1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
|
? my $ctx = $main::context;
? $_mt->wrapper_file("wrapper.mt", "Configure", "Using Mruby")->(sub {
<p>
<a href="https://github.com/mruby/mruby">mruby</a> is a lightweight implementation of the Ruby programming language.
With H2O, users can implement their own request handling logic using mruby, either to generate responses or to fix-up the request / response.
</p>
<h3 id="programming-interface">Rack-based Programming Interface</h3>
<p>
The interface between the mruby program and the H2O server is based on <a href="http://www.rubydoc.info/github/rack/rack/master/file/SPEC">Rack interface specification</a>.
Below is a simple configuration that returns <i>hello world</i>.
</p>
<?= $ctx->{example}->('Hello-world in mruby', <<'EOT')
paths:
"/":
mruby.handler: |
Proc.new do |env|
[200, {'content-type' => 'text/plain'}, ["Hello world\n"]]
end
EOT
?>
<p>
It should be noted that as of H2O version 1.7.0, there are limitations when compared to ordinary web application server with support for Rack such as Unicorn:
<ul>
<li>no libraries provided as part of Rack is available (only the interface is compatible)
</ul>
</p>
<p>
In addition to the Rack interface specification, H2O recognizes status code <code>399</code> which can be used to delegate request to the next handler.
The feature can be used to implement access control and response header modifiers.
</p>
<h3 id="access-control">Access Control</h3>
<p>
By using the <code>399</code> status code, it is possible to implement access control using mruby.
The example below restricts access to requests from <code>192.168.</code> private address.
</p>
<?= $ctx->{example}->('Restricting access to 192.168.', <<'EOT')
paths:
"/":
mruby.handler: |
lambda do |env|
if /\A192\.168\./.match(env["REMOTE_ADDR"])
return [399, {}, []]
end
[403, {'content-type' => 'text/plain'}, ["access forbidden\n"]]
end
EOT
?>
<p>
Support for <a href="configure/basic_auth.html">Basic Authentication</a> is also provided by an mruby script.
</p>
<h3 id="delegating-request">Delegating the Request</h3>
<p>
When enabled using the <a href="configure/reproxy_directives.html#reproxy"><code>reproxy</code></a> directive, it is possible to delegate the request from the mruby handler to any other handler.
</p>
<p>
<?= $ctx->{example}->('Rewriting URL with delegation', <<'EOT')
paths:
"/":
mruby.handler: |
lambda do |env|
if /\/user\/([^\/]+)/.match(env["PATH_INFO"])
return [307, {"x-reproxy-url" => "/user.php?user=#{$1}"}, []]
end
return [399, {}, []]
end
EOT
?>
<h3 id="modifying-response">Modifying the Response</h3>
<p>
When the mruby handler returns status code <code>399</code>, H2O delegates the request to the next handler while preserving the headers emitted by the handler.
The feature can be used to add extra headers to the response.
</p>
<p>
For example, the following example sets <code>cache-control</code> header for requests against <code>.css</code> and <code>.js</code> files.
</p>
<?= $ctx->{example}->('Setting cache-control header for certain types of files', <<'EOT')
paths:
"/":
mruby.handler: |
Proc.new do |env|
headers = {}
if /\.(css|js)\z/.match(env["PATH_INFO"])
headers["cache-control"] = "max-age=86400"
end
[399, headers, []]
end
file.dir: /path/to/doc-root
EOT
?>
<p>
Or in the example below, the handler triggers <a href="configure/http2_directives.html#server-push">HTTP/2 server push</a> with the use of <code>Link: rel=preload</code> headers, and then requests a FastCGI application to process the request.
</p>
<?= $ctx->{example}->('Pushing asset files', <<'EOT')
paths:
"/":
mruby.handler: |
Proc.new do |env|
push_paths = []
# push css and js when request is to dir root or HTML
if /(\/|\.html)\z/.match(env["PATH_INFO"])
push_paths << ["/css/style.css", "style"]
push_paths << ["/js/app.js", "script"]
end
[399, push_paths.empty? ? {} : {"link" => push_paths.map{|p| "<#{p[0]}>; rel=preload; as=#{p[1]}"}.join("\n")}, []]
end
fastcgi.connect: ...
EOT
?>
<h3 id="http-client">Using the HTTP Client</h3>
<p>
Starting from version 1.7, a HTTP client API is provided.
HTTP requests issued through the API will be handled asynchronously; the client does not block the event loop of the HTTP server.
</p>
<?= $ctx->{example}->('Mruby handler returning the response of http://example.com', <<'EOT')
paths:
"/":
mruby.handler: |
Proc.new do |env|
req = http_request("http://example.com")
status, headers, body = req.join
[status, headers, body]
end
EOT
?>
<p>
<code>http_request</code> is the method that issues a HTTP request.
</p>
<p>
The method takes two arguments.
First argument is the target URI.
Second argument is an optional hash; <code>method</code> (defaults to <code>GET</code>), <code>header</code>, <code>body</code> attributes are recognized.
</p>
<p>
The method returns a promise object.
When <code>#join</code> method of the promise is invoked, a three-argument array containing the status code, response headers, and the body is returned.
The response body is also a promise.
Applications can choose from three ways when dealing with the body: a) call <code>#each</code> method to receive the contents, b) call <code>#join</code> to retrieve the body as a string, c) return the object as the response body of the mruby handler.
</p>
<p>
The header and the body object passed to <code>http_request</code> should conform to the requirements laid out by the Rack specification for request header and request body.
The response header and the response body object returned by the <code>#join</code> method of the promise returned by <code>http_request</code> conforms to the requirements of the Rack specification.
</p>
<p>
Since the API provides an asynchronous HTTP client, it is possible to effectively issue multiple HTTP requests concurrently and merge them into a single response.
</p>
<p>
When HTTPS is used, servers are verified using the properties of <a href="configure/proxy_directives.html#proxy.ssl.cafile"><code>proxy.ssl.cafile</code></a> and <a href="configure/proxy_directives.html#proxy.ssl.verify-peer"><code>proxy.ssl.verify-peer</code></a> specified at the global level.
</p>
<h3 id="logging-arbitrary-variable">Logging Arbitrary Variable</h3>
<p>
In version 2.3, it is possible from mruby to set and log an arbitrary-named variable that is associated to a HTTP request.
A HTTP response header that starts with <code>x-fallthru-set-</code> is handled specially by the H2O server. Instead of sending the header downstream, the server accepts the value as a request environment variable, taking the suffix of the header name as the name of the variable.
</p>
<p>
This example shows how to read request data, parse json and then log data from mruby.
</p>
<?= $ctx->{example}->('Logging the content of a POST request via request environment variable', <<'EOT')
paths:
"/":
mruby.handler: |
Proc.new do |env|
input = env["rack.input"] ? env["rack.input"].read : '{"default": "true"}'
parsed_json = JSON.parse(input)
parsed_json["time"] = Time.now.to_i
logdata = parsed_json.to_s
[204, {"x-fallthru-set-POSTDATA" => logdata}, []]
end
access-log:
path: /path/to/access-log.json
escape: json
format: '{"POST": %{POSTDATA}e}'
EOT
?>
? })
|
