Add unit tests

Signed-off-by: akugarg <akanksha.garg2k@gmail.com>

src/licensedcode/models.py

@@ -953,6 +953,10 @@ def validate(self, licensing=None):
                     except InvalidRule as e:
                         yield f'Failed to parse and validate license_expression: {e}'
 
+            if self.referenced_filenames:
+                if len(set(self.referenced_filenames)) != len(self.referenced_filenames):
+                    yield 'referenced_filenames cannot contain duplicates.'
+
     def license_keys(self, unique=True):
         """
         Return a list of license keys for this rule.

src/licensedcode/plugin_license.py

@@ -129,7 +129,9 @@ def process_codebase(self, codebase, **kwargs):
 
 def match_reference_license(resource, codebase):
     """
-    Return the updated license matches having reference to another files
+    Return the ``resource`` Resource updating and saving it in place, after adding new
+    icense matches (licenses and license_expressions) following their Rule
+    ``referenced_filenames`` if any. Return None if this is not a file Resource.
     """
     if not resource.is_file:
         return
@@ -141,29 +143,43 @@ def match_reference_license(resource, codebase):
 
     referenced_licenses = []
     referenced_license_expressions = []
-    referenced_filenames = []
-
-    for license in licenses:
-        referenced_files = license['matched_rule']['referenced_filenames']
-        for referenced_filename in referenced_files:
-            if not referenced_filename in referenced_filenames:
-                referenced_filenames.append(referenced_filename)
+    referenced_filenames = get_referenced_filenames(licenses)
+    modified = False
 
     for referenced_filename in referenced_filenames:
-        new_resource = find_reference_licenses(referenced_filename=referenced_filename, resource=resource, codebase=codebase)
+        new_resource = find_referenced_resource(referenced_filename=referenced_filename, resource=resource, codebase=codebase)
         if new_resource:
+            modified = True
             referenced_licenses.extend(new_resource.licenses)
             referenced_license_expressions.extend(new_resource.license_expressions)
 
     licenses.extend(referenced_licenses)
     license_expressions.extend(referenced_license_expressions)
-    codebase.save_resource(resource)
+
+    if modified:
+        codebase.save_resource(resource)
     return resource
 
 
-def find_reference_licenses(referenced_filename, resource, codebase, **kwargs):
+def get_referenced_filenames(license_matches):
+    """
+    Return a list of unique referenced filenames found in the rules of a list of ``license_matches``
+    """
+    referenced_filenames = []
+    for license_match in license_matches:
+        referenced_files = license_match['matched_rule']['referenced_filenames']
+        for referenced_filename in referenced_files:
+            if not referenced_filename in referenced_filenames:
+                referenced_filenames.append(referenced_filename)
+
+    return referenced_filenames
+
+
+def find_referenced_resource(referenced_filename, resource, codebase, **kwargs):
     """
-    Return the resource object for matching referenced-filename given a resource in codebase
+    Return a Resource matching the ``referenced_filename`` path or filename given a ``resource`` in ``codebase``.
+    Return None if the ``referenced_filename`` cannot be found in the same directory as the base ``resource``.
+    ``referenced_filename`` is the path or filename referenced in a LicenseMatch of ``resource``,
     """
     parent = resource.parent(codebase)
 

tests/licensedcode/test_plugin_license.py

@@ -9,6 +9,10 @@
 
 import os
 
+from click.testing import Result
+from licensedcode.plugin_license import find_referenced_resource, match_reference_license
+from licensedcode.plugin_license import get_referenced_filenames
+
 import pytest
 
 from commoncode.testcase import FileDrivenTesting
@@ -67,6 +71,7 @@ def test_license_option_reports_license_texts_diag_long_lines():
     test_loc = test_env.get_test_loc('plugin_license/text_long_lines/scan-diag.expected.json')
     check_json_scan(test_loc, result_file, regen=False)
 
+
 def test_license_match_reference():
     test_dir = test_env.get_test_loc('plugin_license/license_reference/scan/scan-ref', copy=True)
     result_file = test_env.get_temp_file('json')
@@ -75,6 +80,7 @@ def test_license_match_reference():
     test_loc = test_env.get_test_loc('plugin_license/license_reference/scan-ref.expected.json')
     check_json_scan(test_loc, result_file, regen=False)
 
+
 def test_license_match_without_reference():
     test_dir = test_env.get_test_loc('plugin_license/license_reference/scan/scan-without-ref', copy=True)
     result_file = test_env.get_temp_file('json')
@@ -83,6 +89,49 @@ def test_license_match_without_reference():
     test_loc = test_env.get_test_loc('plugin_license/license_reference/scan-wref.expected.json')
     check_json_scan(test_loc, result_file, regen=False)
 
+
+def test_get_referenced_filenames():
+    license_matches = [
+        {'matched_rule': {'referenced_filenames' : ['LICENSE.txt','COPYING']}},
+        {'matched_rule': {'referenced_filenames' : ['COPYING','LICENSE.txt']}},
+        {'matched_rule': {'referenced_filenames' : ['copying']}},
+        {'matched_rule': {'referenced_filenames' : []}},
+    ]
+    expected = ['LICENSE.txt','COPYING','copying']
+    assert get_referenced_filenames(license_matches) == expected
+
+
+def test_find_referenced_resource():
+    # Setup: Create a new scan to use for a virtual codebase
+    test_dir = test_env.get_test_loc('plugin_license/license_reference/scan/scan-ref', copy=True)
+    scan_loc = test_env.get_temp_file('json')
+    args = ['--license', '--license-text', '--license-text-diagnostics', test_dir, '--json', scan_loc]
+    run_scan_click(args)
+
+    # test proper
+    from commoncode.resource import VirtualCodebase        
+    codebase = VirtualCodebase(scan_loc)
+    resource = codebase.get_resource_from_path('scan-ref/license-notice.txt')
+    result = find_referenced_resource(referenced_filename='LICENSE',resource=resource, codebase=codebase)
+    assert result.path == 'scan-ref/LICENSE'
+
+
+def test_match_reference_license():
+    # Setup: Create a new scan to use for a virtual codebase
+    test_dir = test_env.get_test_loc('plugin_license/license_reference/scan/scan-ref', copy=True)
+    scan_loc = test_env.get_temp_file('json')
+    args = ['--license', '--license-text', '--license-text-diagnostics', test_dir, '--json', scan_loc]
+    run_scan_click(args)
+
+    # test proper
+    from commoncode.resource import VirtualCodebase        
+    codebase = VirtualCodebase(scan_loc)
+    resource = codebase.get_resource_from_path('scan-ref/license-notice.txt')
+    assert len(resource.licenses) == 2
+    result = match_reference_license(resource=resource, codebase=codebase)
+    assert len(result.licenses) == 3
+
+
 def test_reindex_licenses_works():
     args = ['--reindex-licenses']
     run_scan_click(args)