AgentSkillsCN

run-tests

运行完整的测试套件——R testthat与Python pytest——仅报告失败用例。

SKILL.md
--- frontmatter
name: run-tests
description: Run the full test suite — R testthat and Python pytest — reporting only failures
disable-model-invocation: true

Run Tests

Run all unit tests for R and Python code.

Steps

  1. •

    R cache/schema tests (most reliable — no pipeline outputs required):

    code
    Rscript -e 'testthat::test_file("tests/test_cache.R")'
Rscript -e 'testthat::test_file("tests/test_schema_update.R")'
  2. •

    R data loading tests (has known pre-existing failures without results/):

    code
    Rscript -e 'testthat::test_file("tests/test_data_loading.R")'
Rscript -e 'testthat::test_file("tests/test_exclusion_loading.R")'
  3. •

    Python unit tests:

    code
    micromamba run -n q100-smk pytest tests/unit/ -v --tb=short

Known Pre-existing Failures (not caused by recent changes)

  • •parse_benchmark_id() regex captures "5.0q" not "v5.0q" — tests expect "v" prefix
  • •test_data_loading.R references old column names (strat_name, strat_filter) from pre-refactor
  • •Any test that loads real data will fail without results/ directory (pipeline outputs not in repo)

Reporting

Report results grouped by test file. Distinguish between:

  • •New failures: Likely caused by recent changes — investigate these
  • •Pre-existing failures: Listed above — note them but don't treat as regressions
  • •Passes: Confirm count