Merge pull request gjtorikian#247 from oreoshake/filter-email-addresses

jch · jch · commit bd7af4ab7ef3 · 2016-04-11T14:40:43.000-07:00
Optionally filter email addresses
diff --git a/lib/html/pipeline/email_reply_filter.rb b/lib/html/pipeline/email_reply_filter.rb
@@ -27,6 +27,8 @@ class EmailReplyFilter < TextFilter
       EMAIL_SIGNATURE_HEADER = %(<div class="email-signature-reply">).freeze
       EMAIL_FRAGMENT_HEADER  = %(<div class="email-fragment">).freeze
       EMAIL_HEADER_END       = "</div>".freeze
+      EMAIL_REGEX            = /[^@\s.][^@\s]*@\[?[a-z0-9.-]+\]?/
+      HIDDEN_EMAIL_PATTERN   = "***@***.***"
 
       # Scans an email body to determine which bits are quoted and which should
       # be hidden. EmailReplyParser is used to split the comment into an Array
@@ -45,6 +47,11 @@ def call
         paragraphs = EmailReplyParser.read(text.dup).fragments.map do |fragment|
           pieces = [escape_html(fragment.to_s.strip).gsub(/^\s*(>|&gt;)/, '')]
           if fragment.quoted?
+            if context[:hide_quoted_email_addresses]
+              pieces.map! do |piece|
+                piece.gsub!(EMAIL_REGEX, HIDDEN_EMAIL_PATTERN)
+              end
+            end
             pieces.unshift EMAIL_QUOTED_HEADER
             pieces << EMAIL_HEADER_END
           elsif fragment.signature?
diff --git a/test/html/pipeline/email_reply_filter_test.rb b/test/html/pipeline/email_reply_filter_test.rb
@@ -0,0 +1,32 @@
+require "test_helper"
+
+EmailReplyFilter = HTML::Pipeline::EmailReplyFilter
+
+class HTML::Pipeline::EmailReplyFilterTest < Minitest::Test
+  def setup
+    @body = <<-EMAIL
+Hey, don't send email addresses in comments. They aren't filtered.
+
+> On Mar 5, 2016, at 08:05, Boaty McBoatface <boatymcboatface@example.com> wrote:
+>
+> Sup. alreadyleaked@example.com
+>
+> —
+> Reply to this email directly or view it on GitHub.
+EMAIL
+  end
+
+  def test_doesnt_hide_by_default
+    filter = EmailReplyFilter.new(@body)
+    doc = filter.call.to_s
+    assert_match %r(alreadyleaked@example.com), doc
+    assert_match %r(boatymcboatface@example.com), doc
+  end
+
+  def test_hides_email_addresses_when_configured
+    filter = EmailReplyFilter.new(@body, :hide_quoted_email_addresses => true)
+    doc = filter.call.to_s
+    refute_match %r(boatymcboatface@example.com), doc
+    refute_match %r(alreadyleaked@example.com), doc
+  end
+end
