extract-page-structure

To extract page structure metadata from raw HTML, provide the HTML payload as input to receive a structured JSON output containing metadata, blocks, assets, and embeds. The extraction preserves document order and avoids fabrication.

Parsing NYT Birdkit interactive patterns from web-scraping output is handled by applying strict extraction rules aligned to Birdkit and ai2html iframe embeds. This classifies page blocks accurately while preserving the original document order.

Metadata extraction prioritizes Open Graph and structured-data values found in the head and first body elements. This ensures the most accurate metadata is captured for the final PageStructure JSON interface.

Yes, you can use this to discover assets and embeds for design-doc workflows. It assembles a complete PageStructure payload by identifying assets and embeds, enabling consistent downstream layout-analysis tooling.

No external dependencies are required for this HTML metadata extraction. It operates independently to process raw HTML sources and output the structured PageStructure JSON without needing additional libraries.

A key limitation when extracting page blocks from unstructured HTML is the avoidance of fabrication. The extraction applies strict rules for recognized patterns like Birdkit, meaning unrecognized interactive blocks may not be classified.