When Civitai returns 429 (Too Many Requests) during example image
downloads, the previous behavior treated all failures identically and
permanently removed the corresponding images from model metadata —
making them impossible to retry.
This commit adds:
- 429 detection + Retry-After header parsing in download_to_memory
- Exponential backoff retry (up to 3 attempts) in
download_model_images_with_tracking
- Separate tracking of rate-limited vs permanently failed URLs
- rate_limited_models progress tracking persisted to disk
- Rate-limited models are NOT added to failed_models/processed_models
so they are automatically retried on subsequent download runs
- Force mode clears failed_models when rate-limited images exist
The proxy settings allow selecting a SOCKS proxy type, but the SOCKS
URL was passed to aiohttp's per-request `proxy=` argument, which only
supports http(s) proxies. With a SOCKS proxy this opens a plain TCP
connection to the proxy port and sends an HTTP request; the SOCKS
server replies with its handshake bytes (e.g. b"\x05\xff") and aiohttp
fails with "Bad status line ... Expected HTTP/, RTSP/ or ICE/".
Route SOCKS proxy types through an aiohttp-socks ProxyConnector on the
session instead, leaving the `proxy=` kwarg for http(s) proxies only.
trust_env now keys off whether an app-level proxy is active. Adds
aiohttp-socks to requirements.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Add RateLimitError import and _make_request wrapper method to handle rate limiting
- Update API methods to use _make_request wrapper instead of direct downloader calls
- Add explicit RateLimitError handling in API methods to properly propagate rate limit errors
- Add _extract_retry_after method to parse Retry-After headers
- Improve error handling by surfacing rate limit information to callers
These changes ensure that rate limiting from the Civitai API is properly detected and handled, allowing callers to implement appropriate backoff strategies when rate limits are encountered.
- Replaced direct usage of Civitai client with a fallback metadata provider across all recipe parsers.
- Updated metadata service to improve initialization and error handling.
- Enhanced download manager to utilize a downloader service for file operations.
- Improved recipe scanner to fetch model information through the new metadata provider.
- Updated utility functions to streamline image downloading and processing.
- Added comprehensive logging and error handling for better debugging and reliability.
- Introduced `get_default_metadata_provider()` for simplified access to the default provider.
- Ensured backward compatibility with existing APIs and workflows.
- Introduced a new `Downloader` class to centralize HTTP/HTTPS download management.
- Replaced direct `aiohttp` session handling with the unified downloader in `MetadataArchiveManager`, `DownloadManager`, and `ExampleImagesProcessor`.
- Added support for resumable downloads, progress tracking, and error handling in the new downloader.
- Updated methods to utilize the downloader's capabilities for downloading files and images, improving code maintainability and readability.
