Merge pull request #32 from njsmith/missing-reason-phrase

Tolerate missing reason phrases

Author: njsmith

h11/_abnf.py

@@ -93,8 +93,12 @@
     r"{http_version}"
     r" "
     r"(?P<status_code>{status_code})"
-    r" "
-    r"(?P<reason>{reason_phrase})"
+    # However, there are apparently a few too many servers out there that just
+    # leave out the reason phrase:
+    #   https://github.com/scrapy/scrapy/issues/345#issuecomment-281756036
+    #   https://github.com/seanmonstar/httparse/issues/29
+    # so make it optional. ?: is a non-capturing group.
+    r"(?: (?P<reason>{reason_phrase}))?"
     .format(**globals()))
 
 HEXDIG = r"[0-9A-Fa-f]"

h11/_readers.py

@@ -74,6 +74,9 @@ def maybe_read_from_SEND_RESPONSE_server(buf):
     if not lines:
         raise LocalProtocolError("no response line received")
     matches = validate(status_line_re, lines[0])
+    # Tolerate missing reason phrases
+    if matches["reason"] is None:
+        matches["reason"] = b""
     status_code = matches["status_code"] = int(matches["status_code"])
     class_ = InformationalResponse if status_code < 200 else Response
     return class_(headers=list(_decode_header_lines(lines[1:])), **matches)

h11/tests/test_io.py

@@ -142,6 +142,13 @@ def test_readers_unusual():
        Response(status_code=200, headers=[("Foo", "")],
                 http_version="1.0", reason=b"OK"))
 
+    # Tolerate broken servers that leave off the response code
+    tr(READERS[SERVER, SEND_RESPONSE],
+       b"HTTP/1.0 200\r\n"
+       b"Foo: bar\r\n\r\n",
+       Response(status_code=200, headers=[("Foo", "bar")],
+                http_version="1.0", reason=b""))
+
     # obsolete line folding
     tr(READERS[CLIENT, IDLE],
        b"HEAD /foo HTTP/1.1\r\n"