Merge pull request #264 from emedgene/master

jamescasbon · Feb 5, 2017 · 0711a91 · 0711a91
2 parents c15c89d + a2f4a44
commit 0711a91
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 4 deletions.
diff --git a/vcf/cparse.pyx b/vcf/cparse.pyx
@@ -1,8 +1,8 @@
 from model import _Call
 
-cdef _map(func, iterable, bad='.'):
+cdef _map(func, iterable, bad=['.', '']):
     '''``map``, but make bad values None.'''
-    return [func(x) if x != bad else None
+    return [func(x) if x not in bad else None
             for x in iterable]
 
 INTEGER = 'Integer'

diff --git a/vcf/parser.py b/vcf/parser.py
@@ -354,9 +354,9 @@ def _parse_metainfo(self):
         self.samples = fields[9:]
         self._sample_indexes = dict([(x,i) for (i,x) in enumerate(self.samples)])
 
-    def _map(self, func, iterable, bad='.'):
+    def _map(self, func, iterable, bad=['.', '']):
         '''``map``, but make bad values None.'''
-        return [func(x) if x != bad else None
+        return [func(x) if x not in bad else None
                 for x in iterable]
 
     def _parse_filter(self, filt_str):

diff --git a/vcf/test/bad-info-character.vcf b/vcf/test/bad-info-character.vcf
@@ -0,0 +1,14 @@
+##fileformat=VCFv4.1
+##INFO=<ID=EMPTY_1,Number=1,Type=Float,Description="A floating point value">
+##INFO=<ID=EMPTY_3,Number=3,Type=Float,Description="Floating point values">
+##INFO=<ID=EMPTY_N,Number=.,Type=Float,Description="Floating point values">
+##INFO=<ID=DOT_1,Number=1,Type=Character,Description="A character value">
+##INFO=<ID=DOT_3,Number=3,Type=Character,Description="Character values">
+##INFO=<ID=DOT_N,Number=.,Type=Character,Description="Character values">
+##INFO=<ID=NOTEMPTY_1,Number=1,Type=Float,Description="A floating point value">
+##INFO=<ID=NOTEMPTY_3,Number=3,Type=Float,Description="Floating point values">
+##INFO=<ID=NOTEMPTY_N,Number=.,Type=Float,Description="Floating point values">
+##INFO=<ID=FLAG,Number=0,Type=Flag,Description="HapMap2 membership">
+##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	Sample
+chr1	100	id1	G	A	.	.	FLAG;EMPTY_1=;EMPTY_3=;EMPTY_N=;DOT_1=.;DOT_3=.,.,.;DOT_N=.;NOTEMPTY_1=1;NOTEMPTY_3=1,2,3;NOTEMPTY_N=1	GT	0/1
diff --git a/vcf/test/test_vcf.py b/vcf/test/test_vcf.py
@@ -393,6 +393,24 @@ def test_write(self):
             self.assertEquals(l.INFO, r.INFO)
 
 
+class TestBadInfoFields(unittest.TestCase):
+    def test_parse(self):
+        reader = vcf.Reader(fh('bad-info-character.vcf'))
+        record = next(reader)
+        self.assertEquals(record.INFO['DOT_1'], None)
+        self.assertEquals(record.INFO['DOT_3'], [None, None, None])
+        self.assertEquals(record.INFO['DOT_N'], [None])
+        self.assertEquals(record.INFO['EMPTY_1'], None)
+        # Perhaps EMPTY_3 should yield [None, None, None] but this is really a
+        # cornercase of unspecified behaviour.
+        self.assertEquals(record.INFO['EMPTY_3'], [None])
+        self.assertEquals(record.INFO['EMPTY_N'], [None])
+        self.assertEquals(record.INFO['NOTEMPTY_1'], 1)
+        self.assertEquals(record.INFO['NOTEMPTY_3'], [1, 2, 3])
+        self.assertEquals(record.INFO['NOTEMPTY_N'], [1])
+        pass
+
+
 class TestParseMetaLine(unittest.TestCase):
     def test_parse(self):
         reader = vcf.Reader(fh('parse-meta-line.vcf'))
@@ -1724,3 +1742,4 @@ def test_strelka(self):
 suite.addTests(unittest.TestLoader().loadTestsFromTestCase(TestGATKMeta))
 suite.addTests(unittest.TestLoader().loadTestsFromTestCase(TestUncalledGenotypes))
 suite.addTests(unittest.TestLoader().loadTestsFromTestCase(TestStrelka))
+suite.addTests(unittest.TestLoader().loadTestsFromTestCase(TestBadInfoFields))