h2: move header parsing to netlib

4 files changed, 48 insertions, 67 deletions

diff --git a/mitmproxy/protocol/http2.py b/mitmproxy/protocol/http2.py
index b9a30c7e..f6261b6b 100644
--- a/mitmproxy/protocol/http2.py
+++ b/mitmproxy/protocol/http2.py
@@ -330,39 +330,13 @@ class Http2SingleStreamLayer(http._HttpTransmissionLayer, basethread.BaseThread)
         if self.zombie:  # pragma: no cover
             raise exceptions.Http2ProtocolException("Zombie Stream")
 
-        authority = self.request_headers.get(':authority', '')
-        method = self.request_headers.get(':method', 'GET')
-        scheme = self.request_headers.get(':scheme', 'https')
-        path = self.request_headers.get(':path', '/')
-        self.request_headers.clear(":method")
-        self.request_headers.clear(":scheme")
-        self.request_headers.clear(":path")
-        host = None
-        port = None
-
-        if path == '*' or path.startswith("/"):
-            first_line_format = "relative"
-        elif method == 'CONNECT':  # pragma: no cover
-            raise NotImplementedError("CONNECT over HTTP/2 is not implemented.")
-        else:  # pragma: no cover
-            first_line_format = "absolute"
-            # FIXME: verify if path or :host contains what we need
-            scheme, host, port, _ = netlib.http.url.parse(path)
-
-        if authority:
-            host, _, port = authority.partition(':')
-
-        if not host:
-            host = 'localhost'
-        if not port:
-            port = 443 if scheme == 'https' else 80
-        port = int(port)
-
         data = []
         while self.request_data_queue.qsize() > 0:
             data.append(self.request_data_queue.get())
         data = b"".join(data)
 
+        first_line_format, method, scheme, host, port, path = http2.parse_headers(self.request_headers)
+
         return models.HTTPRequest(
             first_line_format,
             method,
diff --git a/netlib/http/http2/__init__.py b/netlib/http/http2/__init__.py
index 6a979a0d..60064190 100644
--- a/netlib/http/http2/__init__.py
+++ b/netlib/http/http2/__init__.py
@@ -1,6 +1,8 @@
 from __future__ import absolute_import, print_function, division
 from netlib.http.http2 import framereader
+from netlib.http.http2.utils import parse_headers
 
 __all__ = [
     "framereader",
+    "parse_headers",
 ]
diff --git a/netlib/http/http2/utils.py b/netlib/http/http2/utils.py
new file mode 100644
index 00000000..4d5f580c
--- /dev/null
+++ b/netlib/http/http2/utils.py
@@ -0,0 +1,37 @@
+from netlib.http import url
+
+
+def parse_headers(headers):
+    authority = headers.get(':authority', b'')
+    method = headers.get(':method', b'GET')
+    scheme = headers.get(':scheme', b'https')
+    path = headers.get(':path', b'/')
+
+    headers.clear(":method")
+    headers.clear(":scheme")
+    headers.clear(":path")
+
+    host = None
+    port = None
+
+    if path == b'*' or path.startswith(b"/"):
+        first_line_format = "relative"
+    elif method == b'CONNECT':  # pragma: no cover
+        raise NotImplementedError("CONNECT over HTTP/2 is not implemented.")
+    else:  # pragma: no cover
+        first_line_format = "absolute"
+        # FIXME: verify if path or :host contains what we need
+        scheme, host, port, _ = url.parse(path)
+
+    if authority:
+        host, _, port = authority.partition(b':')
+
+    if not host:
+        host = b'localhost'
+
+    if not port:
+        port = 443 if scheme == b'https' else 80
+
+    port = int(port)
+
+    return first_line_format, method, scheme, host, port, path
diff --git a/pathod/protocols/http2.py b/pathod/protocols/http2.py
index c8728940..d94cd981 100644
--- a/pathod/protocols/http2.py
+++ b/pathod/protocols/http2.py
@@ -7,8 +7,7 @@ import hyperframe.frame
 from hpack.hpack import Encoder, Decoder
 
 from netlib import utils, strutils
-from netlib.http import url
-from netlib.http.http2 import framereader
+from netlib.http import http2
 import netlib.http.headers
 import netlib.http.response
 import netlib.http.request
@@ -101,46 +100,15 @@ class HTTP2StateProtocol(object):
 
         timestamp_end = time.time()
 
-        authority = headers.get(':authority', b'')
-        method = headers.get(':method', 'GET')
-        scheme = headers.get(':scheme', 'https')
-        path = headers.get(':path', '/')
-
-        headers.clear(":method")
-        headers.clear(":scheme")
-        headers.clear(":path")
-
-        host = None
-        port = None
-
-        if path == '*' or path.startswith("/"):
-            first_line_format = "relative"
-        elif method == 'CONNECT':
-            first_line_format = "authority"
-            if ":" in authority:
-                host, port = authority.split(":", 1)
-            else:
-                host = authority
-        else:
-            first_line_format = "absolute"
-            # FIXME: verify if path or :host contains what we need
-            scheme, host, port, _ = url.parse(path)
-            scheme = scheme.decode('ascii')
-            host = host.decode('ascii')
-
-        if host is None:
-            host = 'localhost'
-        if port is None:
-            port = 80 if scheme == 'http' else 443
-        port = int(port)
+        first_line_format, method, scheme, host, port, path = http2.parse_headers(headers)
 
         request = netlib.http.request.Request(
             first_line_format,
-            method.encode('ascii'),
-            scheme.encode('ascii'),
-            host.encode('ascii'),
+            method,
+            scheme,
+            host,
             port,
-            path.encode('ascii'),
+            path,
             b"HTTP/2.0",
             headers,
             body,
@@ -286,7 +254,7 @@ class HTTP2StateProtocol(object):
 
     def read_frame(self, hide=False):
         while True:
-            frm = framereader.http2_read_frame(self.tcp_handler.rfile)
+            frm = http2.framereader.http2_read_frame(self.tcp_handler.rfile)
             if not hide and self.dump_frames:  # pragma no cover
                 print(frm.human_readable("<<"))