SQUASH: Added test .tsv files added a test method

qiime2 · Aug 9, 2019 · 608dfc6 · 608dfc6
1 parent 51cbb8f
commit 608dfc6
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 7 deletions.
diff --git a/q2_types/feature_data/_format.py b/q2_types/feature_data/_format.py
@@ -105,7 +105,7 @@ def _check_n_records(self, n=None):
             file_ = enumerate(fh) if n is None else zip(range(n), fh)
 
             for i, line in file_:
-                # Checks rows in the file, excludes header row
+                # Tracks line count for error reporting
                 i = i + 1
                 if line == '':
                     # EOF
@@ -122,15 +122,17 @@ def _check_n_records(self, n=None):
                                               "must be the first two header "
                                               "values to be valid.\n\n The "
                                               "first two header values "
-                                              "provided are: {}."
-                                              .format(cells[:2]))
+                                              "provided are: {}.\nIssue on "
+                                              "line {}"
+                                              .format(cells[:2], i))
                     header = cells
                 else:
                     if len(cells) != len(header):
-                        raise ValidationError('Number of columns are not the '
-                                              'same as number of headers in '
-                                              'the file. \nHeader values: '
-                                              '{} \nColumn values: {} '
+                        raise ValidationError("Number of columns are not the "
+                                              "same as number of headers in "
+                                              "the file. \nHeader values: "
+                                              "{} \nColumn values: {}\nIssue "
+                                              "on line: {}"
                                               .format(header, cells[:], i))
 
                     data_lines += 1

diff --git a/q2_types/feature_data/tests/data/taxonomy/greater-column-length.tsv b/q2_types/feature_data/tests/data/taxonomy/greater-column-length.tsv
@@ -0,0 +1,3 @@
+Feature ID	Taxon
+seq1	k__Bacteria; p__Proteobacteria	-1.0
+seq2	k__Bacteria	1.0
diff --git a/q2_types/feature_data/tests/data/taxonomy/greater-header-length.tsv b/q2_types/feature_data/tests/data/taxonomy/greater-header-length.tsv
@@ -0,0 +1,3 @@
+Feature ID	Taxon	Confidence	Random
+seq1	k__Foo; p__Bar	-1.0
+seq2	k__Foo; p__Baz	-42.0
diff --git a/q2_types/feature_data/tests/test_format.py b/q2_types/feature_data/tests/test_format.py
@@ -134,6 +134,18 @@ def test_tsv_taxonomy_directory_format(self):
 
         format.validate()
 
+    def test_tsv_taxonomy_format_column_header_lengths(self):
+        filenames = ['greater-column-length.tsv', 'greater-header-length.tsv']
+
+        filepaths = [self.get_data_path(os.path.join('taxonomy', filename))
+                     for filename in filenames]
+
+        for filepath in filepaths:
+            format = TSVTaxonomyFormat(filepath, mode='r')
+
+            with self.assertRaisesRegex(ValidationError, 'Number of columns'):
+                format.validate()
+
 
 class TestDNAFASTAFormats(TestPluginBase):
     package = 'q2_types.feature_data.tests'