knee_seg/tools/scan_imports.py

# tools/scan_imports.py
import argparse
import ast
import sys
from pathlib import Path
from collections import defaultdict

# Маппинг модуль→PyPI-пакет
MODULE_TO_PYPI = {
    # Научный стек
    "numpy": "numpy",
    "cv2": "opencv-python",
    "skimage": "scikit-image",
    "matplotlib": "matplotlib",
    "PIL": "pillow",
    "pillow": "pillow",
    "pydicom": "pydicom",
    # GUI
    "PyQt5": "PyQt5",
    # Утилиты
    "yaml": "PyYAML",
    "bs4": "beautifulsoup4",
    "lxml": "lxml",
    "dateutil": "python-dateutil",
    "dotenv": "python-dotenv",
    "yattag": "yattag",
    # Сеть
    "requests": "requests",
    "aiohttp": "aiohttp",
}

# Локальные пакеты по префиксам (дополняй при необходимости)
PROJECT_LOCAL_PREFIXES = {"knee"}

try:
    STDLIB = set(sys.stdlib_module_names)  # Python 3.10+
except Exception:
    STDLIB = set()

STDLIB.update({
    "typing", "pathlib", "json", "re", "subprocess", "shutil", "itertools",
    "functools", "collections", "dataclasses", "asyncio", "concurrent",
    "logging", "argparse", "base64", "hashlib", "hmac", "uuid",
    "tempfile", "time", "datetime", "math", "statistics", "http", "urllib",
    "xml", "csv", "sqlite3", "queue", "threading", "multiprocessing",
    "enum", "inspect", "traceback", "glob", "zipfile", "tarfile",
    "importlib", "pkgutil", "venv",
})

IGNORED = {"__future__"}


def top_level(name: str) -> str:
    return name.split(".")[0]


def find_py_files(src_dir: Path):
    for p in src_dir.rglob("*.py"):
        # пропускаем .venv
        if ".venv" in p.parts:
            continue
        yield p


def collect_imports(py_file: Path):
    try:
        tree = ast.parse(py_file.read_text(encoding="utf-8"))
    except Exception:
        return []
    mods = []
    for node in ast.walk(tree):
        if isinstance(node, ast.Import):
            for alias in node.names:
                mods.append(("abs", top_level(alias.name)))
        elif isinstance(node, ast.ImportFrom):
            # относительные импорты считаем локальными и игнорируем
            if getattr(node, "level", 0) and node.level > 0:
                continue
            if node.module:
                mods.append(("abs", top_level(node.module)))
    return mods


def discover_local_top_levels(src_dir: Path):
    """Имена, которые существуют в src как top-level пакет/модуль."""
    local = set()
    # Папки-пакеты
    for pkg_init in src_dir.rglob("__init__.py"):
        try:
            rel = pkg_init.parent.relative_to(src_dir)
        except ValueError:
            continue
        if rel.parts:
            local.add(rel.parts[0])
    # Одиночные модули
    for mod in src_dir.glob("*.py"):
        local.add(mod.stem)
    return local


def map_to_pypi(mods, local_names):
    result = defaultdict(int)
    for kind, m in mods:
        if m in IGNORED or m in STDLIB:
            continue
        if m in PROJECT_LOCAL_PREFIXES:
            continue
        if m in local_names:
            continue
        pkg = MODULE_TO_PYPI.get(m, m)
        result[pkg] += 1
    return result


def merge_with_existing(out_path: Path, counts: dict):
    existing = []
    if out_path.exists():
        existing = [
            line.strip()
            for line in out_path.read_text(encoding="utf-8").splitlines()
            if line.strip() and not line.strip().startswith("#")
        ]
    existing_pkgs = {line.split("==")[0].split(">=")[0] for line in existing}

    lines = list(existing)
    for pkg in sorted(counts.keys()):
        if pkg not in existing_pkgs:
            lines.append(pkg)
    return "\n".join(lines) + "\n"


def main():
    ap = argparse.ArgumentParser(description="Scan imports and update requirements.in")
    ap.add_argument("--src", default="src", help="Source directory to scan")
    ap.add_argument("--out", default="requirements.in", help="Output requirements.in path")
    ap.add_argument("--update", action="store_true", help="Update existing file instead of overwrite")
    args = ap.parse_args()

    src_dir = Path(args.src).resolve()
    out_path = Path(args.out).resolve()

    all_mods = []
    for py in find_py_files(src_dir):
        all_mods.extend(collect_imports(py))

    local_names = discover_local_top_levels(src_dir)
    counts = map_to_pypi(all_mods, local_names)

    if args.update and out_path.exists():
        content = merge_with_existing(out_path, counts)
    else:
        content = "\n".join(sorted(counts.keys())) + "\n"

    out_path.write_text(content, encoding="utf-8")
    print(f"[✓] requirements.in updated at {out_path}")


if __name__ == "__main__":
    main()