MAINT: adjust regex for genome data semantic types (#322)

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

@@ -54,77 +54,29 @@ def test_seed_ortholog_dir_fmt_collection(self):
             obj.validate()
 
     def test_genes_dirfmt_fa_with_suffix(self):
-        dirpath = self.get_data_path('genes-with-suffix')
+        dirpath = self.get_data_path('genes')
         fmt = GenesDirectoryFormat(dirpath, mode='r')
 
         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')
+        dirpath = self.get_data_path('proteins')
         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')
+        filepath = self.get_data_path('loci/loci1.gff')
         fmt = GFF3Format(filepath, mode='r')
 
         fmt.validate()
 
     def test_loci_dirfmt_with_suffix(self):
-        dirpath = self.get_data_path('loci-with-suffix')
-        fmt = LociDirectoryFormat(dirpath, mode='r')
-
-        fmt.validate()
-
-    def test_loci_dirfmt_with_prefix(self):
-        dirpath = self.get_data_path('loci-with-prefix')
+        dirpath = self.get_data_path('loci')
         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(

q2_types/genome_data/tests/test_transformer.py

@@ -64,8 +64,8 @@ def seqs_to_df(seqs):
     def test_genes_to_dataframe(self):
         _, obs = self.transform_format(GenesDirectoryFormat, pd.DataFrame,
                                        filenames=[
-                                           'genes-with-suffix/genes1.fa',
-                                           'genes-with-suffix/genes2.fa'
+                                           'genes/genes1.fa',
+                                           'genes/genes2.fa'
                                        ])
         exp = self.seqs_to_df(self.genes)
         pd.testing.assert_frame_equal(exp, obs)
@@ -82,8 +82,8 @@ def test_proteins_to_dataframe(self):
             ProteinsDirectoryFormat,
             pd.DataFrame,
             filenames=[
-                'proteins-with-suffix/proteins1.faa',
-                'proteins-with-suffix/proteins2.faa'
+                'proteins/proteins1.faa',
+                'proteins/proteins2.faa'
             ])
         exp = self.seqs_to_df(self.proteins)
         pd.testing.assert_frame_equal(exp, obs)
@@ -100,7 +100,7 @@ def test_gff_to_interval_metadata_iterator(self):
         input, obs = self.transform_format(
             GFF3Format,
             IntervalMetadataIterator,
-            filename='loci-with-suffix/loci1.gff')
+            filename='loci/loci1.gff')
         exp = skbio.io.read(str(input), format='gff3')
 
         for o, e in zip(obs, exp):
@@ -109,7 +109,7 @@ def test_gff_to_interval_metadata_iterator(self):
     def test_interval_metadata_iterator_to_gff(self):
         transformer = self.get_transformer(IntervalMetadataIterator,
                                            GFF3Format)
-        filepath = self.get_data_path('loci-with-suffix/loci1.gff')
+        filepath = self.get_data_path('loci/loci1.gff')
         generator = skbio.io.read(filepath, format='gff3')
         input = IntervalMetadataIterator(generator)
 

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/*',
+             'data/loci-invalid/*',
+             'data/loci/*',
+             'data/ortholog/*',
+             'data/proteins/*',
              ],
         'q2_types.kraken2.tests': [
             'data/*',