Updated LanguageFileHandler.file_extension to file_extensions (allows for multiple file extensions for C++)

Author: Andrei-Constantin-Programmer

resources/repositories/c#_repos.txt

@@ -1,5 +1,13 @@
+activemq-nms-amqp
+activemq-nms-api
+activemq-nms-ems
+activemq-nms-msmq
 activemq-nms-openwire
 activemq-nms-stomp
-activemq-nms-ems
+activemq-nms-xms
+logging-log4net
+lucenenet
+pulsar-dotpulsar
 qpid-proton-dotnet
-logging-log4net
+rocketmq-client-csharp
+usergrid-dotnet

resources/repositories/c++_repos.txt

@@ -1 +1,3 @@
-incubator-pegasus
+brpc
+incubator-pegasus
+trafficserver

src/infrastructure/repository_utils.py

@@ -31,12 +31,12 @@ def read_repositories(language: str) -> List[Repository]:
 
         return [apache_repo_from_name(repo_name) for repo_name in repositories]
 
-def read_commits(repository_url: str, file_extension: str, final_date: Optional[datetime] = None) -> Generator[Commit, None, None]:
+def read_commits(repository_url: str, file_extensions: List[str], final_date: Optional[datetime] = None) -> Generator[Commit, None, None]:
     """
     Reads commits from a repository using PyDriller.
 
     @param: repository_url: The URL of the repository.
-    @param: file_extension: The file extension to search for.
+    @param: file_extensions: The file extensions to search for.
     @param: final_date: Date to read commits up until from the given repository.
 
     @return: A generator of Commit objects.
@@ -45,7 +45,7 @@ def read_commits(repository_url: str, file_extension: str, final_date: Optional[
         raise ValueError("Final date must be in the past.")
     to_date = final_date if final_date else datetime.now()
 
-    return DrillerRepo(repository_url, only_modifications_with_file_types=[file_extension], to=to_date).traverse_commits()
+    return DrillerRepo(repository_url, only_modifications_with_file_types=file_extensions, to=to_date).traverse_commits()
 
 def repo_from_url(repo_url: str):
     repo_name = re.search(r"github\.com/[^/]+/([^/.]+)", repo_url).group(1) if re.search(r"github\.com/[^/]+/([^/.]+)", repo_url) else None

src/mining/commit_retrieval.py

@@ -12,7 +12,7 @@ def _retrieve_files(modified_files, file_handler: LanguageFileHandler):
     files = []
 
     for file in modified_files:
-        if file_handler.file_extension in file.filename:
+        if any(file_extension in file.filename for file_extension in file_handler.file_extensions):
             files.append(file.filename)
 
     return files
@@ -21,7 +21,7 @@ def _retrieve_commits(repo_url, file_handler: LanguageFileHandler, final_date =
     commits = []
 
     try:
-        for commit in repository_utils.read_commits(repo_url, file_handler.file_extension, final_date):
+        for commit in repository_utils.read_commits(repo_url, file_handler.file_extensions, final_date):
             files = _retrieve_files(commit.modified_files, file_handler)
             commits.append(CustomCommit(commit.hash, files, commit.author, commit.author_date))
     except Exception as e:

src/models/file_handlers/CPlusPlusFileHandler.py

@@ -2,10 +2,10 @@
 
 class CPlusPlusFileHandler:
     name = "C++"
-    file_extension = ".cpp"
+    file_extensions = [".cpp", ".cc", ".c++"]
 
     def is_test_file(self, file: str) -> bool:
         return "test" in file
 
     def get_implementation_file(self, test_file: str) -> str:
-        return re.sub(r'_test', '', test_file)
+        return re.sub(r'(_test|test_)', '', test_file)

src/models/file_handlers/CSharpFileHandler.py

@@ -2,7 +2,7 @@
 
 class CSharpFileHandler:
     name = "C#"
-    file_extension = ".cs"
+    file_extensions = [".cs"]
 
     def is_test_file(self, file: str) -> bool:
         return "Test" in file

src/models/file_handlers/JavaFileHandler.py

@@ -2,7 +2,7 @@
 
 class JavaFileHandler:
     name = "Java"
-    file_extension = ".java"
+    file_extensions = [".java"]
 
     def is_test_file(self, file: str) -> bool:
         return "Test" in file

src/models/file_handlers/KotlinFileHandler.py

@@ -2,7 +2,7 @@
 
 class KotlinFileHandler:
     name = "Kotlin"
-    file_extension = ".kt"
+    file_extensions = [".kt"]
 
     def is_test_file(self, file: str) -> bool:
         return "Test" in file

src/models/file_handlers/LanguageFileHandler.py

@@ -1,8 +1,8 @@
-from typing import Protocol
+from typing import List, Protocol
 
 class LanguageFileHandler(Protocol):
     name: str
-    file_extension: str
+    file_extensions: List[str]
 
     def is_test_file(self, file: str) -> bool:
         ...

src/models/file_handlers/PythonFileHandler.py

@@ -2,10 +2,10 @@
 
 class PythonFileHandler:
     name = "Python"
-    file_extension = ".py"
+    file_extensions = [".py"]
 
     def is_test_file(self, file: str) -> bool:
         return "test" in file
 
     def get_implementation_file(self, test_file: str) -> str:
-        return re.sub(r'test_', '', test_file)
+        return re.sub(r'(_test|test_)', '', test_file)