Google News with public article content

Requests the live Google News RSS search endpoint over HTTPS, then optionally resolves each source-issued redirect and reads public publisher pages. No browser, proxy rotation, user cookies or challenge solving is used by this implementation. Requests to Google News are serialized with a minimum interval; an access challenge or HTTP 429 stops further requests rather than repeatedly retrying.

Input

Supply query or queries (queries takes precedence), maxItems (global, 1–5000), hl language/locale and gl country. The sample requests technology when:7d, English US, ten records and deepScrape=true. No packaged test input replaces a user query. The legacy proxyConfiguration field is not used by this direct HTTP implementation.

Deep content and provenance

With deepScrape=true, only records with at least 200 characters of public publisher text are emitted. Inaccessible, paywalled or missing-content sources are skipped. Content comes from a matching Article/NewsArticle JSON-LD body or the actual page article paragraphs; contentSource and contentCompleteness describe this. Publicly exposed text is not a claim that hidden/subscriber content was retrieved. Publisher-declared restricted articles are rejected.

Records preserve RSS title/source/date/snippet and the Google News URL, plus actual publisher URL/canonical URL, observed title/text, available author/images/dates and derived wordCount. Missing source fields are omitted. Retrieval time and publication dates remain separate. Headlines alone never count as a successful deep scrape. deepScrape=false returns genuine RSS records with not-requested status.

The public redirect protocol uses parameters issued by the corresponding Google News page; no signatures, timestamps, publisher URLs or article content are invented. Its unofficial interface may change. Protocol reference: https://github.com/greg-randall/google-rss-decode/blob/main/HOW_IT_WORKS.md . Publisher content remains untrusted data and is never executed.

Run apify run --input-file input.json, inspect the Dataset and RUN_SUMMARY. Source failures can yield fewer than the requested maximum; zero matching records fail explicitly. Platform compute usage incurs the account's applicable charges.

The legacy link field is retained as the resolved publisher URL in deep mode or Google News URL in feed mode. Publication/modified dates and authors/images/site/language also use observed publisher meta tags when native article JSON-LD does not expose them. RSS date remains separate: a recent feed/index date does not prove a recently written article, and evergreen pages can appear in source results.