I made the hyper command to run hyper quickly.

To use cli, install again hyper.

$ python setup.py develop

For example, it's easy to run for simple debugging.

$ hyper -v google.com
Selected NPN: h2-14
Sending frame SettingsFrame on stream 0
Received frame SettingsFrame on stream 0
Sending frame SettingsFrame on stream 0
Received frame WindowUpdateFrame on stream 0
HPACK encoding [(':method', 'GET'), (':scheme', 'https'), (':authority', 'google.com'), (':path', '/')]
...

Other options are here. They're encoding or do not show response data, and so on.

$ hyper -h
usage: hyper [-h] [-e ENCODING] [-n] [-m METHOD] [-v] host [path]

positional arguments:
  host                  set host to request
  path                  set path to resource (default: /)

optional arguments:
  -h, --help            show this help message and exit
  -e ENCODING, --encoding ENCODING
                        set charset for content-type
  -n, --nullout         do not show response data
  -m METHOD, --method METHOD
                        set http method (default: GET)
  -v, --verbose         set verbose mode (loglevel=DEBUG)

I'll implement if you want additional function or typical use. I have no idea what kind of functions is useful.

Any information on how to create a multipart body? On the code comments of hyper body is about a file-like object or bytestring

        v2_url = 'service.example.api.com'
        conn = HTTP20Connection(v2_url)
        conn.request('POST', '/endpoint', headers=headers, body='NOTHING_HERE_YET')
        resp = conn.get_response()
        return resp.read()

Im getting this error exception when accessing a external api that expects a multipart body.

{"code":"INVALID_REQUEST_EXCEPTION","description":"No multipart body found in the payload."}

Any help, links to samples or hint would be appreciated.

Besides the frame/settings renumbering, the biggest single change is the new DATA frame padding format.

Unfortunately, it's currently impossible to fully conform to the v10 spec with solely the ssl standard library module at our disposal, so merging this should probably wait until a pyOpenSSL compatibility layer can be written that exposes the necessary TLS options.

I'm sick of Python implementations insisting that headers are dictionaries. They aren't, it's a bad match. We can and should do better. This is a proposed initial implementation that's in the spirit of what Go does, which I think is a substantially better model.

Feedback is encouraged here. Note that this has been strongly influenced by shazow/urllib3#561, shazow/urllib3#562, shazow/urllib3#563, and shazow/urllib3#564.

Thanks for the comprehensive test suite; it made this port a snap. The only externally-facing change I had to make was removing the magic * argument in HTTP20Connection.__init__() - if you like, I can restore it on Python 3 with trickery a la SocketServerThread.

Barring the introduction of pyOpenSSL, supporting NPN is impossible on Python 2.7, so for now, the client assumes the server speaks HTTP/2.0 (referred to in the spec as "with prior knowledge").

An automatic migration tool like 2to3 was unneeded, and I think it's best to keep it that way.

I have problem with running Python hyper library that adds support for HTTP/2 protocol.

On my Fedora machine I installed it with pip and pip3 to use it with Python 2.7.8 and Python 3.4.1. Then I copied test script that connects with twitter:

from hyper import HTTP20Connection

conn = HTTP20Connection('twitter.com:443')
conn.request('GET', '/')
resp = conn.getresponse()

print(resp.read())

I run it using Python 2.7.8 and it ends with error:

[mn:/plib] 1 ‡ python hyper_test.py 
Traceback (most recent call last):
  File "hyper_test.py", line 4, in <module>
    conn.request('GET', '/')
  File "/usr/lib/python2.7/site-packages/hyper/http20/connection.py", line 149, in request
    self.endheaders(message_body=body, final=True, stream_id=stream_id)
  File "/usr/lib/python2.7/site-packages/hyper/http20/connection.py", line 310, in endheaders
    self.connect()
  File "/usr/lib/python2.7/site-packages/hyper/http20/connection.py", line 215, in connect
    self._recv_cb()
  File "/usr/lib/python2.7/site-packages/hyper/http20/connection.py", line 580, in _recv_cb
    self._consume_single_frame()
  File "/usr/lib/python2.7/site-packages/hyper/http20/connection.py", line 496, in _consume_single_frame
    frame, length = Frame.parse_frame_header(header)
  File "/usr/lib/python2.7/site-packages/hyper/http20/frame.py", line 52, in parse_frame_header
    frame = FRAMES[type](stream_id)
KeyError: 80

It ends with different error with Python 3.4.1:

[mn:/plib] 1 ‡ python3 hyper_test.py 
Traceback (most recent call last):
  File "hyper_test.py", line 4, in <module>
    conn.request('GET', '/')
  File "/usr/lib/python3.4/site-packages/hyper/http20/connection.py", line 149, in request
    self.endheaders(message_body=body, final=True, stream_id=stream_id)
  File "/usr/lib/python3.4/site-packages/hyper/http20/connection.py", line 310, in endheaders
    self.connect()
  File "/usr/lib/python3.4/site-packages/hyper/http20/connection.py", line 204, in connect
    sock = wrap_socket(sock, self.host)
  File "/usr/lib/python3.4/site-packages/hyper/http20/tls.py", line 44, in wrap_socket
    assert ssl_sock.selected_npn_protocol() in H2_NPN_PROTOCOLS
AssertionError

What can cause such problems on my environment?

PS I have also asked such question on StackOverflow: http://stackoverflow.com/questions/28990996/python-hyper-library-do-not-work-on-my-environment

Amazon have released a new API for Alexa but its HTTP/2 only, this is my first attempts with http/2 but there seems to be a problem with accessing the amazon API using hyper, other test servers work fine and I can also connect using curl, I get this error using either hyper on the command line or in a script. (Token redacted to XXXXX, message me if you need a token to test with)

[email protected]:~/curl-7.48.0# hyper --debug GET https://avs-alexa-na.amazon.com/v20160207/directives 'Authorization:Bearer XXXXXXXXXXXXXXXX' Url Info: {'secure': True, 'fragment': None, 'netloc': 'avs-alexa-na.amazon.com', 'host': 'avs-alexa-na.amazon.com', 'query': None, 'path': '/v20160207/directives', 'scheme': 'https', 'port': 443} Commandline Argument: Namespace(_url='https://avs-alexa-na.amazon.com/v20160207/directives', body=None, debug=True, h2=False, headers={'Authorization': 'Bearer XXXXXXXXXXXX'}, items=[<hyper.cli.KeyValue object at 0x766c45f0>], method='GET', url=<hyper.cli.UrlInfo object at 0x766c4710>) Selected protocol: None Traceback (most recent call last): File "/usr/local/bin/hyper", line 11, in sys.exit(main()) File "/usr/local/lib/python2.7/dist-packages/hyper/cli.py", line 244, in main data = request(args) File "/usr/local/lib/python2.7/dist-packages/hyper/cli.py", line 234, in request response = conn.get_response() File "/usr/local/lib/python2.7/dist-packages/hyper/common/connection.py", line 124, in get_response return self._conn.get_response(_args, *_kwargs) File "/usr/local/lib/python2.7/dist-packages/hyper/http11/connection.py", line 199, in get_response response = self.parser.parse_response(self._sock.buffer) File "/usr/local/lib/python2.7/dist-packages/hyper/http11/parser.py", line 52, in parse_response version, status, reason = temp_buffer[0:index].split(None, 2) ValueError: need more than 1 value to unpack [email protected]:~/curl-7.48.0#

Fix for #69

• Before processing a frame, check if its length is valid using the range defined by the spec.
• Allow endpoint to update the max frame size setting.

I opened this with the apns2 package maintainer since that is where I discovered the bug but I think this is really a problem inside of hyper:

https://github.com/Pr0Ger/PyAPNs2/issues/11

self._sock is None when it hits this line:

https://github.com/Lukasa/hyper/blob/development/hyper/http20/connection.py#L644

It is a rare occurance but it does happen. Any thoughts on what could be causing it?

Fix for #15

• When shutting down the connection, close all streams and send a GoAway frame to the endpoint.
• If an unexpected stream identifier is received then cancel stream with an error of type PROTOCOL_ERROR.

Could you confirm whether accessing www.google.co.uk would occur as below?

My environment is OS X Yosemite 10.10.1 (x86_64) and use MacPorts to install Python 3.4.2.

(hyper)$ python
Python 3.4.2 (default, Nov 23 2014, 23:10:23) 
[GCC 4.2.1 Compatible Apple LLVM 6.0 (clang-600.0.54)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from hyper import HTTP20Connection
>>> conn = HTTP20Connection('www.google.co.uk')
>>> conn.request('GET', '/')
1
>>> response = conn.getresponse()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".../hyper/http20/connection.py", line 171, in getresponse
    return HTTP20Response(stream.getheaders(), stream)
  File ".../hyper/http20/stream.py", line 293, in getheaders
    self._recv_cb()
  File ".../hyper/http20/connection.py", line 541, in _recv_cb
    self._consume_single_frame()
  File ".../hyper/http20/connection.py", line 483, in _consume_single_frame
    frame, length = Frame.parse_frame_header(header)
  File ".../hyper/http20/frame.py", line 52, in parse_frame_header
    frame = FRAMES[type](stream_id)
KeyError: 97

The buffering issue in BufferedSocket causes this error. In my environment self._sck.recv_into(_buffer) read data only 1135 bytes in spite of _buffer size is 65535 bytes when target length (amt) is 16384 bytes. I'm not sure whether socket.recv() would ensure to read the buffer given length.

$ man 2 recv
     ssize_t
     recv(int socket, void *buffer, size_t length, int flags);

So, I changed to read data from socket until count == 0 (it means no data) or self._bytes_in_buffer == amt.

When I use the method request in hyper.http20.connection,it's finally leading to the block of method named endheaders. When I was debugging the program, the order of packets is like the pic.Data packet is actually after WINDOW_UPDATE[1] packet. But I just let them running without debug.It was running like the pic.The first data packet was sent before the WINDOW_UPDATE[1] packet.And it wasn't recognized to be HTTP2 protocol packet by Wireshark. It's so wired!!Please help me!Thx!!

Hello

I have written, below script to test my code for HTTP/2. But with my current script., I need to send authorization with every request, i.e my connection get closed after every request. Is there any way, we can retain connetion with hyper.

Below is my code:

from hyper import HTTP20Connection

import hyper
import ssl
import time

header = {'Accept': 'application/xhtml+xml;v=2.0', 'Authorization': 'Basic RGVmYXVsdCBVc2VyOnJvYm90aWNz' }

Accept_header = {'Accept': 'application/xhtml+xml;v=2.0'}

context = ssl.SSLContext(ssl.PROTOCOL_SSLv23)
context.options |= ssl.OP_NO_SSLv2
context.options |= ssl.OP_NO_SSLv3

with HTTP20Connection('localhost:443', force_proto='h2', ssl_context=context) as conn:
    conn.request('GET', '/', headers=header)

    resp = conn.get_response()
    
    print(resp.satus)
    
    conn.request('GET', '/ctrl')

    resp = conn.get_response()
    
    print(resp.status)
    print("Sleeping...")
    time.sleep(10)

Hi,

Is there any way to send client side certificates with hyper?

This is what I am talking about: https://2.python-requests.org/en/master/user/advanced/#client-side-certificates

Documenting it for those poor souls who are still stuck with hyper under Python2.7.

Passing a Unicode object (not an "str") in Python2.7 to a hyper.tls.init_context as a cert argument raises an exception IOError: [Errno 21] Is a directory.

This happens because "basestring" is erroneously set to (str, str) here:

https://github.com/python-hyper/hyper/blob/18b629b8487169870235fe13387e2ae03c178b9f/hyper/tls.py#L123-L127

Note that because of the assignment to the basestring on line 127, an attempt to dereference the object before will always raise NameError, because the interpreter will consider basestring as a local variable. The behavior is explained in the Python FAQ, and the solution is to declare basestring as global explicitly.