Merge remote-tracking branch 'origin/main' into feat/326-system-s9-backup-ui
This commit is contained in:
+96
-44
@@ -8,6 +8,7 @@ return empty results rather than raising.
|
||||
"""
|
||||
|
||||
import hashlib
|
||||
import http.client
|
||||
import json
|
||||
import os
|
||||
import re
|
||||
@@ -420,22 +421,35 @@ def get_preview_image_from_metadata(
|
||||
|
||||
_THUMB_MAX_SIZE = 512
|
||||
|
||||
# CivitAI's own domains (#340: civitai.red and civitai.green serve the same
|
||||
# content as civitai.com).
|
||||
CIVITAI_DOMAINS = ("civitai.com", "civitai.red", "civitai.green")
|
||||
# Only these hosts may ever receive the CivitAI API key, and only over HTTPS.
|
||||
CIVITAI_HOSTS = frozenset(
|
||||
{"civitai.com", "www.civitai.com", "api.civitai.com", "image.civitai.com"}
|
||||
f"{prefix}{domain}"
|
||||
for domain in CIVITAI_DOMAINS
|
||||
for prefix in ("", "www.", "api.", "image.")
|
||||
)
|
||||
MAX_CIVITAI_DOWNLOAD_BYTES = 50 * 1024 * 1024
|
||||
_DOWNLOAD_CHUNK_BYTES = 64 * 1024
|
||||
|
||||
|
||||
# A redirect is followed (#337: image.civitai.com now answers with a redirect
|
||||
# to blobs-b2.civitai.com) only over HTTPS to civitai.com or a subdomain of
|
||||
# it, at most this many hops, and never with the API key.
|
||||
# to blobs-b2.civitai.com) only over HTTPS to one of CIVITAI_DOMAINS or a
|
||||
# subdomain of one, at most this many hops, and never with the API key.
|
||||
MAX_CIVITAI_REDIRECTS = 3
|
||||
_REDIRECT_CODES = frozenset({301, 302, 303, 307, 308})
|
||||
_USER_AGENT = "ComfyUI-PromptManager/1.0"
|
||||
|
||||
|
||||
class _DownloadFailed(Exception):
|
||||
"""A download failed for the short, loggable ``reason`` it carries."""
|
||||
|
||||
def __init__(self, reason: str):
|
||||
super().__init__(reason)
|
||||
self.reason = reason
|
||||
|
||||
|
||||
class _RefuseRedirects(urllib.request.HTTPRedirectHandler):
|
||||
"""Never follow a redirect inside urllib: it would re-send the Bearer
|
||||
header. A 3xx surfaces as ``HTTPError``; ``_download_one`` decides."""
|
||||
@@ -467,10 +481,18 @@ def _loggable(url: str) -> str:
|
||||
return f"{parts.scheme}://{host}{parts.path}"
|
||||
|
||||
|
||||
def _on_civitai_domain(hostname: str) -> bool:
|
||||
"""True when ``hostname`` is one of CIVITAI_DOMAINS or a subdomain of one."""
|
||||
return any(
|
||||
hostname == domain or hostname.endswith(f".{domain}")
|
||||
for domain in CIVITAI_DOMAINS
|
||||
)
|
||||
|
||||
|
||||
def is_civitai_redirect_target(url: str) -> bool:
|
||||
"""True when a redirect to ``url`` may be followed: HTTPS on civitai.com
|
||||
or a subdomain of it, matched on the parsed host name, default port and
|
||||
no credentials in the URL."""
|
||||
"""True when a redirect to ``url`` may be followed: HTTPS on one of
|
||||
CIVITAI_DOMAINS or a subdomain of one, matched on the parsed host name,
|
||||
default port and no credentials in the URL."""
|
||||
try:
|
||||
parts = urlsplit(url)
|
||||
port = parts.port
|
||||
@@ -479,7 +501,7 @@ def is_civitai_redirect_target(url: str) -> bool:
|
||||
hostname = (parts.hostname or "").lower()
|
||||
return (
|
||||
parts.scheme == "https"
|
||||
and (hostname == "civitai.com" or hostname.endswith(".civitai.com"))
|
||||
and _on_civitai_domain(hostname)
|
||||
and parts.username is None
|
||||
and parts.password is None
|
||||
and port in (None, 443)
|
||||
@@ -500,8 +522,9 @@ def _open_following_civitai_redirects(url: str, api_key: str):
|
||||
"""Open ``url``, following only allowed CivitAI redirects (keyless).
|
||||
|
||||
The first request carries the key; every redirected one carries the
|
||||
User-Agent alone. Returns the open response, or None when a redirect is
|
||||
refused or the hop limit is reached. Other HTTP errors propagate.
|
||||
User-Agent alone. Returns the open response; raises ``_DownloadFailed``
|
||||
when a redirect is refused or the hop limit is reached. Other HTTP
|
||||
errors propagate.
|
||||
"""
|
||||
headers = {"User-Agent": _USER_AGENT}
|
||||
if api_key:
|
||||
@@ -517,20 +540,13 @@ def _open_following_civitai_redirects(url: str, api_key: str):
|
||||
if target is None:
|
||||
raise
|
||||
if hop == MAX_CIVITAI_REDIRECTS:
|
||||
logger.warning(
|
||||
f"Refusing redirect past {MAX_CIVITAI_REDIRECTS} hops "
|
||||
f"from {_loggable(url)}"
|
||||
)
|
||||
return None
|
||||
raise _DownloadFailed(f"more than {MAX_CIVITAI_REDIRECTS} redirects")
|
||||
if not is_civitai_redirect_target(target):
|
||||
logger.warning(
|
||||
f"Refusing redirect from {_loggable(current)} to {_loggable(target)}"
|
||||
)
|
||||
return None
|
||||
raise _DownloadFailed(f"redirect refused to {_loggable(target)}")
|
||||
logger.debug(f"Following redirect to {_loggable(target)}")
|
||||
current = target
|
||||
headers = {"User-Agent": _USER_AGENT} # never the key across a redirect
|
||||
return None # pragma: no cover - the loop returns on its last hop
|
||||
raise AssertionError("unreachable") # pragma: no cover - loop ends above
|
||||
|
||||
|
||||
def is_civitai_url(url: str) -> bool:
|
||||
@@ -573,40 +589,76 @@ def _read_capped(resp, cap: int) -> Optional[bytes]:
|
||||
return b"".join(chunks)
|
||||
|
||||
|
||||
def _network_reason(error: BaseException) -> str:
|
||||
"""A short reason for a failed request: a timeout or a network error."""
|
||||
cause = getattr(error, "reason", error) # URLError wraps the socket error
|
||||
if isinstance(cause, TimeoutError):
|
||||
return "timed out"
|
||||
return f"network error: {cause}"
|
||||
|
||||
|
||||
def _fetch_civitai_body(url: str, api_key: str) -> bytes:
|
||||
"""The body behind ``url`` (redirects as allowed, size capped), or
|
||||
``_DownloadFailed`` with the reason it could not be fetched."""
|
||||
try:
|
||||
with _open_following_civitai_redirects(url, api_key) as resp:
|
||||
raw = _read_capped(resp, MAX_CIVITAI_DOWNLOAD_BYTES)
|
||||
except urllib.error.HTTPError as error:
|
||||
error.close()
|
||||
raise _DownloadFailed(f"HTTP {error.code}") from None
|
||||
except (urllib.error.URLError, OSError, http.client.HTTPException) as error:
|
||||
raise _DownloadFailed(_network_reason(error)) from None
|
||||
if raw is None:
|
||||
cap = MAX_CIVITAI_DOWNLOAD_BYTES
|
||||
size = f"{cap // (1024 * 1024)} MB" if cap >= 1024 * 1024 else f"{cap} byte"
|
||||
raise _DownloadFailed(f"larger than the {size} cap")
|
||||
return raw
|
||||
|
||||
|
||||
def _thumbnail(raw: bytes):
|
||||
"""``raw`` decoded and shrunk to an RGB thumbnail, or ``_DownloadFailed``."""
|
||||
from io import BytesIO
|
||||
|
||||
from PIL import Image
|
||||
|
||||
try:
|
||||
with Image.open(BytesIO(raw)) as img:
|
||||
img.thumbnail((_THUMB_MAX_SIZE, _THUMB_MAX_SIZE), Image.LANCZOS)
|
||||
return img.convert("RGB")
|
||||
except Exception as error: # PIL raises several types for bad data
|
||||
raise _DownloadFailed(
|
||||
f"image could not be decoded ({type(error).__name__})"
|
||||
) from None
|
||||
|
||||
|
||||
def _download_one(url: str, local_path: Path, api_key: str) -> Optional[str]:
|
||||
"""Download a single CivitAI image, resize to thumbnail, save as JPEG.
|
||||
|
||||
Refuses any URL that is not HTTPS on an allow-listed CivitAI host so the
|
||||
API key is never sent elsewhere, follows a redirect only as
|
||||
``_open_following_civitai_redirects`` allows (without the key), and
|
||||
refuses bodies larger than MAX_CIVITAI_DOWNLOAD_BYTES.
|
||||
refuses bodies larger than MAX_CIVITAI_DOWNLOAD_BYTES. A failure logs
|
||||
one warning with a short reason and the URL without query or
|
||||
credentials (#340), and returns None.
|
||||
"""
|
||||
if not is_civitai_url(url):
|
||||
logger.warning(f"Refusing non-CivitAI download URL: {_loggable(url)}")
|
||||
return None
|
||||
try:
|
||||
resp = _open_following_civitai_redirects(url, api_key)
|
||||
if resp is None:
|
||||
return None
|
||||
with resp:
|
||||
raw = _read_capped(resp, MAX_CIVITAI_DOWNLOAD_BYTES)
|
||||
if raw is None:
|
||||
logger.warning(f"Refusing oversized download: {_loggable(url)}")
|
||||
return None
|
||||
|
||||
# Resize to thumbnail to save disk space
|
||||
from io import BytesIO
|
||||
|
||||
from PIL import Image
|
||||
|
||||
with Image.open(BytesIO(raw)) as img:
|
||||
img.thumbnail((_THUMB_MAX_SIZE, _THUMB_MAX_SIZE), Image.LANCZOS)
|
||||
rgb = img.convert("RGB")
|
||||
_save_atomically(rgb, local_path)
|
||||
if not is_civitai_url(url):
|
||||
raise _DownloadFailed("not an allowed CivitAI URL")
|
||||
image = _thumbnail(_fetch_civitai_body(url, api_key))
|
||||
try:
|
||||
_save_atomically(image, local_path)
|
||||
except OSError as error:
|
||||
raise _DownloadFailed(
|
||||
f"image could not be saved ({error.strerror or type(error).__name__})"
|
||||
) from None
|
||||
except _DownloadFailed as failure:
|
||||
reason = failure.reason
|
||||
except Exception as error: # best effort: an import never fails on one image
|
||||
reason = f"unexpected {type(error).__name__}"
|
||||
else:
|
||||
return str(local_path.resolve())
|
||||
except Exception as e:
|
||||
logger.debug(f"Failed to download {_loggable(url)}: {e}")
|
||||
return None
|
||||
logger.warning(f"CivitAI image download failed: {_loggable(url)}: {reason}")
|
||||
return None
|
||||
|
||||
|
||||
_PART_PREFIX = ".pm-"
|
||||
|
||||
Reference in New Issue
Block a user