Translate HTML to text for update alert emails.

3 files changed, 64 insertions, 3 deletions

diff --git a/perllib/FixMyStreet/Template.pm b/perllib/FixMyStreet/Template.pm
index 7d6798415..35efcc1cf 100644
--- a/perllib/FixMyStreet/Template.pm
+++ b/perllib/FixMyStreet/Template.pm
@@ -7,6 +7,7 @@ use FixMyStreet;
 use mySociety::Locale;
 use Attribute::Handlers;
 use HTML::Scrubber;
+use HTML::TreeBuilder;
 use FixMyStreet::Template::SafeString;
 use FixMyStreet::Template::Context;
 use FixMyStreet::Template::Stash;
@@ -161,6 +162,66 @@ sub sanitize {
 }
 
 
+=head2 email_sanitize_text
+
+Intended for use in the _email_comment_list.txt template to allow HTML
+in updates from staff/superusers. Sanitizes the HTML and then converts
+it all to text.
+
+=cut
+
+sub email_sanitize_text : Fn('email_sanitize_text') {
+    my $update = shift;
+
+    my $text = $update->{item_text};
+    my $extra = $update->{item_extra};
+    $extra = $extra ? RABX::wire_rd(new IO::String($extra)) : {};
+
+    my $staff = $extra->{is_superuser} || $extra->{is_body_user};
+
+    return $text unless $staff;
+
+    $text = FixMyStreet::Template::sanitize($text);
+
+    my $tree = HTML::TreeBuilder->new_from_content($text);
+    _sanitize_elt($tree);
+
+    return $tree->as_text;
+}
+
+my $list_type;
+my $list_num;
+my $sanitize_text_subs = {
+    b => [ '*', '*' ],
+    strong => [ '*', '*' ],
+    i => [ '_', '_' ],
+    em => [ '_', '_' ],
+    p => [ '', "\n\n" ],
+    li => [ '', "\n\n" ],
+};
+sub _sanitize_elt {
+    my $elt = shift;
+    foreach ($elt->content_list) {
+        next unless ref $_;
+        $list_type = $_->tag, $list_num = 1 if $_->tag eq 'ol' || $_->tag eq 'ul';
+        _sanitize_elt($_);
+        $_->replace_with("\n") if $_->tag eq 'br';
+        $_->replace_with('[image: ', $_->attr('alt'), ']') if $_->tag eq 'img';
+        $_->replace_with($_->as_text, ' [', $_->attr('href'), ']') if $_->tag eq 'a';
+        $_->replace_with_content if $_->tag eq 'span' || $_->tag eq 'font';
+        $_->replace_with_content if $_->tag eq 'ul' || $_->tag eq 'ol';
+        if ($_->tag eq 'li') {
+            $sanitize_text_subs->{li}[0] = $list_type eq 'ol' ? "$list_num. " : '* ';
+            $list_num++;
+        }
+        if (my $sub = $sanitize_text_subs->{$_->tag}) {
+            $_->preinsert($sub->[0]);
+            $_->postinsert($sub->[1]);
+            $_->replace_with_content;
+        }
+    }
+}
+
 =head2 email_sanitize_html
 
 Intended for use in the _email_comment_list.html template to allow HTML
diff --git a/t/app/controller/alert_new.t b/t/app/controller/alert_new.t
index 16455c7c4..31d8c52ed 100644
--- a/t/app/controller/alert_new.t
+++ b/t/app/controller/alert_new.t
@@ -878,7 +878,7 @@ subtest 'check staff updates can include sanitized HTML' => sub {
         user => $user1,
     });
 
-    my $update1 = $mech->create_comment_for_problem($report, $user2, 'Staff User', 'This is some update text with <strong>HTML</strong> and *italics*. <script>not allowed</script>', 't', 'confirmed', undef, { confirmed  => $r_dt->clone->add( minutes => 8 ) });
+    my $update1 = $mech->create_comment_for_problem($report, $user2, 'Staff User', '<p>This is some update text with <strong>HTML</strong> and *italics*.</p> <ul><li>Even a list</li><li>Which might work</li><li>In the <a href="https://www.fixmystreet.com/">text</a> part</li></ul> <script>not allowed</script>', 't', 'confirmed', undef, { confirmed  => $r_dt->clone->add( minutes => 8 ) });
     $update1->set_extra_metadata(is_body_user => $user2->from_body->id);
     $update1->update;
 
@@ -896,7 +896,7 @@ subtest 'check staff updates can include sanitized HTML' => sub {
     FixMyStreet::DB->resultset('AlertType')->email_alerts();
     my $email = $mech->get_email;
     my $plain = $mech->get_text_body_from_email($email);
-    like $plain, qr/This is some update text with <strong>HTML<\/strong> and \*italics\*\./, 'plain text part contains exactly what was entered';
+    like $plain, qr/This is some update text with \*HTML\* and \*italics\*\.\r\n\r\n\* Even a list\r\n\r\n\* Which might work\r\n\r\n\* In the text \[https:\/\/www.fixmystreet.com\/\] part/, 'plain text part contains no HTML tags from staff update';
     like $plain, qr/Public users <i>cannot<\/i> use HTML\./, 'plain text part contains exactly what was entered';
 
     my $html = $mech->get_html_body_from_email($email);
diff --git a/templates/email/default/_email_comment_list.txt b/templates/email/default/_email_comment_list.txt
index dbf00640f..3e01580c3 100644
--- a/templates/email/default/_email_comment_list.txt
+++ b/templates/email/default/_email_comment_list.txt
@@ -1,7 +1,7 @@
 [% FOR row IN data -%]
 [% row.item_name _ ' : ' IF row.item_name AND NOT row.item_anonymous -%]
 [% '(' _ cobrand.prettify_dt(row.confirmed) _ ') ' IF cobrand.include_time_in_update_alerts -%]
-[% row.item_text %]
+[% email_sanitize_text(row) %]
 
 ------