Merge 608dfc6 into cf45830

q2_types/feature_data/_format.py

@@ -90,38 +90,59 @@ class TSVTaxonomyFormat(model.TextFileFormat):
 
     Optionally followed by other arbitrary columns.
 
-    This format supports comment lines starting with #, and blank lines. The
-    expected header must be the first non-comment, non-blank line. In addition
-    to the header, there must be at least one line of data.
+    This format supports blank lines. The expected header must be the first
+    non-blank line. In addition to the header, there must be at least one line
+    of data.
 
     """
     HEADER = ['Feature ID', 'Taxon']
 
-    def sniff(self):
+    def _check_n_records(self, n=None):
         with self.open() as fh:
             data_lines = 0
             header = None
-            while data_lines < 10:
-                line = fh.readline()
 
+            file_ = enumerate(fh) if n is None else zip(range(n), fh)
+
+            for i, line in file_:
+                # Tracks line count for error reporting
+                i = i + 1
                 if line == '':
                     # EOF
                     break
                 elif line.lstrip(' ') == '\n':
                     # Blank line
                     continue
 
-                cells = line.rstrip('\n').split('\t')
+                cells = line.strip('\n').split('\t')
+
                 if header is None:
                     if cells[:2] != self.HEADER:
-                        return False
+                        raise ValidationError("['Feature ID' and 'Taxon'] "
+                                              "must be the first two header "
+                                              "values to be valid.\n\n The "
+                                              "first two header values "
+                                              "provided are: {}.\nIssue on "
+                                              "line {}"
+                                              .format(cells[:2], i))
                     header = cells
                 else:
                     if len(cells) != len(header):
-                        return False
+                        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
 
-            return header is not None and data_lines > 0
+            if data_lines == 0:
+                raise ValidationError("No feature records found, only blank "
+                                      "lines and/or a header row.")
+
+    def _validate_(self, level):
+        self._check_n_records(n={'min': 10, 'max': None}[level])
 
 
 TSVTaxonomyDirectoryFormat = model.SingleFileDirectoryFormat(

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

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

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'