Merge branch '0.5.x'

# Conflicts:
#	httpolice/__metadata__.py

CHANGELOG.rst

@@ -7,6 +7,17 @@ This project adheres to `Semantic Versioning <http://semver.org/>`_
 (which means it is unstable until 1.0).
 
 
+0.5.2 - 2017-03-24
+~~~~~~~~~~~~~~~~~~
+- Fixed a few rare crashing bugs found with `american fuzzy lop`_.
+- Fixed a couple cosmetic bugs in HTML reports.
+- When parsing a message with an unknown `transfer coding`_, HTTPolice now
+  correctly skips any checks on its payload body (such as notice `1038`_).
+
+.. _american fuzzy lop: http://lcamtuf.coredump.cx/afl/
+.. _transfer coding: https://tools.ietf.org/html/rfc7230#section-4
+
+
 0.5.1 - 2017-03-15
 ~~~~~~~~~~~~~~~~~~
 - Fixed compatibility with `httpolice-devtool`_ (when you point it to a local

httpolice/framing1.py

@@ -287,9 +287,9 @@ def _decode_transfer_coding(msg, coding):
         except Exception as e:
             msg.complain(1027, coding=coding, error=e)
             msg.body = Unavailable
-    elif okay(coding):
-        msg.complain(1003, coding=coding)
     else:
+        if okay(coding):
+            msg.complain(1003, coding=coding)
         msg.body = Unavailable
 
 

httpolice/message.py

@@ -156,7 +156,8 @@ def xml_data(self):
             except defusedxml.EntitiesForbidden:
                 self.complain(1275)
                 return Unavailable
-            except xml.etree.ElementTree.ParseError as e:
+            # http://bugs.python.org/issue29896
+            except (xml.etree.ElementTree.ParseError, UnicodeError) as e:
                 self.complain(1039, error=e)
                 return Unavailable
         else:

httpolice/parse.py

@@ -652,9 +652,8 @@ def skip(self, n):
     def consume_n_bytes(self, n):
         r = self.peek(n)
         if len(r) < n:
-            raise ParseError(name=self.name, point=self.point,
-                             found=u'%d bytes' % len(r),
-                             expected=[(u'%d bytes' % n,)])
+            raise ParseError(self.name, self.point, found=None,
+                             expected=[(u'%d bytes' % n, None)])
         else:
             self.point += n
             return r
@@ -740,8 +739,8 @@ def collect_annotations(self):
 class ParseError(Exception):
 
     def __init__(self, name, point, found, expected):
-        super(ParseError, self).__init__(u'unexpected %r at byte position %r' %
-                                         (found, point))
+        super(ParseError, self).__init__(
+            u'unexpected input at byte position %r' % point)
         self.name = name
         self.point = point
         self.found = found

httpolice/reports/html.py

@@ -95,7 +95,7 @@ def _common_meta(document):
         H.meta(charset=u'utf-8')
         H.meta(name=u'generator', content=u'HTTPolice %s' % version)
         H.style(type=u'text/css').add_raw_string(css_code)
-        H.base(_target=u'blank')
+        H.base(target=u'_blank')
 
 
 def _render_exchanges(exchanges, trashcan):
@@ -288,7 +288,7 @@ def _render_known(obj):
     text = printable(six.text_type(obj))
     cite = known.citation(obj)
     if cite:
-        with H.a(text, href=cite.url, title=None):
+        with H.a(text, href=cite.url):
             title = known.title(obj, with_citation=True)
             if title:
                 H.attr(title=title)

httpolice/syntax/rfc7231.py

@@ -174,7 +174,10 @@ def _obsolete_date(complain, r):
 def _check_day_of_week(complain, r):
     (claimed_dow, r) = r
     if r is not Unavailable and r.weekday() != claimed_dow:
-        complain(1108, date=r.strftime(u'%Y-%m-%d'),
+        # Don't use `strftime` here because it raises `ValueError`
+        # on years < 1900 in Python 2.7. (Of course, year < 1900
+        # in ``Date`` is pathological, but that's no excuse to crash).
+        complain(1108, date=u'%04d-%02d-%02d' % (r.year, r.month, r.day),
                  claimed=_DAY_NAMES[claimed_dow],
                  actual=_DAY_NAMES[r.weekday()])
     return r

test/combined_data/1005_4

@@ -0,0 +1,21 @@
+1005 1007
+
+# The chunked stream is abruptly cut off before the indicated number of bytes.
+# This used to crash HTTPolice when generating an HTML report.
+
+======== BEGIN INBOUND STREAM ========
+POST / HTTP/1.1
+Host: example.com
+Content-Type: text/plain
+User-Agent: demo
+Transfer-Encoding: chunked
+
+1c
+foo bar foo bar
+======== BEGIN OUTBOUND STREAM ========
+HTTP/1.1 200 OK
+Date: Thu, 31 Dec 2015 18:26:56 GMT
+Content-Type: text/plain
+Content-Length: 14
+
+Hello world!

test/combined_data/1039_2

@@ -0,0 +1,16 @@
+1039
+
+# This used to crash HTTPolice.
+
+======== BEGIN INBOUND STREAM ========
+GET / HTTP/1.1
+Host: example.com
+User-Agent: demo
+
+======== BEGIN OUTBOUND STREAM ========
+HTTP/1.1 200 OK
+Date: Thu, 31 Dec 2015 18:26:56 GMT
+Content-Type: text/xml
+Content-Length: 12
+
+<resu�ts/>

test/combined_data/1108_2

@@ -0,0 +1,16 @@
+1108
+
+# This used to crash HTTPolice on Python 2.7.
+
+======== BEGIN INBOUND STREAM ========
+GET / HTTP/1.1
+Host: example.com
+User-Agent: demo
+
+======== BEGIN OUTBOUND STREAM ========
+HTTP/1.1 200 OK
+Date: Fri, 02 Aug 1015 18:26:56 GMT
+Content-Type: text/plain
+Content-Length: 14
+
+Hello world!

test/test_streams_input.py

@@ -139,6 +139,11 @@ def test_chunked_empty():
     assert exch1.request.body == b''
 
 
+def test_unknown_transfer_encoding():
+    [exch1] = load_from_file('1003_1')
+    assert exch1.responses[0].body is Unavailable
+
+
 def test_implicit_response_framing():
     [exch1] = load_from_file('1025_2')
     assert exch1.responses[0].body == b'Hello world!\r\n'