From 4918ced348318b285ce6b222242b59703b5d8486 Mon Sep 17 00:00:00 2001
From: Andrew Yates <ayates@ebi.ac.uk>
Date: Fri, 24 Feb 2012 09:54:33 +0000
Subject: [PATCH] Letting datafiles support more than just one file extension
---
modules/Bio/EnsEMBL/DBSQL/DataFileAdaptor.pm | 29 +++++++---
modules/Bio/EnsEMBL/DataFile.pm | 57 +++++++++++++++-----
2 files changed, 67 insertions(+), 19 deletions(-)
diff --git a/modules/Bio/EnsEMBL/DBSQL/DataFileAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/DataFileAdaptor.pm
index 626ff6da45..ac3dbfc98d 100644
--- a/modules/Bio/EnsEMBL/DBSQL/DataFileAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/DataFileAdaptor.pm
@@ -54,7 +54,7 @@ use base qw/Bio::EnsEMBL::DBSQL::BaseAdaptor/;
use Bio::EnsEMBL::DataFile;
use Bio::EnsEMBL::DBSQL::BaseAdaptor;
-use Bio::EnsEMBL::Utils::Exception qw/throw warning/;
+use Bio::EnsEMBL::Utils::Exception qw/throw warning deprecate/;
use Bio::EnsEMBL::Utils::Scalar qw/:assert/;
my $GLOBAL_BASE_PATH;
@@ -106,22 +106,37 @@ sub get_base_path {
Description : Returns an expected extension for the given DataFile type
Returntype : Scalar of the expected file extension
Exceptions : Raised if the given file type is not understood
+ Status : Deprecated
=cut
sub DataFile_to_extension {
+ my ($self, $df) = @_;
+ deprecate("Use DataFile_to_extensions() instead");
+ my $extensions = $self->DataFile_to_extensions($df);
+ return $extensions->[0];
+}
+
+=head2
+
+
+
+=cut
+
+sub DataFile_to_extensions {
my ($self, $df) = @_;
my $type = $df->file_type();
- my $ext = {
- BAM => 'bam',
+ my $extensions = {
+ BAM => ['bam', 'bam.bai'],
# BIGBED => 'bb',
- BIGWIG => 'bw',
- VCF => 'vcf',
+ BIGWIG => ['bw'],
+ VCF => ['vcf.gz', 'vcf.gz.tbi'],
}->{$type};
- throw sprintf(q{No extension found for the type '%s'}, $type ) if ! $ext;
- return $ext;
+ throw sprintf(q{No extensions found for the type '%s'}, $type ) if ! $extensions;
+ return $extensions;
}
+
=head2 DataFile_to_adaptor
Arg[1] : Bio::EnsEMBL::DataFile
diff --git a/modules/Bio/EnsEMBL/DataFile.pm b/modules/Bio/EnsEMBL/DataFile.pm
index 230277f57e..7a0d0407c7 100644
--- a/modules/Bio/EnsEMBL/DataFile.pm
+++ b/modules/Bio/EnsEMBL/DataFile.pm
@@ -108,8 +108,16 @@ sub get_ExternalAdaptor {
sub path {
my ($self, $base) = @_;
+ my $all_paths = $self->get_all_paths($base);
+ return $all_paths->[0];
+}
+
+sub get_all_paths {
+ my ($self, $base) = @_;
+
+ return [$self->url()] if $self->absolute();
- return $self->url() if $self->absolute();
+ my @all_paths;
$base = $self->adaptor()->get_base_path($base) if ! $base;
@@ -130,26 +138,33 @@ sub path {
push(@portions, software_version()) if $self->version_lock();
push(@portions, $self->adaptor()->db()->group());
+ #Targets are the files to generate
+ my @targets;
#If URL is populated we assume we need to add this onto the end but removing the /
if($self->url()) {
my @split = split(/\//, $self->url());
- push(@portions, @split);
+ push(@targets, [@split]);
}
else {
- my $ext = $self->adaptor()->DataFile_to_extension($self);
- my $filename = sprintf(q{%s.%s}, $self->name(), $ext);
- push(@portions, $filename);
+ my $extensions = $self->adaptor()->DataFile_to_extensions($self);
+ foreach my $ext (@{$extensions}) {
+ my $filename = sprintf(q{%s.%s}, $self->name(), $ext);
+ push(@targets, [$filename]);
+ }
}
my $is_uri = is_uri($base);
- my $path;
- if($is_uri) {
- $path = join(q{/}, $base, @portions);
+ foreach my $t (@targets) {
+ my $path;
+ if($is_uri) {
+ $path = join(q{/}, $base, @portions, @{$t});
+ }
+ else {
+ $path = File::Spec->catfile($base, @portions, @{$t});
+ }
+ push(@all_paths, $path);
}
- else {
- $path = File::Spec->catfile($base, @portions);
- }
- return $path;
+ return \@all_paths;
}
=head2 coord_system
@@ -273,4 +288,22 @@ sub file_type {
return $self->{'file_type'};
}
+#=head2 files
+#
+# Args :
+# Example : my $files = @{$df->files()};
+# Description : Returns all the file names we expect to cover for a flat file
+# Returntype : type return_description
+# Exceptions :
+# Caller : caller
+# Status : status
+#
+#=cut
+#
+#
+#sub files {
+# my ($self) = @_;
+#
+#}
+
1;
\ No newline at end of file
--
GitLab