cache.py 2.1 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374
  1. import datetime
  2. import lzma
  3. import pathlib
  4. import time
  5. import typing
  6. import cbor2
  7. import httpx
  8. import urllib.parse
  9. if typing.TYPE_CHECKING:
  10. import httpx._types
  11. ONE_DAY = datetime.timedelta(days=1)
  12. CACHE_DIR = pathlib.Path(__file__).parent / 'cache'
  13. client = httpx.Client(transport=httpx.HTTPTransport(http2=True, retries=3), timeout=5) # retry on ConnectError and ConnectTimeout
  14. @typing.overload
  15. def get(url: str, *, json: typing.Literal[True]=True, headers: httpx._types.HeaderTypes|None=None,
  16. expiry: datetime.timedelta=datetime.timedelta(minutes=10)) -> typing.Any:
  17. ...
  18. @typing.overload
  19. def get(url: str, *, json: typing.Literal[False], headers: httpx._types.HeaderTypes|None=None,
  20. expiry: datetime.timedelta=datetime.timedelta(minutes=10)) -> str:
  21. ...
  22. def get(url: str, *, json=True, headers=None, expiry=datetime.timedelta(minutes=10)) -> typing.Any:
  23. parsed = urllib.parse.urlparse(url)
  24. assert parsed.hostname is not None
  25. cache_filename = urllib.parse.quote(parsed.path.removeprefix('/'), safe='')
  26. if parsed.query:
  27. cache_filename += '?' + urllib.parse.quote(parsed.query, safe='')
  28. if json:
  29. cache_filename += '.cbor'
  30. cache_filename += '.xz'
  31. cache_path = CACHE_DIR / parsed.hostname / cache_filename
  32. try:
  33. if cache_path.stat().st_mtime > time.time() - expiry.total_seconds(): # less than 10 minutes old
  34. with lzma.open(cache_path, 'rb') as f:
  35. if json:
  36. return cbor2.load(f)
  37. else:
  38. return f.read().decode('utf-8')
  39. except FileNotFoundError:
  40. pass # fall through
  41. r = get_with_retries(url, headers)
  42. if r.status_code == 204:
  43. raise NoContent
  44. cache_path.parent.mkdir(parents=True, exist_ok=True)
  45. with lzma.open(cache_path, 'wb') as f:
  46. if json:
  47. data = r.json()
  48. cbor2.dump(data, f)
  49. else:
  50. data = r.text
  51. f.write(data.encode('utf-8'))
  52. return data
  53. def get_with_retries(url: str, headers: httpx._types.HeaderTypes|None=None) -> httpx.Response:
  54. for attempt in range(5):
  55. try:
  56. return client.get(url, headers=headers).raise_for_status()
  57. except httpx.ReadTimeout:
  58. if attempt == 4:
  59. raise
  60. else:
  61. print(url, 'attempt', attempt+1, 'timed out; retrying...')
  62. raise AssertionError
  63. class NoContent(Exception):
  64. pass