Merge pull request #66 from andrewjpage/sq_exceeds_80

Sq exceeds 80

Dockerfile

@@ -21,6 +21,6 @@ RUN apt-get update -qq && apt-get install -y python-genometools genometools wget
 #
 # Download build and install gff3toembl python
 #
-RUN wget https://github.com/sanger-pathogens/gff3toembl/archive/v1.1.0.tar.gz && mv v1.1.0.tar.gz /opt && cd /opt && tar xzf v1.1.0.tar.gz
+RUN wget https://github.com/sanger-pathogens/gff3toembl/archive/v1.1.1.tar.gz && mv v1.1.1.tar.gz /opt && cd /opt && tar xzf v1.1.1.tar.gz
 
-RUN cd /opt && tar xzf v1.1.0.tar.gz && rm v1.1.0.tar.gz && cd /opt/gff3toembl-1.1.0 && python setup.py install
+RUN cd /opt && tar xzf v1.1.1.tar.gz && rm v1.1.1.tar.gz && cd /opt/gff3toembl-1.1.1 && python setup.py install

gff3toembl/EMBLContig.py

@@ -20,7 +20,8 @@ def format(self):
     except AttributeError:
       raise ValueError("Could not format contig, no sequence data found")
     formatted_string = header + features + sequence
-    line_lengths = map(len, formatted_string.split('\n'))
+    header_features_formatted_string = header + features
+    line_lengths = map(len, header_features_formatted_string.split('\n'))
     maximum_line_length = max(line_lengths)
     if maximum_line_length > 80:
       raise ValueError("Could not format contig, a line exceeded 80 characters in length")
@@ -409,6 +410,7 @@ def calculate_nucleotide_counts(self, sequence):
     return counts
 
   def format_header(self, nucleotide_counts):
+    # This line can exceed 80 characters
     template = "SQ   Sequence {total} BP; {a} A; {c} C; {g} G; {t} T; {other} other;"
     total_counts = sum(nucleotide_counts.values())
     nucleotide_counts['total'] = total_counts

gff3toembl/tests/EMBLContig_test.py

@@ -47,17 +47,6 @@ def test_format(self):
 """
     self.assertEqual(calculated_string, expected_string)
 
-  def test_format_line_too_long(self):
-    contig = EMBLContig()
-    header_mock = MagicMock()
-    sequence_mock = MagicMock()
-    header_mock.format.return_value = "Header"
-    sequence_mock.format.return_value = "Very long line: " + '*'*100
-    contig.header = header_mock
-    contig.features = {}
-    contig.sequence = sequence_mock
-    self.assertRaises(ValueError, contig.format)
-
   def test_add_feature(self):
     contig = EMBLContig()
     contig.add_feature(

setup.py

@@ -8,7 +8,7 @@ def read(fname):
 
 setup(
     name='gff3toembl',
-    version='1.1.0',
+    version='1.1.1',
     description='Convert a GFF3 file to EMBL format for submission',
     long_description=read('README.md'),
     packages = find_packages(),