Ultimate Web ScraperPage Text Extractor
Extract clean plain text plus title, author, date, and word count from a list of pages. Navigation, ads, and boilerplate are stripped automatically.
The Page Text Extractor pulls clean, readable text and metadata from a list of pages — one row per page. It's built for feeding content to AI: no nav menus, no ads, no cookie banners, just the article.
How it works
- Add URLs by manual paste, CSV upload, or Data Source (a URL column from a previous extraction — for example, article links gathered with the List Extractor).
- Confirm the list; the first URL loads as a preview.
- Start the extraction. Pages are processed in the background with a live progress overlay, and rows stream into the Data Table.
What you get per page
| Column | Content |
|---|---|
| url | Source URL |
| title | Page title, upgraded from social meta tags or the main heading when those are richer |
| description | Meta description |
| content | The main content as plain text |
| word_count | Word count of the content |
| author | Author, from meta tags and common author markup |
| publish_date | Publication date, from article metadata and common date markup |
Content extraction targets the page's main content container and always strips scripts, styles, navigation, headers, footers, sidebars, ads, social-share widgets, and comments. Output is plain, whitespace-normalized text — not Markdown or HTML.
Use cases
- LLM corpora — build clean text datasets for retrieval, fine-tuning, or analysis without writing a scraper per site.
- Content audits — word counts, authors, and publish dates for every page of a blog in one table.
- Research — collect the readable text of dozens of sources into one searchable place.
Options
Sensible defaults: 1000 ms delay between requests, 30 s page timeout, wait-for-page-load on. The Faster Extraction toggle runs several tabs in parallel, and there's no cap on how many URLs you can process.
Recipes are fully supported — save a URL list and configuration, re-run it anytime.
Limitations
- Content that requires interaction to render isn't captured (standard page loads and JavaScript-rendered content are handled via load waits).
- LinkedIn URLs are blocked.
- There's no direct Markdown export — output is plain text, exported via the Data Table.