- Read publisher pages with full browser request headers, and retry a page once in headless Chromium after three failed HTTP attempts or an HTTP 401.
- Report publisher browser attempts and rescues in COVERAGE and the run log.
- Refuse paywalled bodies declared through relative Article URLs; join author-name lists into one string.
- Keep browser cleanup failures from losing captured fields or storage records.
- Rewrite the README for buyers, with dated field-coverage, startup and memory measurements.
- Deliver fast-mode batches through one SDK dataset push with charge clipping; keep uncharged new rows out of SNAPSHOT.
- Preserve confirmed og:image when a JSON-LD image reference has no URL, and prefer confirmed publisher descriptions over feed summaries when publisher text is requested.
- Add resolveUrls=false for Google News links and feed metadata without signatures, decoding or publisher reads; reject text extraction in that mode.
- Raise metadata and feed-only source work to 270 seconds; keep text work at 240 seconds under the recommended 300-second timeout.
- Start Google signature GETs with direct anonymous egress and retry measured blocks through up to ten fresh datacenter sessions.
- Expose signature fallback request and attempt counts in the separate COVERAGE record.
- Resolve up to 100 article URLs per batch with 50 concurrent HTTP reads, avoiding Google's locale redirect.
- Reuse a verified TLS context across fresh HTTP clients and require at least 512 MB memory.
- Keep the 240-second source-work budget, output fields, input modes and per-row charge-limit handling.
Google News mixed query/topic/publisher feeds, batched publisher URL resolution, readable article text and only-new snapshot comparison. Explicit publication windows use date slicing with finite coverage and disclosed body failures.
2026-10-06 repair: publisher metadata requires a self-URL matching the resolved target or recorded redirect chain, with a foreign-host veto; repeated confirming canonicals and revised headlines retain confirmed fields. Unconfirmed metadata keeps Google News fields and records its coverage reason. Budget-only resolution work no longer changes completed-failure status.
![Google News Scraper [only $0.7/1K] avatar](https://images.apifyusercontent.com/Twu1TG04ITI_hwjby-GKjS3ch9YgJ3AaSAGKR67WRFo/rs:fill:76:76/cb:1/aHR0cHM6Ly9hcGlmeS1pbWFnZS11cGxvYWRzLXByb2QuczMudXMtZWFzdC0xLmFtYXpvbmF3cy5jb20vUmVkNnNHSEFOa0x6b3NzdkUtYWN0b3ItemVOd3Z2dk5VNnpWMWFra1MtMTBnTFVnUzdoSS1nb29nbGVfbmV3c19zY3JhcGVyX2Zhc3QucG5n.webp){kind=avatar}
