Merge pull request #92 from nexB/84-improve-scoring-for-added

Add license/copyright score for 'added' files #84

Author: steven-esser

src/deltacode/__init__.py

@@ -204,7 +204,7 @@ def license_diff(self):
         ])
 
         for delta in self.deltas:
-            utils.determine_license_diff(delta, unique_categories)
+            utils.update_from_license_info(delta, unique_categories)
 
     def copyright_diff(self):
         """
@@ -215,7 +215,7 @@ def copyright_diff(self):
         attribute -- if there has been a copyright change.
         """
         for delta in self.deltas:
-            utils.determine_copyright_diff(delta)
+            utils.update_from_copyright_info(delta)
 
     def index_deltas(self, index_key='path', delta_list=[]):
         """
@@ -277,9 +277,16 @@ def is_unmodified(self):
         other than 'unmodified' and return True if all but 'unmodified' are
         ruled out.
         """
-        if (self.old_file and self.new_file and
-                self.old_file.sha1 == self.new_file.sha1 and
-                self.old_file.path == self.new_file.path):
+        if (self.new_file and self.old_file and
+                self.new_file.sha1 == self.old_file.sha1 and
+                self.new_file.path == self.old_file.path):
+            return True
+
+    def is_added(self):
+        """
+        Identify a Delta object reflecting the addition of a File.
+        """
+        if self.new_file and not self.old_file:
             return True
 
     def to_dict(self):

src/deltacode/utils.py

@@ -33,15 +33,43 @@
 from commoncode import paths
 
 
-def determine_license_diff(delta, unique_categories):
+def update_from_license_info(delta, unique_categories):
     """
-    Increase the Delta object's 'score' attribute and add one or more
-    appropriate categories to its 'factors' attribute if there has been a
-    license change and depending on the nature of that change.
+    Increase an 'added' or 'modified' Delta object's 'score' attribute and add
+    one or more appropriate categories to its 'factors' attribute if there has
+    been a license change and depending on the nature of that change.
     """
-    if not delta.is_modified():
+    if delta.is_added():
+        update_added_from_license_info(delta, unique_categories)
+
+    if delta.is_modified():
+        update_modified_from_license_info(delta, unique_categories)
+
+
+def update_added_from_license_info(delta, unique_categories):
+    """
+    Increase an 'added' Delta object's 'score' attribute and add
+    one or more categories to its 'factors' attribute if there has
+    been a license change.
+    """
+    new_licenses = delta.new_file.licenses or []
+    new_categories = set(license.category for license in new_licenses)
+
+    if delta.new_file.has_licenses():
+        delta.update(20, 'license info added')
+        # no license ==> 'Copyleft Limited'or higher
+        for category in new_categories:
+            if category in unique_categories:
+                delta.update(20, category.lower() + ' added')
         return
 
+
+def update_modified_from_license_info(delta, unique_categories):
+    """
+    Increase a 'modified' Delta object's 'score' attribute and add
+    one or more categories to its 'factors' attribute if there has
+    been a license change.
+    """
     if not delta.new_file.has_licenses() and delta.old_file.has_licenses():
         delta.update(15, 'license info removed')
         return
@@ -71,15 +99,36 @@ def determine_license_diff(delta, unique_categories):
                 delta.update(20, category.lower() + ' added')
 
 
-def determine_copyright_diff(delta):
+def update_from_copyright_info(delta):
+    """
+    Increase an 'added' or 'modified' Delta object's 'score' attribute and add
+    one or more appropriate categories to its 'factors' attribute if there has
+    been a copyright change and depending on the nature of that change.
+    """
+    if delta.is_added():
+        update_added_from_copyright_info(delta)
+
+    if delta.is_modified():
+        update_modified_from_copyright_info(delta)
+
+
+def update_added_from_copyright_info(delta):
     """
-    Increase the Delta object's 'score' attribute and add one or more
-    appropriate categories to its 'factors' attribute if there has been a
-    copyright change and depending on the nature of that change.
+    Increase an 'added' Delta object's 'score' attribute and add
+    one or more categories to its 'factors' attribute if there has
+    been a copyright change.
     """
-    if not delta.is_modified():
+    if delta.new_file.has_copyrights():
+        delta.update(10, 'copyright info added')
         return
 
+
+def update_modified_from_copyright_info(delta):
+    """
+    Increase a 'modified' Delta object's 'score' attribute and add
+    one or more categories to its 'factors' attribute if there has
+    been a copyright change.
+    """
     new_copyrights = delta.new_file.copyrights or []
     old_copyrights = delta.old_file.copyrights or []
 

tests/data/cli/1_file_moved_and_1_copy.csv

@@ -1,4 +1,4 @@
 Factors,Score,Path,Name,Type,Size,Old Path
-added,100,b/a4.py,a4.py,file,200,
-added,100,b/a4_copy.py,a4_copy.py,file,200,
+added license info added copyright info added,130,b/a4.py,a4.py,file,200,
+added license info added copyright info added,130,b/a4_copy.py,a4_copy.py,file,200,
 removed,0,a/a4.py,a4.py,file,200,

tests/data/cli/1_file_moved_and_added.csv

@@ -1,4 +1,4 @@
 Factors,Score,Path,Name,Type,Size,Old Path
-added,100,b/a4.py,a4.py,file,200,
-added,100,c/a4.py,a4.py,file,200,
+added license info added copyright info added,130,b/a4.py,a4.py,file,200,
+added license info added copyright info added,130,c/a4.py,a4.py,file,200,
 removed,0,a/a4.py,a4.py,file,200,

tests/data/cli/added1.csv

@@ -1,2 +1,2 @@
 Factors,Score,Path,Name,Type,Size,Old Path
-added,100,a/a5.py,a5.py,file,200,
+added license info added copyright info added,130,a/a5.py,a5.py,file,200,

tests/data/cli/renamed1.csv

@@ -1,3 +1,3 @@
 Factors,Score,Path,Name,Type,Size,Old Path
-added,100,a/a4_renamed_not_modified.py,a4_renamed_not_modified.py,file,200,
+added license info added copyright info added,130,a/a4_renamed_not_modified.py,a4_renamed_not_modified.py,file,200,
 removed,0,a/a4.py,a4.py,file,200,

tests/test_deltacode.py

@@ -1463,7 +1463,7 @@ def test_DeltaCode_sort_order(self):
         deltas_object = deltacode_object.deltas
 
         expected = [
-            ['added'],
+            ['added', 'license info added', 'copyright info added'],
             ['modified'],
             ['moved'],
             ['removed'],