Add new VirtualCodebase creation test #1439

    * Add test that creates a VirtualCodebase from multiple scans
    * Check to see if headers exist before attempting to get them when loading scan_data
    * Raise Exception when importing a scan with no files

Signed-off-by: Jono Yang <jyang@nexb.com>

Author: JonoYang

src/scancode/resource.py

@@ -1409,8 +1409,14 @@ def _get_scan_data(self, location):
             combined_scan_data = OrderedDict(headers=[], files=[])
             for loc in location:
                 scan_data = self._get_scan_data_helper(loc)
-                combined_scan_data['headers'].extend(scan_data['headers'])
-                combined_scan_data['files'].extend(scan_data['files'])
+                headers = scan_data.get('headers')
+                if headers:
+                    combined_scan_data['headers'].extend(headers)
+                files = scan_data.get('files')
+                if files:
+                    combined_scan_data['files'].extend(files)
+                else:
+                    raise Exception('Input file does not have Resources to import: {}'.format(loc))
             combined_scan_data['headers'] = sorted(combined_scan_data['headers'], key=lambda x: x['start_timestamp'])
             return combined_scan_data
         return self._get_scan_data_helper(location)

tests/scancode/data/resource/virtual_codebase/combine-1.json

@@ -0,0 +1,15 @@
+{
+  "files_count": 2,
+  "files": [
+    {
+      "path": "samples",
+      "type": "directory",
+      "summary": []
+    },
+    {
+      "path": "samples/NOTICE",
+      "type": "file",
+      "summary": []
+    }
+  ]
+}

tests/scancode/data/resource/virtual_codebase/combine-2.json

@@ -0,0 +1,15 @@
+{
+  "files_count": 2,
+  "files": [
+    {
+      "path": "thirdparty",
+      "type": "directory",
+      "summary": []
+    },
+    {
+      "path": "thirdparty/example.zip",
+      "type": "file",
+      "summary": []
+    }
+  ]
+}

tests/scancode/test_resource.py

@@ -1195,3 +1195,18 @@ def test_VirtualCodebase_check_that_already_existing_parent_is_updated_properly(
             ])
         ]
         assert expected == results
+
+    def test_VirtualCodebase_create_from_multiple_scans(self):
+        test_file_1 = self.get_test_loc('resource/virtual_codebase/combine-1.json')
+        test_file_2 = self.get_test_loc('resource/virtual_codebase/combine-2.json')
+        input = (test_file_1, test_file_2)
+        codebase = VirtualCodebase(input)
+        results = sorted(r.to_dict() for r in codebase.walk())
+        expected = [
+            OrderedDict([(u'path', u'virtual_root'), (u'type', u'directory'), (u'summary', []), (u'scan_errors', [])]),
+            OrderedDict([(u'path', u'virtual_root/samples'), (u'type', u'directory'), (u'summary', []), (u'scan_errors', [])]),
+            OrderedDict([(u'path', u'virtual_root/samples/NOTICE'), (u'type', u'file'), (u'summary', []), (u'scan_errors', [])]),
+            OrderedDict([(u'path', u'virtual_root/thirdparty'), (u'type', u'directory'), (u'summary', []), (u'scan_errors', [])]),
+            OrderedDict([(u'path', u'virtual_root/thirdparty/example.zip'), (u'type', u'file'), (u'summary', []), (u'scan_errors', [])])
+        ]
+        assert expected == results