The autocompact buffer percentage was hardcoded to 16.5%. Users who set CLAUDE_CODE_AUTO_COMPACT_WINDOW to a custom token count (e.g. 400000 on a 1M-context model) saw a miscalibrated context meter and incorrect warning thresholds in the context-monitor hook (which reads used_pct from the bridge file the statusline writes). Now reads CLAUDE_CODE_AUTO_COMPACT_WINDOW from the hook env and computes: buffer_pct = acw_tokens / total_tokens * 100 Defaults to 16.5% when the var is absent or zero, preserving existing behavior. Also applies the renameDecimalPhases zero-padding fix for clean CI. Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
325 lines
10 KiB
JavaScript
325 lines
10 KiB
JavaScript
/**
|
|
* Tests for gsd-statusline.js GSD state display helpers.
|
|
*
|
|
* Covers:
|
|
* - parseStateMd across YAML-frontmatter, body-fallback, and partial formats
|
|
* - formatGsdState graceful degradation when fields are missing
|
|
* - readGsdState walk-up search with proper bounds
|
|
*/
|
|
|
|
'use strict';
|
|
|
|
const { test, describe } = require('node:test');
|
|
const assert = require('node:assert/strict');
|
|
const fs = require('node:fs');
|
|
const os = require('node:os');
|
|
const path = require('node:path');
|
|
|
|
const { parseStateMd, formatGsdState, readGsdState } = require('../hooks/gsd-statusline.js');
|
|
|
|
// ─── parseStateMd ───────────────────────────────────────────────────────────
|
|
|
|
describe('parseStateMd', () => {
|
|
test('parses full YAML frontmatter', () => {
|
|
const content = [
|
|
'---',
|
|
'status: executing',
|
|
'milestone: v1.9',
|
|
'milestone_name: Code Quality',
|
|
'---',
|
|
'',
|
|
'# State',
|
|
'Phase: 1 of 5 (fix-graphiti-deployment)',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.status, 'executing');
|
|
assert.equal(s.milestone, 'v1.9');
|
|
assert.equal(s.milestoneName, 'Code Quality');
|
|
assert.equal(s.phaseNum, '1');
|
|
assert.equal(s.phaseTotal, '5');
|
|
assert.equal(s.phaseName, 'fix-graphiti-deployment');
|
|
});
|
|
|
|
test('treats literal "null" values as null', () => {
|
|
const content = [
|
|
'---',
|
|
'status: null',
|
|
'milestone: null',
|
|
'milestone_name: null',
|
|
'---',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.status, null);
|
|
assert.equal(s.milestone, null);
|
|
assert.equal(s.milestoneName, null);
|
|
});
|
|
|
|
test('strips surrounding quotes from frontmatter values', () => {
|
|
const content = [
|
|
'---',
|
|
'milestone_name: "Code Quality"',
|
|
"milestone: 'v1.9'",
|
|
'---',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.milestone, 'v1.9');
|
|
assert.equal(s.milestoneName, 'Code Quality');
|
|
});
|
|
|
|
test('parses phase without name', () => {
|
|
const content = [
|
|
'---',
|
|
'status: planning',
|
|
'---',
|
|
'Phase: 3 of 10',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.phaseNum, '3');
|
|
assert.equal(s.phaseTotal, '10');
|
|
assert.equal(s.phaseName, null);
|
|
});
|
|
|
|
test('falls back to body Status when frontmatter is missing', () => {
|
|
const content = [
|
|
'# State',
|
|
'Status: Ready to plan',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.status, 'planning');
|
|
});
|
|
|
|
test('body fallback recognizes executing state', () => {
|
|
const content = 'Status: Executing phase 2';
|
|
assert.equal(parseStateMd(content).status, 'executing');
|
|
});
|
|
|
|
test('body fallback recognizes complete state', () => {
|
|
const content = 'Status: Complete';
|
|
assert.equal(parseStateMd(content).status, 'complete');
|
|
});
|
|
|
|
test('body fallback recognizes archived as complete', () => {
|
|
const content = 'Status: Archived';
|
|
assert.equal(parseStateMd(content).status, 'complete');
|
|
});
|
|
|
|
test('returns empty object for empty content', () => {
|
|
const s = parseStateMd('');
|
|
assert.deepEqual(s, {});
|
|
});
|
|
|
|
test('returns partial state when only some fields present', () => {
|
|
const content = [
|
|
'---',
|
|
'milestone: v2.0',
|
|
'---',
|
|
].join('\n');
|
|
|
|
const s = parseStateMd(content);
|
|
assert.equal(s.milestone, 'v2.0');
|
|
assert.equal(s.status, undefined);
|
|
assert.equal(s.phaseNum, undefined);
|
|
});
|
|
});
|
|
|
|
// ─── formatGsdState ─────────────────────────────────────────────────────────
|
|
|
|
describe('formatGsdState', () => {
|
|
test('formats full state with milestone name, status, and phase name', () => {
|
|
const out = formatGsdState({
|
|
milestone: 'v1.9',
|
|
milestoneName: 'Code Quality',
|
|
status: 'executing',
|
|
phaseNum: '1',
|
|
phaseTotal: '5',
|
|
phaseName: 'fix-graphiti-deployment',
|
|
});
|
|
assert.equal(out, 'v1.9 Code Quality · executing · fix-graphiti-deployment (1/5)');
|
|
});
|
|
|
|
test('skips placeholder "milestone" value in milestoneName', () => {
|
|
const out = formatGsdState({
|
|
milestone: 'v1.0',
|
|
milestoneName: 'milestone',
|
|
status: 'planning',
|
|
});
|
|
assert.equal(out, 'v1.0 · planning');
|
|
});
|
|
|
|
test('uses short phase form when phase name is missing', () => {
|
|
const out = formatGsdState({
|
|
milestone: 'v2.0',
|
|
status: 'executing',
|
|
phaseNum: '3',
|
|
phaseTotal: '7',
|
|
});
|
|
assert.equal(out, 'v2.0 · executing · ph 3/7');
|
|
});
|
|
|
|
test('omits phase entirely when phaseNum/phaseTotal missing', () => {
|
|
const out = formatGsdState({
|
|
milestone: 'v1.0',
|
|
status: 'planning',
|
|
});
|
|
assert.equal(out, 'v1.0 · planning');
|
|
});
|
|
|
|
test('handles milestone version only (no name)', () => {
|
|
const out = formatGsdState({
|
|
milestone: 'v1.9',
|
|
status: 'executing',
|
|
});
|
|
assert.equal(out, 'v1.9 · executing');
|
|
});
|
|
|
|
test('handles milestone name only (no version)', () => {
|
|
const out = formatGsdState({
|
|
milestoneName: 'Foundations',
|
|
status: 'planning',
|
|
});
|
|
assert.equal(out, 'Foundations · planning');
|
|
});
|
|
|
|
test('returns empty string for empty state', () => {
|
|
assert.equal(formatGsdState({}), '');
|
|
});
|
|
|
|
test('returns only available parts when everything else is missing', () => {
|
|
assert.equal(formatGsdState({ status: 'planning' }), 'planning');
|
|
});
|
|
});
|
|
|
|
// ─── readGsdState ───────────────────────────────────────────────────────────
|
|
|
|
describe('readGsdState', () => {
|
|
const tmpRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'gsd-statusline-test-'));
|
|
|
|
test('finds STATE.md in the starting directory', () => {
|
|
const proj = fs.mkdtempSync(path.join(tmpRoot, 'proj-'));
|
|
fs.mkdirSync(path.join(proj, '.planning'), { recursive: true });
|
|
fs.writeFileSync(
|
|
path.join(proj, '.planning', 'STATE.md'),
|
|
'---\nstatus: executing\nmilestone: v1.0\n---\n'
|
|
);
|
|
|
|
const s = readGsdState(proj);
|
|
assert.equal(s.status, 'executing');
|
|
assert.equal(s.milestone, 'v1.0');
|
|
});
|
|
|
|
test('walks up to find STATE.md in a parent directory', () => {
|
|
const proj = fs.mkdtempSync(path.join(tmpRoot, 'proj-'));
|
|
fs.mkdirSync(path.join(proj, '.planning'), { recursive: true });
|
|
fs.writeFileSync(
|
|
path.join(proj, '.planning', 'STATE.md'),
|
|
'---\nstatus: planning\n---\n'
|
|
);
|
|
|
|
const nested = path.join(proj, 'src', 'components', 'deep');
|
|
fs.mkdirSync(nested, { recursive: true });
|
|
|
|
const s = readGsdState(nested);
|
|
assert.equal(s.status, 'planning');
|
|
});
|
|
|
|
test('returns null when no STATE.md exists in the walk-up chain', () => {
|
|
const proj = fs.mkdtempSync(path.join(tmpRoot, 'proj-'));
|
|
const nested = path.join(proj, 'src');
|
|
fs.mkdirSync(nested, { recursive: true });
|
|
|
|
assert.equal(readGsdState(nested), null);
|
|
});
|
|
|
|
test('returns null on malformed STATE.md without crashing', () => {
|
|
const proj = fs.mkdtempSync(path.join(tmpRoot, 'proj-'));
|
|
fs.mkdirSync(path.join(proj, '.planning'), { recursive: true });
|
|
// Valid file (no content to crash on) — parseStateMd returns {}
|
|
fs.writeFileSync(path.join(proj, '.planning', 'STATE.md'), '');
|
|
|
|
const s = readGsdState(proj);
|
|
// Empty file yields an empty state object, not null — the function
|
|
// only returns null when no file is found.
|
|
assert.deepEqual(s, {});
|
|
});
|
|
});
|
|
|
|
// ─── CLAUDE_CODE_AUTO_COMPACT_WINDOW context meter (#2219) ──────────────────
|
|
|
|
describe('context meter respects CLAUDE_CODE_AUTO_COMPACT_WINDOW (#2219)', () => {
|
|
const { execFileSync } = require('node:child_process');
|
|
const hookPath = path.join(__dirname, '..', 'hooks', 'gsd-statusline.js');
|
|
|
|
/**
|
|
* Run the statusline hook with a synthetic context_window payload and
|
|
* return the used_pct written to the bridge file.
|
|
*/
|
|
function runHook(remainingPct, totalTokens, acwEnv) {
|
|
const sessionId = `test-2219-${Date.now()}-${Math.random().toString(36).slice(2)}`;
|
|
const payload = JSON.stringify({
|
|
model: { display_name: 'Claude' },
|
|
workspace: { current_dir: os.tmpdir() },
|
|
session_id: sessionId,
|
|
context_window: {
|
|
remaining_percentage: remainingPct,
|
|
total_tokens: totalTokens,
|
|
},
|
|
});
|
|
|
|
const env = { ...process.env };
|
|
if (acwEnv != null) {
|
|
env.CLAUDE_CODE_AUTO_COMPACT_WINDOW = String(acwEnv);
|
|
} else {
|
|
delete env.CLAUDE_CODE_AUTO_COMPACT_WINDOW;
|
|
}
|
|
|
|
try {
|
|
execFileSync(process.execPath, [hookPath], {
|
|
input: payload,
|
|
env,
|
|
timeout: 4000,
|
|
});
|
|
} catch (e) {
|
|
// Non-zero exit is fine — hook exits 0 on success, but we're reading
|
|
// the bridge file, not the exit code. Ignore exit failures here.
|
|
}
|
|
|
|
const bridgePath = path.join(os.tmpdir(), `claude-ctx-${sessionId}.json`);
|
|
const bridge = JSON.parse(fs.readFileSync(bridgePath, 'utf8'));
|
|
fs.unlinkSync(bridgePath);
|
|
return bridge.used_pct;
|
|
}
|
|
|
|
test('default buffer (no env var): 50% remaining → ~60% used', () => {
|
|
// Default 16.5% buffer: usableRemaining = (50 - 16.5) / (100 - 16.5) * 100 ≈ 40.12%
|
|
// used ≈ 100 - 40.12 = 59.88 → rounded 60
|
|
const used = runHook(50, 1_000_000, null);
|
|
assert.strictEqual(used, 60);
|
|
});
|
|
|
|
test('CLAUDE_CODE_AUTO_COMPACT_WINDOW=400000: 50% remaining → ~83% used', () => {
|
|
// With 1M total, 400k window → buffer = 40%. usableRemaining = (50 - 40) / (100 - 40) * 100 ≈ 16.67%
|
|
// used ≈ 100 - 16.67 = 83.33 → rounded 83
|
|
const used = runHook(50, 1_000_000, 400_000);
|
|
assert.strictEqual(used, 83);
|
|
});
|
|
|
|
test('CLAUDE_CODE_AUTO_COMPACT_WINDOW=0 falls back to default buffer', () => {
|
|
// Explicit "0" means unset — should behave like no env var (16.5% buffer)
|
|
const used = runHook(50, 1_000_000, 0);
|
|
assert.strictEqual(used, 60);
|
|
});
|
|
|
|
test('buffer capped at 100% when ACW exceeds total context', () => {
|
|
// Pathological: ACW > totalCtx → buffer = 100%. With no usable range left,
|
|
// usableRemaining = max(0, (50-100)/(100-100)*100) = max(0, -Inf) = 0,
|
|
// so used = 100 (context reported as completely full).
|
|
const used = runHook(50, 1_000_000, 2_000_000);
|
|
assert.strictEqual(used, 100);
|
|
});
|
|
});
|