Merge pull request #21 from phnmnl/feature/roundoff_comparison

A new roundoff comparison function for csv files

tests/comparators/rounded_comparison_csv/correlation_table.csv

@@ -0,0 +1,10 @@
+"spec","r1","r2","r3","r4"
+"Br",0.538284659385681,0.347380578517914,0.984475374221802,0.858644068241119
+"Cc",0.972778141498566,0.821589231491089,0.667640388011932,0.982909619808197
+"Ft",0.711512565612793,0.79552161693573,0.820532739162445,0.348022520542145
+"Gp",0.916094779968262,0.286471843719482,0.93779468536377,0.787193953990936
+"Hc",0.64589512348175,0.552187442779541,0.625308692455292,0.823695063591003
+"Mp",0.997475743293762,-0.0424592085182667,0.997138321399689,0.989661455154419
+"Pu",0.68191522359848,0.419826716184616,0.923761785030365,0.409022688865662
+"Po",0.900495171546936,0.641049265861511,0.985500574111938,0.607609212398529
+"Rs",0.977232158184052,0,0.677042424678802,0.854979515075684

tests/comparators/rounded_comparison_csv/diff_field.csv

@@ -0,0 +1,10 @@
+"spec","r1","r2","r3","r4"
+"Br",1.538284659385681,0.347380578517914,0.984475374221802,0.858644068241119
+"Cc",0.972778141498566,0.821589231491089,0.667640388011932,0.982909619808197
+"Ft",0.711512565612793,0.79552161693573,0.820532739162445,0.348022520542145
+"Gp",0.916094779968262,0.286471843719482,0.93779468536377,0.787193953990936
+"Hc",0.64589512348175,0.552187442779541,0.625308692455292,0.823695063591003
+"Mp",0.997475743293762,-0.0424592085182667,0.997138321399689,0.989661455154419
+"Pu",0.68191522359848,0.419826716184616,0.923761785030365,0.409022688865662
+"Po",0.900495171546936,0.641049265861511,0.985500574111938,0.607609212398529
+"Rs",0.977232158184052,0,0.677042424678802,0.854979515075684

tests/comparators/rounded_comparison_csv/diff_heading.csv

@@ -0,0 +1,10 @@
+"spec","Wrong","r2","r3","r4"
+"Br",0.538284659385681,0.347380578517914,0.984475374221802,0.858644068241119
+"Cc",0.972778141498566,0.821589231491089,0.667640388011932,0.982909619808197
+"Ft",0.711512565612793,0.79552161693573,0.820532739162445,0.348022520542145
+"Gp",0.916094779968262,0.286471843719482,0.93779468536377,0.787193953990936
+"Hc",0.64589512348175,0.552187442779541,0.625308692455292,0.823695063591003
+"Mp",0.997475743293762,-0.0424592085182667,0.997138321399689,0.989661455154419
+"Pu",0.68191522359848,0.419826716184616,0.923761785030365,0.409022688865662
+"Po",0.900495171546936,0.641049265861511,0.985500574111938,0.607609212398529
+"Rs",0.977232158184052,0,0.677042424678802,0.854979515075684

tests/comparators/rounded_comparison_csv/diff_under_round.csv

@@ -0,0 +1,10 @@
+"spec","r1","r2","r3","r4"
+"Br",0.538284659385689,0.347380578517914,0.984475374221802,0.858644068241119
+"Cc",0.972778141498566,0.821589231491089,0.667640388011932,0.982909619808197
+"Ft",0.711512565612793,0.79552161693573,0.820532739162445,0.348022520542145
+"Gp",0.916094779968262,0.286471843719482,0.93779468536377,0.787193953990936
+"Hc",0.64589512348175,0.552187442779541,0.625308692455292,0.823695063591003
+"Mp",0.997475743293762,-0.0424592085182667,0.997138321399689,0.989661455154419
+"Pu",0.68191522359848,0.419826716184616,0.923761785030365,0.409022688865662
+"Po",0.900495171546936,0.641049265861511,0.985500574111938,0.607609212398529
+"Rs",0.977232158184052,0,0.677042424678802,0.854979515075684

tests/comparators/rounded_comparison_csv/wrong_correlation.csv

@@ -0,0 +1,10 @@
+"spec","l1","r2","r3","r4"
+"Br",0.538284659385682,0.347380578517914,0.984475374221802,0.858644068241119
+"Cc",0.972778141498566,0.821589231491089,0.667640388011932,0.982909619808197
+"Ft",0.711512565612793,0.79552161693573,0.820532739162445,0.348022520542145
+"Gp",0.916094779968262,0.286471843719482,0.93779468536377,0.787193953990936
+"Hc",0.64589512348175,0.552187442779541,0.625308692455292,0.823695063591003
+"Mp",0.997475743293762,-0.0424592085182667,0.997138321399689,0.989661455154419
+"Pu",0.68191522359848,0.419826716184616,0.923761785030365,0.409022688865662
+"Po",0.900495171546936,0.641049265861511,0.985500574111938,0.607609212398529
+"Rs",0.977232158184052,0,0.677042424678802,0.854979515075684

tests/test_wft4galaxy.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python
+
+import os
+import sys
+import unittest
+
+MyDir = os.path.abspath(os.path.dirname(__file__))
+
+class TestWft4Galaxy(unittest.TestCase):
+    def setUp(self):
+        pass
+
+    def tearDown(self):
+        pass
+
+class TestRoundedCsvComparator(unittest.TestCase):
+
+    TestDir = os.path.join(MyDir, 'comparators', 'rounded_comparison_csv')
+    CorrelationFile = os.path.join(TestDir,  'correlation_table.csv')
+
+    def test_identical_files(self):
+        from wft4galaxy.comparators import rounded_comparison_csv
+        self.assertTrue(rounded_comparison_csv(self.CorrelationFile, self.CorrelationFile))
+
+    def test_diff_heading(self):
+        from wft4galaxy.comparators import rounded_comparison_csv
+        different_file = os.path.join(self.TestDir, 'diff_heading.csv')
+        self.assertFalse(rounded_comparison_csv(different_file, self.CorrelationFile))
+
+    def test_diff_field(self):
+        from wft4galaxy.comparators import rounded_comparison_csv
+        different_file = os.path.join(self.TestDir, 'diff_field.csv')
+        self.assertFalse(rounded_comparison_csv(different_file, self.CorrelationFile))
+
+    def test_diff_field_less_than_roundoff(self):
+        from wft4galaxy.comparators import rounded_comparison_csv
+        # A number in a field is different, but at a higher precision than required
+        # by comparison function
+        different_file = os.path.join(self.TestDir, 'diff_under_round.csv')
+        self.assertTrue(rounded_comparison_csv(different_file, self.CorrelationFile))
+
+
+
+def suite():
+    s = unittest.TestLoader().loadTestsFromTestCase(TestWft4Galaxy)
+    s.addTests(unittest.TestLoader().loadTestsFromTestCase(TestRoundedCsvComparator))
+    return s
+
+def main():
+    result = unittest.TextTestRunner(verbosity=2).run(suite())
+    return 0 if result.wasSuccessful() else 1
+
+if __name__ == '__main__':
+    sys.exit(main())

wft4galaxy/comparators/__init__.py

@@ -1,8 +1,11 @@
+
+from __future__ import print_function
+
 import os as _os
 import sys as _sys
 import logging as _logging
-from wft4galaxy import common as _common
 from difflib import unified_diff as _unified_diff
+from wft4galaxy import common as _common
 
 _logger = _common.LoggerManager.get_logger(__name__)
 
@@ -45,7 +48,6 @@ def base_comparator(actual_output_filename, expected_output_filename):
                 out_fp.writelines("%r\n" % item.rstrip('\n') for item in ldiff)
         return len(ldiff) == 0
 
-
 def csv_same_row_and_col_lengths(actual_output_filename, expected_output_filename):
     import csv
 
@@ -63,3 +65,42 @@ def csv_same_row_and_col_lengths(actual_output_filename, expected_output_filenam
         if colsActual and colsExpected:
             return cmp(colsActual, colsExpected) == 0
         return False
+
+def _get_float(s):
+    try:
+        return float(s)
+    except ValueError:
+        return None
+
+def _compare_strings_as_floats(precision, a, b):
+    a_float = _get_float(a)
+    b_float = _get_float(b)
+    if a_float is not None and b_float is not None:
+        return round(a_float, precision) == round(b_float, precision)
+    else:
+        return False
+
+def rounded_comparison_csv(actual_output, expected_output):
+    precision = 2
+    import csv
+    import itertools as it
+    try:
+        with open(actual_output) as actual, open(expected_output) as expected:
+            aout = csv.reader(actual)
+            eout = csv.reader(expected)
+            for expected_row in eout:
+                actual_row = next(aout)
+                for actual_field, expected_field in it.izip(actual_row, expected_row):
+                    if not _compare_strings_as_floats(precision, actual_field, expected_field) \
+                       and actual_field != expected_field:
+                           print("Difference found between expected and actual output", file=_sys.stderr)
+                           print("Expected field text:", expected_field, file=_sys.stderr)
+                           print("Actual field text:", actual_field, file=_sys.stderr)
+                           print("Expected row:", expected_row, file=_sys.stderr)
+                           print("Actual row:", actual_row, file=_sys.stderr)
+                           return False
+    except StopIteration:
+        print("Actual output is shorted than expected output", file=_sys.stderr)
+        return False
+
+    return True