pysfi 0.1.10__py3-none-any.whl → 0.1.12__py3-none-any.whl

sfi/pyembedinstall/pyembedinstall.py

@@ -0,0 +1,629 @@
+"""Download Python embeddable package to a specific directory."""
+
+from __future__ import annotations
+
+import argparse
+import logging
+import platform
+import shutil
+import time
+import zipfile
+from contextlib import suppress
+from dataclasses import dataclass
+from functools import cached_property
+from pathlib import Path
+from typing import Final
+from urllib.error import HTTPError, URLError
+from urllib.request import Request, urlopen
+
+from sfi.pyprojectparse.pyprojectparse import Solution
+
+logging.basicConfig(level=logging.INFO, format="%(message)s")
+
+cwd = Path.cwd()
+logger = logging.getLogger(__name__)
+
+# Default cache directory
+_DEFAULT_CACHE_DIR = Path.home() / ".pysfi" / ".cache" / "embed-python"
+
+
+# Architecture mapping
+ARCH_DICT: Final = {
+    "amd64": "amd64",
+    "x86_64": "amd64",
+    "x86": "amd64",
+    "arm64": "arm64",
+    "aarch64": "arm64",
+}
+
+# User-Agent for HTTP requests
+USER_AGENT: Final = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
+
+
+@dataclass(frozen=True)
+class EmbedDownloader:
+    """Class for downloading Python embeddable package."""
+
+    version: str
+    arch: str
+    cache_dir: Path
+    offline: bool
+    skip_speed_test: bool
+    timeout: int = 5
+
+    @cached_property
+    def cache_file(self) -> Path:
+        """Get the cache file path for the embeddable package."""
+        return self.cache_dir / f"python-{self.version}-embed-{self.arch}.zip"
+
+    @cached_property
+    def download_urls(self) -> list[str]:
+        """Get the list of URLs to try for download."""
+        if self.skip_speed_test:
+            url = self.get_official_url_template(self.arch).format(version=self.version)
+            logger.info(f"Skipping speed test, using official URL: {url}")
+            return [url]
+        return self.find_available_urls(self.version, self.arch, self.timeout)
+
+    def download(self) -> bool:
+        """Download the Python embeddable package to the cache file."""
+        return self.has_cached_file or self.download_python_package(
+            self.cache_file, self.download_urls, self.offline
+        )
+
+    @cached_property
+    def has_cached_file(self) -> bool:
+        """Check if the embeddable package is already cached."""
+        if not self.cache_file.exists():
+            return False
+
+        is_valid = zipfile.is_zipfile(self.cache_file)
+        if not is_valid:
+            logger.warning(f"Corrupted cache file detected: {self.cache_file}")
+            logger.info("Deleting corrupted cache file and re-downloading...")
+            try:
+                self.cache_file.unlink()
+            except OSError as e:
+                logger.warning(f"Failed to delete corrupted cache file: {e}")
+        else:
+            logger.debug(f"Using cached file: {self.cache_file}")
+
+        return is_valid
+
+    @cached_property
+    def latest_version(self) -> str | None:
+        """Get the latest version of Python available for download."""
+        version = self.get_latest_patch_version(self.version, self.arch, self.cache_dir)
+        return version if "." in version else None
+
+    def download_with_progress(self, url: str, dest_path: Path) -> None:
+        """Download file with progress bar and integrity check."""
+        logger.info(f"Downloading from: {url}")
+
+        req = Request(url, headers={"User-Agent": USER_AGENT})
+        try:
+            with urlopen(req, timeout=self.timeout) as response:
+                total_size = int(response.headers.get("content-length", 0))
+                dest_path.parent.mkdir(parents=True, exist_ok=True)
+
+                downloaded = self.download_file_chunks(response, dest_path, total_size)
+                logger.info(f"Download completed: {downloaded / 1024 / 1024:.2f} MB")
+        except HTTPError as e:
+            logger.error(f"HTTP error {e.code} while downloading {url}")
+            raise
+        except URLError as e:
+            logger.error(f"Failed to download {url}: {e}")
+            raise
+
+    def download_file_chunks(self, response, dest_path: Path, total_size: int) -> int:
+        """Download file chunks and return downloaded size."""
+        downloaded = 0
+        last_progress = 0
+        last_log_time = 0
+        chunk_size = 65536
+
+        with dest_path.open("wb") as f:
+            while chunk := response.read(chunk_size):
+                f.write(chunk)
+                downloaded += len(chunk)
+
+                if total_size > 0:
+                    progress = (downloaded / total_size) * 100
+                    current_time = time.time()
+                    # Update progress every 1% or every 0.5 seconds, whichever comes first
+                    if (
+                        int(progress) > last_progress
+                        and current_time - last_log_time >= 0.5
+                    ) or downloaded == total_size:
+                        downloaded_mb = downloaded / 1024 / 1024
+                        total_mb = total_size / 1024 / 1024
+                        logger.info(
+                            f"Progress: {downloaded_mb:.2f} MB / {total_mb:.2f} MB ({progress:.1f}%)"
+                        )
+                        last_progress = int(progress)
+                        last_log_time = current_time
+
+        if total_size > 0 and downloaded != total_size:
+            raise URLError(
+                f"Incomplete download: expected {total_size} bytes, got {downloaded} bytes"
+            )
+        return downloaded
+
+    def try_download_from_url(self, cache_file: Path, url: str) -> bool:
+        """Try to download from a single URL and validate."""
+        try:
+            self.download_with_progress(url, cache_file)
+        except (URLError, HTTPError):
+            # Remove partial file on download failure
+            if cache_file.exists():
+                with suppress(OSError):
+                    cache_file.unlink()
+            raise
+
+        is_valid = zipfile.is_zipfile(cache_file)
+
+        if is_valid:
+            logger.info("Downloaded file is valid")
+        else:
+            logger.warning("Downloaded file is corrupted, trying next URL...")
+            if cache_file.exists():
+                with suppress(OSError):
+                    cache_file.unlink()
+
+        return is_valid
+
+    def download_python_package(
+        self, cache_file: Path, urls: list[str], offline: bool
+    ) -> bool:
+        """Download Python package from available URLs."""
+        if offline:
+            logger.error("Offline mode: no cached file found")
+            return False
+
+        if not urls:
+            logger.error("No available URLs to download from")
+            return False
+
+        # Attempt download from each URL
+        for i, url in enumerate(urls, 1):
+            try:
+                logger.info(f"Attempting download {i}/{len(urls)}: {url}")
+                if self.try_download_from_url(cache_file, url):
+                    return True
+            except (URLError, HTTPError) as e:
+                logger.warning(f"Download failed from {url}: {e}")
+                if i < len(urls):
+                    logger.info("Retrying with next available URL...")
+
+        # All downloads failed
+        logger.error(
+            "Failed to download Python from all available URLs. "
+            "Please check your internet connection and version."
+        )
+        return False
+
+    def get_official_url_template(self, arch: str | None = None) -> str:
+        """Get official URL template based on architecture."""
+        arch = arch or self.arch
+        return f"https://www.python.org/ftp/python/{{version}}/python-{{version}}-embed-{arch}.zip"
+
+    def get_mirror_url_templates(self, arch: str | None = None) -> list[str]:
+        """Get mirror URL templates based on architecture."""
+        arch = arch or self.arch
+        return [
+            f"https://mirrors.huaweicloud.com/python/{{version}}/python-{{version}}-embed-{arch}.zip",
+            f"https://mirrors.aliyun.com/python-release/{{version}}/python-{{version}}-embed-{arch}.zip",
+            f"https://mirrors.tuna.tsinghua.edu.cn/python/{{version}}/python-{{version}}-embed-{arch}.zip",
+            f"https://mirrors.pku.edu.cn/python/{{version}}/python-{{version}}-embed-{arch}.zip",
+        ]
+
+    def test_url_speed(
+        self, url: str, timeout: int | None = None, use_get: bool = False
+    ) -> float:
+        """Test URL speed by making a HEAD or GET request."""
+        timeout = timeout or self.timeout
+        start_time = time.time()
+
+        # First attempt HEAD request
+        if not use_get:
+            req = Request(url, headers={"User-Agent": USER_AGENT}, method="HEAD")
+            try:
+                with urlopen(req, timeout=timeout) as response:
+                    if response.status == 200:
+                        elapsed = time.time() - start_time
+                        logger.debug(f"URL {url}: {elapsed:.3f}s")
+                        return elapsed
+                    logger.debug(f"URL {url}: HTTP {response.status}")
+            except HTTPError as e:
+                if e.code in (405, 403):
+                    use_get = True  # Fall back to GET request
+                else:
+                    logger.debug(f"URL {url}: HTTP {e.code}")
+                    return float("inf")
+            except (URLError, TimeoutError) as e:
+                logger.debug(f"URL {url}: {type(e).__name__}")
+                return float("inf")
+
+        # Fallback to GET request if needed
+        if use_get:
+            req = Request(
+                url,
+                headers={"User-Agent": USER_AGENT, "Range": "bytes=0-1023"},
+                method="GET",
+            )
+            try:
+                with urlopen(req, timeout=timeout) as response:
+                    if response.status in (206, 200):
+                        elapsed = time.time() - start_time
+                        logger.debug(f"URL {url}: {elapsed:.3f}s (GET fallback)")
+                        return elapsed
+                    logger.debug(f"URL {url}: HTTP {response.status} (GET fallback)")
+            except (URLError, TimeoutError) as e:
+                logger.debug(f"URL {url}: {type(e).__name__} (GET fallback)")
+
+        return float("inf")
+
+    def get_urls_to_test(
+        self, version: str | None = None, arch: str | None = None
+    ) -> list[str]:
+        """Get list of URLs to test for speed."""
+        version = version or self.version
+        arch = arch or self.arch
+        return [
+            template.format(version=version)
+            for template in [
+                *self.get_mirror_url_templates(arch),
+                self.get_official_url_template(arch),
+            ]
+        ]
+
+    def find_available_urls(
+        self,
+        version: str | None = None,
+        arch: str | None = None,
+        timeout: int | None = None,
+    ) -> list[str]:
+        """Find all available URLs for downloading Python embeddable package, sorted by speed."""
+        version = version or self.version
+        arch = arch or self.arch
+        timeout = timeout or self.timeout
+        official_url = self.get_official_url_template(arch).format(version=version)
+
+        logger.debug("Testing mirror speeds...")
+        all_tests = [
+            (url, self.test_url_speed(url, timeout))
+            for url in self.get_urls_to_test(version, arch)
+        ]
+        speed_results = [
+            (speed, url) for url, speed in all_tests if speed != float("inf")
+        ]
+
+        # Log results
+        for speed, url in speed_results:
+            logger.debug(f"  ✓ {url} ({speed:.3f}s)")
+        for url, speed in all_tests:
+            if speed == float("inf"):
+                logger.debug(f"  ✗ {url} (unavailable)")
+
+        if not speed_results:
+            logger.warning("All mirrors failed, falling back to official URL")
+            return [official_url]
+
+        # Sort URLs by speed and return
+        available_urls = [url for _, url in sorted(speed_results)]
+        if available_urls and logger.isEnabledFor(logging.DEBUG):
+            logger.debug(
+                f"Available URLs (sorted by speed): {len(available_urls)} found"
+            )
+            for i, url in enumerate(available_urls, 1):
+                logger.debug(f"  {i}. {url}")
+        return available_urls
+
+    # Methods moved from EmbedInstallConfig
+    def is_version_complete(self, version: str) -> bool:
+        """Check if version string is complete (has 3 or more parts)."""
+        version_parts = version.split(".")
+        return len(version_parts) >= 3
+
+    def validate_version_format(self, version: str) -> bool:
+        """Validate if version has correct major.minor format."""
+        version_parts = version.split(".")
+        return "." in version and len(version_parts) == 2
+
+    def extract_version_from_filename(self, filename: str, arch: str) -> str:
+        """Extract version from cache filename."""
+        return filename.replace("python-", "").replace(f"-embed-{arch}", "")
+
+    def get_cached_version(
+        self, major: str, minor: str, arch: str, cache_dir: Path
+    ) -> str | None:
+        """Get the highest cached version for a given major.minor."""
+        if not cache_dir.exists():
+            return None
+
+        pattern = f"python-{major}.{minor}.*-embed-{arch}.zip"
+        cached_versions = [
+            self.extract_version_from_filename(cache_file.stem, arch)
+            for cache_file in cache_dir.glob(pattern)
+            if cache_file.stem and zipfile.is_zipfile(cache_file)
+        ]
+
+        return max(cached_versions, default=None) if cached_versions else None
+
+    def check_version_url(self, version: str, arch: str, timeout: int) -> bool:
+        """Check if version URL is accessible."""
+        base_url = self.get_official_url_template(arch)
+        test_url = base_url.format(version=version)
+        req = Request(test_url, headers={"User-Agent": USER_AGENT}, method="HEAD")
+
+        try:
+            with urlopen(req, timeout=timeout) as response:
+                if response.status == 200:
+                    return True
+                logger.debug(f"Version {version}: HTTP {response.status}")
+                return False
+        except HTTPError as e:
+            log_level = logger.info if e.code == 404 else logger.debug
+            log_level(f"Version {version}: HTTP {e.code}")
+        except (URLError, TimeoutError) as e:
+            logger.debug(f"Version {version}: {type(e).__name__}")
+
+        return False
+
+    def find_latest_patch_online(
+        self, major: str, minor: str, arch: str, timeout: int
+    ) -> str | None:
+        """Find latest patch version by checking online."""
+        # Check patches from highest to lowest (50 down to 0)
+        for patch in range(50, -1, -1):
+            version = f"{major}.{minor}.{patch}"
+            if self.check_version_url(version, arch, timeout):
+                logger.info(f"Latest version found: {version}")
+                return version
+        return None
+
+    def resolve_patch_version(
+        self, major_minor: str, arch: str, cache_dir: Path, timeout: int
+    ) -> str:
+        """Resolve version to latest patch or return original."""
+        major, minor = major_minor.split(".")
+
+        # Check for cached version first
+        if cached_version := self.get_cached_version(major, minor, arch, cache_dir):
+            logger.info(f"Using cached version: {cached_version}")
+            return cached_version
+
+        # Otherwise find latest online
+        logger.info(f"Checking latest version for {major_minor}...")
+        latest_version = self.find_latest_patch_online(major, minor, arch, timeout)
+
+        if latest_version:
+            return latest_version
+
+        logger.warning(
+            f"Could not find any patch version for {major_minor}, using as-is"
+        )
+        return major_minor
+
+    def get_latest_patch_version(
+        self, major_minor: str, arch: str, cache_dir: Path, timeout: int = 5
+    ) -> str:
+        """Get the latest patch version for a given major.minor version.
+
+        Args:
+            major_minor: Major.minor version (e.g., '3.13') or full version (e.g., '3.13.1')
+            arch: Architecture (amd64 or arm64)
+            cache_dir: Cache directory to check for cached versions
+            timeout: Request timeout in seconds
+
+        Returns:
+            Full version string (e.g., '3.13.1') or original if already complete
+        """
+        # Return early if version is already complete or invalid format
+        if self.is_version_complete(major_minor) or not self.validate_version_format(
+            major_minor
+        ):
+            log_msg = (
+                f"Version already complete: {major_minor}, skipping auto-completion"
+                if self.is_version_complete(major_minor)
+                else f"Invalid version format: {major_minor}, using as-is"
+            )
+            logger.debug(log_msg)
+            return major_minor
+
+        return self.resolve_patch_version(major_minor, arch, cache_dir, timeout)
+
+
+@dataclass(frozen=True)
+class EmbedInstaller:
+    """Class for installing Python embeddable package."""
+
+    root_dir: Path
+    version: str
+    cache_dir: Path
+    offline: bool
+    skip_speed_test: bool
+    keep_cache: bool
+    timeout: int = 5
+
+    def __post_init__(self):
+        pass
+
+    @cached_property
+    def solution(self) -> Solution:
+        """Get the solution from the target directory."""
+        return Solution.from_directory(self.root_dir)
+
+    @cached_property
+    def downloader(self) -> EmbedDownloader:
+        """Get the embed downloader for the given version and architecture."""
+        return EmbedDownloader(
+            version=self.version,
+            arch=self.arch,
+            cache_dir=self.cache_dir,
+            offline=self.offline,
+            skip_speed_test=self.skip_speed_test,
+            timeout=self.timeout,
+        )
+
+    @cached_property
+    def arch(self) -> str:
+        """Get the architecture of the machine."""
+        machine_arch = platform.machine().lower()
+        return ARCH_DICT.get(machine_arch, "amd64")
+
+    @cached_property
+    def runtime_dir(self) -> Path:
+        """Get the runtime directory."""
+        runtime_path = self.root_dir / "dist" / "runtime"
+        runtime_path.mkdir(parents=True, exist_ok=True)
+        return runtime_path
+
+    @cached_property
+    def python_exe_path(self) -> Path:
+        """Get the path to python.exe in the target directory."""
+        return self.runtime_dir / "python.exe"
+
+    @cached_property
+    def cache_file(self) -> Path:
+        """Get the path to the cache file."""
+        return self.downloader.cache_file
+
+    def extract_package(self) -> bool:
+        """Extract package to target directory."""
+        if not zipfile.is_zipfile(self.cache_file):
+            logger.error(f"Invalid zip file: {self.cache_file}")
+            return False
+
+        try:
+            logger.info(f"Extracting to: {self.runtime_dir}")
+            self.runtime_dir.mkdir(parents=True, exist_ok=True)
+
+            # First get file count for progress
+            with zipfile.ZipFile(self.cache_file, "r") as zip_ref:
+                file_count = len(zip_ref.namelist())
+
+            # Extract the archive
+            shutil.unpack_archive(str(self.cache_file), str(self.runtime_dir), "zip")
+            logger.info(f"Extracted {file_count} files")
+            return True
+        except (zipfile.BadZipFile, OSError) as e:
+            logger.error(f"Failed to extract: {e}")
+            return False
+
+    def cleanup_cache(self) -> None:
+        """Clean up cache file if not keeping it."""
+        if not self.keep_cache and not self.offline:
+            try:
+                if self.cache_file.exists():
+                    logger.debug(f"Cleaning up cache: {self.cache_file}")
+                    self.cache_file.unlink()
+            except OSError as e:
+                logger.warning(f"Failed to clean up cache file: {e}")
+
+    def install_package(self) -> bool:
+        """Install Python embeddable package to target directory using embedded methods."""
+        # Download the package (latest_version is resolved automatically during download)
+        if not self.downloader.download():
+            logger.error("Failed to download Python embeddable package")
+            return False
+
+        # Extract the package to runtime directory
+        if not self.extract_package():
+            logger.error("Failed to extract Python embeddable package")
+            return False
+
+        # Clean up cache if not keeping it
+        self.cleanup_cache()
+
+        # Verify installation by checking for python.exe
+        success = self.python_exe_path.exists()
+        if success:
+            logger.info(f"Installation completed, python.exe: {self.python_exe_path}")
+        else:
+            logger.error("Installation failed, python.exe not found")
+
+        return success
+
+
+def parse_args() -> argparse.Namespace:
+    """Create and configure argument parser for CLI."""
+    parser = argparse.ArgumentParser(
+        prog="pyembedinstall",
+        description="Download and install Python embeddable package to a specific directory.",
+    )
+    parser.add_argument(
+        "directory",
+        type=str,
+        nargs="?",
+        default=str(cwd),
+        help="Directory to install Python embeddable package (default: current directory)",
+    )
+    parser.add_argument("--debug", "-d", action="store_true", help="Enable debug mode")
+    parser.add_argument(
+        "--offline",
+        "-o",
+        action="store_true",
+        help="Offline mode (use cached files only)",
+    )
+    parser.add_argument(
+        "--version",
+        "-v",
+        type=str,
+        default="3.8.10",
+        help="Python version to install (default: 3.8.10)",
+    )
+    parser.add_argument(
+        "--cache-dir",
+        "-C",
+        type=str,
+        default=str(_DEFAULT_CACHE_DIR),
+        help="Cache directory for downloaded files",
+    )
+    parser.add_argument(
+        "--keep-cache",
+        "-k",
+        action="store_true",
+        default=True,
+        help="Keep downloaded cache files",
+    )
+    parser.add_argument(
+        "--skip-speed-test",
+        "-s",
+        action="store_true",
+        help="Skip speed test and use official URL directly",
+    )
+    parser.add_argument(
+        "--timeout",
+        "-t",
+        type=int,
+        default=5,
+        help="Timeout in seconds for URL speed test (default: 5)",
+    )
+    return parser.parse_args()
+
+
+def main() -> None:
+    """Main entry point for pyembedinstall CLI."""
+    args = parse_args()
+
+    if args.debug:
+        logger.setLevel(logging.DEBUG)
+
+    cache_dir = Path(args.cache_dir)
+    cache_dir.mkdir(parents=True, exist_ok=True)
+
+    t0 = time.perf_counter()
+
+    if not EmbedInstaller(
+        root_dir=Path(args.directory),
+        version=args.version,
+        cache_dir=cache_dir,
+        offline=args.offline,
+        skip_speed_test=args.skip_speed_test,
+        keep_cache=args.keep_cache,
+        timeout=args.timeout,
+    ).install_package():
+        exit(1)
+
+    logger.info(f"Installation completed in {time.perf_counter() - t0:.4f} seconds")