Add retry logic to enrichSTRING_API for transient HTTP errors

[Copilot]

The STRING API can fail transiently (rate limits, server errors, gateway timeouts), causing enrichment analysis to fail with no recovery path.

Changes

• R/WrapperFuncs.R: Added max_retries (default 3) and retry_delay (default 5s) parameters to enrichSTRING_API(). Wraps the httr::POST call in a repeat loop that retries on HTTP 429, 500, 502, 503, 504; sleeps between attempts; verbose mode logs each retry.

• tests/testthat/test-enrichSTRING_API.R: Three new tests covering retry exhaustion, no-retry on non-retryable errors (e.g. 400), and success after transient failures — using local_mocked_bindings with explicit status-code sequences to avoid fragile call-count comparisons.

Usage

# Retry up to 5 times with 10s delay; default is max_retries=3, retry_delay=5
enrichSTRING_API(genes, max_retries = 5, retry_delay = 10)

Original prompt

Done, try again

Created from VS Code.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[veitveit]

@copilot THere are still some faling tests (see your pull request), please investigate.

[Copilot]

@copilot THere are still some faling tests (see your pull request), please investigate.

Investigated the CI failures in commit 5ee89bf. Two root causes:

1. runFuncEnrich() using shiny::validate() outside Shiny context (Ubuntu/macOS failures): When the STRING API returns no results in CI, validate(need(!is.null(x), "...")) threw a shiny.silent.error that propagated as an uncaught error in the vignette. Fixed by using stop() when outside a Shiny reactive domain, and validate() only within Shiny.

2. Vignette not handling STRING API unavailability (Ubuntu/macOS failures): The vignette chunk calling runFuncEnrich() had no error handling. Fixed by wrapping it in tryCatch and making the visualization chunk conditional on the enrichment result being non-NULL.

The Windows failure (could not find function "run_fuzzy_cmeans") and oldrel-1 failure (ggtree build error) appear to be pre-existing infrastructure issues unrelated to this PR.

[veitveit]

@copilot Still plenthy of failing CI runs.

[Copilot]

@copilot Still plenthy of failing CI runs.

Found and fixed the remaining CI failures in commit e202660. All three failing jobs (macOS, Ubuntu release, Ubuntu devel) had the same root cause: man/enrichSTRING_API.Rd was missing the max_retries and retry_delay parameters in its \usage{} and \arguments{} sections, causing R CMD check to fail with a "codoc mismatch" WARNING. Updated the .Rd file to match the function signature.

The Windows failure (could not find function "run_fuzzy_cmeans") and oldrel-1 failure (ggtree build error) are pre-existing infrastructure issues unrelated to this PR.