X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FCatalyst%2FEngine.pm;h=7a98187f01a5a90111a1ff4e3f50a4022bfda13b;hb=518576166dfa8607e2a0384ac46fa061dd931884;hp=8d1bdbe26012275a4eed9aaa9cb742baedee6166;hpb=6e5b548ecc3b047f6297daf874b65752cfba5909;p=catagits%2FCatalyst-Runtime.git diff --git a/lib/Catalyst/Engine.pm b/lib/Catalyst/Engine.pm index 8d1bdbe..7a98187 100644 --- a/lib/Catalyst/Engine.pm +++ b/lib/Catalyst/Engine.pm @@ -10,13 +10,36 @@ use HTML::Entities; use HTTP::Body; use HTTP::Headers; use URI::QueryParam; -use Scalar::Util (); +use Moose::Util::TypeConstraints; +use Plack::Loader; +use Plack::Middleware::Conditional; +use Plack::Middleware::ReverseProxy; +use Encode (); +use utf8; + +use namespace::clean -except => 'meta'; + +has env => (is => 'ro', writer => '_set_env', clearer => '_clear_env'); # input position and length has read_length => (is => 'rw'); has read_position => (is => 'rw'); -no Moose; +has _prepared_write => (is => 'rw'); + +has _response_cb => ( + is => 'ro', + isa => 'CodeRef', + writer => '_set_response_cb', + clearer => '_clear_response_cb', +); + +has _writer => ( + is => 'ro', + isa => duck_type([qw(write close)]), + writer => '_set_writer', + clearer => '_clear_writer', +); # Amount of data to read from input on each pass our $CHUNKSIZE = 64 * 1024; @@ -44,16 +67,24 @@ sub finalize_body { my ( $self, $c ) = @_; my $body = $c->response->body; no warnings 'uninitialized'; - if ( Scalar::Util::blessed($body) && $body->can('read') or ref($body) eq 'GLOB' ) { - while ( !eof $body ) { - read $body, my ($buffer), $CHUNKSIZE; - last unless $self->write( $c, $buffer ); - } + if ( blessed($body) && $body->can('read') or ref($body) eq 'GLOB' ) { + my $got; + do { + $got = read $body, my ($buffer), $CHUNKSIZE; + $got = 0 unless $self->write( $c, $buffer ); + } while $got > 0; + close $body; } else { $self->write( $c, $body ); } + + $self->_writer->close; + $self->_clear_writer; + $self->_clear_env; + + return; } =head2 $self->finalize_cookies($c) @@ -74,7 +105,7 @@ sub finalize_cookies { my $val = $response->cookies->{$name}; my $cookie = ( - Scalar::Util::blessed($val) + blessed($val) ? $val : CGI::Simple::Cookie->new( -name => $name, @@ -82,7 +113,8 @@ sub finalize_cookies { -expires => $val->{expires}, -domain => $val->{domain}, -path => $val->{path}, - -secure => $val->{secure} || 0 + -secure => $val->{secure} || 0, + -httponly => $val->{httponly} || 0, ) ); @@ -102,11 +134,37 @@ is in debug mode, or a `please come back later` message otherwise. =cut +sub _dump_error_page_element { + my ($self, $i, $element) = @_; + my ($name, $val) = @{ $element }; + + # This is fugly, but the metaclass is _HUGE_ and demands waaay too much + # scrolling. Suggestions for more pleasant ways to do this welcome. + local $val->{'__MOP__'} = "Stringified: " + . $val->{'__MOP__'} if ref $val eq 'HASH' && exists $val->{'__MOP__'}; + + my $text = encode_entities( dump( $val )); + sprintf <<"EOF", $name, $text; +

%s

+
+ 
%s
+
+EOF +} + sub finalize_error { my ( $self, $c ) = @_; $c->res->content_type('text/html; charset=utf-8'); - my $name = $c->config->{name} || join(' ', split('::', ref $c)); + my $name = ref($c)->config->{name} || join(' ', split('::', ref $c)); + + # Prevent Catalyst::Plugin::Unicode::Encoding from running. + # This is a little nasty, but it's the best way to be clean whether or + # not the user has an encoding plugin. + + if ($c->can('encoding')) { + $c->{encoding} = ''; + } my ( $title, $error, $infos ); if ( $c->debug ) { @@ -123,23 +181,16 @@ sub finalize_error { $name = "

$name

"; # Don't show context in the dump - delete $c->req->{_context}; - delete $c->res->{_context}; + $c->req->_clear_context; + $c->res->_clear_context; # Don't show body parser in the dump - delete $c->req->{_body}; + $c->req->_clear_body; my @infos; my $i = 0; for my $dump ( $c->dump_these ) { - my $name = $dump->[0]; - my $value = encode_entities( dump( $dump->[1] )); - push @infos, sprintf <<"EOF", $name, $value; -

%s

-
- 
%s
-
-EOF + push @infos, $self->_dump_error_page_element($i, $dump); $i++; } $infos = join "\n", @infos; @@ -157,6 +208,8 @@ EOF (dk) Venligst prov igen senere (pl) Prosze sprobowac pozniej (pt) Por favor volte mais tarde +(ru) Попробуйте еще раз позже +(ua) Спробуйте ще раз пізніше $name = ''; @@ -242,7 +295,7 @@ EOF } /* from http://users.tkk.fi/~tkarvine/linux/doc/pre-wrap/pre-wrap-css3-mozilla-opera-ie.html */ /* Browser specific (not valid) styles to make preformatted text wrap */ - pre { + pre { white-space: pre-wrap; /* css-3 */ white-space: -moz-pre-wrap; /* Mozilla, since 1999 */ white-space: -pre-wrap; /* Opera 4-6 */ @@ -260,10 +313,12 @@ EOF - - # Trick IE + # Trick IE. Old versions of IE would display their own error page instead + # of ours if we'd give it less than 512 bytes. $c->res->{body} .= ( ' ' x 512 ); + $c->res->{body} = Encode::encode("UTF-8", $c->res->{body}); + # Return 500 $c->res->status(500); } @@ -274,7 +329,17 @@ Abstract method, allows engines to write headers to response =cut -sub finalize_headers { } +sub finalize_headers { + my ($self, $ctx) = @_; + + my @headers; + $ctx->response->headers->scan(sub { push @headers, @_ }); + + $self->_set_writer($self->_response_cb->([ $ctx->response->status, \@headers ])); + $self->_clear_response_cb; + + return; +} =head2 $self->finalize_read($c) @@ -291,6 +356,8 @@ Clean up after uploads, deleting temp files. sub finalize_uploads { my ( $self, $c ) = @_; + # N.B. This code is theoretically entirely unneeded due to ->cleanup(1) + # on the HTTP::Body object. my $request = $c->request; foreach my $key (keys %{ $request->uploads }) { my $upload = $request->uploads->{$key}; @@ -309,16 +376,19 @@ sets up the L object body using L sub prepare_body { my ( $self, $c ) = @_; + my $appclass = ref($c) || $c; if ( my $length = $self->read_length ) { my $request = $c->request; - unless ( $request->{_body} ) { + unless ( $request->_body ) { my $type = $request->header('Content-Type'); - $request->{_body} = HTTP::Body->new( $type, $length ); - $request->{_body}->tmpdir( $c->config->{uploadtmp} ) - if exists $c->config->{uploadtmp}; + $request->_body(HTTP::Body->new( $type, $length )); + $request->_body->cleanup(1); # Make extra sure! + $request->_body->tmpdir( $appclass->config->{uploadtmp} ) + if exists $appclass->config->{uploadtmp}; } - - while ( my $buffer = $self->read($c) ) { + + # Check for definedness as you could read '0' + while ( defined ( my $buffer = $self->read($c) ) ) { $c->prepare_body_chunk($buffer); } @@ -332,7 +402,7 @@ sub prepare_body { } else { # Defined but will cause all body code to be skipped - $c->request->{_body} = 0; + $c->request->_body(0); } } @@ -345,21 +415,21 @@ Add a chunk to the request body. sub prepare_body_chunk { my ( $self, $c, $chunk ) = @_; - $c->request->{_body}->add($chunk); + $c->request->_body->add($chunk); } =head2 $self->prepare_body_parameters($c) -Sets up parameters from body. +Sets up parameters from body. =cut sub prepare_body_parameters { my ( $self, $c ) = @_; - - return unless $c->request->{_body}; - - $c->request->body_parameters( $c->request->{_body}->param ); + + return unless $c->request->_body; + + $c->request->body_parameters( $c->request->_body->param ); } =head2 $self->prepare_connection($c) @@ -368,7 +438,22 @@ Abstract method implemented in engines. =cut -sub prepare_connection { } +sub prepare_connection { + my ($self, $ctx) = @_; + + my $env = $self->env; + my $request = $ctx->request; + + $request->address( $env->{REMOTE_ADDR} ); + $request->hostname( $env->{REMOTE_HOST} ) + if exists $env->{REMOTE_HOST}; + $request->protocol( $env->{SERVER_PROTOCOL} ); + $request->remote_user( $env->{REMOTE_USER} ); + $request->method( $env->{REQUEST_METHOD} ); + $request->secure( $env->{'psgi.url_scheme'} eq 'https' ? 1 : 0 ); + + return; +} =head2 $self->prepare_cookies($c) @@ -388,7 +473,19 @@ sub prepare_cookies { =cut -sub prepare_headers { } +sub prepare_headers { + my ($self, $ctx) = @_; + + my $env = $self->env; + my $headers = $ctx->request->headers; + + for my $header (keys %{ $env }) { + next unless $header =~ /^(HTTP|CONTENT|COOKIE)/i; + (my $field = $header) =~ s/^HTTPS?_//; + $field =~ tr/_/-/; + $headers->header($field => $env->{$header}); + } +} =head2 $self->prepare_parameters($c) @@ -426,7 +523,61 @@ abstract method, implemented by engines. =cut -sub prepare_path { } +sub prepare_path { + my ($self, $ctx) = @_; + + my $env = $self->env; + + my $scheme = $ctx->request->secure ? 'https' : 'http'; + my $host = $env->{HTTP_HOST} || $env->{SERVER_NAME}; + my $port = $env->{SERVER_PORT} || 80; + my $base_path = $env->{SCRIPT_NAME} || "/"; + + # set the request URI + my $path; + if (!$ctx->config->{use_request_uri_for_path}) { + my $path_info = $env->{PATH_INFO}; + if ( exists $env->{REDIRECT_URL} ) { + $base_path = $env->{REDIRECT_URL}; + $base_path =~ s/\Q$path_info\E$//; + } + $path = $base_path . $path_info; + $path =~ s{^/+}{}; + $path =~ s/([^$URI::uric])/$URI::Escape::escapes{$1}/go; + $path =~ s/\?/%3F/g; # STUPID STUPID SPECIAL CASE + } + else { + my $req_uri = $env->{REQUEST_URI}; + $req_uri =~ s/\?.*$//; + $path = $req_uri; + $path =~ s{^/+}{}; + } + + # Using URI directly is way too slow, so we construct the URLs manually + my $uri_class = "URI::$scheme"; + + # HTTP_HOST will include the port even if it's 80/443 + $host =~ s/:(?:80|443)$//; + + if ($port !~ /^(?:80|443)$/ && $host !~ /:/) { + $host .= ":$port"; + } + + my $query = $env->{QUERY_STRING} ? '?' . $env->{QUERY_STRING} : ''; + my $uri = $scheme . '://' . $host . '/' . $path . $query; + + $ctx->request->uri( (bless \$uri, $uri_class)->canonical ); + + # set the base URI + # base must end in a slash + $base_path .= '/' unless $base_path =~ m{/$}; + + my $base_uri = $scheme . '://' . $host . $base_path; + + $ctx->request->base( bless \$base_uri, $uri_class ); + + return; +} =head2 $self->prepare_request($c) @@ -437,8 +588,12 @@ process the query string and extract query parameters. =cut sub prepare_query_parameters { - my ( $self, $c, $query_string ) = @_; - + my ($self, $c) = @_; + + my $query_string = exists $self->env->{QUERY_STRING} + ? $self->env->{QUERY_STRING} + : ''; + # Check for keywords (no = signs) # (yes, index() is faster than a regex :)) if ( index( $query_string, '=' ) < 0 ) { @@ -450,17 +605,17 @@ sub prepare_query_parameters { # replace semi-colons $query_string =~ s/;/&/g; - + my @params = grep { length $_ } split /&/, $query_string; for my $item ( @params ) { - - my ($param, $value) + + my ($param, $value) = map { $self->unescape_uri($_) } split( /=/, $item, 2 ); - + $param = $self->unescape_uri($item) unless defined $param; - + if ( exists $query{$param} ) { if ( ref $query{$param} ) { push @{ $query{$param} }, $value; @@ -488,7 +643,7 @@ sub prepare_read { # Initialize the read position $self->read_position(0); - + # Initialize the amount of data we think we need to read $self->read_length( $c->request->header('Content-Length') || 0 ); } @@ -499,7 +654,10 @@ Populate the context object from the request object. =cut -sub prepare_request { } +sub prepare_request { + my ($self, $ctx, %args) = @_; + $self->_set_env($args{env}); +} =head2 $self->prepare_uploads($c) @@ -509,9 +667,9 @@ sub prepare_uploads { my ( $self, $c ) = @_; my $request = $c->request; - return unless $request->{_body}; + return unless $request->_body; - my $uploads = $request->{_body}->upload; + my $uploads = $request->_body->upload; my $parameters = $request->parameters; foreach my $name (keys %$uploads) { my $files = $uploads->{$name}; @@ -521,7 +679,7 @@ sub prepare_uploads { my $u = Catalyst::Request::Upload->new ( size => $upload->{size}, - type => $headers->content_type, + type => scalar $headers->content_type, headers => $headers, tempname => $upload->{tempname}, filename => $upload->{filename}, @@ -557,6 +715,10 @@ sub prepare_write { } =head2 $self->read($c, [$maxlength]) +Reads from the input stream by calling C<< $self->read_chunk >>. + +Maintains the read_length and read_position counters as data is read. + =cut sub read { @@ -574,6 +736,11 @@ sub read { my $readlen = ( $remaining > $maxlength ) ? $maxlength : $remaining; my $rc = $self->read_chunk( $c, my $buffer, $readlen ); if ( defined $rc ) { + if (0 == $rc) { # Nothing more to read even though Content-Length + # said there should be. + $self->finalize_read; + return; + } $self->read_position( $self->read_position + $rc ); return $buffer; } @@ -585,12 +752,16 @@ sub read { =head2 $self->read_chunk($c, $buffer, $length) -Each engine inplements read_chunk as its preferred way of reading a chunk -of data. +Each engine implements read_chunk as its preferred way of reading a chunk +of data. Returns the number of bytes read. A return of 0 indicates that +there is no more data to be read. =cut -sub read_chunk { } +sub read_chunk { + my ($self, $ctx) = (shift, shift); + return $self->env->{'psgi.input'}->read(@_); +} =head2 $self->read_length @@ -601,13 +772,55 @@ header. The amount of input data that has already been read. -=head2 $self->run($c) +=head2 $self->run($app, $server) -Start the engine. Implemented by the various engine classes. +Start the engine. Builds a PSGI application and calls the +run method on the server passed in.. =cut -sub run { } +sub run { + my ($self, $app, $psgi, @args) = @_; + # FIXME - Do something sensible with the options we're passed + my $server = pop @args if blessed $args[-1]; + $server ||= Plack::Loader->auto(); # We're not being called from a script, + # so auto detect what backend to run on. + # This does *NOT* cover mod_perl. + $server->run($psgi); +} + +=head2 build_psgi_app ($app, @args) + +Builds and returns a PSGI application closure, wrapping it in the reverse proxy +middleware if the using_frontend_proxy config setting is set. + +=cut + +sub build_psgi_app { + my ($self, $app, @args) = @_; + + my $psgi_app = sub { + my ($env) = @_; + + return sub { + my ($respond) = @_; + $self->_set_response_cb($respond); + $app->handle_request(env => $env); + }; + }; + + $psgi_app = Plack::Middleware::Conditional->wrap( + $psgi_app, + builder => sub { Plack::Middleware::ReverseProxy->wrap($_[0]) }, + condition => sub { + my ($env) = @_; + return if $app->config->{ignore_frontend_proxy}; + return $env->{REMOTE_ADDR} eq '127.0.0.1' || $app->config->{using_frontend_proxy}; + }, + ); + + return $psgi_app; +} =head2 $self->write($c, $buffer) @@ -618,36 +831,17 @@ Writes the buffer to the client. sub write { my ( $self, $c, $buffer ) = @_; - unless ( $self->{_prepared_write} ) { + unless ( $self->_prepared_write ) { $self->prepare_write($c); - $self->{_prepared_write} = 1; - } - - my $len = length($buffer); - my $wrote = syswrite STDOUT, $buffer; - - if ( !defined $wrote && $! == EWOULDBLOCK ) { - # Unable to write on the first try, will retry in the loop below - $wrote = 0; + $self->_prepared_write(1); } - - if ( defined $wrote && $wrote < $len ) { - # We didn't write the whole buffer - while (1) { - my $ret = syswrite STDOUT, $buffer, $CHUNKSIZE, $wrote; - if ( defined $ret ) { - $wrote += $ret; - } - else { - next if $! == EWOULDBLOCK; - return; - } - - last if $wrote >= $len; - } - } - - return $wrote; + + return 0 if !defined $buffer; + + my $len = length($buffer); + $self->_writer->write($buffer); + + return $len; } =head2 $self->unescape_uri($uri) @@ -669,13 +863,25 @@ sub unescape_uri { , see finalize_body +=head2 $self->env + +Hash containing environment variables including many special variables inserted +by WWW server - like SERVER_*, REMOTE_*, HTTP_* ... + +Before accessing environment variables consider whether the same information is +not directly available via Catalyst objects $c->request, $c->engine ... + +BEWARE: If you really need to access some environment variable from your Catalyst +application you should use $c->engine->env->{VARNAME} instead of $ENV{VARNAME}, +as in some enviroments the %ENV hash does not contain what you would expect. + =head1 AUTHORS Catalyst Contributors, see Catalyst.pm =head1 COPYRIGHT -This program is free software, you can redistribute it and/or modify it under +This library is free software. You can redistribute it and/or modify it under the same terms as Perl itself. =cut