Commit | Line | Data |
b2e1304d |
1 | package HTTP::Request::AsCGI; |
f1dd0792 |
2 | # ABSTRACT: Set up a CGI environment from an HTTP::Request |
b2e1304d |
3 | use strict; |
4 | use warnings; |
090cc060 |
5 | use bytes; |
b2e1304d |
6 | use base 'Class::Accessor::Fast'; |
7 | |
b332ee65 |
8 | use Carp; |
9 | use HTTP::Response; |
10 | use IO::Handle; |
11 | use IO::File; |
bb6e4a92 |
12 | use URI (); |
11454d47 |
13 | use URI::Escape (); |
b2e1304d |
14 | |
5e458489 |
15 | __PACKAGE__->mk_accessors(qw[ environment request stdin stdout stderr ]); |
16 | |
17 | # old typo |
18 | =begin Pod::Coverage |
19 | |
20 | enviroment |
21 | |
22 | =end Pod::Coverage |
23 | |
24 | =cut |
25 | |
26 | *enviroment = \&environment; |
5889d784 |
27 | |
bb6e4a92 |
28 | my %reserved = map { sprintf('%02x', ord($_)) => 1 } split //, $URI::reserved; |
29 | sub _uri_safe_unescape { |
30 | my ($s) = @_; |
8596f5a5 |
31 | $s =~ s/%([a-fA-F0-9]{2})/$reserved{lc($1)} ? "%$1" : pack('C', hex($1))/ge; |
32 | $s |
bb6e4a92 |
33 | } |
34 | |
b2e1304d |
35 | sub new { |
b332ee65 |
36 | my $class = shift; |
37 | my $request = shift; |
26e3d92b |
38 | |
b332ee65 |
39 | unless ( @_ % 2 == 0 && eval { $request->isa('HTTP::Request') } ) { |
40 | croak(qq/usage: $class->new( \$request [, key => value] )/); |
26e3d92b |
41 | } |
42 | |
b332ee65 |
43 | my $self = $class->SUPER::new( { restored => 0, setuped => 0 } ); |
44 | $self->request($request); |
45 | $self->stdin( IO::File->new_tmpfile ); |
46 | $self->stdout( IO::File->new_tmpfile ); |
b2e1304d |
47 | |
30efa07d |
48 | my $host = $request->header('Host'); |
49 | my $uri = $request->uri->clone; |
50 | $uri->scheme('http') unless $uri->scheme; |
51 | $uri->host('localhost') unless $uri->host; |
52 | $uri->port(80) unless $uri->port; |
a3875fc6 |
53 | $uri->host_port($host) unless !$host || ( $host eq $uri->host_port ); |
4e0afe7d |
54 | |
b3996e05 |
55 | # Get it before canonicalized so REQUEST_URI can be as raw as possible |
56 | my $request_uri = $uri->path_query; |
57 | |
ca38286c |
58 | $uri = $uri->canonical; |
30efa07d |
59 | |
5e458489 |
60 | my $environment = { |
b2e1304d |
61 | GATEWAY_INTERFACE => 'CGI/1.1', |
30efa07d |
62 | HTTP_HOST => $uri->host_port, |
a3875fc6 |
63 | HTTPS => ( $uri->scheme eq 'https' ) ? 'ON' : 'OFF', # not in RFC 3875 |
8596f5a5 |
64 | PATH_INFO => $uri->path, |
a3875fc6 |
65 | QUERY_STRING => $uri->query || '', |
66 | SCRIPT_NAME => '/', |
67 | SERVER_NAME => $uri->host, |
68 | SERVER_PORT => $uri->port, |
69 | SERVER_PROTOCOL => $request->protocol || 'HTTP/1.1', |
70 | SERVER_SOFTWARE => "HTTP-Request-AsCGI/$VERSION", |
71 | REMOTE_ADDR => '127.0.0.1', |
72 | REMOTE_HOST => 'localhost', |
73 | REMOTE_PORT => int( rand(64000) + 1000 ), # not in RFC 3875 |
b3996e05 |
74 | REQUEST_URI => $request_uri, # not in RFC 3875 |
b332ee65 |
75 | REQUEST_METHOD => $request->method, |
76 | @_ |
77 | }; |
26e3d92b |
78 | |
4d6b96d8 |
79 | # RFC 3875 says PATH_INFO is not URI-encoded. That's really |
80 | # annoying for applications that you can't tell "%2F" vs "/", but |
81 | # doing the partial decoding then makes it impossible to tell |
82 | # "%252F" vs "%2F". Encoding everything is more compatible to what |
83 | # web servers like Apache or lighttpd do, anyways. |
84 | $environment->{PATH_INFO} = URI::Escape::uri_unescape($environment->{PATH_INFO}); |
8596f5a5 |
85 | |
9c216915 |
86 | foreach my $field ( $request->headers->header_field_names ) { |
b2e1304d |
87 | |
ca38286c |
88 | my $key = uc("HTTP_$field"); |
2aaf55bc |
89 | $key =~ tr/-/_/; |
ca38286c |
90 | $key =~ s/^HTTP_// if $field =~ /^Content-(Length|Type)$/; |
b2e1304d |
91 | |
5e458489 |
92 | unless ( exists $environment->{$key} ) { |
93 | $environment->{$key} = $request->headers->header($field); |
b2e1304d |
94 | } |
95 | } |
96 | |
5e458489 |
97 | unless ( $environment->{SCRIPT_NAME} eq '/' && $environment->{PATH_INFO} ) { |
98 | $environment->{PATH_INFO} =~ s/^\Q$environment->{SCRIPT_NAME}\E/\//; |
99 | $environment->{PATH_INFO} =~ s/^\/+/\//; |
26e3d92b |
100 | } |
101 | |
5e458489 |
102 | $self->environment($environment); |
b332ee65 |
103 | |
104 | return $self; |
26e3d92b |
105 | } |
106 | |
107 | sub setup { |
108 | my $self = shift; |
109 | |
5e458489 |
110 | $self->{restore}->{environment} = {%ENV}; |
26e3d92b |
111 | |
b332ee65 |
112 | binmode( $self->stdin ); |
26e3d92b |
113 | |
b332ee65 |
114 | if ( $self->request->content_length ) { |
26e3d92b |
115 | |
bd90442e |
116 | $self->stdin->print($self->request->content) |
b332ee65 |
117 | or croak("Can't write request content to stdin handle: $!"); |
b2e1304d |
118 | |
bd90442e |
119 | $self->stdin->seek(0, SEEK_SET) |
b332ee65 |
120 | or croak("Can't seek stdin handle: $!"); |
bd90442e |
121 | |
122 | $self->stdin->flush |
123 | or croak("Can't flush stdin handle: $!"); |
b2e1304d |
124 | } |
26e3d92b |
125 | |
b2c02cd0 |
126 | open( $self->{restore}->{stdin}, '<&'. STDIN->fileno ) |
b332ee65 |
127 | or croak("Can't dup stdin: $!"); |
b2e1304d |
128 | |
b2c02cd0 |
129 | open( STDIN, '<&='. $self->stdin->fileno ) |
b332ee65 |
130 | or croak("Can't open stdin: $!"); |
ca38286c |
131 | |
b332ee65 |
132 | binmode( STDIN ); |
26e3d92b |
133 | |
b332ee65 |
134 | if ( $self->stdout ) { |
26e3d92b |
135 | |
b2c02cd0 |
136 | open( $self->{restore}->{stdout}, '>&'. STDOUT->fileno ) |
b332ee65 |
137 | or croak("Can't dup stdout: $!"); |
26e3d92b |
138 | |
b2c02cd0 |
139 | open( STDOUT, '>&='. $self->stdout->fileno ) |
b332ee65 |
140 | or croak("Can't open stdout: $!"); |
76391122 |
141 | |
a3875fc6 |
142 | binmode( $self->stdout ); |
143 | binmode( STDOUT); |
144 | } |
30efa07d |
145 | |
b332ee65 |
146 | if ( $self->stderr ) { |
090cc060 |
147 | |
b2c02cd0 |
148 | open( $self->{restore}->{stderr}, '>&'. STDERR->fileno ) |
b332ee65 |
149 | or croak("Can't dup stderr: $!"); |
441eeb04 |
150 | |
b2c02cd0 |
151 | open( STDERR, '>&='. $self->stderr->fileno ) |
b332ee65 |
152 | or croak("Can't open stderr: $!"); |
26e3d92b |
153 | |
154 | binmode( $self->stderr ); |
155 | binmode( STDERR ); |
156 | } |
b2e1304d |
157 | |
9c216915 |
158 | { |
159 | no warnings 'uninitialized'; |
903274b1 |
160 | %ENV = (%ENV, %{ $self->environment }); |
5889d784 |
161 | } |
162 | |
9c216915 |
163 | if ( $INC{'CGI.pm'} ) { |
164 | CGI::initialize_globals(); |
165 | } |
166 | |
b332ee65 |
167 | $self->{setuped}++; |
9c216915 |
168 | |
169 | return $self; |
5889d784 |
170 | } |
171 | |
780060e5 |
172 | sub response { |
b332ee65 |
173 | my ( $self, $callback ) = @_; |
780060e5 |
174 | |
b332ee65 |
175 | return undef unless $self->stdout; |
780060e5 |
176 | |
b332ee65 |
177 | seek( $self->stdout, 0, SEEK_SET ) |
178 | or croak("Can't seek stdout handle: $!"); |
14f243e8 |
179 | |
b332ee65 |
180 | my $headers; |
181 | while ( my $line = $self->stdout->getline ) { |
182 | $headers .= $line; |
183 | last if $headers =~ /\x0d?\x0a\x0d?\x0a$/; |
780060e5 |
184 | } |
15e50838 |
185 | |
b332ee65 |
186 | unless ( defined $headers ) { |
187 | $headers = "HTTP/1.1 500 Internal Server Error\x0d\x0a"; |
780060e5 |
188 | } |
189 | |
b332ee65 |
190 | unless ( $headers =~ /^HTTP/ ) { |
191 | $headers = "HTTP/1.1 200 OK\x0d\x0a" . $headers; |
4e0afe7d |
192 | } |
decf17dc |
193 | |
b332ee65 |
194 | my $response = HTTP::Response->parse($headers); |
195 | $response->date( time() ) unless $response->date; |
decf17dc |
196 | |
b332ee65 |
197 | my $message = $response->message; |
198 | my $status = $response->header('Status'); |
780060e5 |
199 | |
b332ee65 |
200 | if ( $message && $message =~ /^(.+)\x0d$/ ) { |
201 | $response->message($1); |
780060e5 |
202 | } |
26e3d92b |
203 | |
b332ee65 |
204 | if ( $status && $status =~ /^(\d\d\d)\s?(.+)?$/ ) { |
780060e5 |
205 | |
b332ee65 |
206 | my $code = $1; |
207 | my $message = $2 || HTTP::Status::status_message($code); |
4e0afe7d |
208 | |
b332ee65 |
209 | $response->code($code); |
210 | $response->message($message); |
4e0afe7d |
211 | } |
15e50838 |
212 | |
b332ee65 |
213 | my $length = ( stat( $self->stdout ) )[7] - tell( $self->stdout ); |
780060e5 |
214 | |
b332ee65 |
215 | if ( $response->code == 500 && !$length ) { |
4e0afe7d |
216 | |
b332ee65 |
217 | $response->content( $response->error_as_HTML ); |
218 | $response->content_type('text/html'); |
4e0afe7d |
219 | |
4d6e304a |
220 | return $response; |
780060e5 |
221 | } |
4e0afe7d |
222 | |
b332ee65 |
223 | if ($callback) { |
4d6e304a |
224 | |
b332ee65 |
225 | my $handle = $self->stdout; |
4d6e304a |
226 | |
b332ee65 |
227 | $response->content( sub { |
4e0afe7d |
228 | |
b332ee65 |
229 | if ( $handle->read( my $buffer, 4096 ) ) { |
230 | return $buffer; |
474703e4 |
231 | } |
b332ee65 |
232 | |
233 | return undef; |
234 | }); |
4d6e304a |
235 | } |
b332ee65 |
236 | else { |
4d6e304a |
237 | |
b332ee65 |
238 | my $length = 0; |
4d6e304a |
239 | |
b332ee65 |
240 | while ( $self->stdout->read( my $buffer, 4096 ) ) { |
241 | $length += length($buffer); |
242 | $response->add_content($buffer); |
243 | } |
4e0afe7d |
244 | |
b332ee65 |
245 | if ( $length && !$response->content_length ) { |
246 | $response->content_length($length); |
decf17dc |
247 | } |
780060e5 |
248 | } |
249 | |
780060e5 |
250 | return $response; |
251 | } |
252 | |
b2e1304d |
253 | sub restore { |
254 | my $self = shift; |
4e0afe7d |
255 | |
9c216915 |
256 | { |
257 | no warnings 'uninitialized'; |
5e458489 |
258 | %ENV = %{ $self->{restore}->{environment} }; |
9c216915 |
259 | } |
12852959 |
260 | |
b2c02cd0 |
261 | open( STDIN, '<&'. fileno($self->{restore}->{stdin}) ) |
b332ee65 |
262 | or croak("Can't restore stdin: $!"); |
30efa07d |
263 | |
b332ee65 |
264 | sysseek( $self->stdin, 0, SEEK_SET ) |
265 | or croak("Can't seek stdin: $!"); |
30efa07d |
266 | |
b332ee65 |
267 | if ( $self->{restore}->{stdout} ) { |
26e3d92b |
268 | |
b332ee65 |
269 | STDOUT->flush |
270 | or croak("Can't flush stdout: $!"); |
26e3d92b |
271 | |
b2c02cd0 |
272 | open( STDOUT, '>&'. fileno($self->{restore}->{stdout}) ) |
b332ee65 |
273 | or croak("Can't restore stdout: $!"); |
090cc060 |
274 | |
b332ee65 |
275 | sysseek( $self->stdout, 0, SEEK_SET ) |
276 | or croak("Can't seek stdout: $!"); |
a3875fc6 |
277 | } |
26e3d92b |
278 | |
b332ee65 |
279 | if ( $self->{restore}->{stderr} ) { |
12852959 |
280 | |
b332ee65 |
281 | STDERR->flush |
282 | or croak("Can't flush stderr: $!"); |
30efa07d |
283 | |
b2c02cd0 |
284 | open( STDERR, '>&'. fileno($self->{restore}->{stderr}) ) |
b332ee65 |
285 | or croak("Can't restore stderr: $!"); |
090cc060 |
286 | |
b332ee65 |
287 | sysseek( $self->stderr, 0, SEEK_SET ) |
288 | or croak("Can't seek stderr: $!"); |
6f5fb9a7 |
289 | } |
090cc060 |
290 | |
b332ee65 |
291 | $self->{restored}++; |
9c216915 |
292 | |
293 | return $self; |
b2e1304d |
294 | } |
295 | |
296 | sub DESTROY { |
297 | my $self = shift; |
b332ee65 |
298 | $self->restore if $self->{setuped} && !$self->{restored}; |
b2e1304d |
299 | } |
300 | |
301 | 1; |
302 | |
303 | __END__ |
304 | |
b2e1304d |
305 | =head1 SYNOPSIS |
306 | |
bd7813ac |
307 | use CGI; |
308 | use HTTP::Request; |
309 | use HTTP::Request::AsCGI; |
15e50838 |
310 | |
bd7813ac |
311 | my $request = HTTP::Request->new( GET => 'http://www.host.com/' ); |
312 | my $stdout; |
15e50838 |
313 | |
bd7813ac |
314 | { |
315 | my $c = HTTP::Request::AsCGI->new($request)->setup; |
316 | my $q = CGI->new; |
15e50838 |
317 | |
bd7813ac |
318 | print $q->header, |
319 | $q->start_html('Hello World'), |
320 | $q->h1('Hello World'), |
321 | $q->end_html; |
15e50838 |
322 | |
bd7813ac |
323 | $stdout = $c->stdout; |
15e50838 |
324 | |
5e458489 |
325 | # environment and descriptors will automatically be restored |
2d51e42f |
326 | # when $c is destructed. |
bd7813ac |
327 | } |
15e50838 |
328 | |
bd7813ac |
329 | while ( my $line = $stdout->getline ) { |
330 | print $line; |
331 | } |
15e50838 |
332 | |
b2e1304d |
333 | =head1 DESCRIPTION |
334 | |
9a0e0b26 |
335 | Provides a convenient way of setting up an CGI environment from an HTTP::Request. |
2d51e42f |
336 | |
b2e1304d |
337 | =head1 METHODS |
338 | |
15e50838 |
339 | =over 4 |
14f243e8 |
340 | |
b332ee65 |
341 | =item new ( $request [, key => value ] ) |
14f243e8 |
342 | |
9a0e0b26 |
343 | Constructor. The first argument must be a instance of HTTP::Request, followed |
344 | by optional pairs of environment key and value. |
14f243e8 |
345 | |
5e458489 |
346 | =item environment |
14f243e8 |
347 | |
15e50838 |
348 | Returns a hashref containing the environment that will be used in setup. |
b332ee65 |
349 | Changing the hashref after setup has been called will have no effect. |
14f243e8 |
350 | |
b332ee65 |
351 | =item setup |
9c216915 |
352 | |
9a0e0b26 |
353 | Sets up the environment and descriptors. |
9c216915 |
354 | |
b332ee65 |
355 | =item restore |
9c216915 |
356 | |
5e458489 |
357 | Restores the environment and descriptors. Can only be called after setup. |
9c216915 |
358 | |
b332ee65 |
359 | =item request |
9c216915 |
360 | |
b332ee65 |
361 | Returns the request given to constructor. |
9c216915 |
362 | |
b332ee65 |
363 | =item response |
9c216915 |
364 | |
b332ee65 |
365 | Returns a HTTP::Response. Can only be called after restore. |
9c216915 |
366 | |
b332ee65 |
367 | =item stdin |
9c216915 |
368 | |
b332ee65 |
369 | Accessor for handle that will be used for STDIN, must be a real seekable |
370 | handle with an file descriptor. Defaults to a tempoary IO::File instance. |
14f243e8 |
371 | |
b332ee65 |
372 | =item stdout |
14f243e8 |
373 | |
b332ee65 |
374 | Accessor for handle that will be used for STDOUT, must be a real seekable |
375 | handle with an file descriptor. Defaults to a tempoary IO::File instance. |
9c216915 |
376 | |
b332ee65 |
377 | =item stderr |
14f243e8 |
378 | |
b332ee65 |
379 | Accessor for handle that will be used for STDERR, must be a real seekable |
380 | handle with an file descriptor. |
14f243e8 |
381 | |
382 | =back |
b2e1304d |
383 | |
74fbb9dd |
384 | =head1 SEE ALSO |
385 | |
386 | =over 4 |
387 | |
388 | =item examples directory in this distribution. |
389 | |
390 | =item L<WWW::Mechanize::CGI> |
391 | |
392 | =item L<Test::WWW::Mechanize::CGI> |
393 | |
394 | =back |
395 | |
2d51e42f |
396 | =head1 THANKS TO |
17b370b0 |
397 | |
398 | Thomas L. Shinnick for his valuable win32 testing. |
399 | |
b2e1304d |
400 | =cut |