config.py

"""
config.py
Configuration and constants
"""

# ---------------------------------------------------------------------------
# Directories ignored during scan
# ---------------------------------------------------------------------------

IGNORE_DIRS_DEFAULT = (
    ".git,.idea,.vscode,__pycache__,.pytest_cache,node_modules,"
    "venv,.venv,env,build,dist,.next,.cache,"
    "site-packages,.mypy_cache,target,.gradle,"
    "bin,obj"
)

# ---------------------------------------------------------------------------
# Binary / ignored extensions
# ---------------------------------------------------------------------------

IGNORE_EXTENSIONS = {
    ".exe",
    ".dll",
    ".so",
    ".dylib",
    ".png",
    ".jpg",
    ".jpeg",
    ".gif",
    ".ico",
    ".svg",
    ".webp",
    ".zip",
    ".rar",
    ".7z",
    ".tar",
    ".gz",
    ".pdf",
    ".mp4",
    ".mp3",
    ".wav",
    ".mov",
    ".woff",
    ".woff2",
    ".ttf",
    ".eot",
    ".pyc",
    ".class",
    ".jar",
    ".db",
    ".sqlite3",
    ".lock",
    ".bin",
    ".dat",
}

# ---------------------------------------------------------------------------
# Extension → Language
# ---------------------------------------------------------------------------

LANGUAGE_MAP = {
    ".py": "Python",
    ".js": "JavaScript",
    ".jsx": "JavaScript (JSX)",
    ".ts": "TypeScript",
    ".tsx": "TypeScript (TSX)",
    ".html": "HTML",
    ".css": "CSS",
    ".scss": "SCSS",
    ".json": "JSON",
    ".yaml": "YAML",
    ".yml": "YAML",
    ".toml": "TOML",
    ".xml": "XML",
    ".md": "Markdown",
    ".sql": "SQL",
    ".java": "Java",
    ".go": "Go",
    ".rb": "Ruby",
    ".php": "PHP",
    ".c": "C",
    ".cpp": "C++",
    ".cs": "C#",
    ".swift": "Swift",
    ".kt": "Kotlin",
    ".sh": "Shell",
    ".txt": "Text",
    ".env": "Environment",
    ".ini": "INI",
    ".cfg": "Config",
    ".rst": "reStructuredText",
}

# ---------------------------------------------------------------------------
# Categories
# ---------------------------------------------------------------------------

EXTENSION_CATEGORIES = {
    "Python": {
        ".py",
    },
    "JavaScript / TypeScript": {
        ".js",
        ".jsx",
        ".ts",
        ".tsx",
    },
    "Web": {
        ".html",
        ".css",
        ".scss",
    },
    "Configuration": {
        ".json",
        ".yaml",
        ".yml",
        ".toml",
        ".xml",
        ".ini",
        ".cfg",
        ".env",
    },
    "Documentation": {
        ".md",
        ".txt",
        ".rst",
    },
    "Other Languages": {
        ".java",
        ".go",
        ".rb",
        ".php",
        ".c",
        ".cpp",
        ".cs",
        ".swift",
        ".kt",
        ".sql",
        ".sh",
    },
}

# ---------------------------------------------------------------------------
# Token estimation
# ---------------------------------------------------------------------------

CHARS_PER_TOKEN = 4.0

# ---------------------------------------------------------------------------
# Output defaults
# ---------------------------------------------------------------------------

DEFAULT_MAX_FILE_SIZE_MB = 2.0
DEFAULT_MAX_TOKENS = 0

OUTPUT_MARKDOWN = "Markdown"
OUTPUT_XML = "XML"

OUTPUT_MD_EXT = "md"
OUTPUT_XML_EXT = "xml"

PREVIEW_LIMIT = 3000

MAX_WORKERS = 16