* may attempt to download and install SDK components (platforms, build-tools, platform-tools) simultaneously.
 * This can cause race conditions leading to:
 * - `FileAlreadyExistsException` when extracting files
 * - Corrupted installations (e.g., "Build-tool X has corrupt source.properties")
 * - Missing SDK components after installation
 *

 *   <li>Extracting child URLs from the HTML content based on configured rules.</li>
 *   <li>Handling redirect URLs specified in the response headers.</li>
 * </ol>
 * <p>
 * The class also provides methods for configuring features and properties of the
 * underlying DOM parser, as well as defining rules for extracting child URLs
 * from specific HTML tags and attributes.
 * </p>
 *

    private static final String X_ROBOTS_TAG = "X-Robots-Tag";

    /** XPath expression for extracting thumbnail content from meta tags */
    private static final String META_NAME_THUMBNAIL_CONTENT = "//META[@name=\"thumbnail\" or @name=\"THUMBNAIL\"]/@content";

    /** XPath expression for extracting Open Graph image content from meta tags */

                || url.startsWith("storage:") || url.startsWith("s3:") || url.startsWith("gcs:");
    }

    /**
     * Checks if the given URL should skip URL decoding when extracting file names.
     * Some protocols (like SMB, FTP, S3, GCS) should preserve the original URL encoding.
     *
     * @param url the URL to check
     * @return true if URL decoding should be skipped for this protocol

                else:
                    yield {**(maybe_nested_state or {}), **(maybe_original_state or {})}

    return merged_lifespan  # type: ignore[return-value]


# Cache for endpoint context to avoid re-extracting on every request
_endpoint_context_cache: dict[int, EndpointContext] = {}


def _extract_endpoint_context(func: Any) -> EndpointContext:
    """Extract endpoint context with caching to avoid repeated file I/O."""