app-store/server/scanner.py

import os
import re
import subprocess
import zipfile
import xml.etree.ElementTree as ET
from typing import Dict, List, Optional
from pathlib import Path


def parse_manifest(apk_path: str) -> Dict:
    """Extract metadata from an APK by reading AndroidManifest.xml directly."""
    metadata = {
        "package_name": None,
        "version_name": None,
        "version_code": None,
        "label": None,
        "icon": None,
        "min_sdk": None,
        "target_sdk": None,
        "permissions": [],
    }

    try:
        with zipfile.ZipFile(apk_path) as z:
            # Check for compiled manifest (binary XML)
            if "AndroidManifest.xml" in z.namelist():
                # Binary XML - use aapt for reliable parsing
                metadata = _parse_binary_manifest(apk_path, metadata)
            else:
                # No manifest found
                return metadata

            # Extract package name from filename if not found in manifest
            if not metadata["package_name"]:
                apk_name = Path(apk_path).stem
                metadata["package_name"] = apk_name

            # Look for icon resource (fallback if aapt didn't find one)
            if not metadata.get("icon"):
                icon_res = _find_icon_resource(apk_path)
                if icon_res:
                    metadata["icon"] = icon_res

    except Exception as e:
        print(f"Error parsing {apk_path}: {e}")

    return metadata


def parse_manifest_bytes(data: bytes) -> Dict:
    """Extract metadata from APK bytes (for in-memory validation)."""
    metadata = {
        "package_name": None,
        "version_name": None,
        "version_code": None,
        "label": None,
        "icon": None,
        "min_sdk": None,
        "target_sdk": None,
        "permissions": [],
    }

    try:
        import tempfile
        with tempfile.NamedTemporaryFile(suffix=".apk", delete=False) as tmp:
            tmp.write(data)
            tmp_path = tmp.name
        try:
            with zipfile.ZipFile(tmp_path) as z:
                if "AndroidManifest.xml" in z.namelist():
                    metadata = _parse_binary_manifest(tmp_path, metadata)
        finally:
            os.unlink(tmp_path)
    except Exception as e:
        print(f"Error parsing APK bytes: {e}")

    return metadata


def _parse_binary_manifest(data_or_path, metadata: Dict) -> Dict:
    """Parse binary XML manifest using aapt for reliable extraction.

    Accepts either a file path (str) or raw APK bytes.
    """
    import tempfile

    if isinstance(data_or_path, str):
        # Already a file path, use directly with aapt
        result = _parse_with_aapt(data_or_path)
        if result:
            metadata.update(result)
        return metadata

    # Raw bytes - write temp APK for aapt
    try:
        with tempfile.NamedTemporaryFile(suffix=".apk", delete=False) as tmp:
            tmp.write(data_or_path)
            tmp_path = tmp.name

        result = _parse_with_aapt(tmp_path)
        os.unlink(tmp_path)

        if result:
            metadata.update(result)
    except Exception as e:
        print(f"Error parsing binary manifest with aapt: {e}")
        # Fallback: basic string extraction if aapt fails
        _parse_fallback(data_or_path, metadata)

    return metadata


def _parse_with_aapt(apk_path: str) -> Optional[Dict]:
    """Use aapt or aapt2 to extract manifest info reliably."""
    for tool in ["aapt2", "aapt"]:
        try:
            result = subprocess.run(
                [tool, "dump", "badging", apk_path],
                capture_output=True, text=True, timeout=30
            )
            if result.returncode == 0:
                return _parse_aapt_output(result.stdout)
        except (FileNotFoundError, subprocess.TimeoutExpired):
            continue
    return None


def _parse_aapt_output(output: str) -> Dict:
    """Parse aapt dump badging output."""
    result = {
        "package_name": None,
        "version_name": None,
        "version_code": None,
        "label": None,
        "icon": None,
        "min_sdk": None,
        "target_sdk": None,
        "permissions": [],
    }

    for line in output.splitlines():
        # Package: name='com.example.app'
        m = re.search(r"package: name='([^']+)'", line)
        if m:
            result["package_name"] = m.group(1)
        # versionName='1.0.0'
        m = re.search(r"versionName='([^']+)'", line)
        if m:
            result["version_name"] = m.group(1)
        # versionCode='1'
        m = re.search(r"versionCode='([^']+)'", line)
        if m:
            result["version_code"] = m.group(1)
        # application-label:'My App'
        m = re.search(r"application-label:'([^']*)'", line)
        if m:
            result["label"] = m.group(1)
        # application-icon-128:'res/path'
        m = re.search(r"application-icon-\d+:'([^']+)'", line)
        if m:
            result["icon"] = m.group(1)
        # sdkVersion:'21' / minSdkVersion:'21' / targetSdkVersion:'33'
        m = re.search(r"minSdkVersion='([^']+)'", line)
        if m:
            result["min_sdk"] = m.group(1)
        m = re.search(r"targetSdkVersion='([^']+)'", line)
        if m:
            result["target_sdk"] = m.group(1)
        # uses-permission: name='android.permission.INTERNET'
        m = re.search(r"uses-permission: name='([^']+)'", line)
        if m:
            result["permissions"].append(m.group(1))

    return result


def _parse_fallback(data: bytes, metadata: Dict) -> None:
    """Fallback string-based extraction when aapt is unavailable."""
    try:
        if len(data) < 20:
            return

        if data[0:4] != b'\x03\x00\x08\x00' and data[0:4] != b'\x00\x08\x00\x03':
            return

        strings = _extract_strings_fallback(data)

        for s in strings:
            if s and "." in s and not s.startswith("android"):
                if re.match(r'^[a-zA-Z][a-zA-Z0-9_.]*$', s) and len(s.split('.')) >= 2:
                    if metadata["package_name"] is None:
                        metadata["package_name"] = s
                        break

        for s in strings:
            if s and re.match(r'^\d+\.\d+', s):
                if metadata["version_name"] is None:
                    metadata["version_name"] = s
                    break

    except Exception as e:
        print(f"Error in fallback manifest parsing: {e}")


def _extract_strings_fallback(data: bytes) -> List[str]:
    """Extract printable strings >= 4 chars from binary data."""
    strings = []
    current = []
    for byte in data:
        if 32 <= byte < 127:
            current.append(chr(byte))
        else:
            if len(current) >= 4:
                strings.append(''.join(current))
            current = []
    if len(current) >= 4:
        strings.append(''.join(current))
    return strings


def _find_icon_resource(apk_path: str) -> Optional[str]:
    """Find icon resource path in APK."""
    try:
        with zipfile.ZipFile(apk_path) as z:
            names = z.namelist()
            # Look for icon in res/drawable-* folders
            for name in names:
                if "ic_launcher" in name and name.endswith(".png"):
                    return name
                if "icon" in name and name.endswith(".png"):
                    return name
    except Exception:
        pass
    return None


def scan_repository(repo_path: str) -> List[Dict]:
    """Scan repository directory for APKs and return metadata."""
    apps = []
    repo_dir = Path(repo_path)

    if not repo_dir.exists():
        return apps

    for apk_file in sorted(repo_dir.glob("*.apk")):
        metadata = parse_manifest(str(apk_file))

        if not metadata["package_name"]:
            continue

        app = {
            "id": apk_file.stem,
            "package_name": metadata["package_name"],
            "version_name": str(metadata["version_name"]) if metadata["version_name"] else "unknown",
            "version_code": str(metadata["version_code"]) if metadata["version_code"] else "0",
            "name": metadata["label"] or apk_file.stem,
            "description": "",
            "icon": metadata["icon"],
            "size": apk_file.stat().st_size,
            "file_path": str(apk_file),
            "min_sdk": str(metadata["min_sdk"]) if metadata["min_sdk"] else None,
            "target_sdk": str(metadata["target_sdk"]) if metadata["target_sdk"] else None,
            "permissions": metadata["permissions"],
        }

        # Check for README or info file alongside APK
        info_file = repo_dir / f"{apk_file.stem}.json"
        if info_file.exists():
            import json
            with open(info_file) as f:
                info = json.load(f)
            for key in ("name", "description", "icon", "package_name",
                        "version_name", "version_code", "min_sdk",
                        "target_sdk", "screenshots"):
                if key in info:
                    app[key] = str(info[key]) if key in ("version_name", "version_code", "min_sdk", "target_sdk") else info[key]
            if "permissions" in info:
                app["permissions"] = info["permissions"]

        apps.append(app)

    return apps