HTTP::Proxy::BodyFilter::save - A filter that saves transfered data to a file
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;
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 will create directories if it needs to!
Note: Remember that the default
mime parameter for
text/* and that you may need to change it for other MIME types.
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:
The template option uses the following options:
The file name is build from the
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)
/ in the URI path are replaced by the separator used by File::Spec.
The result of the template is modified by the no_host, no_dirs and cut_dirs.
The default template is the local equivalent of the
%h/%P Unix path.
no_host option makes
%h empty. Default is false.
no_dirs option removes all directories from
%d. Default is false.
cut_dirs options removes the first n directories from the content of
%d. Default is
The prefix option prepends the given prefix to the filename created from the template. Default is
Using your own subroutine is also possible, with the following parameter:
filename option is used, the
template option and the other template-related options (
prefix) are ignored.
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).
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
Other options help the filter decide where and when to save:
With the 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 file.1. If that a file is saved yet again with the same name, the third copy will be named file.2, and so on.
Default is true.
If multiple is set to false then a file will be overwritten by the next one with the same name.
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
Last-Modified is more recent than the local file's timestamp.
Default is false.
This option is not implemented.
keep_old option will prevent the file to be saved if a file with the same name already exists. Default is false.
No matter if multiple is set or not, the file will not be saved if keep_old is set to true.
status option limits the status codes for which a response body will be saved. The default is
[ 200 ], which prevent saving error pages (for 404 codes).
Given a request for the 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
This filter implements several methods, which are all called atuomatically:
Handle all the parameters passed to the constructor to define the filter behaviour.
Open the file to which the data will be saved.
Save all the data that goes through to the opened file.
Close the file when the whole message body has been processed.
This method returns a false value, thus indicating to the system that it will not modify data passing through.
Philippe "BooK" Bruhat, <email@example.com>.
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
Thanks to Rafa�l Garcia-Suarez and David Rigaudiere for their help on irc while coding the nasty
Thanks to Howard Jones for the inspiration and initial patch for the
filename option. Lucas Gonze provided a patch to make
status actually work.
Thanks to Max Maischein for detecting a bug in the parameter validation for
Thanks to Mark Tilford, who found out that the
filename option was incorrectly used internally (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 (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=433951, http://rt.cpan.org/Ticket/Display.html?id=33018).
Copyright 2004-2013, Philippe Bruhat.
This module is free software; you can redistribute it or modify it under the same terms as Perl itself.