HTTP-Upload-FlowJs/lib/HTTP/Upload/FlowJs.pm
package HTTP::Upload::FlowJs;
use strict;
use Carp qw(croak);
use Filter::signatures;
no warnings 'experimental::signatures';
use feature 'signatures';
use Text::CleanFragment 'clean_fragment';
use Data::Dumper;
use MIME::Detect;
our $VERSION = '0.02';
use JSON qw(encode_json decode_json);
=head1 NAME
HTTP::Upload::FlowJs - handle resumable multi-part HTTP uploads with flowjs
=head1 SYNOPSIS
This synopsis assumes a L<Plack>/L<PSGI>-like environment. There are
plugins for L<Dancer> and L<Mojolicious> planned. See
L<HTTP::Upload::FlowJs::Examples> for longer examples.
The C<flow.js> workflow assumes that your application handles two kinds of
requests, POST requests for storing the payload data and GET requests for
retrieving information about uploaded parts. You will have to make various calls
to the HTTP::Upload::FlowJs object to validate the incoming request at every
stage.
use HTTP::Upload::FlowJs;
my $uploads = '/tmp/flowjs_uploads/';
my $flowjs = HTTP::Upload::FlowJs->new(
incomingDirectory => $uploads,
allowedContentType => sub { $_[0] =~ m!^image/! },
);
my @parameter_names = $flowjs->parameter_names();
# In your POST handler for /upload:
sub POST_upload {
my $params = params();
my %info;
@info{ @parameter_names } = @{$params}{@parameter_names};
$info{ localChunkSize } = -s $params{ file };
# or however you get the size of the uploaded chunk
# you might want to set this so users don't clobber each others upload
my $session_id = '';
my @invalid = $flowjs->validateRequest( 'POST', \%info, $session_id );
if( @invalid ) {
warn 'Invalid flow.js upload request:';
warn $_ for @invalid;
return [500,[],["Invalid request"]];
return;
};
if( $flowjs->disallowedContentType( \%info, $session_id )) {
# We can determine the content type, and it's not an image
return [415,[],["File type disallowed"]];
};
my $chunkname = $flowjs->chunkName( \%info, undef );
# Save or copy the uploaded file
upload('file')->copy_to($chunkname);
# Now check if we have received all chunks of the file
if( $flowjs->uploadComplete( \%info, undef )) {
# Combine all chunks to final name
my $digest = Digest::SHA256->new();
my( $content_type, $ext ) = $flowjs->sniffContentType();
my $final_name = "file1.$ext";
open( my $fh, '>', $final_name )
or die $!;
binmode $fh;
my( $ok, @unlink_chunks )
= $flowjs->combineChunks( \%info, undef, $fh, $digest );
unlink @unlink_chunks;
# Notify backend that a file arrived
print sprintf "File '%s' upload complete\n", $final_name;
};
# Signal OK
return [200,[],[]]
};
# This checks whether a file has been received completely or
# needs to be uploaded again
sub GET_upload {
my $params = params();
my %info;
@info{ @parameter_names} = @{$params}{@parameter_names};
my @invalid = $flowjs->validateRequest( 'GET', \%info, session->{connid} );
if( @invalid ) {
warn 'Invalid flow.js upload request:';
warn $_ for @invalid;
return [500, [], [] ];
} elsif( $flowjs->disallowedContentType( \%info, $session_id)) {
# We can determine the content type, and it's not an image
return [415,[],["File type disallowed"]];
} else {
my( $status, @messages )
= $flowjs->chunkOK( $uploads, \%info, $session_id );
if( $status != 500 ) {
# 200 or 416
return [$status, [], [] ];
} else {
warn $_ for @messages;
return [$status, [], [] ];
};
};
};
=head1 OVERVIEW
L<flow.js|https://github.com/flowjs/flow.js> is a client-side Javascript upload
library that uploads
a file in multiple parts. It requires two API points on the server side,
one C<GET> API point to check whether a part already has been uploaded
completely and one C<POST> API point to send the data of each partial
upload to. This Perl module implements the backend functionality for
both endpoints. It does not implement the handling of the HTTP requests
themselves, but you likely already use a framework like L<Mojolicious>
or L<Dancer> for that.
=head1 METHODS
=head2 C<< HTTP::Upload::FlowJs->new >>
my $flowjs = HTTP::Upload::FlowJs->new(
maxChunkCount => 1000,
maxFileSize => 10_000_000,
maxChunkSize => 1024*1024,
simultaneousUploads => 3,
allowedContentType => sub {
my($type) = @_;
$type =~ m!^image/!; # we only allow for cat images
},
);
=over 4
B<incomingDirectory> - path for the temporary upload parts
Required
B<maxChunkCount> - hard maximum chunks allowed for a single upload
Default 1000
B<maxFileSize> - hard maximum total file size for a single upload
Default 10_000_000
B<maxChunkSize> - hard maximum chunk size for a single chunk
Default 1048576
B<minChunkSize> - hard minimum chunk size for a single chunk
Default 1024
The minimum chunk size is required since the file type detection
works on the first chunk. If the first chunk is too small, its file type
cannot be checked.
B<forceChunkSize> - force all chunks to be less or equal than C<maxChunkSize>
Default: true
Otherwise, the last chunk will be greater than or equal to C<maxChunkSize>
(the last uploaded chunk will be at least this size and up to two the size).
Note: when C<forceChunkSize> is C<false> it only make C<chunkSize> value in
L</jsConfig> equal to C<maxChunkSize/2>.
B<simultaneousUploads> - simultaneously allowed uploads per file
Default 3
This is just an indication to the Javascript C<flow.js> client
if you pass it the configuration from this object. This is not enforced
in any way yet.
B<allowedContentType> - subroutine to check the MIME type
The default is to allow any kind of file
If you need more advanced checking, do so after you've determined a file
upload as complete with C<< $flowjs->uploadComplete >>.
B<fileParameterName> - The name of the multipart POST parameter to use for the
file chunk
Default C<file>
=back
More interesting limits would be hard maxima for the number of pending
uploads or the number of outstanding chunks per user/session. Checking
these would entail a call to C<glob> for each check and thus would be
fairly disk-intensive on some systems.
=cut
sub new( $class, %options ) {
croak "Need a directory name for the temporary upload parts"
unless $options{ incomingDirectory };
$options{ maxChunkCount } ||= 1000;
$options{ maxFileSize } ||= 10_000_000;
$options{ maxChunkSize } ||= 1024*1024;
$options{ minChunkSize } //= 1024;
$options{ forceChunkSize } //= 1;
$options{ simultaneousUploads } ||= 3;
$options{ mime } ||= MIME::Detect->new();
$options{ fileParameterName } ||= 'file';
$options{ allowedContentType } ||= sub { 1 };
bless \%options => $class;
};
=head2 C<< $flowjs->incomingDirectory >>
Return the incoming directory name.
=cut
sub incomingDirectory( $self ) {
$self->{incomingDirectory};
};
=head2 C<< $flowjs->mime >>
Return the L<MIME::Detect> instance.
=cut
sub mime($self) {
$self->{mime}
};
=head2 C<< $flowjs->jsConfig >>
=head2 C<< $flowjs->jsConfigStr >>
# Perl HASH
my $config = $flowjs->jsConfig(
target => '/upload',
);
# JSON string
my $config = $flowjs->jsConfigStr(
target => '/upload',
);
Create a JSON string that encapsulates the configuration of the Perl
object for inclusion with the JS side of the world.
=cut
sub jsConfig( $self, %override ) {
# The last uploaded chunk will be at least this size and up to two the size
# when forceChunkSize is false
my $chunkSize = $self->{maxChunkSize};
$chunkSize = $chunkSize/2 unless $self->{forceChunkSize}; # / placate Filter::Simple
+{
(
map { $_ => $self->{$_} } (qw(
simultaneousUploads
forceChunkSize
))
),
chunkSize => $chunkSize,
testChunks => 1,
withCredentials => 1,
uploadMethod => 'POST',
%override,
};
}
sub jsConfigStr( $self, %override ) {
encode_json($self->jsConfig(%override))
}
=head2 C<< $flowjs->parameter_names >>
my $params = params(); # request params
my @parameter_names = $flowjs->parameter_names; # params needed by Flowjs
my %info;
@info{ @parameter_names } = @{$params}{@parameter_names};
$info{ file } = $params{ file };
$info{ localChunkSize } = -s $params{ file };
my @invalid = $flowjs->validateRequest( 'POST', \%info );
Returns needed params for validating request.
=cut
sub parameter_names( $self, $required_params ) {
my $params = $self->{parameter_names} ||= {
flowChunkNumber => 1,
flowTotalChunks => 1,
flowChunkSize => 1,
flowCurrentChunkSize => 1,
flowTotalSize => 1,
flowIdentifier => 1,
flowFilename => 1,
flowRelativePath => 0,
};
if ( $required_params ) {
return grep { $params->{$_} } keys( %{$params} );
}
return keys( %{$params} );
}
=head2 C<< $flowjs->validateRequest >>
my $session_id = '';
my @invalid = $flowjs->validateRequest( 'POST', \%info, $session_id );
if( @invalid ) {
warning 'Invalid flow.js upload request:';
warning $_ for @invalid;
status 500;
return;
};
Does formal validation of the request HTTP parameters. It does not
check previously stored information.
B<Note> when C<POST> there are addition required params C<localChunkSize>
and C<$self->{fileParameterName}> (default 'file').
=cut
sub validateRequest( $self, $method, $info, $sessionId=undef ) {
# Validate the input somewhat
local $Data::Dumper::Useqq = 1;
my @invalid;
my @required = $self->parameter_names('required');
if( $method eq 'POST') {
push @required, $self->{fileParameterName}, 'localChunkSize'
;
};
for my $param (@required) {
if( ! exists $info->{ $param } or !defined $info->{$param}) {
push @invalid, sprintf 'Parameter [%s] is required but is missing',
$param,
;
};
};
if( @invalid ) {
return @invalid;
};
# Numbers should be numbers
for my $param (qw(flowChunkNumber flowTotalChunks flowChunkSize flowTotalSize flowCurrentChunkSize)) {
if( exists $info->{ $param } and $info->{ $param } !~ /^[0-9]+$/) {
push @invalid, sprintf 'Parameter [%s] should be numeric, but is [%s]; set to 0',
$param,
Dumper $info->{$param}
;
$info->{ $param } = 0;
};
};
# Check maximum chunk count
for my $param (qw(flowChunkNumber flowTotalChunks )) {
if( exists $info->{ $param } and not $info->{ $param } <= $self->{maxChunkCount}) {
push @invalid, sprintf 'Parameter [%s] should be less than %d, but is [%s]',
$param,
$self->{maxChunkCount},
$info->{$param},
;
};
};
# The chunk number needs to be less than or equal to the total chunks
if( ($info->{ flowChunkNumber } || 0) > ($info->{ flowTotalChunks } || 0)) {
push @invalid, sprintf 'Flow chunk number [%s] is greater than the number of total chunks [%s]',
$info->{ flowChunkNumber },
$info->{ flowTotalChunks },
;
};
# Filenames should contain no path fragments
# This will interact badly with directory uploads, but oh well
for my $param (qw(flowFilename)) {
# Sanitize the filename
if( exists $info->{ $param } and $info->{ $param } =~ m![/\\]! ) {
push @invalid, sprintf 'Parameter [%s] contains invalid path segments',
$param,
Dumper $info->{$param}
;
};
};
# Filenames and paths should not contain upward directory references
for my $param (qw(flowFilename flowRelativePath)) {
# Sanitize the filename
if( exists $info->{ $param } and $info->{ $param } =~ m![/\\]\.\.[/\\]! ) {
push @invalid, sprintf 'Parameter [%s] contains invalid upward path segments [%s]',
$param,
Dumper $info->{$param}
;
};
};
# The filename shouldn't contain control characters
for my $param (qw(flowFilename flowRelativePath)) {
if( exists $info->{ $param } and $info->{ $param } =~ m![\x00-\x1f]! ) {
push @invalid, sprintf 'Parameter [%s] contains control characters [%s]',
$param,
Dumper $info->{$param}
;
};
};
my $min_max_error = 0;
for my $param (qw(flowChunkSize flowCurrentChunkSize)) {
if( exists $info->{ $param } and $info->{ $param } > $self->{ maxChunkSize } ) {
$min_max_error = 1;
push @invalid, sprintf 'Uploaded chunk [%d] of file [%s] is too large [%d], allowed is [%d]',
$info->{flowChunkNumber},
$info->{flowFilename},
$info->{$param},
$self->{maxChunkSize},
;
}
}
for my $param (qw(flowChunkSize flowCurrentChunkSize)) {
if( exists $info->{ $param } and $info->{ $param } < $self->{ minChunkSize }
and ( $info->{flowChunkNumber} < $info->{flowTotalChunks} # only last chunk could be smaller
or $info->{flowTotalChunks} <= 1 # when total chunks > 1
)
) {
$min_max_error = 1;
push @invalid, sprintf 'Uploaded chunk [%d] of file [%s] is too small [%d], allowed is [%d]',
$info->{flowChunkNumber},
$info->{flowFilename},
$info->{$param},
$self->{minChunkSize},
;
}
}
if( ! $min_max_error and ($info->{ flowTotalSize } || 0) > $self->{ maxFileSize } ) {
# Uploaded file would be too large
push @invalid, sprintf 'Uploaded file [%s] would be too large ([%d]) allowed is [%d]',
$info->{flowFilename},
$info->{flowTotalSize},
$self->{maxFileSize},
;
} elsif( ! $min_max_error and $method eq 'POST' and $info->{ localChunkSize } > $info->{flowChunkSize} ) {
# Uploaded chunk is larger than the maximum chunk upload size
push @invalid, sprintf 'Uploaded chunk [%d] of file [%s] is larger than it should be ([%d], allowed is [%d])',
$info->{flowChunkNumber},
$info->{flowFilename},
$info->{localChunkSize},
$self->{maxChunkSize},
;
} elsif( ! $min_max_error and $info->{ flowCurrentChunkSize } < $self->expectedChunkSize( $info ) ) {
# Uploaded chunk is a middle or end chunk but is too small
push @invalid, sprintf 'Uploaded chunk [%s] is too small ([%d]) expect [%d]',
$info->{flowChunkNumber},
$info->{flowCurrentChunkSize},
$self->expectedChunkSize( $info ),
;
} elsif( ! $min_max_error and $method eq 'POST' and $info->{ localChunkSize } < $info->{ flowCurrentChunkSize } ) {
# Real uploaded chunk is smaller than provided chunk upload size
push @invalid, sprintf 'Uploaded chunk [%s] is too small ([%d]) expect [%d]',
$info->{flowChunkNumber},
$info->{localChunkSize},
$info->{flowCurrentChunkSize},
;
} elsif( ! $min_max_error and $info->{ flowCurrentChunkSize } > $self->expectedChunkSize( $info ) ) {
# Uploaded chunk is a middle or end chunk but is too large
push @invalid, sprintf 'Uploaded chunk [%s] is too large ([%d]) expect [%d]',
$info->{flowChunkNumber},
$info->{flowCurrentChunkSize},
$self->expectedChunkSize( $info ),
;
} else {
# Everything is OK with the chunk size and file size, I guess.
};
@invalid
};
=head2 C<< $flowJs->expectedChunkSize >>
my $expectedSize = $flowJs->expectedChunkSize( $info, $chunkIndex );
Returns the file size we expect for the chunk C<$chunkIndex>. The index
starts at 1, if it is not passed in or zero, we assume it is for the current
chunk as indicated by C<$info>.
=cut
sub expectedChunkSize( $self, $info, $index=0 ) {
# If we are not the last chunk, we need to be what the information says:
$index ||= $info->{flowChunkNumber};
if( ! $info->{flowTotalChunks}) {
# Some kind of invalid request, it'll be zero
return 0
} elsif( $index != $info->{flowTotalChunks}) {
return $info->{flowChunkSize}
} elsif( ! $info->{flowChunkSize} ) {
# No size, we guess it'll be zero:
return 0
} elsif( ! $info->{flowTotalSize} ) {
# Total size is zero
return 0;
} else {
# The last chunk can be smaller or sized just like all the chunks
# if the file size happens to be divided by the chunk size
if( $info->{flowTotalSize} % $info->{flowChunkSize}) {
return $info->{flowTotalSize} % $info->{flowChunkSize}
} else {
return $info->{flowChunkSize}
};
}
}
=head2 C<< $flowjs->resetUploadDirectories >>
if( $firstrun or $wipe ) {
$flowJs->resetUploadDirectories( $wipe )
};
Creates the directory for incoming uploads. If C<$wipe>
is passed, it will remove all partial files from the directory.
=cut
sub resetUploadDirectories( $self, $wipe=undef ) {
my $dir = $self->{incomingDirectory};
if( ! -d $dir ) {
mkdir $dir
or return $!;
};
if( $wipe ) {
unlink glob( $dir . "/*.part" );
};
}
=head2 C<< $flowjs->chunkName >>
my $target = $flowjs->chunkName( $info, $sessionid );
Returns the local filename of the chunk described by C<$info> and
the C<$sessionid> if given. An optional index can be passed in as
the third parameter to get the filename of another chunk than
the current chunk.
my $target = $flowjs->chunkName( $info, $sessionid, 1 );
# First chunk
=cut
sub chunkName( $self, $info, $sessionPrefix=undef, $index=0 ) {
my $dir = $self->{incomingDirectory};
$sessionPrefix = '' unless defined $sessionPrefix;
my $chunkname = sprintf "%s/%s%s.part%03d",
$dir,
$sessionPrefix,
clean_fragment($info->{ flowIdentifier }),
$index || $info->{ flowChunkNumber },
;
$chunkname
}
=head2 C<< $flowjs->chunkOK >>
my( $status, @messages ) = $flowjs->chunkOK( $info, $sessionPrefix );
if( $status == 500 ) {
warn $_ for @messages;
return [ 500, [], [] ]
} elsif( $status == 200 ) {
# That chunk exists and has the right size
return [ 200, [], [] ]
} else {
# That chunk does not exist and should be uploaded
return [ 416, [],[] ]
}
=cut
sub chunkOK($self, $info, $sessionPrefix=undef, $index=0) {
my @messages = $self->validateRequest( 'GET', $info, $sessionPrefix );
if( @messages ) {
return 500, @messages
};
my $chunkname = $self->chunkName( $info, $sessionPrefix, $index );
my $exists = -f $chunkname && -s $chunkname == $self->expectedChunkSize( $info, $index );
if( $exists ) {
return 200
} else {
return 416
}
}
=head2 C<< $flowjs->uploadComplete( $info, $sessionPrefix=undef ) >>
if( $flowjs->uploadComplete($info, $sessionPrefix) ) {
# do something with the chunks
}
=cut
sub uploadComplete( $self, $info, $sessionPrefix=undef ) {
my $complete = 1;
for( 1.. $info->{ flowTotalChunks }) {
my( $status, @messages ) = $self->chunkOK( $info, $sessionPrefix, $_ ) ;
$complete = $complete && $status == 200 && !@messages;
if( ! $complete ) {
# No need to check the rest
last;
};
};
!!$complete
}
=head2 C<< $flowjs->chunkFh >>
my $fh = $flowjs->chunkFh( $info, $sessionid, $index );
Returns an opened filehandle to the chunk described by C<$info>. The session
and the index are optional.
=cut
sub chunkFh( $self, $info, $sessionPrefix=undef, $index=0 ) {
my %info = %$info;
$info{ chunkNumber } = $index if $index;
my $chunkname = $self->chunkName( \%info, $sessionPrefix, $index );
open my $chunk, '<', $chunkname
or croak "Can't open chunk '$chunkname': $!";
binmode $chunk;
$chunk
}
=head2 C<< $flowjs->chunkContent >>
my $content = $flowjs->chunkContent( $info, $sessionid, $index );
Returns the content of a chunk described by C<$info>. The session
and the index are optional.
=cut
sub chunkContent( $self, $info, $sessionPrefix=undef, $index=0 ) {
my $chunk = $self->chunkFh( $info, $sessionPrefix, $index );
local $/; # / placate Filter::Simple
<$chunk>
}
=head2 C<< $flowjs->disallowedContentType( $info, $session ) >>
if( $flowjs->disallowedContentType( $info, $session )) {
return 415, "This type of file is not allowed";
};
Checks that the subroutine validator passed in the constructor allows
this MIME type. Unrecognized files will be blocked.
=cut
sub disallowedContentType( $self, $info, $session=undef ) {
my( $content_type, $image_ext ) = $self->sniffContentType($info,$session);
if( !defined $content_type ) {
# we need more chunks uploaded to check the content type
return
} elsif( $content_type eq '' ) {
# We couldn't determine what the content type is?!
return 1
} elsif( !$self->{allowedContentType}->( $content_type )) {
return $content_type || 1
} else {
return
};
};
=head2 C<< $flowjs->sniffContentType( $info, $session ) >>
my( $content_type, $image_ext ) = $flowjs->sniffContentType( $info, $session );
if( !defined $content_type ) {
# we need more chunks uploaded to check the content type
} elsif( $content_type eq '' ) {
# We couldn't determine what the content type is?!
return 415, "This type of upload is not allowed";
} elsif( $content_type !~ m!^image/(jpeg|png|gif)$!i ) {
return 415, "This type of upload is not allowed";
} else {
# We allow this upload to continue, as it seems to have
# an appropriate content type
};
This allows for finer-grained checking of the MIME-type. See also
the C<allowedContentType> argument in the constructor and
L<< ->disallowedContentType >> for a more convenient way to quickly
check the upload type.
=cut
sub sniffContentType( $self, $info, $sessionPrefix=undef ) {
my( $content_type, $image_ext );
my( $status, @messages ) = $self->chunkOK( $info, $sessionPrefix, 1 );
if( 200 == $status ) {
my $fh = $self->chunkFh( $info, $sessionPrefix, 1 );
my $t = $self->mime->mime_type($fh);
if( $t ) {
$content_type = $t->mime_type;
$image_ext = $t->extension;
} else {
$content_type = '';
$image_ext = '';
};
} else {
# Chunk 1 not uploaded/complete yet
}
return $content_type, $image_ext;
};
=head2 C<< $flowjs->combineChunks $info, $sessionPrefix, $target_fh, $digest ) >>
if( not $flowjs->uploadComplete($info, $sessionPrefix) ) {
print "Upload not yet completed\n";
return;
};
open my $file, '>', 'user_upload.jpg'
or die "Couldn't create final file 'user_upload.jpg': $!";
binmode $file;
my $digest = Digest::SHA256->new();
my($ok,@unlink) = $flowjs->combineChunks( $info, undef, $file, $digest );
close $file;
if( $ok ) {
print "Received upload OK, removing temporary upload files\n";
unlink @unlink;
print "Checksum: " . $digest->md5hex;
} else {
# whoops
print "Received upload failed, removing target file\n";
unlink 'user_upload.jpg';
};
=cut
sub combineChunks( $self, $info, $sessionPrefix, $target_fh, $digest=undef ) {
my @unlink_chunks;
my $ok = 1;
for( 1.. $info->{ flowTotalChunks }) {
my $chunkname = $self->chunkName( $info, $sessionPrefix, $_ );
push @unlink_chunks, $chunkname;
my $content = $self->chunkContent( $info, $sessionPrefix, $_ );
$digest->add( $content )
if $digest;
print { $target_fh } $content;
};
return $ok, @unlink_chunks
}
=head2 C<< $flowjs->pendingUploads >>
my $uploading = $flowjs->pendingUploads();
In scalar context, returns the number of pending uploads. In list context,
returns the list of filenames that belong to the pending uploads. This list
can be larger than the number of pending uploads, as one upload can have more
than one chunk.
=cut
sub pendingUploads( $self ) {
my @files;
my %uploads;
my $incoming = $self->incomingDirectory;
opendir my $dir, $incoming
or croak sprintf "Couldn't read incoming directory '%s': %s",
$self->incomingDirectory, $!;
@files = sort
map {
(my $upload = $_) =~ s!\.part\d+$!!;
$uploads{ $upload }++;
$_
}
grep { -f }
map {
"$incoming/$_"
} readdir $dir;
wantarray ? @files : scalar keys %uploads;
}
=head2 C<< $flowjs->staleUploads( $timeout, $now ) >>
my @stale_files = $flowjs->staleUploads(3600);
In scalar context, returns the number of stale uploads. In list context,
returns the list of filenames that belong to the stale uploads.
An upload is considered stale if no part of it has been written to since
C<$timeout> seconds ago.
The optional C<$timeout> parameter is the minimum age of an incomplete upload
before it is considered stale.
The optional C<$now> parameter is the point of reference for C<$timeout>.
It defaults to C<time>.
=cut
sub staleUploads( $self, $timeout = 3600, $now = time ) {
my $cutoff = $now - $timeout;
my %mtime;
my @files = reverse sort $self->pendingUploads();
for ( @files ) {
(my $upload = $_) =~ s!\.part\d+$!!;
if( ! exists $mtime{ $upload } or $mtime{ $upload } < $cutoff ) {
my @stat = stat( $_ );
# We want to remember the newest instance for this upload
$mtime{ $upload } ||= 0;
$mtime{ $upload } = $stat[9]
if $stat[9] > $mtime{ $upload };
#warn "$upload: $mtime{ $upload } ($stat[9])";
} else {
#warn "$upload has already younger known participant, is not stale";
};
};
my %stale;
@files = grep {
(my $upload = $_) =~ s!\.part\d+$!!;
if( exists $mtime{ $upload } and $mtime{ $upload } < $cutoff ) {
$stale{ $upload } = 1;
};
} @files;
wantarray ? @files : scalar keys %stale;
}
=head2 C<< $flowjs->purgeStaleOrInvalid( $timeout, $now ) >>
my @errors = $flowjs->purgeStaleOrInvalid();
Routine to delete all stale uploads and uploads with an invalid
file type.
This is mostly a helper routine to run from a cron job.
Note that if you allow uploads of multiple flowJs instances into the same
directory, they need to all have the same allowed file types or this method
will delete files from another instance.
=cut
sub purgeStaleOrInvalid($self, $timeout = 3600, $now = time ) {
# First, kill off all stale files
my @errors;
for my $f ($self->staleUploads( $timeout, $now )) {
unlink $f or push @errors, [$f => "$!"];
};
for my $f ($self->pendingUploads()) {
# Hmm - here we need to synthesize session info from a filename
# not really easy, isn't it?!
};
@errors
};
1;
=head1 REPOSITORY
The public repository of this module is
L<https://github.com/Corion/HTTP-Upload-FlowJs>.
=head1 SUPPORT
The public support forum of this module is
L<https://perlmonks.org/>.
=head1 BUG TRACKER
Please report bugs in this module via the RT CPAN bug queue at
L<https://rt.cpan.org/Public/Dist/Display.html?Name=HTTP-Upload-FlowJs>
or via mail to L<bug-http-upload-flowjs@rt.cpan.org>.
=head1 AUTHOR
Max Maischein C<corion@cpan.org>
=head1 COPYRIGHT (c)
Copyright 2009-2018 by Max Maischein C<corion@cpan.org>.
=head1 LICENSE
This module is released under the same terms as Perl itself.
=cut