# Changelog of PDF to Markdown Scraper | Text, Tables & OCR (`thescrapelab/apify-pdf-url-scraper`) Actor

- **URL**: https://apify.com/thescrapelab/apify-pdf-url-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/thescrapelab/apify-pdf-url-scraper.md

## Changelog

### Candidate — PDF uploads, selected pages and mixed-document OCR

- Added direct PDF uploads and optional page selection, including open-ended ranges and copied en dashes.
- Fixed scanned-page reading in documents that mix readable text and scans. Original page numbers are retained in tables, downloads, page images and search chunks.
- Blank and unreadable pages are reported in diagnostics without an extracted-page fee.
- Reused the table parser within each PDF and refreshed the upload and API instructions.

### 2026-10-28 — Simpler PDF extraction and document downloads

- Exposed page limits, tables, local English OCR and AI-search chunks independently of paid AI cleanup.
- Added document download manifests and run summaries without extra charged dataset rows.
- Reserved base delivery fees before optional AI work and charged AI-enhanced pages together with their saved page rows.
- Added confirmed delivery handling, clear partial statuses and failure on uncertain writes without retry.
- Updated the customer README, title, SEO and logo. Preserved the existing Actor URL, input aliases and page output contract.
- Reduced basic extraction prices to $1.00–$1.25 per 1,000 pages after bounded cloud cost checks. AI cleanup and start rates are unchanged.
- Replaced a deprecated Standard cleanup model with Qwen3.7 Plus, enforced explicit budgets when provider pricing is unknown, and reduced the default allocation to 512 MB for every mode.
- Verified local tables/OCR, source-cited chunks, clean-page no-charge paths and both paid cleanup tiers.