use ordered dict for samples, fixes #2

setup.py

@@ -1,4 +1,12 @@
 from setuptools import setup
+import sys
+
+requires = []
+
+# install ordereddict for python < 2.7
+py_version = sys.version_info
+if py_version.major == 2 and py_version.minor < 7:
+    requires.append('ordereddict')
 
 setup(
     name='PyVCF',
@@ -7,5 +15,6 @@
     author='James Casbon',
     author_email='[email protected]',
     description='Variant Call Format (VCF) parser for python',
-    test_suite='test'
+    test_suite='test',
+    requires=requires
 )

test/test_vcf.py

@@ -40,6 +40,9 @@ def testCalls(self):
             n += 1
             self.assertEqual(len(site.samples), len(self.samples))
 
+            # check sample ordering is preserved:
+            self.assertEqual([x['name'] for x in site.samples.values()], self.samples)
+
         self.assertEqual(n,  self.n_calls)
 
 

vcf.py

@@ -93,6 +93,12 @@
 import re
 import csv
 
+try:
+    from ordereddict import OrderedDict
+except ImportError:
+    from collections import OrderedDict
+
+
 
 # Metadata parsers/constants
 RESERVED_INFO = {
@@ -387,9 +393,9 @@ def _parse_info(self, info_str):
     def _parse_samples(self, samples, samp_fmt):
         '''Parse a sample entry according to the format specified in the FORMAT
         column.'''
-        samp_data = []
+        samp_data = OrderedDict()
         samp_fmt = samp_fmt.split(':')
-        for sample in samples:
+        for name, sample in zip(self.samples, samples):
             sampdict = dict(zip(samp_fmt, sample.split(':')))
             for fmt in sampdict:
                 vals = sampdict[fmt].split(',')
@@ -408,14 +414,8 @@ def _parse_samples(self, samples, samp_fmt):
                 elif sampdict[fmt] == './.' and self.aggro:
                     sampdict[fmt] = None
 
-            samp_data.append(sampdict)
-
-        for name, data in zip(self.samples, samp_data):
-            data['name'] = name
-            try:
-                data['_GT'] = int(data['GT'][0]), int(data['GT'][-1])
-            except ValueError:
-                pass
+            sampdict['name'] = name
+            samp_data[name] = sampdict
 
         return samp_data