Fixed #89 to handle duplicated keys in the input.

update the wording from "doesn't" to "does not"
Added test cases/code for #89

Author: chinyeungli

about_code_tool/genabout.py

@@ -69,6 +69,13 @@ def __init__(self):
         self.errors = []
         self.extract_dje_license_error = False
 
+    @staticmethod
+    def get_duplicated_keys(input_file):
+        csv_context = csv.reader(open(input_file))
+        keys_row = csv_context.next()
+        lower_case_keys_row = [k.lower() for k in keys_row]
+        return ([key for key in keys_row if lower_case_keys_row.count(key.lower()) > 1])
+
     @staticmethod
     def get_input_list(input_file):
         csvfile = csv.DictReader(open(input_file, 'rU'))
@@ -203,7 +210,7 @@ def verify_license_files(self, input_list, project_dir, license_in_project):
                     if _exists(license_file_path):
                         license_files_list.append((license_file_path, about_parent_dir))
                     else:
-                        self.warnings.append(Warn('license_text_file', license_file_path, "License doesn't exist."))
+                        self.warnings.append(Warn('license_text_file', license_file_path, "License does not exist."))
             except Exception as e:
                 print(repr(e))
                 print("The input does not have the 'license_text_file' key which is required.")
@@ -301,7 +308,7 @@ def get_dje_license_list(self, gen_location, input_list, gen_license):
                     about_parent_dir = dirname(file_location)
                     license_file = gen_location.rpartition('/')[0] + join(about_parent_dir, line['license_text_file'])
                     if not _exists(license_file):
-                        self.errors.append(Error('license_text_file', license_file, "The 'license_text_file' doesn't exist."))
+                        self.errors.append(Error('license_text_file', license_file, "The 'license_text_file' does not exist."))
                 else:
                     if gen_license:
                         if line['dje_license_key']:
@@ -514,17 +521,17 @@ def main(parser, options, args):
 
     if copy_license_path:
         if not _exists(copy_license_path):
-            print("The project path doesn't exist.")
+            print("The project path does not exist.")
             sys.exit(errno.EINVAL)
 
     if license_text_path:
         if not _exists(license_text_path):
-            print("The license text path doesn't exist.")
+            print("The license text path does not exist.")
             sys.exit(errno.EINVAL)
 
     if mapping_config:
         if not _exists('MAPPING.CONFIG'):
-            print("The file 'MAPPING.CONFIG' doesn't exist.")
+            print("The file 'MAPPING.CONFIG' does not exist.")
             sys.exit(errno.EINVAL)
 
     if extract_license:
@@ -566,6 +573,13 @@ def main(parser, options, args):
 
     gen = GenAbout()
 
+    dup_keys = gen.get_duplicated_keys(input_path)
+    if dup_keys:
+        print('The input file contains duplicated keys. Duplicated keys are not allowed.')
+        print(dup_keys)
+        print('\nPlease fix the input file and re-run the tool.')
+        sys.exit(errno.EINVAL)
+
     # Clear the log file
     with open(output_path + LOG_FILENAME, 'w'):
         pass

about_code_tool/tests/test_genabout.py

@@ -113,6 +113,18 @@ def test_validate_duplication_no_dup(self):
                        'name': 'ABOUT tool', 'version': ''}]
         self.assertFalse(gen.validate_duplication(input_list), "The list has no duplication.")
 
+    def test_get_duplicated_keys_have_dup(self):
+        gen = genabout.GenAbout()
+        test_input = join(TESTDATA_PATH, "test_files_for_genabout/dup_keys.csv")
+        expected_list = ['copyright', 'copyright']
+        self.assertTrue(gen.get_duplicated_keys(test_input))
+
+    def test_get_duplicated_keys_have_dup_diff_case(self):
+        gen = genabout.GenAbout()
+        test_input = join(TESTDATA_PATH, "test_files_for_genabout/dup_keys_with_diff_case.csv")
+        expected_list = ['copyright', 'Copyright']
+        self.assertTrue(gen.get_duplicated_keys(test_input))
+
     def test_validate_mandatory_fields_no_missing(self):
         gen = genabout.GenAbout()
         input_list = [{'about_file': '/about.ABOUT', 'about_resource': '.',

about_code_tool/tests/testdata/test_files_for_genabout/dup_keys.csv

@@ -0,0 +1,2 @@
+about_file,about_resource,copyright,name,version,copyright
+about.ABOUT,.,nexB,ABOUT tool,0.8.1,someone

about_code_tool/tests/testdata/test_files_for_genabout/dup_keys_with_diff_case.csv

@@ -0,0 +1,2 @@
+about_file,about_resource,copyright,name,version,Copyright
+about.ABOUT,.,nexB,ABOUT tool,0.8.1,someone