package HTTP::Proxy::BodyFilter::save;
use strict;
use HTTP::Proxy;
use HTTP::Proxy::BodyFilter;
use vars qw( @ISA );
@ISA = qw( HTTP::Proxy::BodyFilter );
use Fcntl;
use File::Spec;
use File::Path;
use Carp;
sub init {
my $self = shift;
my %args = (
template => File::Spec->catfile( '%h', '%P' ),
no_host => 0,
no_dirs => 0,
cut_dirs => 0,
prefix => '',
filename => undef,
multiple => 1,
keep_old => 0, timestamp => 0,
status => [ 200 ],
@_
);
croak "Can't timestamp and keep older files at the same time"
if $args{keep_old} && $args{timestamp};
croak "status must be an array reference"
unless ref($args{status}) eq 'ARRAY';
croak "status must contain only HTTP codes"
if grep { !/^[12345]\d\d$/ } @{ $args{status} };
croak "filename must be a code reference"
if defined $args{filename} && !UNIVERSAL::isa( $args{filename}, 'CODE' );
$self->{"_hpbf_save_filename_code"} = $args{filename};
$self->{"_hpbf_save_$_"} = $args{$_}
for qw( template no_host no_dirs cut_dirs prefix
multiple keep_old timestamp status );
}
sub begin {
my ( $self, $message ) = @_;
delete @{$self}{qw( _hpbf_save_filename _hpbf_save_fh )};
my $uri = $message->isa( 'HTTP::Request' )
? $message->uri : $message->request->uri;
if( $message->isa( 'HTTP::Response' ) ) {
my $code = $message->code;
return unless grep { $code eq $_ } @{ $self->{_hpbf_save_status} };
}
my $file = '';
if( defined $self->{_hpbf_save_filename_code} ) {
$file = $self->{_hpbf_save_filename_code}->($message);
unless ( defined $file and $file ne '' ) {
$self->proxy->log( HTTP::Proxy::FILTERS, "HTBF::save",
"Filter will not save $uri" );
return;
}
}
else {
my @segs = $uri->path_segments; shift @segs;
splice(@segs, 0, $self->{_hpbf_save_cut_dirs} >= @segs
? @segs - 1 : $self->{_hpbf_save_cut_dirs} );
my %vars = (
'%' => '%',
h => $self->{_hpbf_save_no_host} ? '' : $uri->host,
f => $segs[-1] || 'index.html', p => $self->{_hpbf_save_no_dirs} ? $segs[-1] || 'index.html'
: File::Spec->catfile(@segs),
q => $uri->query,
);
pop @segs;
$vars{d}
= $self->{_hpbf_save_no_dirs} ? ''
: @segs ? File::Spec->catfile(@segs)
: '';
$vars{P} = $vars{p} . ( $vars{q} ? "?$vars{q}" : '' );
$file = File::Spec->catfile( $self->{_hpbf_save_prefix} || (),
$self->{_hpbf_save_template} );
$file =~ s/%(.)/$vars{$1}/g;
}
$file = File::Spec->rel2abs( $file );
my $dir = File::Spec->catpath( (File::Spec->splitpath($file))[ 0, 1 ], '' );
if( ! -e $dir ) {
eval { mkpath( $dir ) };
if ($@) {
$self->proxy->log( HTTP::Proxy::ERROR, "HTBF::save",
"Unable to create directory $dir" );
return;
}
$self->proxy->log( HTTP::Proxy::FILTERS, "HTBF::save",
"Created directory $dir" );
}
if ( -e $file ) {
if ( $self->{_hpbf_save_timestamp} ) {
}
elsif ( $self->{_hpbf_save_keep_old} ) {
$self->proxy->log( HTTP::Proxy::FILTERS, "HPBF::save",
"Skip saving $uri" );
delete $self->{_hpbf_save_fh}; return;
}
}
my ( $ext, $n, $i ) = ( "", 0 );
my $flags = O_WRONLY | O_EXCL | O_CREAT;
while( ! sysopen( $self->{_hpbf_save_fh}, "$file$ext", $flags ) ) {
$self->proxy->log( HTTP::Proxy::ERROR, "HPBF::save",
"Too many errors opening $file$ext" ), return
if $i++ - $n == 10; if( $self->{_hpbf_save_multiple} ) {
$ext = "." . ++$n while -e $file.$ext;
next;
}
else {
$flags = O_WRONLY | O_CREAT;
}
}
$self->{_hpbf_save_filename} = $file.$ext;
binmode( $self->{_hpbf_save_fh} ); $self->proxy->log( HTTP::Proxy::FILTERS, "HPBF::save",
"Saving $uri to $file$ext" );
}
sub filter {
my ( $self, $dataref ) = @_;
return unless exists $self->{_hpbf_save_fh};
my $res = $self->{_hpbf_save_fh}->syswrite( $$dataref );
$self->proxy->log( HTTP::Proxy::ERROR, "HPBF::save", "syswrite() error: $!")
if ! defined $res; }
sub end {
my ($self) = @_;
if( $self->{_hpbf_save_fh} ) {
$self->{_hpbf_save_fh}->close; delete $self->{_hpbf_save_fh};
}
}
sub will_modify { 0 }
1;
__END__
=head1 NAME
HTTP::Proxy::BodyFilter::save - A filter that saves transfered data to a file
=head1 SYNOPSIS
use HTTP::Proxy;
use HTTP::Proxy::BodyFilter::save;
my $proxy = HTTP::Proxy->new;
# save RFC files as we browse them
$proxy->push_filter(
path => qr!/rfc\d+.txt!,
mime => 'text/plain',
response => HTTP::Proxy::BodyFilter::save->new(
template => '%f',
prefix => 'rfc',
keep_old => 1,
)
);
$proxy->start;
=head1 DESCRIPTION
The HTTP::Proxy::BodyFilter::save filter can save HTTP messages (responses
or request) bodies to files. The name of the file is determined by a
template and the URI of the request.
Simply insert this filter in a filter stack, and it will save the data
as it flows through the proxy. Depending on where the filter is located
in the stack, the saved data can be more or less modified.
This filter I<will> create directories if it needs to!
I<Note:> Remember that the default C<mime> parameter for C<push_filter()>
is C<text/*> and that you may need to change it for other MIME types.
=head2 Constructor
The constructor accepts quite a few options. Most of them control
the construction of the filename that will be used to save the
response body. There are two options to compute this filename:
=over 4
=item *
use a template
=item *
use your own filename creation routine
=back
The template option uses the following options:
=over 4
=item B<template> => I<string>
The file name is build from the C<template> option. The following
placeholders are available:
%% a percent sign
%h the host
%p the path (no leading separator)
%d the path (filename removed)
%f the filename (or 'index.html' if absent)
%q the query string
%P the path and the query string,
separated by '?' (if the query string is not empty)
C</> in the URI path are replaced by the separator used by File::Spec.
The result of the template is modified by the B<no_host>, B<no_dirs>
and B<cut_dirs>.
The default template is the local equivalent of the C<%h/%P> Unix path.
=item B<no_host> => I<boolean>
The C<no_host> option makes C<%h> empty. Default is I<false>.
=item B<no_dirs> => I<boolean>
The C<no_dirs> option removes all directories from C<%p>, C<%P> and C<%d>.
Default is I<false>.
=item B<cut_dirs> => I<number>
The C<cut_dirs> options removes the first I<n> directories from the
content of C<%p>, C<%P> and C<%d>. Default is C<0>.
=item B<prefix> => I<string>
The B<prefix> option prepends the given prefix to the filename
created from the template. Default is C<"">.
=back
Using your own subroutine is also possible, with the following parameter:
=over 4
=item B<filename> => I<coderef>
When the C<filename> option is used, the C<template> option and the
other template-related options (C<no_host>, C<no_dirs>, C<cut_dirs>
and C<prefix>) are ignored.
The C<filename> option expects a reference to a subroutine. The subroutine
will receive the HTTP::Message object and must return a string which
is the path of the file to be created (an absolute path is recommended,
but a relative path is accepted).
Returning C<""> or C<undef> will prevent the creation of the file.
This lets a filter decide even more precisely what to save or not,
even though this should be done in the match subroutine (see
HTTP::Proxy's C<pushè_filte()> method).
=back
Other options help the filter decide where and when to save:
=over 4
=item B<multiple> => I<boolean>
With the B<multiple> option, saving the same file in the same directory
will result in the original copy of file being preserved and the second
copy being named F<file.1>. If that a file is saved yet again with the same
name, the third copy will be named F<file.2>, and so on.
Default is I<true>.
If B<multiple> is set to I<false> then a file will be overwritten
by the next one with the same name.
=item B<timestamp> => I<boolean>
With the C<timestamp> option, the decision as to whether or not to save
a newer copy of a file depends on the local and remote timestamp and
size of the file.
The file is saved only if the date given in the C<Last-Modified> is more
recent than the local file's timestamp.
Default is I<false>.
B<This option is not implemented.>
=item B<keep_old> => I<boolean>
The C<keep_old> option will prevent the file to be saved if a file
with the same name already exists. Default is I<false>.
No matter if B<multiple> is set or not, the file will I<not> be saved
if B<keep_old> is set to true.
=item B<status> => \@codes
The C<status> option limits the status codes for which a response body
will be saved. The default is C<[ 200 ]>, which prevent saving error
pages (for 404 codes).
=back
=head2 Examples
Given a request for the L<http://search.cpan.org/dist/HTTP-Proxy/> URI,
the filename is computed as follows, depending on the constructor
options:
No options -> search.cpan.org/dist/HTTP-Proxy/index.html
no_host => 1 -> dist/HTTP-Proxy/index.html
no_dirs => 1 -> search.cpan.org/index.html
no_host => 1,
no_dirs => 1,
prefix => 'data' -> data/index.html
cut_dirs => 1 -> search.cpan.org/HTTP-Proxy/index.html
cut_dirs => 2 -> search.cpan.org/index.html
=head1 METHODS
This filter implements several methods, which are all called atuomatically:
=over 4
=item init()
Handle all the parameters passed to the constructor to define the
filter behaviour.
=item begin()
Open the file to which the data will be saved.
=item filter()
Save all the data that goes through to the opened file.
=item end()
Close the file when the whole message body has been processed.
=item will_modify()
This method returns a I<false> value, thus indicating to the system
that it will not modify data passing through.
=back
=head1 SEE ALSO
L<HTTP::Proxy>, L<HTTP::Proxy::BodyFilter>.
=head1 AUTHOR
Philippe "BooK" Bruhat, E<lt>book@cpan.orgE<gt>.
=head1 ACKNOWLEDGMENTS
Thanks to Mat Proud for asking how to store all pages which go through
the proxy to disk, without any processing. The further discussion we
had led to the writing of this class.
Wget(1) provided the inspiration for many of the file naming options.
Thanks to Nicolas Chuche for telling me about C<O_EXCL>.
Thanks to Rafaël Garcia-Suarez and David Rigaudiere for their help on
irc while coding the nasty C<begin()> method. C<;-)>
Thanks to Howard Jones for the inspiration and initial patch for the
C<filename> option. Lucas Gonze provided a patch to make C<status>
actually work.
Thanks to Max Maischein for detecting a bug in the parameter validation
for C<filename> (L<http://rt.cpan.org/Ticket/Display.html?id=14548>).
Thanks to Mark Tilford, who found out that the
C<filename> option was incorrectly used internally
(L<http://rt.cpan.org/Ticket/Display.html?id=18644>).
Thanks to Roland Stigge and Gunnar Wolf for
reporting and forwarding Debian bug #433951 to CPAN RT
(L<http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=433951>,
L<http://rt.cpan.org/Ticket/Display.html?id=33018>).
=head1 COPYRIGHT
Copyright 2004-2008, Philippe Bruhat.
=head1 LICENSE
This module is free software; you can redistribute it or modify it under
the same terms as Perl itself.
=cut