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 |
cfe19df0 |
18 | |
5e458489 |
19 | =begin Pod::Coverage |
20 | |
21 | enviroment |
22 | |
23 | =end Pod::Coverage |
24 | |
25 | =cut |
26 | |
27 | *enviroment = \&environment; |
5889d784 |
28 | |
bb6e4a92 |
29 | my %reserved = map { sprintf('%02x', ord($_)) => 1 } split //, $URI::reserved; |
30 | sub _uri_safe_unescape { |
31 | my ($s) = @_; |
8596f5a5 |
32 | $s =~ s/%([a-fA-F0-9]{2})/$reserved{lc($1)} ? "%$1" : pack('C', hex($1))/ge; |
33 | $s |
bb6e4a92 |
34 | } |
35 | |
b2e1304d |
36 | sub new { |
b332ee65 |
37 | my $class = shift; |
38 | my $request = shift; |
26e3d92b |
39 | |
b332ee65 |
40 | unless ( @_ % 2 == 0 && eval { $request->isa('HTTP::Request') } ) { |
41 | croak(qq/usage: $class->new( \$request [, key => value] )/); |
26e3d92b |
42 | } |
43 | |
b332ee65 |
44 | my $self = $class->SUPER::new( { restored => 0, setuped => 0 } ); |
45 | $self->request($request); |
46 | $self->stdin( IO::File->new_tmpfile ); |
47 | $self->stdout( IO::File->new_tmpfile ); |
b2e1304d |
48 | |
30efa07d |
49 | my $host = $request->header('Host'); |
50 | my $uri = $request->uri->clone; |
51 | $uri->scheme('http') unless $uri->scheme; |
52 | $uri->host('localhost') unless $uri->host; |
53 | $uri->port(80) unless $uri->port; |
a3875fc6 |
54 | $uri->host_port($host) unless !$host || ( $host eq $uri->host_port ); |
4e0afe7d |
55 | |
b3996e05 |
56 | # Get it before canonicalized so REQUEST_URI can be as raw as possible |
57 | my $request_uri = $uri->path_query; |
58 | |
ca38286c |
59 | $uri = $uri->canonical; |
30efa07d |
60 | |
5e458489 |
61 | my $environment = { |
b2e1304d |
62 | GATEWAY_INTERFACE => 'CGI/1.1', |
30efa07d |
63 | HTTP_HOST => $uri->host_port, |
a3875fc6 |
64 | HTTPS => ( $uri->scheme eq 'https' ) ? 'ON' : 'OFF', # not in RFC 3875 |
8596f5a5 |
65 | PATH_INFO => $uri->path, |
a3875fc6 |
66 | QUERY_STRING => $uri->query || '', |
67 | SCRIPT_NAME => '/', |
68 | SERVER_NAME => $uri->host, |
69 | SERVER_PORT => $uri->port, |
70 | SERVER_PROTOCOL => $request->protocol || 'HTTP/1.1', |
cfe19df0 |
71 | SERVER_SOFTWARE => 'HTTP-Request-AsCGI/' . our $VERSION, |
a3875fc6 |
72 | REMOTE_ADDR => '127.0.0.1', |
73 | REMOTE_HOST => 'localhost', |
74 | REMOTE_PORT => int( rand(64000) + 1000 ), # not in RFC 3875 |
b3996e05 |
75 | REQUEST_URI => $request_uri, # not in RFC 3875 |
b332ee65 |
76 | REQUEST_METHOD => $request->method, |
77 | @_ |
78 | }; |
26e3d92b |
79 | |
4d6b96d8 |
80 | # RFC 3875 says PATH_INFO is not URI-encoded. That's really |
81 | # annoying for applications that you can't tell "%2F" vs "/", but |
82 | # doing the partial decoding then makes it impossible to tell |
83 | # "%252F" vs "%2F". Encoding everything is more compatible to what |
84 | # web servers like Apache or lighttpd do, anyways. |
85 | $environment->{PATH_INFO} = URI::Escape::uri_unescape($environment->{PATH_INFO}); |
8596f5a5 |
86 | |
9c216915 |
87 | foreach my $field ( $request->headers->header_field_names ) { |
b2e1304d |
88 | |
ca38286c |
89 | my $key = uc("HTTP_$field"); |
2aaf55bc |
90 | $key =~ tr/-/_/; |
ca38286c |
91 | $key =~ s/^HTTP_// if $field =~ /^Content-(Length|Type)$/; |
b2e1304d |
92 | |
5e458489 |
93 | unless ( exists $environment->{$key} ) { |
94 | $environment->{$key} = $request->headers->header($field); |
b2e1304d |
95 | } |
96 | } |
97 | |
5e458489 |
98 | unless ( $environment->{SCRIPT_NAME} eq '/' && $environment->{PATH_INFO} ) { |
99 | $environment->{PATH_INFO} =~ s/^\Q$environment->{SCRIPT_NAME}\E/\//; |
100 | $environment->{PATH_INFO} =~ s/^\/+/\//; |
26e3d92b |
101 | } |
102 | |
5e458489 |
103 | $self->environment($environment); |
b332ee65 |
104 | |
105 | return $self; |
26e3d92b |
106 | } |
107 | |
108 | sub setup { |
109 | my $self = shift; |
110 | |
5e458489 |
111 | $self->{restore}->{environment} = {%ENV}; |
26e3d92b |
112 | |
b332ee65 |
113 | binmode( $self->stdin ); |
26e3d92b |
114 | |
b332ee65 |
115 | if ( $self->request->content_length ) { |
26e3d92b |
116 | |
bd90442e |
117 | $self->stdin->print($self->request->content) |
b332ee65 |
118 | or croak("Can't write request content to stdin handle: $!"); |
b2e1304d |
119 | |
bd90442e |
120 | $self->stdin->seek(0, SEEK_SET) |
b332ee65 |
121 | or croak("Can't seek stdin handle: $!"); |
bd90442e |
122 | |
123 | $self->stdin->flush |
124 | or croak("Can't flush stdin handle: $!"); |
b2e1304d |
125 | } |
26e3d92b |
126 | |
b2c02cd0 |
127 | open( $self->{restore}->{stdin}, '<&'. STDIN->fileno ) |
b332ee65 |
128 | or croak("Can't dup stdin: $!"); |
b2e1304d |
129 | |
b2c02cd0 |
130 | open( STDIN, '<&='. $self->stdin->fileno ) |
b332ee65 |
131 | or croak("Can't open stdin: $!"); |
ca38286c |
132 | |
b332ee65 |
133 | binmode( STDIN ); |
26e3d92b |
134 | |
b332ee65 |
135 | if ( $self->stdout ) { |
26e3d92b |
136 | |
b2c02cd0 |
137 | open( $self->{restore}->{stdout}, '>&'. STDOUT->fileno ) |
b332ee65 |
138 | or croak("Can't dup stdout: $!"); |
26e3d92b |
139 | |
b2c02cd0 |
140 | open( STDOUT, '>&='. $self->stdout->fileno ) |
b332ee65 |
141 | or croak("Can't open stdout: $!"); |
76391122 |
142 | |
a3875fc6 |
143 | binmode( $self->stdout ); |
144 | binmode( STDOUT); |
145 | } |
30efa07d |
146 | |
b332ee65 |
147 | if ( $self->stderr ) { |
090cc060 |
148 | |
b2c02cd0 |
149 | open( $self->{restore}->{stderr}, '>&'. STDERR->fileno ) |
b332ee65 |
150 | or croak("Can't dup stderr: $!"); |
441eeb04 |
151 | |
b2c02cd0 |
152 | open( STDERR, '>&='. $self->stderr->fileno ) |
b332ee65 |
153 | or croak("Can't open stderr: $!"); |
26e3d92b |
154 | |
155 | binmode( $self->stderr ); |
156 | binmode( STDERR ); |
157 | } |
b2e1304d |
158 | |
9c216915 |
159 | { |
160 | no warnings 'uninitialized'; |
903274b1 |
161 | %ENV = (%ENV, %{ $self->environment }); |
5889d784 |
162 | } |
163 | |
9c216915 |
164 | if ( $INC{'CGI.pm'} ) { |
165 | CGI::initialize_globals(); |
166 | } |
167 | |
b332ee65 |
168 | $self->{setuped}++; |
9c216915 |
169 | |
170 | return $self; |
5889d784 |
171 | } |
172 | |
780060e5 |
173 | sub response { |
b332ee65 |
174 | my ( $self, $callback ) = @_; |
780060e5 |
175 | |
b332ee65 |
176 | return undef unless $self->stdout; |
780060e5 |
177 | |
b332ee65 |
178 | seek( $self->stdout, 0, SEEK_SET ) |
179 | or croak("Can't seek stdout handle: $!"); |
14f243e8 |
180 | |
b332ee65 |
181 | my $headers; |
182 | while ( my $line = $self->stdout->getline ) { |
183 | $headers .= $line; |
184 | last if $headers =~ /\x0d?\x0a\x0d?\x0a$/; |
780060e5 |
185 | } |
15e50838 |
186 | |
b332ee65 |
187 | unless ( defined $headers ) { |
188 | $headers = "HTTP/1.1 500 Internal Server Error\x0d\x0a"; |
780060e5 |
189 | } |
190 | |
b332ee65 |
191 | unless ( $headers =~ /^HTTP/ ) { |
192 | $headers = "HTTP/1.1 200 OK\x0d\x0a" . $headers; |
4e0afe7d |
193 | } |
decf17dc |
194 | |
b332ee65 |
195 | my $response = HTTP::Response->parse($headers); |
196 | $response->date( time() ) unless $response->date; |
decf17dc |
197 | |
b332ee65 |
198 | my $message = $response->message; |
199 | my $status = $response->header('Status'); |
780060e5 |
200 | |
b332ee65 |
201 | if ( $message && $message =~ /^(.+)\x0d$/ ) { |
202 | $response->message($1); |
780060e5 |
203 | } |
26e3d92b |
204 | |
b332ee65 |
205 | if ( $status && $status =~ /^(\d\d\d)\s?(.+)?$/ ) { |
780060e5 |
206 | |
b332ee65 |
207 | my $code = $1; |
208 | my $message = $2 || HTTP::Status::status_message($code); |
4e0afe7d |
209 | |
b332ee65 |
210 | $response->code($code); |
211 | $response->message($message); |
4e0afe7d |
212 | } |
15e50838 |
213 | |
b332ee65 |
214 | my $length = ( stat( $self->stdout ) )[7] - tell( $self->stdout ); |
780060e5 |
215 | |
b332ee65 |
216 | if ( $response->code == 500 && !$length ) { |
4e0afe7d |
217 | |
b332ee65 |
218 | $response->content( $response->error_as_HTML ); |
219 | $response->content_type('text/html'); |
4e0afe7d |
220 | |
4d6e304a |
221 | return $response; |
780060e5 |
222 | } |
4e0afe7d |
223 | |
b332ee65 |
224 | if ($callback) { |
4d6e304a |
225 | |
b332ee65 |
226 | my $handle = $self->stdout; |
4d6e304a |
227 | |
b332ee65 |
228 | $response->content( sub { |
4e0afe7d |
229 | |
b332ee65 |
230 | if ( $handle->read( my $buffer, 4096 ) ) { |
231 | return $buffer; |
474703e4 |
232 | } |
b332ee65 |
233 | |
234 | return undef; |
235 | }); |
4d6e304a |
236 | } |
b332ee65 |
237 | else { |
4d6e304a |
238 | |
b332ee65 |
239 | my $length = 0; |
4d6e304a |
240 | |
b332ee65 |
241 | while ( $self->stdout->read( my $buffer, 4096 ) ) { |
242 | $length += length($buffer); |
243 | $response->add_content($buffer); |
244 | } |
4e0afe7d |
245 | |
b332ee65 |
246 | if ( $length && !$response->content_length ) { |
247 | $response->content_length($length); |
decf17dc |
248 | } |
780060e5 |
249 | } |
250 | |
780060e5 |
251 | return $response; |
252 | } |
253 | |
b2e1304d |
254 | sub restore { |
255 | my $self = shift; |
4e0afe7d |
256 | |
9c216915 |
257 | { |
258 | no warnings 'uninitialized'; |
5e458489 |
259 | %ENV = %{ $self->{restore}->{environment} }; |
9c216915 |
260 | } |
12852959 |
261 | |
b2c02cd0 |
262 | open( STDIN, '<&'. fileno($self->{restore}->{stdin}) ) |
b332ee65 |
263 | or croak("Can't restore stdin: $!"); |
30efa07d |
264 | |
b332ee65 |
265 | sysseek( $self->stdin, 0, SEEK_SET ) |
266 | or croak("Can't seek stdin: $!"); |
30efa07d |
267 | |
b332ee65 |
268 | if ( $self->{restore}->{stdout} ) { |
26e3d92b |
269 | |
b332ee65 |
270 | STDOUT->flush |
271 | or croak("Can't flush stdout: $!"); |
26e3d92b |
272 | |
b2c02cd0 |
273 | open( STDOUT, '>&'. fileno($self->{restore}->{stdout}) ) |
b332ee65 |
274 | or croak("Can't restore stdout: $!"); |
090cc060 |
275 | |
b332ee65 |
276 | sysseek( $self->stdout, 0, SEEK_SET ) |
277 | or croak("Can't seek stdout: $!"); |
a3875fc6 |
278 | } |
26e3d92b |
279 | |
b332ee65 |
280 | if ( $self->{restore}->{stderr} ) { |
12852959 |
281 | |
b332ee65 |
282 | STDERR->flush |
283 | or croak("Can't flush stderr: $!"); |
30efa07d |
284 | |
b2c02cd0 |
285 | open( STDERR, '>&'. fileno($self->{restore}->{stderr}) ) |
b332ee65 |
286 | or croak("Can't restore stderr: $!"); |
090cc060 |
287 | |
b332ee65 |
288 | sysseek( $self->stderr, 0, SEEK_SET ) |
289 | or croak("Can't seek stderr: $!"); |
6f5fb9a7 |
290 | } |
090cc060 |
291 | |
b332ee65 |
292 | $self->{restored}++; |
9c216915 |
293 | |
294 | return $self; |
b2e1304d |
295 | } |
296 | |
297 | sub DESTROY { |
298 | my $self = shift; |
b332ee65 |
299 | $self->restore if $self->{setuped} && !$self->{restored}; |
b2e1304d |
300 | } |
301 | |
302 | 1; |
303 | |
304 | __END__ |
305 | |
b2e1304d |
306 | =head1 SYNOPSIS |
307 | |
bd7813ac |
308 | use CGI; |
309 | use HTTP::Request; |
310 | use HTTP::Request::AsCGI; |
15e50838 |
311 | |
bd7813ac |
312 | my $request = HTTP::Request->new( GET => 'http://www.host.com/' ); |
313 | my $stdout; |
15e50838 |
314 | |
bd7813ac |
315 | { |
316 | my $c = HTTP::Request::AsCGI->new($request)->setup; |
317 | my $q = CGI->new; |
15e50838 |
318 | |
bd7813ac |
319 | print $q->header, |
320 | $q->start_html('Hello World'), |
321 | $q->h1('Hello World'), |
322 | $q->end_html; |
15e50838 |
323 | |
bd7813ac |
324 | $stdout = $c->stdout; |
15e50838 |
325 | |
5e458489 |
326 | # environment and descriptors will automatically be restored |
2d51e42f |
327 | # when $c is destructed. |
bd7813ac |
328 | } |
15e50838 |
329 | |
bd7813ac |
330 | while ( my $line = $stdout->getline ) { |
331 | print $line; |
332 | } |
15e50838 |
333 | |
b2e1304d |
334 | =head1 DESCRIPTION |
335 | |
9a0e0b26 |
336 | Provides a convenient way of setting up an CGI environment from an HTTP::Request. |
2d51e42f |
337 | |
b2e1304d |
338 | =head1 METHODS |
339 | |
15e50838 |
340 | =over 4 |
14f243e8 |
341 | |
b332ee65 |
342 | =item new ( $request [, key => value ] ) |
14f243e8 |
343 | |
9a0e0b26 |
344 | Constructor. The first argument must be a instance of HTTP::Request, followed |
345 | by optional pairs of environment key and value. |
14f243e8 |
346 | |
5e458489 |
347 | =item environment |
14f243e8 |
348 | |
15e50838 |
349 | Returns a hashref containing the environment that will be used in setup. |
b332ee65 |
350 | Changing the hashref after setup has been called will have no effect. |
14f243e8 |
351 | |
b332ee65 |
352 | =item setup |
9c216915 |
353 | |
9a0e0b26 |
354 | Sets up the environment and descriptors. |
9c216915 |
355 | |
b332ee65 |
356 | =item restore |
9c216915 |
357 | |
5e458489 |
358 | Restores the environment and descriptors. Can only be called after setup. |
9c216915 |
359 | |
b332ee65 |
360 | =item request |
9c216915 |
361 | |
b332ee65 |
362 | Returns the request given to constructor. |
9c216915 |
363 | |
b332ee65 |
364 | =item response |
9c216915 |
365 | |
b332ee65 |
366 | Returns a HTTP::Response. Can only be called after restore. |
9c216915 |
367 | |
b332ee65 |
368 | =item stdin |
9c216915 |
369 | |
b332ee65 |
370 | Accessor for handle that will be used for STDIN, must be a real seekable |
371 | handle with an file descriptor. Defaults to a tempoary IO::File instance. |
14f243e8 |
372 | |
b332ee65 |
373 | =item stdout |
14f243e8 |
374 | |
b332ee65 |
375 | Accessor for handle that will be used for STDOUT, must be a real seekable |
376 | handle with an file descriptor. Defaults to a tempoary IO::File instance. |
9c216915 |
377 | |
b332ee65 |
378 | =item stderr |
14f243e8 |
379 | |
b332ee65 |
380 | Accessor for handle that will be used for STDERR, must be a real seekable |
381 | handle with an file descriptor. |
14f243e8 |
382 | |
383 | =back |
b2e1304d |
384 | |
74fbb9dd |
385 | =head1 SEE ALSO |
386 | |
387 | =over 4 |
388 | |
389 | =item examples directory in this distribution. |
390 | |
391 | =item L<WWW::Mechanize::CGI> |
392 | |
393 | =item L<Test::WWW::Mechanize::CGI> |
394 | |
395 | =back |
396 | |
2d51e42f |
397 | =head1 THANKS TO |
17b370b0 |
398 | |
399 | Thomas L. Shinnick for his valuable win32 testing. |
400 | |
b2e1304d |
401 | =cut |