MAINT: adjust regex for genome data semantic types

q2_types/genome_data/_format.py

@@ -19,21 +19,21 @@ def _validate_(self, level):
 
 
 class GenesDirectoryFormat(model.DirectoryFormat):
-    genes = model.FileCollection(r'(.*\_)?genes[0-9]*\.(fa|fna|fasta)$',
+    genes = model.FileCollection(r'.+\.(fa|fna|fasta)$',
                                  format=DNAFASTAFormat)
 
     @genes.set_path_maker
     def genes_path_maker(self, genome_id):
-        return '%s_genes.fasta' % genome_id
+        return '%s.fasta' % genome_id
 
 
 class ProteinsDirectoryFormat(model.DirectoryFormat):
-    proteins = model.FileCollection(r'(.*\_)?proteins[0-9]*\.(fa|faa|fasta)$',
+    proteins = model.FileCollection(r'.+\.(fa|faa|fasta)$',
                                     format=ProteinFASTAFormat)
 
     @proteins.set_path_maker
     def proteins_path_maker(self, genome_id):
-        return '%s_proteins.fasta' % genome_id
+        return '%s.fasta' % genome_id
 
 
 class GFF3Format(model.TextFileFormat):
@@ -160,12 +160,12 @@ def _validate_(self, level):
 
 
 class LociDirectoryFormat(model.DirectoryFormat):
-    loci = model.FileCollection(r'(.*\_)?loci[0-9]*\.gff$',
+    loci = model.FileCollection(r'.+\.gff$',
                                 format=GFF3Format)
 
     @loci.set_path_maker
     def loci_path_maker(self, genome_id):
-        return '%s_loci.gff' % genome_id
+        return '%s.gff' % genome_id
 
 
 plugin.register_formats(

q2_types/genome_data/tests/test_format.py

@@ -59,44 +59,12 @@ def test_genes_dirfmt_fa_with_suffix(self):
 
         fmt.validate()
 
-    def test_genes_dirfmt_fa_with_prefix(self):
-        dirpath = self.get_data_path('genes-with-prefix')
-        fmt = GenesDirectoryFormat(dirpath, mode='r')
-
-        fmt.validate()
-
-    def test_genes_dirfmt_fa_with_wrong_prefix(self):
-        dirpath = self.get_data_path('genes-with-wrong-prefix')
-        fmt = GenesDirectoryFormat(dirpath, mode='r')
-
-        with self.assertRaisesRegex(
-                ValidationError,
-                'Missing one or more files for GenesDirectoryFormat'
-        ):
-            fmt.validate()
-
     def test_proteins_dirfmt_fa_with_suffix(self):
         dirpath = self.get_data_path('proteins-with-suffix')
         fmt = ProteinsDirectoryFormat(dirpath, mode='r')
 
         fmt.validate()
 
-    def test_proteins_dirfmt_fa_with_prefix(self):
-        dirpath = self.get_data_path('proteins-with-prefix')
-        fmt = ProteinsDirectoryFormat(dirpath, mode='r')
-
-        fmt.validate()
-
-    def test_proteins_dirfmt_fa_with_wrong_prefix(self):
-        dirpath = self.get_data_path('proteins-with-wrong-prefix')
-        fmt = ProteinsDirectoryFormat(dirpath, mode='r')
-
-        with self.assertRaisesRegex(
-            ValidationError,
-            'Missing one or more files for ProteinsDirectoryFormat'
-        ):
-            fmt.validate()
-
     def test_gff_format_positive_with_suffix(self):
         filepath = self.get_data_path('loci-with-suffix/loci1.gff')
         fmt = GFF3Format(filepath, mode='r')
@@ -109,22 +77,6 @@ def test_loci_dirfmt_with_suffix(self):
 
         fmt.validate()
 
-    def test_loci_dirfmt_with_prefix(self):
-        dirpath = self.get_data_path('loci-with-prefix')
-        fmt = LociDirectoryFormat(dirpath, mode='r')
-
-        fmt.validate()
-
-    def test_loci_dirfmt_with_wrong_prefix(self):
-        dirpath = self.get_data_path('loci-with-wrong-prefix')
-        fmt = LociDirectoryFormat(dirpath, mode='r')
-
-        with self.assertRaisesRegex(
-                ValidationError,
-                'Missing one or more files for LociDirectoryFormat'
-        ):
-            fmt.validate()
-
     def test_gff_format_wrong_version(self):
         filepath = self.get_data_path('loci-invalid/loci-wrong-version.gff')
         with self.assertRaisesRegex(

setup.py

@@ -56,13 +56,12 @@
             ['data/*', 'data/*/*',
              'data/mags-fa/*', 'data/mags-fasta/*'],
         'q2_types.genome_data.tests':
-            ['data/*/', 'data/genes-with-prefix/*',
-             'data/genes-with-suffix/*', 'data/genes-with-wrong-prefix/*',
-             'data/loci-invalid/*',  'data/loci-with-prefix/*',
-             'data/loci-with-suffix/*', 'data/loci-with-wrong-prefix/*',
-             'data/ortholog/*', 'data/proteins-with-suffix/*',
-             'data/proteins-with-prefix/*',
-             'data/proteins-with-wrong-prefix/*',
+            ['data/*/',
+             'data/genes-with-suffix/*',
+             'data/loci-invalid/*',
+             'data/loci-with-suffix/*',
+             'data/ortholog/*',
+             'data/proteins-with-suffix/*',
              ],
         'q2_types.kraken2.tests': [
             'data/*',