https://github.com/kenichi/h2
an http/2 client & server based on http-2 and modern ruby
https://github.com/kenichi/h2
http2 http2-push ruby
Last synced: 11 months ago
JSON representation
an http/2 client & server based on http-2 and modern ruby
- Host: GitHub
- URL: https://github.com/kenichi/h2
- Owner: kenichi
- License: mit
- Created: 2016-11-27T02:54:03.000Z (over 9 years ago)
- Default Branch: master
- Last Pushed: 2018-08-22T22:50:44.000Z (over 7 years ago)
- Last Synced: 2025-04-15T03:20:26.385Z (12 months ago)
- Topics: http2, http2-push, ruby
- Language: Ruby
- Homepage:
- Size: 238 KB
- Stars: 25
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE.txt
- Code of conduct: CODE_OF_CONDUCT.md
Awesome Lists containing this project
README
# H2
[](https://travis-ci.org/kenichi/h2)
H2 is an HTTP/2 client and server based on the [http-2](https://github.com/igrigorik/http-2) gem.
H2 uses:
* keyword arguments (>=2.0)
* exception-less socket IO (>=2.3).
## Server Usage
Server API is currently optional, so `h2/server` must be required separately.
The server uses [Celluloid::IO](https://github.com/celluloid/celluloid-io), but
h2's gemspec does not require it, so celluloid-io must be separately added to
`Gemfile`. It is currently based on `celluloid-io-0.17.3`.
```ruby
require 'h2/server'
server = H2::Server::HTTP.new host: addr, port: port do |connection|
connection.each_stream do |stream|
stream.respond status: 200, body: "hello, world!\n"
stream.connection.goaway
end
end
stream = H2.get url: "http://#{addr}:#{port}", tls: false
stream.body #=> "hello, world!\n"
```
See more server examples:
* [HTTPS Hello World](https://github.com/kenichi/h2/blob/master/examples/server/https_hello_world.rb)
* [Push Promises](https://github.com/kenichi/h2/blob/master/examples/server/push_promise.rb)
* [SSE/EventSource](https://github.com/kenichi/h2/blob/master/examples/server/sse.rb)
## Client Usage
```ruby
require 'h2'
#
# --- one-shot convenience
#
stream = H2.get url: 'https://example.com'
stream.ok? #=> true
stream.headers #=> Hash
stream.body #=> String
stream.closed? #=> true
client = stream.client #=> H2::Client
client.closed? #=> true
#
# --- normal connection
#
client = H2::Client.new host: 'example.com', port: 443
stream = client.get path: '/'
stream.ok? #=> true
stream.headers #=> Hash, method blocks until stream is closed
stream.body #=> String, method blocks until stream is closed
stream.closed? #=> true
client.closed? #=> false unless server sent GOAWAY
client.on :promise do |p| # check/cancel a promise
p.on :headers do |h|
if h['etag'] == 'some_value'
p.cancel! # already have
end
end
end
stream.block! # blocks until this stream and any associated push streams are closed
stream.ok? #=> true
stream.headers #=> Hash
stream.body #=> String
stream.closed? #=> true
stream.pushes #=> Set
stream.pushes.each do |pp|
pp.parent == stream #=> true
pp.headers #=> Hash
pp.body #=> String
end
client.goaway!
```
## Client CLI
For more info on using the CLI `h2` installed with this gem:
`$ h2 --help`
## Using TLS CA Certificates with the Client
If you're running on macOS and using Homebrew's openssl package, you may need to
specify the CA file in the TLS options:
```ruby
client = H2::Client.new host: 'example.com', port: 443, tls: { ca_file: '/usr/local/etc/openssl/cert.pem' }
```
or when using the CLI:
`$ h2 --cafile /usr/local/etc/openssl/cert.pem https://example.com/`
## Alternate Concurrency Models
Right now, h2 uses one new thread per connection. This is hardly ideal, so a
couple other models are tentatively supported out of the box:
* [celluloid](https://github.com/celluloid/celluloid)
* [concurrent-ruby](https://github.com/ruby-concurrency/concurrent-ruby)
Neither of these gems are hard dependencies. If you want to use either one, you must
have it available to your Ruby VM, most likely via Bundler, *and* require the
sub-component of h2 that will prepend and extend `H2::Client`. They are also intended
to be mutually exclusive: you can have both in your VM, but you can only use one at a
time with h2's client.
#### Celluloid Pool
To use a celluloid actor pool for reading from `H2::Client` connections:
```ruby
require 'h2/client/celluloid'
```
This will lazily fire up a celluloid pool, with defaults defined by Celluloid.
NOTE: if you've added celluloid-io and required the 'h2/server' API, Celluloid
will be loaded in your Ruby VM already; however, you must still require this to
have the client use Celluloid actor pools.
#### Concurrent-Ruby ThreadPoolExecutor
To use a concurrent-ruby thread pool executor for reading from `H2::Client` connections:
```ruby
require 'h2/client/concurrent'
```
This will lazily fire up a `Concurrent::ThreadPoolExecutor` with the following settings:
```ruby
procs = ::Concurrent.processor_count
min_threads: 0,
max_threads: procs,
max_queue: procs * 5
```
## TODO
* [x] HTTPS / TLS
* [x] push promise cancellation
* [x] alternate concurrency models
* [ ] fix up CLI to be more curlish
* [ ] update server API
## Contributing
Bug reports and pull requests are welcome on GitHub at https://github.com/kenichi/h2. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the [Contributor Covenant](http://contributor-covenant.org) code of conduct.
## License
The gem is available as open source under the terms of the [MIT License](http://opensource.org/licenses/MIT).