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

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

Author: johnmhoran

src/deltacode/__init__.py

@@ -282,6 +282,13 @@ def is_unmodified(self):
                 self.old_file.path == self.new_file.path):
             return True
 
+    def is_added(self):
+        """
+        Identify a Delta object reflecting the addition of a File.
+        """
+        if not self.old_file and self.new_file:
+            return True
+
     def to_dict(self):
         """
         Return an OrderedDict comprising the 'factors', 'score' and new and old

src/deltacode/utils.py

@@ -35,10 +35,22 @@
 
 def determine_license_diff(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 delta.is_added():
+        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
+
     if not delta.is_modified():
         return
 
@@ -73,10 +85,15 @@ def determine_license_diff(delta, unique_categories):
 
 def determine_copyright_diff(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' 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():
+        if delta.new_file.has_copyrights():
+            delta.update(10, 'copyright info added')
+            return
+
     if not delta.is_modified():
         return
 

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'],

tests/test_utils.py

@@ -654,6 +654,73 @@ def test_determine_license_diff_copyleft_to_copyleft_limited(self):
         assert len(test_delta.factors) == 1
         assert 'license change' in test_delta.factors
 
+    def test_determine_license_diff_file_added_permissive_license(self):
+        test_file_new = models.File({
+            'path':'/test/path.txt',
+            'name': 'path.txt',
+            'sha1': 'a',
+            'original_path': '',
+            "licenses": [
+                {
+                    "key": "mit",
+                    "score": 80.0,
+                    "short_name": "MIT License",
+                    "category": "Permissive"
+                }
+            ]
+        })
+
+        test_delta = deltacode.Delta(100, test_file_new, None)
+
+        utils.determine_license_diff(test_delta, unique_categories)
+
+        assert test_delta.score == 120
+        assert len(test_delta.factors) == 1
+
+        assert 'license info added' in test_delta.factors
+
+    def test_determine_license_diff_file_added_commercial_and_copyleft_licenses(self):
+        test_file_new = models.File({
+            'path':'/test/path.txt',
+            'name': 'path.txt',
+            'sha1': 'a',
+            'original_path': '',
+            "licenses": [
+                {
+                    "key": "commercial-license",
+                    "score": 55.0,
+                    "short_name": "Commercial License",
+                    "category": "Commercial",
+                    "owner": "Unspecified"
+                },
+                {
+                    "key": "adapt-1.0",
+                    "score": 15.0,
+                    "short_name": "APL 1.0",
+                    "category": "Copyleft",
+                    "owner": "OSI - Open Source Initiative"
+                }
+            ]
+        })
+
+        test_delta = deltacode.Delta(100, test_file_new, None)
+
+        utils.determine_license_diff(test_delta, unique_categories)
+
+        assert test_delta.score == 160
+        assert len(test_delta.factors) == 3
+
+        assert 'license info added' in test_delta.factors
+
+        expected_factors = [
+            'license info added',
+            'commercial added',
+            'copyleft added'
+        ]
+
+        for factor in expected_factors:
+            assert factor in test_delta.factors
+
     def test_determine_copyright_diff_empty(self):
         test_delta = deltacode.Delta()
 
@@ -1016,6 +1083,32 @@ def test_determine_copyright_diff_one_copyright_removed(self):
         assert len(test_delta.factors) == 1
         assert 'copyright change' in test_delta.factors
 
+    def test_determine_copyright_diff_file_added_one_copyright(self):
+        test_file_new = models.File({
+            'path':'/test/path.txt',
+            'name': 'path.txt',
+            'sha1': 'a',
+            'original_path': '',
+            "copyrights": [
+                {
+                    "statements": [
+                        "Copyright (c) 2017-2018 Francois Hennebique and others."
+                    ],
+                    "holders": [
+                        "Francois Hennebique and others."
+                    ]
+                }
+            ]
+        })
+
+        test_delta = deltacode.Delta(100, test_file_new, None)
+
+        utils.determine_copyright_diff(test_delta)
+
+        assert test_delta.score == 110
+        assert len(test_delta.factors) == 1
+        assert 'copyright info added' in test_delta.factors
+
     def test_determine_lic_copy_diffs_copyright_and_license_info_added(self):
         test_file_new = models.File({
             'path':'/test/path.txt',
@@ -1352,6 +1445,98 @@ def test_determine_lic_copy_diffs_license_change_no_copyright_change(self):
         for factor in expected_factors:
             assert factor in test_delta.factors
 
+    def test_determine_lic_copy_diffs_file_added_copyright_and_permissive_license(self):
+        test_file_new = models.File({
+            'path':'/test/path.txt',
+            'name': 'path.txt',
+            'sha1': 'a',
+            'original_path': '',
+            "licenses": [
+                {
+                    "key": "mit",
+                    "score": 80.0,
+                    "short_name": "MIT License",
+                    "category": "Permissive"
+                }
+            ],
+            "copyrights": [
+                {
+                    "statements": [
+                        "Copyright (c) 2017-2018 Francois Hennebique and others."
+                    ],
+                    "holders": [
+                        "Francois Hennebique and others."
+                    ]
+                }
+            ]
+        })
+
+        test_delta = deltacode.Delta(100, test_file_new, None)
+
+        utils.determine_license_diff(test_delta, unique_categories)
+        utils.determine_copyright_diff(test_delta)
+
+        expected_factors = [
+            'license info added',
+            'copyright info added'
+        ]
+
+        assert test_delta.score == 130
+        assert len(test_delta.factors) == 2
+        for factor in expected_factors:
+            assert factor in test_delta.factors
+
+    def test_determine_lic_copy_diffs_file_added_copyright_and_commercial_and_copyleft_licenses(self):
+        test_file_new = models.File({
+            'path':'/test/path.txt',
+            'name': 'path.txt',
+            'sha1': 'a',
+            'original_path': '',
+            "licenses": [
+                {
+                    "key": "commercial-license",
+                    "score": 55.0,
+                    "short_name": "Commercial License",
+                    "category": "Commercial",
+                    "owner": "Unspecified"
+                },
+                {
+                    "key": "adapt-1.0",
+                    "score": 15.0,
+                    "short_name": "APL 1.0",
+                    "category": "Copyleft",
+                    "owner": "OSI - Open Source Initiative"
+                }
+            ],
+            "copyrights": [
+                {
+                    "statements": [
+                        "Copyright (c) 2017-2018 Francois Hennebique and others."
+                    ],
+                    "holders": [
+                        "Francois Hennebique and others."
+                    ]
+                }
+            ]
+        })
+
+        test_delta = deltacode.Delta(100, test_file_new, None)
+
+        utils.determine_license_diff(test_delta, unique_categories)
+        utils.determine_copyright_diff(test_delta)
+
+        expected_factors = [
+            'license info added',
+            'commercial added',
+            'copyleft added',
+            'copyright info added'
+        ]
+
+        assert test_delta.score == 170
+        assert len(test_delta.factors) == 4
+        for factor in expected_factors:
+            assert factor in test_delta.factors
+
     def test_align_trees_simple(self):
         test_scan_new = self.get_test_loc('utils/align-trees-simple-new.json')
         # Our old scan uses --full-root option in scancode