py3++

9 files changed, 56 insertions, 48 deletions

diff --git a/mitmproxy/filt.py b/mitmproxy/filt.py
index d98e3749..b1b72aa7 100644
--- a/mitmproxy/filt.py
+++ b/mitmproxy/filt.py
@@ -35,6 +35,7 @@ from __future__ import absolute_import, print_function, division
 
 import re
 import sys
+from netlib import strutils
 
 import pyparsing as pp
 
@@ -78,38 +79,43 @@ class FResp(_Action):
     help = "Match response"
 
     def __call__(self, f):
-        return True if f.response else False
+        return bool(f.response)
 
 
 class _Rex(_Action):
     flags = 0
+    is_binary = True
 
     def __init__(self, expr):
         self.expr = expr
+        if self.is_binary:
+            expr = strutils.escaped_str_to_bytes(expr)
         try:
-            self.re = re.compile(self.expr, self.flags)
+            self.re = re.compile(expr, self.flags)
         except:
             raise ValueError("Cannot compile expression.")
 
 
-def _check_content_type(expr, o):
-    val = o.headers.get("content-type")
-    if val and re.search(expr, val):
-        return True
-    return False
+def _check_content_type(rex, message):
+    return any(
+        name.lower() == b"content-type" and
+        rex.search(value)
+        for name, value in message.headers.fields
+    )
 
 
 class FAsset(_Action):
     code = "a"
     help = "Match asset in response: CSS, Javascript, Flash, images."
     ASSET_TYPES = [
-        "text/javascript",
-        "application/x-javascript",
-        "application/javascript",
-        "text/css",
-        "image/.*",
-        "application/x-shockwave-flash"
+        b"text/javascript",
+        b"application/x-javascript",
+        b"application/javascript",
+        b"text/css",
+        b"image/.*",
+        b"application/x-shockwave-flash"
     ]
+    ASSET_TYPES = [re.compile(x) for x in ASSET_TYPES]
 
     def __call__(self, f):
         if f.response:
@@ -124,9 +130,9 @@ class FContentType(_Rex):
     help = "Content-type header"
 
     def __call__(self, f):
-        if _check_content_type(self.expr, f.request):
+        if _check_content_type(self.re, f.request):
             return True
-        elif f.response and _check_content_type(self.expr, f.response):
+        elif f.response and _check_content_type(self.re, f.response):
             return True
         return False
 
@@ -136,7 +142,7 @@ class FRequestContentType(_Rex):
     help = "Request Content-Type header"
 
     def __call__(self, f):
-        return _check_content_type(self.expr, f.request)
+        return _check_content_type(self.re, f.request)
 
 
 class FResponseContentType(_Rex):
@@ -145,7 +151,7 @@ class FResponseContentType(_Rex):
 
     def __call__(self, f):
         if f.response:
-            return _check_content_type(self.expr, f.response)
+            return _check_content_type(self.re, f.response)
         return False
 
 
@@ -222,7 +228,7 @@ class FMethod(_Rex):
     flags = re.IGNORECASE
 
     def __call__(self, f):
-        return bool(self.re.search(f.request.method))
+        return bool(self.re.search(f.request.data.method))
 
 
 class FDomain(_Rex):
@@ -231,12 +237,13 @@ class FDomain(_Rex):
     flags = re.IGNORECASE
 
     def __call__(self, f):
-        return bool(self.re.search(f.request.host))
+        return bool(self.re.search(f.request.data.host))
 
 
 class FUrl(_Rex):
     code = "u"
     help = "URL"
+    is_binary = False
     # FUrl is special, because it can be "naked".
 
     @classmethod
@@ -252,6 +259,7 @@ class FUrl(_Rex):
 class FSrc(_Rex):
     code = "src"
     help = "Match source address"
+    is_binary = False
 
     def __call__(self, f):
         return f.client_conn.address and self.re.search(repr(f.client_conn.address))
@@ -260,6 +268,7 @@ class FSrc(_Rex):
 class FDst(_Rex):
     code = "dst"
     help = "Match destination address"
+    is_binary = False
 
     def __call__(self, f):
         return f.server_conn.address and self.re.search(repr(f.server_conn.address))
diff --git a/mitmproxy/protocol/http_replay.py b/mitmproxy/protocol/http_replay.py
index e804eba9..986de845 100644
--- a/mitmproxy/protocol/http_replay.py
+++ b/mitmproxy/protocol/http_replay.py
@@ -49,7 +49,7 @@ class RequestReplayThread(basethread.BaseThread):
                     server = models.ServerConnection(server_address, (self.config.host, 0))
                     server.connect()
                     if r.scheme == "https":
-                        connect_request = models.make_connect_request((r.host, r.port))
+                        connect_request = models.make_connect_request((r.data.host, r.port))
                         server.wfile.write(http1.assemble_request(connect_request))
                         server.wfile.flush()
                         resp = http1.read_response(
diff --git a/netlib/http/headers.py b/netlib/http/headers.py
index 14888ea9..f052a53b 100644
--- a/netlib/http/headers.py
+++ b/netlib/http/headers.py
@@ -156,8 +156,10 @@ class Headers(multidict.MultiDict):
         Returns:
             The number of replacements made.
         """
-        pattern = _always_bytes(pattern)
-        repl = _always_bytes(repl)
+        if isinstance(pattern, six.text_type):
+            pattern = strutils.escaped_str_to_bytes(pattern)
+        if isinstance(repl, six.text_type):
+            repl = strutils.escaped_str_to_bytes(repl)
         pattern = re.compile(pattern, flags)
         replacements = 0
 
diff --git a/netlib/http/message.py b/netlib/http/message.py
index b633b671..0583c246 100644
--- a/netlib/http/message.py
+++ b/netlib/http/message.py
@@ -1,5 +1,6 @@
 from __future__ import absolute_import, print_function, division
 
+import re
 import warnings
 
 import six
@@ -196,11 +197,14 @@ class Message(basetypes.Serializable):
         Returns:
             The number of replacements made.
         """
-        # TODO: Proper distinction between text and bytes.
+        if isinstance(pattern, six.text_type):
+            pattern = strutils.escaped_str_to_bytes(pattern)
+        if isinstance(repl, six.text_type):
+            repl = strutils.escaped_str_to_bytes(repl)
         replacements = 0
         if self.content:
             with decoded(self):
-                self.content, replacements = strutils.safe_subn(
+                self.content, replacements = re.subn(
                     pattern, repl, self.content, flags=flags
                 )
         replacements += self.headers.replace(pattern, repl, flags)
diff --git a/netlib/http/request.py b/netlib/http/request.py
index b64ccc51..ff057b79 100644
--- a/netlib/http/request.py
+++ b/netlib/http/request.py
@@ -65,10 +65,14 @@ class Request(message.Message):
             Returns:
                 The number of replacements made.
         """
-        # TODO: Proper distinction between text and bytes.
+        if isinstance(pattern, six.text_type):
+            pattern = strutils.escaped_str_to_bytes(pattern)
+        if isinstance(repl, six.text_type):
+            repl = strutils.escaped_str_to_bytes(repl)
+
         c = super(Request, self).replace(pattern, repl, flags)
-        self.path, pc = strutils.safe_subn(
-            pattern, repl, self.path, flags=flags
+        self.path, pc = re.subn(
+            pattern, repl, self.data.path, flags=flags
         )
         c += pc
         return c
diff --git a/netlib/strutils.py b/netlib/strutils.py
index 5ad41c7e..ca6eaa42 100644
--- a/netlib/strutils.py
+++ b/netlib/strutils.py
@@ -1,4 +1,3 @@
-import re
 import unicodedata
 import codecs
 
@@ -56,15 +55,6 @@ def clean_bin(s, keep_spacing=True):
         )
 
 
-def safe_subn(pattern, repl, target, *args, **kwargs):
-    """
-        There are Unicode conversion problems with re.subn. We try to smooth
-        that over by casting the pattern and replacement to strings. We really
-        need a better solution that is aware of the actual content ecoding.
-    """
-    return re.subn(str(pattern), str(repl), target, *args, **kwargs)
-
-
 def bytes_to_escaped_str(data):
     """
     Take bytes and return a safe string that can be displayed to the user.
diff --git a/test/mitmproxy/test_server.py b/test/mitmproxy/test_server.py
index 5415b660..c5caa21f 100644
--- a/test/mitmproxy/test_server.py
+++ b/test/mitmproxy/test_server.py
@@ -518,7 +518,7 @@ class TestTransparent(tservers.TransparentProxyTest, CommonMixin, TcpMixin):
         d = self.pathod('200:b"foo"')
         self._tcpproxy_off()
 
-        assert d.content == "bar"
+        assert d.content == b"bar"
 
         self.master.unload_scripts()
 
@@ -641,7 +641,7 @@ class MasterRedirectRequest(tservers.TestMaster):
 
     @controller.handler
     def response(self, f):
-        f.response.content = str(f.client_conn.address.port)
+        f.response.content = bytes(f.client_conn.address.port)
         f.response.headers["server-conn-id"] = str(f.server_conn.source_address.port)
         super(MasterRedirectRequest, self).response(f)
 
@@ -723,7 +723,7 @@ class TestStreamRequest(tservers.HTTPProxyTest):
     def test_stream_chunked(self):
         connection = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
         connection.connect(("127.0.0.1", self.proxy.port))
-        fconn = connection.makefile()
+        fconn = connection.makefile("rb")
         spec = '200:h"Transfer-Encoding"="chunked":r:b"4\\r\\nthis\\r\\n11\\r\\nisatest__reachhex\\r\\n0\\r\\n\\r\\n"'
         connection.send(
             b"GET %s/p/%s HTTP/1.1\r\n" %
@@ -736,7 +736,7 @@ class TestStreamRequest(tservers.HTTPProxyTest):
         assert resp.status_code == 200
 
         chunks = list(http1.read_body(fconn, None))
-        assert chunks == ["this", "isatest__reachhex"]
+        assert chunks == [b"this", b"isatest__reachhex"]
 
         connection.close()
 
@@ -848,7 +848,7 @@ class TestUpstreamProxy(tservers.HTTPUpstreamProxyTest, CommonMixin, AppMixin):
 
         p = self.pathoc()
         req = p.request("get:'%s/p/418:b\"foo\"'" % self.server.urlbase)
-        assert req.content == "ORLY"
+        assert req.content == b"ORLY"
         assert req.status_code == 418
 
 
diff --git a/test/netlib/test_strutils.py b/test/netlib/test_strutils.py
index 84a0dded..a50fc40a 100644
--- a/test/netlib/test_strutils.py
+++ b/test/netlib/test_strutils.py
@@ -29,10 +29,6 @@ def test_clean_bin():
     assert strutils.clean_bin(u"\u2605") == u"\u2605"
 
 
-def test_safe_subn():
-    assert strutils.safe_subn("foo", u"bar", "\xc2foo")
-
-
 def test_bytes_to_escaped_str():
     assert strutils.bytes_to_escaped_str(b"foo") == "foo"
     assert strutils.bytes_to_escaped_str(b"\b") == r"\x08"
diff --git a/tox.ini b/tox.ini
index 8469ccae..899fffc6 100644
--- a/tox.ini
+++ b/tox.ini
@@ -2,7 +2,7 @@
 envlist = py27, py35, docs, lint
 
 [testenv]
-usedevelop=True
+usedevelop = True
 deps =
   {env:CI_DEPS:}
   -rrequirements.txt
@@ -16,7 +16,7 @@ commands =
 
 [testenv:py35]
 setenv =
-  TESTS = test/netlib test/pathod/ test/mitmproxy/script test/mitmproxy/test_contentview.py test/mitmproxy/test_custom_contentview.py test/mitmproxy/test_app.py test/mitmproxy/test_controller.py test/mitmproxy/test_fuzzing.py test/mitmproxy/test_script.py test/mitmproxy/test_web_app.py test/mitmproxy/test_utils.py test/mitmproxy/test_stateobject.py test/mitmproxy/test_cmdline.py test/mitmproxy/test_contrib_tnetstring.py test/mitmproxy/test_proxy.py test/mitmproxy/test_protocol_http1.py test/mitmproxy/test_platform_pf.py test/mitmproxy/test_server.py
+  TESTS = test/netlib test/pathod/ test/mitmproxy/script test/mitmproxy/test_contentview.py test/mitmproxy/test_custom_contentview.py test/mitmproxy/test_app.py test/mitmproxy/test_controller.py test/mitmproxy/test_fuzzing.py test/mitmproxy/test_script.py test/mitmproxy/test_web_app.py test/mitmproxy/test_utils.py test/mitmproxy/test_stateobject.py test/mitmproxy/test_cmdline.py test/mitmproxy/test_contrib_tnetstring.py test/mitmproxy/test_proxy.py test/mitmproxy/test_protocol_http1.py test/mitmproxy/test_platform_pf.py test/mitmproxy/test_server.py test/mitmproxy/test_filt.py
   HOME = {envtmpdir}
 
 [testenv:docs]
@@ -25,4 +25,7 @@ commands = sphinx-build -W -b html -d {envtmpdir}/doctrees .  {envtmpdir}/html
 
 [testenv:lint]
 deps = flake8>=2.6.2, <3
+usedevelop = False
+skip_install = True
+skipsdist = True
 commands = flake8 --jobs 8 --count mitmproxy netlib pathod examples test