improve display of non-ascii contents

fixes #283

libmproxy/contentviews.py

@@ -27,7 +27,7 @@ import six
 
 from netlib.odict import ODict
 from netlib import encoding
-import netlib.utils
+from netlib.utils import clean_bin, hexdump, urldecode, multipartdecode, parse_content_type
 
 from . import utils
 from .exceptions import ContentViewException
@@ -121,12 +121,14 @@ class ViewAuto(View):
         headers = metadata.get("headers", {})
         ctype = headers.get("content-type")
         if ctype:
-            ct = netlib.utils.parse_content_type(ctype) if ctype else None
+            ct = parse_content_type(ctype) if ctype else None
             ct = "%s/%s" % (ct[0], ct[1])
             if ct in content_types_map:
                 return content_types_map[ct][0](data, **metadata)
             elif utils.isXML(data):
                 return get("XML")(data, **metadata)
+        if utils.isMostlyBin(data):
+            return get("Hex")(data)
         return get("Raw")(data)
 
 
@@ -146,7 +148,7 @@ class ViewHex(View):
 
     @staticmethod
     def _format(data):
-        for offset, hexa, s in netlib.utils.hexdump(data):
+        for offset, hexa, s in hexdump(data):
             yield [
                 ("offset", offset + " "),
                 ("text", hexa + "   "),
@@ -251,7 +253,7 @@ class ViewURLEncoded(View):
     content_types = ["application/x-www-form-urlencoded"]
 
     def __call__(self, data, **metadata):
-        d = netlib.utils.urldecode(data)
+        d = urldecode(data)
         return "URLEncoded form", format_dict(ODict(d))
 
 
@@ -268,7 +270,7 @@ class ViewMultipart(View):
 
     def __call__(self, data, **metadata):
         headers = metadata.get("headers", {})
-        v = netlib.utils.multipartdecode(headers, data)
+        v = multipartdecode(headers, data)
         if v:
             return "Multipart form", self._format(v)
 
@@ -519,6 +521,21 @@ def get(name):
             return i
 
 
+def safe_to_print(lines, encoding="utf8"):
+    """
+    Wraps a content generator so that each text portion is a *safe to print* unicode string.
+    """
+    for line in lines:
+        clean_line = []
+        for (style, text) in line:
+            try:
+                text = clean_bin(text.decode(encoding, "strict"))
+            except UnicodeDecodeError:
+                text = clean_bin(text).decode(encoding, "strict")
+            clean_line.append((style, text))
+        yield clean_line
+
+
 def get_content_view(viewmode, data, **metadata):
     """
         Args:
@@ -527,6 +544,7 @@ def get_content_view(viewmode, data, **metadata):
 
         Returns:
             A (description, content generator) tuple.
+            In contrast to calling the views directly, text is always safe-to-print unicode.
 
         Raises:
             ContentViewException, if the content view threw an error.
@@ -556,4 +574,4 @@ def get_content_view(viewmode, data, **metadata):
         msg.append("Couldn't parse: falling back to Raw")
     else:
         msg.append(ret[0])
-    return " ".join(msg), ret[1]
+    return " ".join(msg), safe_to_print(ret[1])

libmproxy/dump.py

@@ -57,12 +57,8 @@ class Options(object):
                 setattr(self, i, None)
 
 
-_contentview_auto = contentviews.get("Auto")
-_contentview_raw = contentviews.get("Raw")
-
-
 class DumpMaster(flow.FlowMaster):
-    def __init__(self, server, options, outfile=sys.stdout):
+    def __init__(self, server, options, outfile=None):
         flow.FlowMaster.__init__(self, server, flow.State())
         self.outfile = outfile
         self.o = options
@@ -91,7 +87,7 @@ class DumpMaster(flow.FlowMaster):
         if options.outfile:
             path = os.path.expanduser(options.outfile[0])
             try:
-                f = file(path, options.outfile[1])
+                f = open(path, options.outfile[1])
                 self.start_stream(f, self.filt)
             except IOError as v:
                 raise DumpError(v.strerror)
@@ -185,7 +181,7 @@ class DumpMaster(flow.FlowMaster):
 
                 try:
                     type, lines = contentviews.get_content_view(
-                        _contentview_auto,
+                        contentviews.get("Auto"),
                         message.body,
                         headers=message.headers
                     )
@@ -193,7 +189,7 @@ class DumpMaster(flow.FlowMaster):
                     s = "Content viewer failed: \n" + traceback.format_exc()
                     self.add_event(s, "debug")
                     type, lines = contentviews.get_content_view(
-                        _contentview_raw,
+                        contentviews.get("Raw"),
                         message.body,
                         headers=message.headers
                     )
@@ -206,17 +202,19 @@ class DumpMaster(flow.FlowMaster):
                 )
 
                 def colorful(line):
-                    yield "    "  # we can already indent here
+                    yield u"    "  # we can already indent here
                     for (style, text) in line:
                         yield click.style(text, **styles.get(style, {}))
 
                 if self.o.flow_detail == 3:
-                    lines_to_echo = itertools.islice(lines, contentviews.VIEW_CUTOFF)
+                    lines_to_echo = itertools.islice(lines, 70)
                 else:
                     lines_to_echo = lines
 
-                content = "\r\n".join(
+                lines_to_echo = list(lines_to_echo)
-                    "".join(colorful(line)) for line in lines_to_echo
+
+                content = u"\r\n".join(
+                    u"".join(colorful(line)) for line in lines_to_echo
                 )
 
                 self.echo(content)
@@ -302,6 +300,7 @@ class DumpMaster(flow.FlowMaster):
         if f.error:
             self.echo(" << {}".format(f.error.msg), bold=True, fg="red")
 
+        if self.outfile:
             self.outfile.flush()
 
     def _process_flow(self, f):

libmproxy/protocol/http.py

@@ -1,6 +1,7 @@
 from __future__ import (absolute_import, print_function, division)
 import itertools
 import sys
+import traceback
 
 import six
 
@@ -384,9 +385,13 @@ class HttpLayer(Layer):
                     return
 
             except (HttpErrorConnClosed, NetLibError, HttpError, ProtocolException) as e:
+                error_propagated = False
                 if flow.request and not flow.response:
-                    flow.error = Error(repr(e))
+                    flow.error = Error(str(e))
                     self.channel.ask("error", flow)
+                    self.log(traceback.format_exc(), "debug")
+                    error_propagated = True
+
                 try:
                     self.send_response(make_error_response(
                         getattr(e, "code", 502),
@@ -394,6 +399,8 @@ class HttpLayer(Layer):
                     ))
                 except NetLibError:
                     pass
+
+                if not error_propagated:
                     if isinstance(e, ProtocolException):
                         six.reraise(ProtocolException, e, sys.exc_info()[2])
                     else:

libmproxy/protocol/rawtcp.py

@@ -7,7 +7,7 @@ import sys
 from OpenSSL import SSL
 
 from netlib.tcp import NetLibError, ssl_read_select
-from netlib.utils import cleanBin
+from netlib.utils import clean_bin
 from ..exceptions import ProtocolException
 from .base import Layer
 
@@ -58,7 +58,7 @@ class RawTCPLayer(Layer):
                             direction = "-> tcp -> {}".format(repr(self.server_conn.address))
                         else:
                             direction = "<- tcp <- {}".format(repr(self.server_conn.address))
-                        data = cleanBin(buf[:size].tobytes())
+                        data = clean_bin(buf[:size].tobytes())
                         self.log(
                             "{}\r\n{}".format(direction, data),
                             "info"

libmproxy/utils.py

@@ -1,4 +1,4 @@
-from __future__ import absolute_import
+from __future__ import (absolute_import, print_function, division)
 import os
 import datetime
 import re
@@ -30,15 +30,16 @@ def isBin(s):
     """
     for i in s:
         i = ord(i)
-        if i < 9:
+        if i < 9 or 13 < i < 32 or 126 < i:
-            return True
-        elif i > 13 and i < 32:
-            return True
-        elif i > 126:
             return True
     return False
 
 
+def isMostlyBin(s):
+    s = s[:100]
+    return sum(isBin(ch) for ch in s)/len(s) > 0.3
+
+
 def isXML(s):
     for i in s:
         if i in "\n \t":