Feature/add file history query

src/app/container.py

@@ -17,7 +17,7 @@ class Container(containers.DeclarativeContainer):
     config = providers.Configuration()
 
     wiring_config = containers.WiringConfiguration(
-        packages=["app.repos", "app.commits", "core.repos", "core.commits"],
+        packages=["app.repos", "app.commits", "app.files"],
     )
 
     # ==== AppConfig =====

src/app/files/files.py

@@ -0,0 +1,36 @@
+import click
+from loguru import logger
+
+from app.container import Container
+
+from core.files.queries.queries import FilesQueries
+
+from dependency_injector.wiring import Provide, inject
+
+
+@click.option("--name", type=str, required=True)
+@click.option("--file", type=str, required=True)
+@click.command()
+@inject
+def history(
+    name: str,
+    file: str,
+    file_queries: FilesQueries = Provide[Container.core.file_queries],
+) -> None:
+    """
+    Lists the history of a file in a repository.
+
+    Args:
+        name (str): The name of the repository.
+        file (str): The path to the file within the repository.
+        file_queries (FilesQueries): The queries to fetch repositories.
+    """
+    files = file_queries.get_file_history(file, name)
+
+    if not files:
+        logger.info("No file history found.")
+        return
+
+    logger.info(f"File history for '{file}' in repository '{name}':")
+    for file in files:
+        logger.info(file)

src/app/files/group.py

@@ -0,0 +1,14 @@
+import click
+
+from app.files.files import history
+
+
+@click.group()
+def files():
+    """
+    File Management Commands
+    """
+    pass
+
+
+files.add_command(history)

src/core/container.py

@@ -1,12 +1,17 @@
 from dependency_injector import containers, providers
 
 from app.config import AppConfig
-from core.commits.queries.queries import CommitQueries
+
+from core.interfaces.git_client import GitClient
 from core.interfaces.query_client import QueryClient
 from core.interfaces.search_client import SearchClient
+
 from core.repos.operations.add_repo_operation import AddRepoOperation
 from core.repos.operations.delete_repo_operation import DeleteRepoOperation
 from core.repos.operations.update_repo_operation import UpdateRepoOperation
+
+from core.commits.queries.queries import CommitQueries
+from core.files.queries.queries import FilesQueries
 from core.repos.queries.queries import RepoQueries
 
 
@@ -19,6 +24,10 @@ class Container(containers.DeclarativeContainer):
 
     app_config = providers.Singleton(AppConfig)
 
+    wiring_config = containers.WiringConfiguration(
+        packages=["core.repos", "core.commits", "core.files"],
+    )
+
     query_client: providers.Provider[QueryClient] = providers.AbstractSingleton(
         instance_of=QueryClient,
         app_config=app_config,
@@ -29,6 +38,10 @@ class Container(containers.DeclarativeContainer):
         app_config=app_config,
     )
 
+    git_client: providers.Provider[GitClient] = providers.AbstractSingleton(
+        instance_of=GitClient,
+    )
+
     # Repo
     add_repo_operation = providers.Factory(
         AddRepoOperation,
@@ -52,11 +65,15 @@ class Container(containers.DeclarativeContainer):
     repo_queries = providers.Factory(
         RepoQueries,
         query_client=query_client,
-        search_client=search_client,
     )
 
     commit_queries = providers.Factory(
         CommitQueries,
         query_client=query_client,
-        search_client=search_client,
+    )
+
+    file_queries = providers.Factory(
+        FilesQueries,
+        git_client=git_client,
+        query_client=query_client,
     )

src/core/files/queries/models.py

@@ -0,0 +1,9 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class FileQueryModel:
+    commit_hash: str
+    message: str
+    author: str
+    date: str

src/core/files/queries/queries.py

@@ -0,0 +1,50 @@
+from typing import List
+
+from core.interfaces.git_client import GitClient
+from core.interfaces.query_client import QueryClient
+from entities.repos import Repo
+
+from core.files.queries.models import FileQueryModel
+
+
+class FilesQueries:
+    def __init__(self, git_client: GitClient, query_client: QueryClient) -> None:
+        """
+        Initialize the FilesQueries class.
+        This class is responsible for executing queries related to repositories.
+
+        Args:
+            git_client (GitClient): An instance of GitClient for Git operations.
+            query_client (QueryClient): An instance of QueryClient for database operations.
+        """
+        self.git_client = git_client
+        self.query_client = query_client
+
+    def get_file_history(self, file_path: str, repo_name: str) -> List[FileQueryModel]:
+        """
+        Returns the history of a file in the repository.
+
+        Args:
+            file_path (str): The path to the file within the repository.
+            repo_name (str): The path to the repository.
+        """
+        with self.query_client.session() as session:
+            repo: Repo | None = (
+                session.query(Repo).filter(Repo.name == repo_name).one_or_none()
+            )
+
+            if not repo:
+                return []
+
+            results = []
+            for commit in self.git_client.iter_file_history(file_path, repo.path):
+                results.append(
+                    FileQueryModel(
+                        commit_hash=commit.hexsha,
+                        message=commit.message.strip(),
+                        author=commit.author.name,
+                        date=commit.committed_datetime.isoformat(),
+                    )
+                )
+
+            return results

src/core/interfaces/git_client.py

@@ -0,0 +1,17 @@
+from abc import ABC, abstractmethod
+from typing import Iterator
+
+from git import Commit
+
+
+class GitClient(ABC):
+    @abstractmethod
+    def iter_file_history(self, file_name: str, repo_path: str) -> Iterator[Commit]:
+        """
+        Retrieve the history of a file in the repository.
+
+        Args:
+            file_name (str): The path to the file relative to the repository root.
+            repo_path (str): The path to the repository.
+        """
+        pass

src/libs/container.py

@@ -3,6 +3,7 @@
 from app.config import AppConfig
 from libs.chromadb.providers import ChromaClient
 from libs.duckdb.provider import DuckDbClient
+from libs.git.providers import PythonGitClient
 
 
 class Container(containers.DeclarativeContainer):
@@ -19,6 +20,7 @@ class Container(containers.DeclarativeContainer):
         AppConfig,
     )
 
+    # note these override the dependencies in core.container.Container
     # Chroma Client
     search_client = providers.Singleton(
         ChromaClient,
@@ -30,3 +32,8 @@ class Container(containers.DeclarativeContainer):
         DuckDbClient,
         app_config=app_config,
     )
+
+    # Git Client
+    git_client = providers.Singleton(
+        PythonGitClient,
+    )

src/libs/git/providers.py

@@ -0,0 +1,22 @@
+from typing import Iterator
+
+from git import Repo, Commit
+
+from core.interfaces.git_client import GitClient
+
+
+class PythonGitClient(GitClient):
+    def iter_file_history(self, file_name: str, repo_path: str) -> Iterator[Commit]:
+        """
+        Retrieve the history of a file in the repository.
+
+        Args:
+            file_name (str): The path to the file within the repository.
+            repo_path (str): The path to the repository.
+
+        Returns:
+            Dict: A dictionary containing the file history.
+        """
+        git_repo = Repo(repo_path)
+
+        return git_repo.iter_commits(paths=file_name)

src/main.py

@@ -4,6 +4,7 @@
 
 from app.repos.group import repos
 from app.commits.group import commits
+from app.files.group import files
 
 
 @click.group()
@@ -20,6 +21,7 @@ def cli():
 
 cli.add_command(repos)
 cli.add_command(commits)
+cli.add_command(files)
 
 if __name__ == "__main__":
     cli()

tests/core/repos/operations/test_update_repo_operation.py

@@ -62,7 +62,7 @@ def test_execute_should_add_commits_when_commits_do_not_exist_in_db():
     # Arrange
     client = MagicMock()
     session = MagicMock()
-    repo = Repo(id=1,name="test-repo", path="/old/path/to/repo")
+    repo = Repo(id=1, name="test-repo", path="/old/path/to/repo")
     commit1 = Commit(id=1, commit_hash="abc123", repo_id=repo.id)
     session.query.return_value.filter.return_value.one_or_none.side_effect = [
         repo,
@@ -83,7 +83,7 @@ def mock_assign_id_that_flush_would_assign(commit):
     mock_git_repo = MagicMock()
     mock_git_repo.iter_commits.return_value = [
         MagicMock(hexsha="abc123"),
-        MagicMock(hexsha="def456", message="New commit message")
+        MagicMock(hexsha="def456", message="New commit message"),
     ]
     with patch(
         "core.repos.operations.update_repo_operation.GitRepo",
@@ -92,19 +92,15 @@ def mock_assign_id_that_flush_would_assign(commit):
         op = UpdateRepoOperation(query_client=client, search_client=search_client)
         op.execute("test-repo")
 
-     # Assert
+    # Assert
     session.commit.assert_called()
     session.add.assert_called()
     mock_git_repo.iter_commits.assert_called_once()
     search_client.add_to_collection.assert_called_with(
         collection_name="commits",
         data="New commit message",
-        id=f"rep1_com1",
-        metadata={
-            "repo_id": repo.id,
-            "commit_id": "1",
-            "commit_hash": "def456"
-        },
+        id="rep1_com1",
+        metadata={"repo_id": repo.id, "commit_id": "1", "commit_hash": "def456"},
     )