Files
libretro/scripts/scraper/targets/__init__.py
T

85 lines
2.9 KiB
Python

"""Target scraper plugin discovery module.
Auto-detects *_targets_scraper.py files and exposes their scrapers.
"""
from __future__ import annotations
import importlib
import pkgutil
import re
from abc import ABC, abstractmethod
from pathlib import Path
class BaseTargetScraper(ABC):
"""Base class for target scrapers."""
def __init__(self, url: str = "") -> None:
self.url = url
@abstractmethod
def fetch_targets(self) -> dict:
"""Fetch targets and their core lists. Returns dict matching target YAML format."""
...
@staticmethod
def pinned_revision(platform: str, platforms_dir: str | None = None) -> str:
"""The upstream revision platforms/<platform>.yml was scraped at.
A target list describes the same release as the platform's BIOS
list: read on its own "latest stable tag", a Recalbox 10.2 board list
filtered 10.1.1 systems. Moving to a new release is a rescrape of the
platform first.
"""
import yaml
directory = Path(platforms_dir) if platforms_dir else (
Path(__file__).resolve().parents[3] / "platforms"
)
with open(directory / f"{platform}.yml", encoding="utf-8") as handle:
source = str((yaml.safe_load(handle) or {}).get("source", ""))
for pattern in (
r"^https://raw\.githubusercontent\.com/[^/]+/[^/]+/([^/]+)/",
r"^https://github\.com/[^/]+/[^/]+/(?:blob|tree)/([^/]+)/",
r"^https://gitlab\.com/.+?/-/(?:raw|blob|tree)/([^/]+)/",
):
match = re.match(pattern, source)
if match:
return match.group(1)
raise RuntimeError(
f"platforms/{platform}.yml names no pinned revision in its source "
f"({source!r}): rescrape the platform first"
)
def write_output(self, data: dict, output_path: str) -> None:
"""Write target data to YAML file."""
try:
import yaml
except ImportError:
raise ImportError("PyYAML required: pip install pyyaml")
from common import write_text_atomic
write_text_atomic(
output_path, yaml.dump(data, default_flow_style=False, sort_keys=False)
)
_scrapers: dict[str, type] = {}
def discover_target_scrapers() -> dict[str, type]:
"""Auto-discover all *_targets_scraper.py modules."""
if _scrapers:
return _scrapers
package_dir = Path(__file__).parent
for finder, name, ispkg in pkgutil.iter_modules([str(package_dir)]):
if not name.endswith("_targets_scraper"):
continue
module = importlib.import_module(f".{name}", package=__package__)
platform_name = getattr(module, "PLATFORM_NAME", None)
scraper_class = getattr(module, "Scraper", None)
if platform_name and scraper_class:
_scrapers[platform_name] = scraper_class
return _scrapers