testguard-cli 0.6.0 → 0.7.0

package/src/probe/cost.mjs

@@ -0,0 +1,145 @@
+/**
+ * What a probe costs, derived from evidence that already exists (issue #67).
+ *
+ * A probe's wall clock is not a function of how many claims there are. It is
+ * `(baseline runs + faults × confirm runs) × the cost of the claim's defender
+ * SET`, per claim. One slow acceptance test named as a defender by five claims
+ * is therefore paid for thirty times, and nothing in the output says so — the
+ * only symptom is that the gate takes twenty-four minutes and nobody knows
+ * which line to blame.
+ *
+ * Every number here is read back out of `durationMs` on the runs the probe
+ * already recorded. Nothing re-measures, nothing spawns, nothing guesses: a
+ * cost report costs the price of reading one JSON file.
+ *
+ * ATTRIBUTION, HONESTLY. Runs execute a claim's whole defender set together,
+ * so the runner never says how much of a run belonged to which file. Per-file
+ * `ms` is therefore the time of every run that *included* that file: an upper
+ * bound on what removing it could save, and, when a claim names several
+ * defenders, a number that overlaps with its siblings. The per-file figures do
+ * not sum to the total, by construction. `totalMs` is the only additive one.
+ */
+
+/** Every run a record paid for, in the order the probe made them. */
+export const runsOf = (record) => [
+  ...(record.detail?.baselineRuns ?? []),
+  ...(record.detail?.probeRuns ?? []),
+  ...(record.detail?.escalationRuns ?? []),
+];
+
+/**
+ * What one record cost. A reused record reports the duration of the runs it
+ * reused, which is the right number for planning — it is what the record costs
+ * whenever reuse does not apply — and `reused` marks it so a reader does not
+ * read it as wall clock just spent.
+ */
+export function recordCost(record) {
+  const runs = runsOf(record);
+  return {
+    ms: runs.reduce((a, r) => a + (r.durationMs ?? 0), 0),
+    runs: runs.length,
+    reused: Boolean(record.reusedFrom),
+  };
+}
+
+/** Per claim: what its faults cost in total, most expensive first. */
+export function claimCosts(records) {
+  const by = new Map();
+  for (const r of records) {
+    const c = recordCost(r);
+    const e = by.get(r.claim.id) ?? { claimId: r.claim.id, ms: 0, runs: 0, faults: 0, reused: 0, defenders: new Set() };
+    e.ms += c.ms;
+    e.runs += c.runs;
+    e.faults += 1;
+    if (c.reused) e.reused += 1;
+    for (const d of r.defenders?.resolved ?? []) e.defenders.add(d);
+    by.set(r.claim.id, e);
+  }
+  return [...by.values()]
+    .map((e) => ({ ...e, defenders: [...e.defenders].sort() }))
+    .sort((a, b) => b.ms - a.ms || a.claimId.localeCompare(b.claimId));
+}
+
+/**
+ * Per defender file: the time of every run that included it, and the claims
+ * that named it. See the attribution note above — this is a ceiling, not a
+ * share, and the figures overlap.
+ */
+export function defenderCosts(records) {
+  const by = new Map();
+  for (const r of records) {
+    const c = recordCost(r);
+    for (const file of r.defenders?.resolved ?? []) {
+      const e = by.get(file) ?? { file, ms: 0, runs: 0, claims: new Set() };
+      e.ms += c.ms;
+      e.runs += c.runs;
+      e.claims.add(r.claim.id);
+      by.set(file, e);
+    }
+  }
+  return [...by.values()]
+    .map((e) => ({ ...e, claims: [...e.claims].sort() }))
+    .sort((a, b) => b.ms - a.ms || a.file.localeCompare(b.file));
+}
+
+/**
+ * The whole cost model for one evidence document.
+ *
+ * `sharedDefenders` is the actionable part: a file named by more than one
+ * claim, whose cost is multiplied by every claim that names it. Splitting the
+ * behaviour it defends into a unit test that runs in milliseconds is what
+ * turns a twenty-four-minute gate into a one-minute one, and until a report
+ * names the file nobody knows which test to split.
+ */
+export function costReport(records) {
+  const claims = claimCosts(records);
+  const defenders = defenderCosts(records);
+  return {
+    totalMs: claims.reduce((a, c) => a + c.ms, 0),
+    totalRuns: claims.reduce((a, c) => a + c.runs, 0),
+    reusedRecords: records.filter((r) => r.reusedFrom).length,
+    records: records.length,
+    claims,
+    defenders,
+    sharedDefenders: defenders.filter((d) => d.claims.length > 1),
+  };
+}
+
+const secs = (ms) => (ms >= 10_000 ? `${Math.round(ms / 1000)}s` : ms >= 1000 ? `${(ms / 1000).toFixed(1)}s` : `${ms}ms`);
+/** Exported so a caller can format a duration the same way the report does. */
+export const formatMs = secs;
+
+/**
+ * The cost report as text. `limit` caps each list; the totals always describe
+ * everything, so a truncated list never misrepresents the whole.
+ */
+export function renderCost(report, { limit = 10 } = {}) {
+  const out = [];
+  if (!report.records) return 'no evidence records — run `testguard probe` first; cost is read back from the runs it records.';
+  out.push(`${report.records} fault record${report.records === 1 ? '' : 's'} cost ${secs(report.totalMs)} across ${report.totalRuns} defender runs${report.reusedRecords ? ` (${report.reusedRecords} reused; their runs are what they cost when reuse does not apply)` : ''}.`);
+  out.push('');
+  out.push('most expensive claims');
+  for (const c of report.claims.slice(0, limit)) {
+    out.push(`  ${secs(c.ms).padStart(7)}  ${c.claimId.padEnd(34)} ${c.faults} fault${c.faults === 1 ? ' ' : 's'} · ${c.runs} runs${c.reused ? ` · ${c.reused} reused` : ''}`);
+  }
+  if (report.claims.length > limit) out.push(`  … ${report.claims.length - limit} more`);
+  out.push('');
+  out.push('defender files, by the time of the runs that included them');
+  out.push('  (a claim runs its whole defender set at once, so these overlap and do not sum to the total)');
+  for (const d of report.defenders.slice(0, limit)) {
+    out.push(`  ${secs(d.ms).padStart(7)}  ${d.file.padEnd(44)} ${d.claims.length} claim${d.claims.length === 1 ? '' : 's'}`);
+  }
+  if (report.defenders.length > limit) out.push(`  … ${report.defenders.length - limit} more`);
+  if (report.sharedDefenders.length) {
+    out.push('');
+    out.push('shared defenders — each claim pays the file\'s full cost again');
+    for (const d of report.sharedDefenders.slice(0, limit)) {
+      out.push(`  ${secs(d.ms).padStart(7)}  ${d.file}`);
+      out.push(`           named by ${d.claims.join(', ')}`);
+    }
+    out.push('');
+    out.push('  To make a probe cheaper, split the behaviour a shared defender proves into a');
+    out.push('  test that does not need the expensive setup, and point the claim at that.');
+  }
+  return out.join('\n');
+}

package/src/probe/discover.mjs

@@ -1,18 +1,26 @@
 import { join } from 'node:path';
 import { listTestFiles } from './runners/shared.mjs';
+import { testGlobs as PY_TEST_GLOBS } from './runners/python.mjs';
 import { fileImports } from './rank.mjs';
+import { pyFileImports } from './pyimports.mjs';
 import { classifyDefenders } from './mocks.mjs';
 
+/** A target's language decides how its importers are found. Nothing else about it matters here. */
+export const isPython = (rel) => rel.endsWith('.py');
+
 /**
  * When a claim declares no defenders: the test files that import the fault's
- * target file — directly, by relative path or through a resolved alias — and
- * do NOT mock it. A file that mocks the target cannot detect any fault in it;
- * counting it would make `nocover` under-report and waste probe runs on files
- * that cannot fail. `nocover` therefore means exactly "no test file imports
- * this source without mocking it".
+ * target file — directly, by relative path or through a resolved alias in
+ * JavaScript, by module name in Python — and do NOT replace it wholesale. A
+ * file that mocks the target cannot detect any fault in it; counting it would
+ * make `nocover` under-report and waste probe runs on files that cannot fail.
+ * `nocover` therefore means exactly "no test file imports this source without
+ * mocking it".
  */
 export function discoverDefendersDetailed(projectDir, targetRel) {
-  const importing = listTestFiles(projectDir).filter((t) => fileImports(projectDir, join(projectDir, t), targetRel));
+  const importing = isPython(targetRel)
+    ? listTestFiles(projectDir, PY_TEST_GLOBS).filter((t) => pyFileImports(projectDir, join(projectDir, t), targetRel, t))
+    : listTestFiles(projectDir).filter((t) => fileImports(projectDir, join(projectDir, t), targetRel));
   return { importing, ...classifyDefenders(projectDir, targetRel, importing) };
 }
 

package/src/probe/inject.mjs

@@ -63,24 +63,56 @@ export function mutate(source, fault) {
 }
 
 /**
- * Apply a fault to a file on disk. Returns a handle whose `restore()` puts the
- * original back; restore is idempotent and also runs on process exit/signal.
+ * Replace a file's contents on disk. Returns a handle whose `restore()` puts
+ * the original back; restore is idempotent and also runs on process
+ * exit/signal. `content` is the new text, or a function of the original.
+ *
+ * `inPlace` decides what a *missing* target means at restore time, and the two
+ * modes are opposites. In worktree mode the mutation only ever existed inside a
+ * scratch worktree that is discarded at the end of the run, so a target that has
+ * vanished is already restored in every sense that matters — a parallel session
+ * pruning its own leaked worktrees is enough to cause it, and throwing there
+ * converts a harmless cleanup race into a probe that dies with no evidence at
+ * all. In `--in-place` mode the same condition means the user's own file is
+ * gone, and silence would be the worst possible answer.
  */
-export function applyFault(projectDir, fault) {
+export function applyContent(projectDir, file, content, { inPlace = false } = {}) {
   installHandlers();
-  const path = join(projectDir, fault.file);
+  const path = join(projectDir, file);
   const original = readFileSync(path, 'utf8');
-  const anchor = locate(original, fault);
-  if (anchor.status !== 'ok') return { anchor, applied: false, restore: () => {} };
 
   let restored = false;
+  const handle = { applied: true, restore: null };
   const restore = () => {
     if (restored) return;
     restored = true;
     live.delete(restore);
-    writeFileSync(path, original);
+    try {
+      writeFileSync(path, original);
+    } catch (err) {
+      // Only a target that is no longer there. A permission failure or a
+      // directory where a file should be can still mean a mutated file left on
+      // disk, and those must stay loud in either mode.
+      if (inPlace || err.code !== 'ENOENT') throw err;
+      handle.restoreSkipped = 'target-missing';
+    }
   };
+  handle.restore = restore;
   live.add(restore);
-  writeFileSync(path, mutate(original, fault));
-  return { anchor, applied: true, restore };
+  writeFileSync(path, typeof content === 'function' ? content(original) : content);
+  return handle;
+}
+
+/**
+ * Apply a fault to a file on disk. Returns a handle whose `restore()` puts the
+ * original back; `applied` is false and nothing is written when the anchor does
+ * not hit exactly as declared.
+ */
+export function applyFault(projectDir, fault, opts = {}) {
+  const anchor = locate(readFileSync(join(projectDir, fault.file), 'utf8'), fault);
+  if (anchor.status !== 'ok') return { anchor, applied: false, restore: () => {} };
+  // The same object, never a spread: `restoreSkipped` is set on it later.
+  const handle = applyContent(projectDir, fault.file, (src) => mutate(src, fault), opts);
+  handle.anchor = anchor;
+  return handle;
 }

package/src/probe/mocks.mjs

@@ -1,6 +1,7 @@
 import { readFileSync } from 'node:fs';
 import { join } from 'node:path';
 import { specifierResolvesTo } from './rank.mjs';
+import { classifyPythonDefenders } from './pyimports.mjs';
 
 /**
  * Mock-awareness for defenders (issues #20, #21).
@@ -104,6 +105,7 @@ function matchesTargetAnyImport(source, matchesTarget) {
  * the annotation is recorded (`unasserted-annotated`) — silenced, never hidden.
  */
 export function classifyDefenders(projectDir, targetRel, candidates) {
+  if (targetRel.endsWith('.py')) return classifyPythonDefenders(projectDir, targetRel, candidates);
   const canDetect = [];
   const mocking = [];
   const signals = [];