salim0986/graph-bug-ai

1"""2Security utilities for GraphBug AI Service3Provides path traversal protection, input sanitization, and validation helpers4"""5 6import os7import re8from pathlib import Path9from typing import Optional10from pydantic import BaseModel, field_validator, Field11 12 13def safe_join(base_path: str, *paths: str) -> str:14    """15    Safely join paths and prevent directory traversal attacks16    17    Args:18        base_path: The base directory that must contain the result19        *paths: Path components to join20        21    Returns:22        str: Safe joined path23        24    Raises:25        ValueError: If path traversal is detected26        27    Example:28        >>> safe_join("/app/repos", "user_repo", "file.py")29        "/app/repos/user_repo/file.py"30        31        >>> safe_join("/app/repos", "../etc/passwd")32        ValueError: Path traversal detected33    """34    base = Path(base_path).resolve()35    target = (base / Path(*paths)).resolve()36    37    # Ensure target is within base38    try:39        target.relative_to(base)40    except ValueError:41        raise ValueError(f"Path traversal detected: {target} is outside {base}")42    43    return str(target)44 45 46def sanitize_repo_id(repo_id: str) -> str:47    """48    Sanitize repository ID to prevent injection attacks49    50    Args:51        repo_id: Repository identifier52        53    Returns:54        str: Sanitized repo ID (alphanumeric, dash, underscore only)55        56    Raises:57        ValueError: If repo_id contains invalid characters58    """59    # Allow only alphanumeric, dash, and underscore60    if not re.match(r'^[a-zA-Z0-9_-]+$', repo_id):61        raise ValueError(f"Invalid repo_id: {repo_id}. Only alphanumeric, dash, and underscore allowed.")62    63    if len(repo_id) > 200:64        raise ValueError(f"repo_id too long: {len(repo_id)} characters (max 200)")65    66    return repo_id67 68 69def sanitize_owner_repo(owner: str, repo: str) -> tuple[str, str]:70    """71    Sanitize GitHub owner and repo names72    73    Args:74        owner: GitHub username/org75        repo: Repository name76        77    Returns:78        tuple: (sanitized_owner, sanitized_repo)79        80    Raises:81        ValueError: If inputs contain invalid characters82    """83    # GitHub usernames: alphanumeric and dash84    if not re.match(r'^[a-zA-Z0-9-]+$', owner):85        raise ValueError(f"Invalid owner: {owner}")86    87    # GitHub repo names: alphanumeric, dash, underscore, dot88    if not re.match(r'^[a-zA-Z0-9._-]+$', repo):89        raise ValueError(f"Invalid repo: {repo}")90    91    if len(owner) > 100 or len(repo) > 100:92        raise ValueError("Owner or repo name too long (max 100 characters)")93    94    return owner, repo95 96 97def sanitize_file_path(file_path: str, max_length: int = 500) -> str:98    """99    Sanitize file path for safe operations100    101    Args:102        file_path: File path to sanitize103        max_length: Maximum allowed path length104        105    Returns:106        str: Sanitized file path107        108    Raises:109        ValueError: If path is invalid110    """111    # Check for null bytes112    if '\0' in file_path:113        raise ValueError("Null byte detected in file path")114    115    # Check length116    if len(file_path) > max_length:117        raise ValueError(f"File path too long: {len(file_path)} > {max_length}")118    119    # Normalize path to prevent tricks like "///" or "\.\"120    normalized = os.path.normpath(file_path)121    122    # Check for parent directory references123    if normalized.startswith('..') or '/..' in normalized or '\\..' in normalized:124        raise ValueError(f"Parent directory reference detected: {file_path}")125    126    return normalized127 128 129# Pydantic Models for Input Validation130 131class GitHubIdentifierModel(BaseModel):132    """Base model for GitHub identifiers with validation"""133    owner: str = Field(..., min_length=1, max_length=100, pattern=r'^[a-zA-Z0-9-]+$')134    repo: str = Field(..., min_length=1, max_length=100, pattern=r'^[a-zA-Z0-9._-]+$')135 136 137class PRNumberModel(BaseModel):138    """Model for PR number validation"""139    pr_number: int = Field(..., gt=0, lt=1000000, description="PR number must be between 1 and 999999")140 141 142class RepoIdModel(BaseModel):143    """Model for repository ID validation"""144    repo_id: str = Field(..., min_length=1, max_length=200, pattern=r'^[a-zA-Z0-9_-]+$')145 146 147class FilePathModel(BaseModel):148    """Model for file path validation"""149    file_path: str = Field(..., min_length=1, max_length=500)150    151    @field_validator('file_path')152    @classmethod153    def validate_file_path(cls, v: str) -> str:154        """Validate file path for security"""155        return sanitize_file_path(v)156 157 158class CommitSHAModel(BaseModel):159    """Model for Git commit SHA validation"""160    sha: str = Field(..., min_length=7, max_length=40, pattern=r'^[a-fA-F0-9]+$')161 162 163def validate_installation_id(installation_id: str) -> str:164    """165    Validate GitHub App installation ID166    167    Args:168        installation_id: Installation ID to validate169        170    Returns:171        str: Validated installation ID172        173    Raises:174        ValueError: If installation_id is invalid175    """176    # Installation IDs should be numeric strings177    if not installation_id.isdigit():178        raise ValueError(f"Invalid installation_id: {installation_id}")179    180    if len(installation_id) > 20:181        raise ValueError("Installation ID too long")182    183    return installation_id184 185 186def sanitize_error_message(error: Exception, include_details: bool = False) -> dict:187    """188    Sanitize error messages to prevent information leakage189    190    Args:191        error: Exception to sanitize192        include_details: Whether to include stack trace (dev mode only)193        194    Returns:195        dict: Sanitized error response196    """197    # Generic error types198    error_type = type(error).__name__199    200    # Don't expose internal paths or sensitive details201    if include_details:202        return {203            "error": str(error),204            "type": error_type205        }206    else:207        # Production: generic error messages208        if "permission" in str(error).lower() or "access" in str(error).lower():209            return {"error": "Access denied", "code": "ACCESS_DENIED"}210        elif "not found" in str(error).lower():211            return {"error": "Resource not found", "code": "NOT_FOUND"}212        else:213            return {"error": "An error occurred", "code": "INTERNAL_ERROR"}214 215 216# Request size limits217MAX_REQUEST_SIZE = 10 * 1024 * 1024  # 10 MB218MAX_CODE_SNIPPET_SIZE = 100 * 1024   # 100 KB219MAX_DIFF_SIZE = 500 * 1024            # 500 KB220 221 222def validate_request_size(content_length: Optional[int]) -> None:223    """224    Validate request size to prevent DOS attacks225    226    Args:227        content_length: Content-Length header value228        229    Raises:230        ValueError: If request is too large231    """232    if content_length and content_length > MAX_REQUEST_SIZE:233        raise ValueError(f"Request too large: {content_length} bytes (max {MAX_REQUEST_SIZE})")234