From a37af77eb2a343f9af72846af573eeec8adf1ab8 Mon Sep 17 00:00:00 2001
From: "Joshua C. Randall" <jcrandall@alum.mit.edu>
Date: Wed, 29 Nov 2017 16:04:25 +0000
Subject: [PATCH] add test fixture to download example data

---
 gatkcwlgenerator/tests/test.py | 31 +++++++++++++++++++++++--------
 test_requirements.txt          |  4 +++-
 2 files changed, 26 insertions(+), 9 deletions(-)

diff --git a/gatkcwlgenerator/tests/test.py b/gatkcwlgenerator/tests/test.py
index 97b33a3..46bca99 100644
--- a/gatkcwlgenerator/tests/test.py
+++ b/gatkcwlgenerator/tests/test.py
@@ -13,6 +13,21 @@
 
 import requests
 
+"""
+Download example data to be used in CWL integration tests
+"""
+@pytest.fixture(scope="module")
+def example_data():
+    if not os.path.isfile("cwl-example-data/chr22_cwl_test.cram"):
+        from six.moves.urllib.request import urlopen
+        import tarfile
+        print("Downloading and extracting cwl-example-data")
+        tgz = urlopen("https://cwl-example-data.cog.sanger.ac.uk/chr22_cwl_test.tgz")
+        tar = tarfile.open(fileobj=tgz, mode="r|gz")
+        tar.extractall(path="cwl-example-data")
+        tar.close()
+        tgz.close()
+
 """
 Runs the specified command and reports it as an AssertionError if it fails (can override this with
 expect_failure)
@@ -127,22 +142,22 @@ def test_are_cwl_files_valid(self, version):
         if exceptions:
             raise AssertionError("Not all cwl files are valid:\n" + "\n\n".join(exceptions))
 
-    def test_haplotype_caller(self, version):
+    def test_haplotype_caller(self, version, example_data):
         run_command("cwl-runner gatk_cmdline_tools/{}/cwl/HaplotypeCaller.cwl examples/HaplotypeCaller_inputs.yml".format(version))
 
     # Test if the haplotype caller accepts all the correct types
 
-    def test_boolean_type(self, version):
+    def test_boolean_type(self, version, example_data):
         assert "ThreadEfficiencyMonitor" in run_haplotype_caller("monitorThreadEfficiency: True", version).stderr
 
-    def test_integers_type(self, version):
+    def test_integers_type(self, version, example_data):
         assert "42 data thread" in run_haplotype_caller("num_threads: 42", version, expect_failure=True).stderr
 
-    def test_string_type(self, version):
+    def test_string_type(self, version, example_data):
         assert "Specified name does not exist in input bam files" in \
             run_haplotype_caller("sample_name: invalid_sample_name", version, expect_failure=True).stderr
 
-    def test_file_type(self, version):
+    def test_file_type(self, version, example_data):
         BQSR_arg = """
 BQSR:
    class: File
@@ -151,11 +166,11 @@ def test_file_type(self, version):
         assert "Bad input: The GATK report has an unknown/unsupported version in the header" in \
             run_haplotype_caller(BQSR_arg, version, expect_failure=True).stderr
 
-    def test_enum_type(self, version):
+    def test_enum_type(self, version, example_data):
         assert "Strictness is LENIENT" in run_haplotype_caller("validation_strictness: LENIENT", version).stderr
 
-    def test_list_type(self, version):
+    def test_list_type(self, version, example_data):
         run_with_larger_intervals = run_haplotype_caller(extra_info="", version=version,
             filetext="analysis_type: HaplotypeCaller\n" + default_args + "\nintervals: [chr22:10591400-10591500, chr22:10591500-10591645]")
 
-        assert "Processing 246 bp from intervals" in run_with_larger_intervals.stderr
\ No newline at end of file
+        assert "Processing 246 bp from intervals" in run_with_larger_intervals.stderr
diff --git a/test_requirements.txt b/test_requirements.txt
index 9c4cd94..6781ea0 100644
--- a/test_requirements.txt
+++ b/test_requirements.txt
@@ -1,3 +1,5 @@
 pytest
 pytest-xdist
-requests_cache
\ No newline at end of file
+requests_cache
+six
+cwlref-runner