tests{$_},
+               "Can parse non-strict C-T <$_>"
+       );
+}
+
+done_testing;