refactor(specs): update operations for new domain architecture (3b5c3c8c) · Commits · Jan Reimes / 3gpp-crawler

src/tdoc_crawler/specs/downloads.py

+1 −1

Original line number	Diff line number	Diff line
		@@ -8,7 +8,7 @@ from pathlib import Path
		from zipinspect import HTTPZipReader

		from tdoc_crawler.config import resolve_cache_manager
		from tdoc_crawler.crawlers.constants import SPEC_URL_TEMPLATE
		from tdoc_crawler.constants.urls import SPEC_URL_TEMPLATE
		from tdoc_crawler.http_client import download_to_path
		from tdoc_crawler.specs.database import SpecDatabase
		from tdoc_crawler.specs.sources.base import SpecSource

src/tdoc_crawler/specs/operations/init.py

0 → 100644

+15 −0

Original line number	Diff line number	Diff line
		"""Spec operations."""

		from __future__ import annotations

		from tdoc_crawler.specs.operations.checkout import (
		build_default_spec_sources,
		checkout_specs,
		clear_checkout_specs,
		)

		__all__ = [
		"build_default_spec_sources",
		"checkout_specs",
		"clear_checkout_specs",
		]

src/tdoc_crawler/specs/operations/checkout.py

0 → 100644

+81 −0

Original line number	Diff line number	Diff line
		"""Spec checkout operations for downloading specification documents.

		This module provides functionality to download 3GPP specification documents
		from various sources and extract them to a local checkout folder.
		"""

		from __future__ import annotations

		import shutil
		from pathlib import Path
		from typing import cast

		from tdoc_crawler.specs.database import SpecDatabase
		from tdoc_crawler.specs.downloads import SpecDownloads
		from tdoc_crawler.specs.sources.base import FunctionSpecSource, SpecSource
		from tdoc_crawler.specs.sources.threegpp import fetch_threegpp_metadata
		from tdoc_crawler.specs.sources.whatthespec import fetch_whatthespec_metadata


		def clear_checkout_specs(checkout_dir: Path) -> int:
		"""Clear spec checkout entries from the checkout directory.

		Args:
		checkout_dir: Base checkout directory

		Returns:
		Number of entries removed (always 1 if Specs directory existed)
		"""
		specs_dir = checkout_dir / "Specs"
		if not specs_dir.exists():
		return 0
		shutil.rmtree(specs_dir)
		return 1


		def build_default_spec_sources(cache_manager_name: str \| None = None) -> list[SpecSource]:
		"""Build the default list of spec sources.

		Args:
		cache_manager_name: Optional cache manager name for HTTP caching

		Returns:
		List of SpecSource instances for fetching spec metadata
		"""
		return [
		cast("SpecSource", FunctionSpecSource("3gpp", fetch_threegpp_metadata, fetcher_kwargs={"cache_manager_name": cache_manager_name})),
		cast("SpecSource", FunctionSpecSource("whatthespec", fetch_whatthespec_metadata, fetcher_kwargs={"cache_manager_name": cache_manager_name})),
		]


		def checkout_specs(
		spec_numbers: list[str],
		checkout_dir: Path,
		database: SpecDatabase,
		release: str = "latest",
		doc_only: bool = False,
		cache_manager_name: str \| None = None,
		) -> list[Path]:
		"""Checkout spec documents to the checkout directory.

		Args:
		spec_numbers: List of spec numbers to checkout
		checkout_dir: Base checkout directory
		database: SpecDatabase instance for metadata lookup
		release: Release version to checkout
		doc_only: If True, download only document files instead of full zip
		cache_manager_name: Optional cache manager name for HTTP caching

		Returns:
		List of paths to checked out specs
		"""
		sources = build_default_spec_sources(cache_manager_name=cache_manager_name)
		downloader = SpecDownloads(database, cache_manager_name=cache_manager_name)
		return downloader.checkout_specs(spec_numbers, doc_only, checkout_dir, release, sources=sources)


		__all__ = [
		"build_default_spec_sources",
		"checkout_specs",
		"clear_checkout_specs",
		]