Resolve dependabot alerts #3: refactor git clone implementation.

CHANGELOG.md

@@ -1,7 +1,7 @@
 <!--
 ## [Unreleased]
-### ➕ Added
-### 🛠 Changed
+### 🚀 New Features and Enhancements
+### 🛠 Changes
 ### ⚙️ Deprecated
 ### 🗑 Removed
 ### 🐛 Bug Fixes
@@ -14,6 +14,39 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ---
 
+## [v0.0.7] - *2023-08-30*
+
+⚠️ Release v0.0.7 addresses a security vulnerability cloning git repositories via the *GitPython* package on Windows systems. This vulnerability could allow arbitrary command execution if code is run from a directory containing a malicious `git.exe` or `git` executable.
+
+### 🔐 Security Fixes
+#### *Arbitrary Command Execution Mitigation*
+
+- Dependabot Alert [#3](https://github.com/eli64s/readme-ai/security/dependabot/3): GitPython untrusted search path on Windows systems leading to arbitrary code execution.
+- The previous git clone implementation sets the `env` argument to the path of the git executable in the current working directory. This poses a security risk as the code is susceptible to running arbitrary `git` commands from a malicious repository.
+    ```python
+    git.Repo.clone_from(repo_path, temp_dir, depth=1)
+    ```
+- Updated the `env` argument to explicitly set the absolute path of the git executable. This ensures that the git executable used to clone the repository is the one thats installed in the system path, and not the one located in the current working directory.
+    ```python
+    git.Repo.clone_from(repo_path, temp_dir, depth=1, env=git_exec_path)
+    ```
+### 🚀 New Features and Enhancements
+
+#### *Code Modularity*
+
+- Introduced three methods to help isolate the Git executable discovery and validation logic.
+  - `find_git_executable()`: Determines the absolute path of the Git executable.
+  - `validate_git_executable()`: Validates the found Git executable path.
+  - `validate_file_permissions()`: Validates the file permissions of the cloned repository.
+
+#### *File Permission Checks*
+
+- For Unix systems, added checks to ensure the permissions of the cloned repository are set to `0o700`. This is a best practice for secure temporary directories and prevents unauthorized users from accessing the directory.
+
+⚠️ These updates aim to mitigate the vulnerbility raised in Dependabot alert [#3](https://github.com/eli64s/readme-ai/security/dependabot/3). Users are advised to update *readme-ai* to the latest version, i.e ```pip install --upgrade readmeai```. Please be mindful of this vulnerability and use caution when cloning repositories from untrusted sources, especially for Windows users.
+
+---
+
 ## [v0.0.6] - *2023-08-29*
 
 ### 🐛 Bug Fixes
@@ -26,7 +59,7 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ## [v0.0.5] - *2023-07-31*
 
-### ➕ Added
+### 🚀 New Features and Enhancements
 
 - Add [.dockerignore](./.dockerignore) file to exclude unnecessary files from the Docker image.
 
@@ -48,7 +81,7 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ## [v0.0.4] - *2023-07-30*
 
-### ➕ Added
+### 🚀 New Features and Enhancements
 
 - Publish *readme-ai* CLI to PyPI under the module name [readmeai](https://pypi.org/project/readmeai/).
   - Refactored the codebase to use [Click](https://click.palletsprojects.com/en/8.1.x/), migrating from argparse.
@@ -75,7 +108,7 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ## [v0.0.3] - *2023-06-29*
 
-### ➕ Added
+### 🚀 New Features and Enhancements
 
 - Add [pydantic](https://pydantic-docs.helpmanual.io/) to validate the user's repository and api key inputs.
   - Validation was moved from *main.py* to *conf.py*.
@@ -91,12 +124,12 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ## [v0.0.2] - *2023-06-28*
 
-### ➕ Added
+### 🚀 New Features and Enhancements
 
 - Add [CHANGELOG.md](./CHANGELOG.md) to track changes to the project.
 - Add new directory [examples/video](./examples/video) to store mp4 videos to demonstrate the *readme-ai* tool.
 
-### 🛠 Changed
+### 🛠 Changes
 
 - Update [Makefile](./Makefile) and [setup.sh](./setup/setup.sh) to use *poetry* for dependency management.
 
@@ -109,10 +142,10 @@ All notable changes to the *readme-ai* project will be documented in this file.
 
 ## [v0.0.1] - *2023-06-28*
 
-### ➕ Added
+### 🚀 New Features and Enhancements
 - Initial release of *readme-ai* v0.0.1
 
-### 🛠 Changed
+### 🛠 Changes
 
 - Refine the markdown template structure to be more readable.
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "readmeai"
-version = "0.3.1"
+version = "0.3.015"
 description = "🚀 Generate awesome README.md files from the terminal, powered by OpenAI's GPT language model APIs 💫"
 authors = ["Eli <[email protected]>"]
 license = "MIT"

readmeai/preprocess.py

@@ -48,13 +48,13 @@ def __init__(
         self.language_setup = language_setup
         self.encoding_name = config.api.encoding
 
-    def analyze(self, root_path: str, is_remote: bool = False) -> List[Dict]:
+    def analyze(self, repo_path: str, is_remote: bool = False) -> List[Dict]:
         """Analyzes a local or remote git repository."""
         with tempfile.TemporaryDirectory() as temp_dir:
             if is_remote:
-                utils.clone_repository(root_path, temp_dir)
-                root_path = temp_dir
-            contents = self.generate_contents(root_path)
+                utils.clone_repository(repo_path, temp_dir)
+                repo_path = temp_dir
+            contents = self.generate_contents(repo_path)
             contents = self.tokenize_content(contents)
             contents = self.process_language_mapping(contents)
         return contents

readmeai/utils.py

@@ -1,21 +1,81 @@
 """Utility methods for the readme-ai application."""
 
+import os
+import platform
 import re
 from pathlib import Path
-from typing import List
+from typing import List, Optional
 
 import git
 from tiktoken import get_encoding
 
-from . import conf
+from . import conf, logger
 
+logger = logger.Logger(__name__)
 
-def clone_repository(url: str, repo_path: Path) -> None:
+
+def clone_repository(repo_path: str, temp_dir: Path) -> None:
     """Clone a repository to a temporary directory."""
+    git_exec_path = find_git_executable()
+
+    validate_git_executable(git_exec_path)
+
+    env = os.environ.copy()
+    env["GIT_PYTHON_GIT_EXECUTABLE"] = str(git_exec_path)
+
     try:
-        git.Repo.clone_from(url, repo_path, depth=1)
-    except git.exc.GitCommandError as exc:
-        raise ValueError(f"Error cloning repository: {exc}") from exc
+        git.Repo.clone_from(repo_path, temp_dir, depth=1, env=env)
+        logger.info(f"Successfully cloned {repo_path} to {temp_dir}.")
+
+    except git.GitCommandError as excinfo:
+        raise ValueError(f"Git clone error: {excinfo}") from excinfo
+
+    except Exception as excinfo:
+        raise (f"Error cloning git repository: {excinfo}")
+
+    validate_file_permissions(temp_dir)
+
+
+def find_git_executable() -> Optional[Path]:
+    """Find the path to the git executable, if available."""
+
+    git_exec_path = os.environ.get("GIT_PYTHON_GIT_EXECUTABLE")
+
+    if git_exec_path:
+        return Path(git_exec_path)
+
+    # For Windows, set default known location for git executable
+    if platform.system() == "Windows":
+        default_windows_path = Path("C:\\Program Files\\Git\\cmd\\git.EXE")
+        if default_windows_path.exists():
+            return default_windows_path
+
+    # For other OS (including Linux), set executable by looking into PATH
+    paths = os.environ["PATH"].split(os.pathsep)
+    for path in paths:
+        git_path = Path(path) / "git"
+        if git_path.exists():
+            return git_path
+
+    return None
+
+
+def validate_git_executable(git_exec_path: Optional[str]) -> None:
+    """Validate the path to the git executable."""
+    if not git_exec_path or not Path(git_exec_path).exists():
+        raise ValueError(f"Git executable not found at {git_exec_path}")
+
+
+def validate_file_permissions(temp_dir: Path) -> None:
+    """Validates file permissions of the cloned repository."""
+    if platform.system() != "Windows":
+        if isinstance(temp_dir, str):
+            temp_dir = Path(temp_dir)
+        permissions = temp_dir.stat().st_mode & 0o777
+        if permissions != 0o700:
+            raise ValueError(
+                "Error: file permissions of cloned repository must be set to 0o700."
+            )
 
 
 def get_github_file_link(file: str, user_repo_name: str) -> str:
@@ -32,7 +92,7 @@ def get_user_repository_name(url) -> str:
         username, reponame = match.groups()
         return f"{username}/{reponame}"
     else:
-        return "Invalid remote git URL."
+        raise ("Error: invalid remote repository URL.")
 
 
 def adjust_max_tokens(max_tokens: int, prompt: str, target: str = "Hello!") -> int: