Custom HTTP GET with proper User-Agent header and a bounded timeout

Wraps httr::GET() with a descriptive User-Agent to avoid being blocked by services like Figshare that reject default libcurl requests, especially from CI runner IPs (e.g., GitHub Actions).

Usage

codecheck_GET(url, ..., timeout_seconds = 30)

Arguments

url: The URL to request
...: Additional arguments passed to httr::GET() - a caller's own httr::timeout() here overrides the default.
timeout_seconds: Maximum time for the whole request/response cycle.

Value

An httr response object

Details

Also bounds the request to timeout_seconds: plain httr::GET() has no timeout at all, so a server that accepts the connection but never answers (rather than erroring or rate-limiting, which codecheck_GET_retry() already handles) hangs the request - and with it, the whole single-threaded render - indefinitely. A render touches many certificates across several external APIs in one run, so any one of them having a bad moment stalls everything after it with no log output to say why. Bounding it here means the caller gets an error back to handle (`codecheck_GET_retry()` retries it; a bare `codecheck_GET()` caller sees a `try`/`tryCatch`-able failure) instead of a silent, unrecoverable hang.