Skip to content

[Performance] Hybrid Content extraction with Jina AI & Direct Scraper #8

@Smilin01

Description

@Smilin01

Label: performance, backend

Description: Content extraction (fetching full page text) is a bottleneck. Using r.jina.ai is highly effective but can be slow or encounter rate limits. Our direct scraper (

extractTextFromHtml
) is faster but often blocked.

Suggested Approach:

Implement a "Hybrid Race" strategy: Trigger Jina AI and Direct Fetching simultaneously, and take the first one that returns 1000+ characters of clean text.
Implement a simple LRU cache for extracted content to avoid re-fetching the same URLs in follow-up queries.
Optimize the boilerplateSignals list to better clean up modern cookie banners and "Subscribe" overlays.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions