JAR-430: wire make test and cover the two parsing-heavy files - #52
Open
Jartans-Familiar wants to merge 2 commits into
Open
JAR-430: wire make test and cover the two parsing-heavy files#52Jartans-Familiar wants to merge 2 commits into
Jartans-Familiar wants to merge 2 commits into
Conversation
…e65f) Implements the convention decided in CONTRIBUTING.md's Tests section: the gate itself, plus the first two suites behind it. Both languages, because each half of `make test` asserts its file list is non-empty and a one-language first PR would leave the other half red. Four of the Node tests are named `regression:` and each pins a defect the duplicate-comment finder accrued over the grimoire#33 review rounds, until now held only by a code comment. Each was confirmed to fail with its fix reverted. The duplicate-comment seam widens by two functions the suite consumes, parseArgs and trackedPaths, which is the praxis version bump. Co-authored-by: multica-agent <github@multica.ai>
![github-code-quality[bot]](https://avatars.githubusercontent.com/u/9919?s=60&v=4){kind=small}
| import json | ||
| import pathlib | ||
| import tempfile | ||
| import unittest |
This was referenced Aug 17, 2026
…5f) (#54) Second of two, purely additive: no Makefile or CI change, since PR 1 wired the gate. Adds 113 Node tests over the token accounting and setting resolvers in lib/transcript-context.js, the sweep and session-id sanitising in lib/session-state.js, the cooldown split in check-console-log.js, and the regex-flag and glob handling in lib/utils.js. Process contracts for all four shipped hook entry points: exit 0 and either no stdout or exactly one well-formed JSON object, whatever arrives on stdin. sync-codex-agents.js is spawned rather than imported because hooks.json invokes it via require and it works at require time, so a require.main guard would make the hook a no-op; both plugins' copies are covered. The seam widens by three functions the suite consumes -- readFileTail and extractUsageTokens in transcript-context, filterByPatterns in utils -- which is the praxis version bump. Co-authored-by: multica-agent <github@multica.ai>
| }); | ||
|
|
||
| test('applyCooldown stops decrementing at zero', () => { | ||
| const { report, state } = applyCooldown([A], { seen: [A.key], cooldown: 0 }, COOLDOWN); |
Contributor
|
resolve conflicts |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Lands the test gate decided in #34 and specified in
CONTRIBUTING.md's## Tests, plus the first two suites behind it. First of two: the hook-script suites follow as a stacked PR, purely additive.Both languages in one change, because each half of
make testasserts its file list is non-empty -- a one-language first PR would leave the other half red.The gate
Makefilegains atesttarget: two recipe lines, Node then Python, each building an explicit file list withgit ls-filesand refusing to run on an empty one.git ls-filesexits 0 on no match andnode --testwith no arguments walks the whole tree instead of failing, so an unguarded list is a gate that reports success having checked nothing. The assertion idiom is the JSON check's, inverify..github/workflows/ci.ymlgains atestjob pinning Node to 20 (the floor the hook scripts have to run on, not the runner's default) and Python to 3.12 as the other jobs do. Thecheckaggregator namestestin both places -- theneeds:list and theresultsarray -- which is the footgun that file warns about: miss the array and the required status check goes green while the suite is red.The suites
tests/plugins/praxis/skills/code-hygiene/find-duplicate-comments.test.js-- 26 tests overparseArgs,buildSkipMatcher,prose,trackedPaths,buildCommentIndex,findRetoldInDiffanddedupePairs. The two rejection paths inparseArgscallprocess.exit, so they are exercised as a spawned process rather than an import.tests/scripts/test_generate_codex.py-- 40 tests overparse_frontmatter,basic_string,render_agent_toml(including the'''rejection),display_name,short_description,qualify,plugin_manifest's name-mismatch raise, and both drift finders. The module is loaded withimportlib.util.spec_from_file_location, since the hyphen in its filename is not a valid module name.Fixtures are throwaway repositories built with
git initunder the OS temp directory, and temp directories withROOTpatched to point at them. Never this checkout -- a fixture that is this repository lets a rule which suppresses a finding here look correct.Regression tests
Four defects
find-duplicate-comments.jsaccrued over the #33 review rounds were held only by a code comment. Each now has a test namedregression:, and each was confirmed to fail with its fix reverted and nothing else:\rstrip -> 2 failures--skipvalue is matched literally, not compiled as a patternescapeRegExp-> 1 failure\guard -> 1 failuretrackedPathsdrops a submodule gitlinkOne correction to the record on the gitlink case.
find-duplicate-comments.js's header comment says a gitlink fed tocat-file --batch"answers<sha> submodule, which would abort the whole batch." That is current git's documented behaviour but not git 2.43's, which answersmissingwhen the gitlink's commit is absent locally and<sha> commit <size>when it is present -- neither of which aborts. So the test pins the version-independent contract instead: the entry never reaches the batch. The filter is right either way, and on 2.43 the pre-fix cost is a commit object's bytes indexed as if they were source. No code change; the test comment records it.Seam
find-duplicate-comments.jsexports two more functions,parseArgsandtrackedPaths, both consumed by the suite -- that is the praxis1.4.0 -> 1.4.1bump, and the regenerated.codex-plugin/plugin.jsonalongside it.Also
CONTRIBUTING.mdandCLAUDE.mddrop their "once the suite lands" clauses for the target itself.Verification
make test(26 Node + 40 Python, all pass) andmake verifypass locally.make lintpasses every hook that can run here; the three container-based hooks (lychee, actionlint, zizmor) cannot -- Docker's socket is not reachable from this runtime -- so thelintjob is what covers them. The workflow addition reuses the existing jobs' shapes: SHA-pinned actions andpersist-credentials: false.Closes JAR-430 is deliberately absent -- the issue also carries PR 2.
Refs #34
🤖 Generated with Claude Code