From 3aea631831fd00254db1aa32c262dee61001f360 Mon Sep 17 00:00:00 2001
From: Andrew Yates <ayates@ebi.ac.uk>
Date: Wed, 21 Sep 2011 10:53:58 +0000
Subject: [PATCH] Modifications to ensure that user input is not
mis-interpreted as wanting whole chromosomes.
---
modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm | 7 ++++++-
modules/t/sliceAdaptor.t | 10 +++++++++-
2 files changed, 15 insertions(+), 2 deletions(-)
diff --git a/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm b/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
index 07eb2491e8..701d7cbed5 100644
--- a/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
+++ b/modules/Bio/EnsEMBL/DBSQL/SliceAdaptor.pm
@@ -439,11 +439,15 @@ sub fetch_by_region {
=head2 fetch_by_toplevel_location
Arg [1] : string $location
+ Ensembl formatted location. Can be a format like
+ C<name:start-end>, C<name:start..end>, C<name:start> and
+ C<name>.
Example : my $slice = $sa->fetch_by_toplevel_location('X:1-10000')
Description : Converts an Ensembl location/region into the sequence region
name, start and end and passes them onto C<fetch_by_region()>.
The code assumes that the required slice is on the top level
- coordinate system.
+ coordinate system. The code assumes that location formatting
+ is not perfect and will perform basic cleanup before parsing.
Returntype : Bio::EnsEMBL::Slice
Exceptions : If $location is false otherwise see C<fetch_by_region()>
Caller : General
@@ -455,6 +459,7 @@ sub fetch_by_toplevel_location {
my ($self, $location) = @_;
throw 'You must specify a location' if ! $location;
my $regex = qr/^(\w+) :? (\d+)? (?:-|[.]{2})? (\d+)?/xms;
+ $location =~ s/\s+|,//g;
my ($seq_region_name, $start, $end) = $location =~ $regex;
my $coord_system_name = 'toplevel';
return $self->fetch_by_region($coord_system_name, $seq_region_name, $start, $end, undef, undef, 0);
diff --git a/modules/t/sliceAdaptor.t b/modules/t/sliceAdaptor.t
index fe627ea37f..bd97d8315e 100644
--- a/modules/t/sliceAdaptor.t
+++ b/modules/t/sliceAdaptor.t
@@ -1,7 +1,7 @@
use strict;
use warnings;
-use Test::More tests => 129;
+use Test::More tests => 159;
use Bio::EnsEMBL::Test::MultiTestDB;
use Bio::EnsEMBL::DBSQL::SliceAdaptor;
@@ -458,6 +458,14 @@ test_toplevel_location('1:..10', 'chromosome', '1', 1, 10);
test_toplevel_location('1:100', 'chromosome', '1', 100, 246874334);
test_toplevel_location('1:', 'chromosome', '1', 1, 246874334);
test_toplevel_location('1', 'chromosome', '1', 1, 246874334);
+
+test_toplevel_location('1: 1-1,000', 'chromosome', '1', 1, 1000);
+test_toplevel_location('1: 1-1,000,000', 'chromosome', '1', 1, 1000000);
+test_toplevel_location('1: 1-1 000 000', 'chromosome', '1', 1, 1000000);
+test_toplevel_location('1: 1', 'chromosome', '1', 1, 246874334);
+test_toplevel_location('1: -10', 'chromosome', '1', 1, 10);
+test_toplevel_location('1: 100', 'chromosome', '1', 100, 246874334);
+
dies_ok { $slice_adaptor->fetch_by_toplevel_location(); } 'Checking calling without a location fails';
dies_ok { $slice_adaptor->fetch_by_toplevel_location(''); } 'Checking calling with a blank location fails';
ok(!defined $slice_adaptor->fetch_by_toplevel_location('wibble'), 'Checking with a bogus region returns undef');
--
GitLab