Remove cutoff_score from license_diff #62

  * Also fixed failing tests, deleted one duplicative test.

Signed-off-by: John M. Horan <johnmhoran@gmail.com>

Author: johnmhoran

src/deltacode/__init__.py

@@ -56,7 +56,6 @@ def __init__(self, new_path, old_path, options):
         if self.new.path != '' and self.old.path != '':
             self.determine_delta()
             self.determine_moved()
-            # TODO: Need to account for possibly passing in a 'cutoff_score'.
             self.license_diff()
             # Sort deltas by score, descending, i.e., high > low.
             self.deltas.sort(key=lambda Delta: Delta.score, reverse=True)
@@ -180,7 +179,7 @@ def update_deltas(self, added, removed):
         self.deltas.remove(added)
         self.deltas.remove(removed)
 
-    def license_diff(self, cutoff_score=50):
+    def license_diff(self):
         """
         Compare the license details for a pair of 'new' and 'old' File objects
         in a Delta object and change the Delta object's 'score' attribute --
@@ -205,8 +204,8 @@ def license_diff(self, cutoff_score=50):
                     i.score += 15
                     return
 
-                new_keys = set(l.key for l in new_licenses if l.score >= cutoff_score)
-                old_keys = set(l.key for l in old_licenses if l.score >= cutoff_score)
+                new_keys = set(l.key for l in new_licenses)
+                old_keys = set(l.key for l in old_licenses)
 
                 if new_keys != old_keys:
                     i.factors.append('license change')
@@ -225,6 +224,7 @@ def index_deltas(self, index_key='path', delta_list=[]):
         index = {}
 
         for delta in delta_list:
+            # FIXME: This is an ugly way to do this.
             if delta.score == 10:
                 key = getattr(delta.old_file, index_key)
             else:

tests/data/cli/modified_new_license_added_low_score.csv

@@ -1,3 +1,3 @@
 Factors,Score,Path,Name,Type,Size,Old Path
-modified,20,some/path/a/a1.py,a1.py,file,350,
-modified,20,some/path/b/b1.py,b1.py,file,290,
+modified license change,30,some/path/a/a1.py,a1.py,file,350,
+modified license change,30,some/path/b/b1.py,b1.py,file,290,

tests/test_deltacode.py

@@ -250,11 +250,11 @@ def test_DeltaCode_license_modified_low_score(self):
 
         deltas = result.deltas
 
-        assert len([i for i in deltas if i.score == 30]) == 0
-        assert len([i for i in deltas if i.score == 20]) == 2
+        assert len([i for i in deltas if i.score == 30]) == 2
+        assert len([i for i in deltas if i.score == 20]) == 0
 
-        assert [d.score for d in deltas if d.new_file.path == 'some/path/a/a1.py'] == [20]
-        assert [d.score for d in deltas if d.new_file.path == 'some/path/b/b1.py'] == [20]
+        assert [d.score for d in deltas if d.new_file.path == 'some/path/a/a1.py'] == [30]
+        assert [d.score for d in deltas if d.new_file.path == 'some/path/b/b1.py'] == [30]
 
     def test_DeltaCode_license_modified(self):
         new_scan = self.get_test_loc('deltacode/scan_modified_new_license_added.json')
@@ -329,21 +329,6 @@ def test_Delta_modified_license_added(self):
         assert [d.score for d in deltas if d.new_file.path == 'some/path/b/b1.py'] == [30]
         assert [d.score for d in deltas if d.new_file.path == 'some/path/c/c1.py'] == [20]
 
-    def test_Delta_modified_license_added_low_score(self):
-        new_scan = self.get_test_loc('deltacode/scan_modified_new_license_added_low_score.json')
-        old_scan = self.get_test_loc('deltacode/scan_modified_old_license_added_low_score.json')
-
-        options = OrderedDict([
-            ('--all-delta-types', False)
-        ])
-
-        result = DeltaCode(new_scan, old_scan, options)
-
-        deltas = result.deltas
-
-        assert [d.score for d in deltas if d.new_file.path == 'some/path/a/a1.py'] == [20]
-        assert [d.score for d in deltas if d.new_file.path == 'some/path/b/b1.py'] == [20]
-
     def test_Delta_modified_no_license_changes(self):
         new_scan = self.get_test_loc('deltacode/scan_modified_new_no_license_changes.json')
         old_scan = self.get_test_loc('deltacode/scan_modified_old_no_license_changes.json')
@@ -847,8 +832,8 @@ def test_score_new_lic_below_cutoff_score(self):
         assert [d.old_file.sha1 for d in deltas_object if d.old_file.path == 'path.txt'] == ['b']
         assert [d.new_file.sha1 for d in deltas_object if d.new_file.path == 'path.txt'] == ['b_modified']
 
-        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [30]
-        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified', 'license change']
+        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [20]
+        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified']
         assert [d.to_dict().get('old').get('licenses') for d in deltas_object if d.new_file.path == 'path.txt'].pop() == [
             OrderedDict([
                 ('key', 'mit'),
@@ -888,8 +873,8 @@ def test_score_old_lic_below_cutoff_score(self):
         assert [d.old_file.sha1 for d in deltas_object if d.old_file.path == 'path.txt'] == ['b']
         assert [d.new_file.sha1 for d in deltas_object if d.new_file.path == 'path.txt'] == ['b_modified']
 
-        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [30]
-        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified', 'license change']
+        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [20]
+        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified']
         assert [d.to_dict().get('old').get('licenses') for d in deltas_object if d.new_file.path == 'path.txt'].pop() == [
             OrderedDict([
                 ('key', 'mit'),
@@ -1018,8 +1003,8 @@ def test_score_multiple_lic_keys_new_below_cutoff_score(self):
         assert [d.old_file.sha1 for d in deltas_object if d.old_file.path == 'path.txt'] == ['b']
         assert [d.new_file.sha1 for d in deltas_object if d.new_file.path == 'path.txt'] == ['b_modified']
 
-        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [20]
-        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified']
+        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [30]
+        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified', 'license change']
         assert [d.to_dict().get('old').get('licenses') for d in deltas_object if d.new_file.path == 'path.txt'].pop() == [
             OrderedDict([
                 ('key', 'mit'),
@@ -1073,8 +1058,8 @@ def test_score_multiple_lic_keys_old_below_cutoff_score(self):
         assert [d.old_file.sha1 for d in deltas_object if d.old_file.path == 'path.txt'] == ['b']
         assert [d.new_file.sha1 for d in deltas_object if d.new_file.path == 'path.txt'] == ['b_modified']
 
-        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [20]
-        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified']
+        assert [d.score for d in deltas_object if d.new_file.path == 'path.txt'] == [30]
+        assert [d.factors for d in deltas_object if d.new_file.path == 'path.txt'].pop() == ['modified', 'license change']
         assert [d.to_dict().get('old').get('licenses') for d in deltas_object if d.new_file.path == 'path.txt'].pop() == [
             OrderedDict([
                 ('key', 'gpl-2.0'),