parameterize the languages

data_filter.py

@@ -3,14 +3,14 @@
 
 class DataFilter:
 
-    def __init__(self, file_extensions: List[str], path_prefix: str, keyword: str):
-        self.file_extensions = file_extensions
+    def __init__(self, languages: List[str], path_prefix: str, keyword: str):
+        self.languages = languages
         self.path_prefix = path_prefix
         self.keyword = keyword
 
     def matches_file_extensions(self, path: str) -> bool:
-        for ext in self.file_extensions:
-            if path.endswith(ext):
+        for ext in self.languages:
+            if path.endswith(f".{ext.lower()}"):
                 return True
         return False
 

main.py

@@ -17,9 +17,9 @@
 
 
 class App:
-    def __init__(self, file_extensions: List[str], path_prefix: str, keyword: str):
+    def __init__(self, languages: List[str], path_prefix: str, keyword: str):
         self.client = GithubClient()
-        self.data_filter = DataFilter(file_extensions=file_extensions,
+        self.data_filter = DataFilter(languages=languages,
                                       path_prefix=path_prefix, keyword=keyword)
         self.commit_cache = SingleFileCache(location=COMMIT_CACHE_FILE)
         self.repo_cache = MultiFileCache(location=REPO_CACHE_FILE)
@@ -45,13 +45,12 @@ def get_repository_by_commit(self, repository, commit):
 
 
 def main(args):
-    file_extensions = [
-        ".java",
-        ".groovy"
+    languages = [
+        "groovy"
     ]
 
     app = App(
-        file_extensions=file_extensions,
+        languages=languages,
         path_prefix="instrumentation/",
         keyword="test"
     )
@@ -67,30 +66,29 @@ def main(args):
                 commit=commit
             )
             count = count_by_file_extension(files=repo_files["files"],
-                                            file_extensions=file_extensions)
+                                            languages=languages)
             if count:
-                result[snapshot] = {
-                    "date": snapshot,
-                    "java": count[".java"],
-                    "groovy": count[".groovy"]
-                }
+                result[snapshot]["date"] = snapshot
+                for language in languages:
+                    result[snapshot][language] = count[language]
         except Exception as e:
             print(f"Error for {snapshot}, {e}")
 
     dates = []
-    java_counts = []
-    groovy_counts = []
+
+    language_counts = {}
 
     for item in result.values():
-        date = item["date"][:10]
-        java_count = item["java"]
-        groovy_count = item["groovy"]
-        dates.append(date)
-        java_counts.append(java_count)
-        groovy_counts.append(groovy_count)
-
-    plt.plot(dates, java_counts, label='Java')
-    plt.plot(dates, groovy_counts, label='Groovy')
+        dates.append(item["date"][:10])
+        for language in languages:
+            try:
+                language_counts[language].append(item[language])
+            except KeyError:
+                language_counts[language] = [item[language]]
+
+    for lang, counts in language_counts.items():
+        plt.plot(dates, counts, label=lang.capitalize())
+
     plt.xlabel('Date')
     plt.ylabel('Count')
     plt.title('Test Classes by Lang in Instrumentation Directory')

tests/test_data_filter.py

@@ -7,9 +7,9 @@
 class ParseDataTestCase(unittest.TestCase):
 
     def __init__(self, *args, **kwargs):
-        self.file_extensions = [
-            ".java",
-            ".groovy"
+        self.languages = [
+            "java",
+            "groovy"
         ]
         super(ParseDataTestCase, self).__init__(*args, **kwargs)
 
@@ -84,7 +84,7 @@ def test_filters_payload_to_expected_files_based_on_criteria(self):
         """
         data = json.loads(test_json)
 
-        data_filter = DataFilter(file_extensions=self.file_extensions,
+        data_filter = DataFilter(languages=self.languages,
                                  path_prefix="instrumentation/", keyword="test")
 
         test = data_filter.parse_data(payload=data)
@@ -111,7 +111,7 @@ def test_given_url_with_different_case_than_keyword_still_filters_correctly(self
         """
         data = json.loads(payload)
 
-        data_filter = DataFilter(file_extensions=self.file_extensions,
+        data_filter = DataFilter(languages=self.languages,
                                  path_prefix="instrumentation/", keyword="test")
 
         test = data_filter.parse_data(payload=data)

tests/test_utilities.py

@@ -45,14 +45,14 @@ def test_count_by_file_extension(self):
             "test6.txt"
         ]
 
-        extensions = [
-            ".java",
-            ".groovy",
-            ".txt"
+        languages = [
+            "java",
+            "groovy",
+            "txt"
         ]
 
-        result = count_by_file_extension(files, extensions)
-        self.assertEqual(3, result['.java'])
-        self.assertEqual(2, result['.groovy'])
-        self.assertEqual(1, result['.txt'])
+        result = count_by_file_extension(files, languages)
+        self.assertEqual(3, result['java'])
+        self.assertEqual(2, result['groovy'])
+        self.assertEqual(1, result['txt'])
 

utilities.py

@@ -27,10 +27,11 @@ def get_dates_between(start_date_str, end_date, interval):
     return date_list
 
 
-def count_by_file_extension(files: List[str], file_extensions: List[str]) -> dict:
+def count_by_file_extension(files: List[str], languages: List[str]) -> dict:
     file_counts = defaultdict(int)
     for file in files:
-        for ext in file_extensions:
-            if file.endswith(ext):
+        for ext in languages:
+            extension = f".{ext.lower()}"
+            if file.endswith(extension):
                 file_counts[ext] += 1
     return file_counts