From 8f3fe2b3992f14d95d08a27cca0d415be0c470f5 Mon Sep 17 00:00:00 2001
From: Martin Edenhofer <me@edenhofer.de>
Date: Tue, 21 Mar 2017 09:02:15 +0100
Subject: [PATCH] Fixed endless loop in HtmlSanitizer.cleanup_structure in
 certain conditions.

---
 lib/core_ext/string.rb       |  2 +-
 lib/html_sanitizer.rb        | 25 ++++++++++++++++++++-----
 test/unit/aaa_string_test.rb | 31 +++++++++++++++++++++++++++++++
 3 files changed, 52 insertions(+), 6 deletions(-)

diff --git a/lib/core_ext/string.rb b/lib/core_ext/string.rb
index 7564c067c..d7dd013f0 100644
--- a/lib/core_ext/string.rb
+++ b/lib/core_ext/string.rb
@@ -335,7 +335,7 @@ class String
         '(<br(|\/)>|<p>|<div>)[[:space:]]*<b>(Von|From|De|от|Z|Od|Ze|Fra|Van|Mistä|Από|Dal|から|Из|од|iz|Från|จาก|з|Từ):[[:space:]]*</b>',
         '(<br>|<div>)[[:space:]]*<br>[[:space:]]*(Von|From|De|от|Z|Od|Ze|Fra|Van|Mistä|Από|Dal|から|Из|од|iz|Från|จาก|з|Từ):[[:space:]]+',
         '<blockquote(|.+?)>[[:space:]]*<div>[[:space:]]*(On|Am|Le|El|Den|Dňa|W dniu|Il|Op|Dne|Dana)[[:space:]]',
-        '<div(|.+?)>[[:space:]]*<br>[[:space:]]*(On|Am|Le|El|Den|Dňa|W dniu|Il|Op|Dne|Dana)[[:space:]].+?<blockquote',
+        '<div(|.+?)>[[:space:]]*<br>[[:space:]]*(On|Am|Le|El|Den|Dňa|W dniu|Il|Op|Dne|Dana)[[:space:]].{1,500}<blockquote',
       ]
       map.each { |regexp|
         string.sub!(/#{regexp}/m) { |placeholder|
diff --git a/lib/html_sanitizer.rb b/lib/html_sanitizer.rb
index 27a34d81f..42fa5cbf2 100644
--- a/lib/html_sanitizer.rb
+++ b/lib/html_sanitizer.rb
@@ -237,15 +237,30 @@ cleanup html string:
     remove_empty_nodes = %w(p div span small)
     remove_empty_last_nodes = %w(b i u small)
 
-    scrubber = Loofah::Scrubber.new do |node|
+    # remove last empty nodes and empty -not needed- parrent nodes
+    scrubber_structure = Loofah::Scrubber.new do |node|
       if remove_empty_last_nodes.include?(node.name) && node.children.size.zero?
         node.remove
         Loofah::Scrubber::STOP
       end
-
-      if remove_empty_nodes.include?(node.name) && node.children.size == 1 && remove_empty_nodes.include?(node.children.first.name) # && node.children.first.text.blank?
-        node.replace cleanup_structure(node.children.to_s)
+      if remove_empty_nodes.include?(node.name) && node.children.size == 1 && remove_empty_nodes.include?(node.children.first.name)
+        node.replace node.children.to_s
+        Loofah::Scrubber::STOP
       end
+    end
+    string = Loofah.fragment(string).scrub!(scrubber_structure).to_s
+
+    new_string = ''
+    done = true
+    while done
+      new_string = Loofah.fragment(string).scrub!(scrubber_structure).to_s
+      if string == new_string
+        done = false
+      end
+      string = new_string
+    end
+
+    scrubber_cleanup = Loofah::Scrubber.new do |node|
 
       # remove mailto links
       if node['href']
@@ -298,7 +313,7 @@ cleanup html string:
         end
       end
     end
-    Loofah.fragment(string).scrub!(scrubber).to_s
+    Loofah.fragment(string).scrub!(scrubber_cleanup).to_s
   end
 
   def self.add_link(content, urls, node)
diff --git a/test/unit/aaa_string_test.rb b/test/unit/aaa_string_test.rb
index abe4ff95f..287caa8bf 100644
--- a/test/unit/aaa_string_test.rb
+++ b/test/unit/aaa_string_test.rb
@@ -1088,6 +1088,37 @@ christian.schaefer@example.com'
 </div>
 </blockquote>'
     assert_equal(result, html.html2html_strict)
+
+    html = "<div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div style=\"line-height:1.7;color:#000000;font-size:14px;font-family:Arial\"><div>Dear Bob<span style=\"line-height: 23.8px;\">:</span><span style=\"color: rgb(255, 255, 255); line-height: 1.7;\">Mr/Mrs</span></div><div><br></div><div><span style=\"line-height: 1.7;\">We&nbsp;are&nbsp;one&nbsp;of&nbsp;the&nbsp;leading&nbsp;manufacturer&nbsp;and&nbsp;supplier&nbsp;of&nbsp;</span>conduits and cars since 3000.</div><div><br></div><div>Could you inform me the specification you need?</div><div><br></div><div>May I sent you our products catalogues for your reference?</div><div><br></div><div><img src=\"cid:5cb2783c$1$15ae9b384c8$Coremail$zhanabcdzhao$example.com\" orgwidth=\"1101\" orgheight=\"637\" data-image=\"1\" style=\"width: 722.7px; height: 418px; border: none;\"></div><div>Best regards!</div><div><br></div><div><b style=\"line-height: 1.7;\"><i><u><span lang=\"EL\" style=\"font-size:11.0pt;font-family:&quot;Calibri&quot;,sans-serif;color:#17365D;\nmso-ansi-language:EL\">Welcome to our booth B11/1 Hall 13 during SOMEWHERE\n9999.</span></u></i></b></div><div style=\"position:relative;zoom:1\"><div>Bob Smith</div><div><div>Exp. &amp; Imp.</div><div>Town Example Electric Co., Ltd.</div><div>Tel: 0000-11-12345678 (Ext-220) &nbsp;Fax: 0000-11-12345678&nbsp;</div><div>Room1234, NO. 638, Smith Road, Town, 200000, Somewhere</div><div>Web: www.example.com</div></div><div style=\"clear:both\"></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div>"
+    result = '<div>
+<div>Dear Bob:Mr/Mrs</div>
+<div> </div>
+<div>We are one of the leading manufacturer and supplier of conduits and cars since 3000.</div>
+<div> </div>
+<div>Could you inform me the specification you need?</div>
+<div> </div>
+<div>May I sent you our products catalogues for your reference?</div>
+<div> </div>
+<div><img src="cid:5cb2783c%241%2415ae9b384c8%24Coremail%24zhanabcdzhao%24example.com" style="width: 722.7px; height: 418px;"></div>
+<div>Best regards!</div>
+<div> </div>
+<div><b><i><u>Welcome to our booth B11/1 Hall 13 during SOMEWHERE 9999.</u></i></b></div>
+<div>
+<div>Bob Smith</div>
+<div>
+<div>Exp. &amp; Imp.</div>
+<div>Town Example Electric Co., Ltd.</div>
+<div>Tel: 0000-11-12345678 (Ext-220) Fax: 0000-11-12345678</div>
+<div>Room1234, NO. 638, Smith Road, Town, 200000, Somewhere</div>
+<div>Web: <a href="http://www.example.com" rel="nofollow" target="_blank">http://www.example.com</a>
+</div>
+</div>
+<div></div>
+</div>
+</div>'
+    assert_equal(result, html.html2html_strict)
+
+
   end
 
 end