Split parsing of intial line into separate protocols.

2 files changed, 108 insertions, 0 deletions

diff --git a/libmproxy/proxy.py b/libmproxy/proxy.py
index 2481ed12..2c0fa57b 100644
--- a/libmproxy/proxy.py
+++ b/libmproxy/proxy.py
@@ -127,6 +127,61 @@ def read_http_body(rfile, connection, headers, all, limit):
     return content
 
 
+def parse_http_protocol(s):
+    if not s.startswith("HTTP/"):
+        return None
+    major, minor = s.split('/')[1].split('.')
+    major = int(major)
+    minor = int(minor)
+    return major, minor
+
+
+def parse_init_connect(line):
+    try:
+        method, url, protocol = string.split(line)
+    except ValueError:
+        return None
+    if method != 'CONNECT':
+        return None
+    try:
+        host, port = url.split(":")
+    except ValueError:
+        return None
+    port = int(port)
+    mm = parse_http_protocol(protocol)
+    if not mm:
+        return None
+    return host, port, mm[0], mm[1]
+
+
+def parse_init_proxy(line):
+    try:
+        method, url, protocol = string.split(line)
+    except ValueError:
+        return None
+    parts = utils.parse_url(url)
+    if not parts:
+        return None
+    scheme, host, port, path = parts
+    mm = parse_http_protocol(protocol)
+    if not mm:
+        return None
+    return method, scheme, host, port, path, mm[0], mm[1]
+
+
+def parse_init_http(line):
+    try:
+        method, url, protocol = string.split(line)
+    except ValueError:
+        return None
+    if not (url.startswith("/") or url == "*"):
+        return None
+    mm = parse_http_protocol(protocol)
+    if not mm:
+        return None
+    return method, url, mm[0], mm[1]
+
+
 #FIXME: Return full HTTP version specification from here. Allow non-HTTP
 #protocol specs, and make it all editable.
 def parse_request_line(request):
@@ -441,6 +496,7 @@ class ProxyHandler(SocketServer.StreamRequestHandler):
                 # We should gather up everything read from the socket, and specify it all.
                 raise ProxyError(400, 'Invalid request: %s'%line)
         if "expect" in headers:
+            # FIXME: Should be forwarded upstream
             expect = ",".join(headers['expect'])
             if expect == "100-continue" and httpminor >= 1:
                 self.wfile.write('HTTP/1.1 100 Continue\r\n')
diff --git a/test/test_proxy.py b/test/test_proxy.py
index fccef977..c67e93a0 100644
--- a/test/test_proxy.py
+++ b/test/test_proxy.py
@@ -135,7 +135,59 @@ class u_read_headers(libpry.AutoTree):
         assert headers["header"] == ['one\r\n two']
 
 
+class u_parse_http_protocol(libpry.AutoTree):
+    def test_simple(self):
+        assert proxy.parse_http_protocol("HTTP/1.1") == (1, 1)
+        assert proxy.parse_http_protocol("HTTP/0.0") == (0, 0)
+        assert not proxy.parse_http_protocol("foo/0.0")
+
+
+class u_parse_init_connect(libpry.AutoTree):
+    def test_simple(self):
+        assert proxy.parse_init_connect("CONNECT host.com:443 HTTP/1.0")
+        assert not proxy.parse_init_connect("bogus")
+        assert not proxy.parse_init_connect("GET host.com:443 HTTP/1.0")
+        assert not proxy.parse_init_connect("CONNECT host.com443 HTTP/1.0")
+        assert not proxy.parse_init_connect("CONNECT host.com:443 foo/1.0")
+
+
+class u_parse_init_proxy(libpry.AutoTree):
+    def test_simple(self):
+        u = "GET http://foo.com:8888/test HTTP/1.1"
+        m, s, h, po, pa, major, minor = proxy.parse_init_proxy(u)
+        assert m == "GET"
+        assert s == "http"
+        assert h == "foo.com"
+        assert po == 8888
+        assert pa == "/test"
+        assert major == 1
+        assert minor == 1
+
+        assert not proxy.parse_init_proxy("invalid")
+        assert not proxy.parse_init_proxy("GET invalid HTTP/1.1")
+        assert not proxy.parse_init_proxy("GET http://foo.com:8888/test foo/1.1")
+
+
+class u_parse_init_http(libpry.AutoTree):
+    def test_simple(self):
+        u = "GET /test HTTP/1.1"
+        m, u, major, minor = proxy.parse_init_http(u)
+        assert m == "GET"
+        assert u == "/test"
+        assert major == 1
+        assert minor == 1
+
+        assert not proxy.parse_init_http("invalid")
+        assert not proxy.parse_init_http("GET invalid HTTP/1.1")
+        assert not proxy.parse_init_http("GET /test foo/1.1")
+
+
+
 tests = [
+    u_parse_http_protocol(),
+    u_parse_init_connect(),
+    u_parse_init_proxy(),
+    u_parse_init_http(),
     uProxyError(),
     uFileLike(),
     u_parse_request_line(),