try to make ftp tests more reliable

eva_submission/vep_utils.py

@@ -32,7 +32,6 @@
 ]
 
 
-
 def vep_path(version):
     return os.path.join(cfg['vep_path'], f'ensembl-vep-release-{version}/vep')
 
@@ -110,19 +109,19 @@ def get_vep_cache_version_from_ftp(assembly_accession, ensembl_assembly_name=Non
     # otherwise we just search the most recent release.
 
     # First try main Ensembl release
-    ftp = get_ftp_connection(ensembl_ftp_url)
-    all_releases = get_releases(ftp, '/pub', current_release_only)
-    release = search_releases(ftp, all_releases, species_name, assembly_name, taxonomy_id)
-    if release:
-        return release, 'ensembl'
+    with get_ftp_connection(ensembl_ftp_url) as ftp:
+        all_releases = get_releases(ftp, '/pub', current_release_only)
+        release = search_releases(ftp, all_releases, species_name, assembly_name, taxonomy_id)
+        if release:
+            return release, 'ensembl'
 
     # Then try all Ensembl genomes
-    genome_ftp = get_ftp_connection(ensembl_genome_ftp_url)
-    for subdir in ensembl_genome_dirs:
-        genome_releases = get_releases(genome_ftp, subdir, current_release_only)
-        release = search_releases(genome_ftp, genome_releases, species_name, assembly_name, taxonomy_id)
-        if release:
-            return release, 'genomes'
+    with get_ftp_connection(ensembl_genome_ftp_url) as genome_ftp:
+        for subdir in ensembl_genome_dirs:
+            genome_releases = get_releases(genome_ftp, subdir, current_release_only)
+            release = search_releases(genome_ftp, genome_releases, species_name, assembly_name, taxonomy_id)
+            if release:
+                return release, 'genomes'
 
     logger.info(f'No VEP cache found anywhere on Ensembl FTP for {species_name} and {assembly_name}')
     return None, None

tests/test_vep_utils.py

@@ -122,11 +122,12 @@ def test_get_vep_versions(self):
     def test_download_and_extract_vep_cache(self):
         with patch('eva_submission.vep_utils.retrieve_species_scientific_name_from_tax_id_ncbi') as m_get_scf_name:
             m_get_scf_name.return_value = 'whatever_species_name'
-            download_and_extract_vep_cache(
-                get_ftp_connection('ftp.ensembl.org'),
-                '/pub/release-105/variation/indexed_vep_cache/papio_anubis_refseq_vep_105_Panubis1.0.tar.gz', 1001
-            )
-            assert os.path.exists(os.path.join(cfg['vep_cache_path'], 'whatever_species_name', '105_Panubis1.0'))
+            with get_ftp_connection('ftp.ensembl.org') as ftp_conn:
+                download_and_extract_vep_cache(
+                    ftp_conn,
+                    '/pub/release-105/variation/indexed_vep_cache/papio_anubis_refseq_vep_105_Panubis1.0.tar.gz', 1001
+                )
+                assert os.path.exists(os.path.join(cfg['vep_cache_path'], 'whatever_species_name', '105_Panubis1.0'))
 
     def test_get_species_and_assembly(self):
         assemblies2results = {