published-language/instrument/run.mjs

#!/usr/bin/env node
/**
 * Instrument — where each guard actually lives (temoin/ADR-018,
 * prismagram-corpus/ADR-061 §2.6 and §2.8).
 *
 * Runs INSIDE the observed jurisdiction's CI: the jurisdiction measures
 * itself; the registry publishes the instrument and receives the findings.
 * The instrument reads the invariants the jurisdiction declares in its
 * journey documents, finds every file that names them, and states, per
 * invariant, at which stratum its verification is observed:
 *
 *   domicilie      named by a test at its due stratum, nowhere above
 *   renforce       named by tests at its due stratum AND above — defence in
 *                  depth, reported positively, never blamed
 *   mal_domicilie  named by tests only OUTSIDE its due stratum — the one gap
 *   aucun_locus    named by no test at all
 *
 * The instrument states where guards live; it does not run them, and it does
 * not fail on gaps: the gap IS the product. It exits non-zero only when a
 * prerequisite is broken — loud failure, never a silent success.
 *
 * Findings are appended (never rewritten) to the jurisdiction's own
 * repository, and only when they change: an unchanged finding produces no
 * event. The run journal is rewritten at every run, so that staleness can be
 * judged against the declared cadence by whoever reads the repository.
 *
 * Usage:
 *   node run.mjs --config docs/parcours/_meta/strates.yaml [--repo DIR] [--dry-run]
 */
import { appendFileSync, existsSync, mkdirSync, readFileSync, readdirSync, statSync, writeFileSync } from 'node:fs';
import { dirname, isAbsolute, join, relative, sep } from 'node:path';
import { execFileSync } from 'node:child_process';
import { createHash } from 'node:crypto';
import { parse as parseYaml, stringify as stringifyYaml } from 'yaml';

const JOURNEY_FILE = /^P[MU]-\d{3}.*\.md$/;
const IGNORED_DIRS = new Set(['node_modules', 'target', 'dist', 'build', 'coverage', 'out']);
const MAX_FILE_BYTES = 2_000_000;
const MAX_PIECES_PER_STRATUM = 5;

function refuse(message) {
  console.error(`Instrument — REFUS : ${message}`);
  process.exit(1);
}

function parseArgs(argv) {
  const args = { repo: process.cwd(), dryRun: false, config: null };
  for (let i = 0; i < argv.length; i += 1) {
    if (argv[i] === '--config') args.config = argv[i + 1] ?? null, i += 1;
    else if (argv[i] === '--repo') args.repo = argv[i + 1] ?? args.repo, i += 1;
    else if (argv[i] === '--dry-run') args.dryRun = true;
    else refuse(`argument inconnu « ${argv[i]} » — attendus : --config <réglage>, --repo <dépôt>, --dry-run`);
  }
  if (args.config === null)
    refuse('--config <réglage> est requis. Le réglage appartient à la juridiction : voir instrument/README.md pour sa forme.');
  return args;
}

/** The tuning file belongs to the jurisdiction; the instrument only knows its shape. */
function loadConfig(repo, path) {
  const full = isAbsolute(path) ? path : join(repo, path);
  if (!existsSync(full))
    refuse(`le réglage « ${path} » n'existe pas dans ${repo}. Rien n'est créé à la volée : la juridiction pose son réglage (voir instrument/README.md), puis relance.`);
  let cfg;
  try {
    cfg = parseYaml(readFileSync(full, 'utf8'));
  } catch (e) {
    refuse(`le réglage « ${path} » n'est pas un YAML lisible : ${e instanceof Error ? e.message.split('\n')[0] : e}`);
  }
  for (const key of ['juridiction', 'parcours', 'strates', 'epreuves', 'sortie', 'journal_execution'])
    if (cfg?.[key] === undefined) refuse(`le réglage « ${path} » ne déclare pas « ${key} »`);
  if (!Array.isArray(cfg.strates) || cfg.strates.length === 0)
    refuse(`le réglage « ${path} » doit déclarer au moins une strate`);
  for (const s of cfg.strates) {
    if (typeof s?.nom !== 'string' || !Array.isArray(s?.chemins))
      refuse(`chaque strate du réglage porte « nom » et « chemins » — l'une des deux manque`);
    if (s.rang !== undefined && typeof s.rang !== 'number')
      refuse(`la strate « ${s.nom} » déclare un rang qui n'est pas un nombre — le rang est une correspondance vers une échelle tierce, ou rien`);
  }
  const due = cfg.strates.find((s) => s.due === true) ?? cfg.strates[0];
  return { ...cfg, due: due.nom, raw: readFileSync(full, 'utf8'), path };
}

/** Glob → RegExp. Supports **, * and ? — nothing else, on purpose. */
function globToRegExp(glob) {
  let out = '^';
  for (let i = 0; i < glob.length; i += 1) {
    const c = glob[i];
    if (c === '*' && glob[i + 1] === '*') {
      out += glob[i + 2] === '/' ? '(?:.*/)?' : '.*';
      i += glob[i + 2] === '/' ? 2 : 1;
    } else if (c === '*') out += '[^/]*';
    else if (c === '?') out += '[^/]';
    else out += /[A-Za-z0-9_\-/]/.test(c) ? c : `\\${c}`;
  }
  return new RegExp(`${out}$`);
}

function walk(root, dir, acc) {
  for (const name of readdirSync(dir)) {
    if (name.startsWith('.')) continue;
    const full = join(dir, name);
    const st = statSync(full, { throwIfNoEntry: false });
    if (st === undefined) continue;
    if (st.isDirectory() && !IGNORED_DIRS.has(name)) walk(root, full, acc);
    else if (st.isFile() && st.size <= MAX_FILE_BYTES) acc.push(relative(root, full).split(sep).join('/'));
  }
  return acc;
}

function frontMatter(text) {
  const m = /^---\r?\n([\s\S]*?)\r?\n---(\r?\n|$)/.exec(text);
  if (m === null) return null;
  try { return parseYaml(m[1]); } catch { return null; }
}

/**
 * Invariants as the jurisdiction declares them, in its journeys' front matter.
 * The same id declared with two different statements is an homonymy: it is
 * carried into the finding, never resolved here — the instrument exposes, it
 * does not reconcile.
 */
function collectInvariants(repo, dir) {
  let files;
  try {
    files = readdirSync(join(repo, dir)).filter((f) => JOURNEY_FILE.test(f)).sort();
  } catch {
    refuse(`le répertoire des parcours « ${dir} » est introuvable — rien à mesurer n'est pas un succès`);
  }
  const byId = new Map();
  for (const f of files) {
    const fm = frontMatter(readFileSync(join(repo, dir, f), 'utf8'));
    const journey = typeof fm?.id === 'string' ? fm.id : (f.match(/^(P[MU]-\d{3})/)?.[1] ?? f);
    const declared = Array.isArray(fm?.invariants) ? fm.invariants : [];
    for (const inv of declared) {
      if (typeof inv?.id !== 'string') continue;
      const cur = byId.get(inv.id) ?? { id: inv.id, statements: new Map(), declaredBy: [] };
      cur.declaredBy.push(journey);
      const s = typeof inv.enonce === 'string' ? inv.enonce.trim() : '';
      if (s !== '' && !cur.statements.has(s)) cur.statements.set(s, journey);
      byId.set(inv.id, cur);
    }
  }
  if (byId.size === 0)
    refuse(`aucun invariant déclaré dans « ${dir} » — rien à mesurer n'est pas un succès`);
  return { invariants: byId, journeys: files.length };
}

const escapeRe = (s) => s.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');

/** Every file covered by a stratum, with its stratum and its test-ness. */
function classifyFiles(repo, cfg) {
  const strata = cfg.strates.map((s) => ({ nom: s.nom, res: s.chemins.map(globToRegExp) }));
  const testRes = cfg.epreuves.map(globToRegExp);
  const out = [];
  for (const path of walk(repo, repo, [])) {
    const stratum = strata.find((s) => s.res.some((re) => re.test(path)));
    if (stratum === undefined) continue;
    out.push({ path, stratum: stratum.nom, isTest: testRes.some((re) => re.test(path)) });
  }
  return out;
}

/** Where each invariant is NAMED — the instrument observes naming, not execution. */
function scanCitations(repo, files, ids) {
  const pattern = new RegExp(`\\b(${ids.map(escapeRe).join('|')})\\b`);
  const hits = new Map(ids.map((id) => [id, []]));
  for (const f of files) {
    const text = readFileSync(join(repo, f.path), 'utf8');
    if (!pattern.test(text)) continue;
    const each = new RegExp(pattern.source, 'g');
    text.split('\n').forEach((line, i) => {
      for (const m of line.matchAll(each))
        hits.get(m[1]).push({ path: f.path, line: i + 1, stratum: f.stratum, isTest: f.isTest });
    });
  }
  return hits;
}

function judge(hits, cfg) {
  const order = cfg.strates.map((s) => s.nom);
  const strata = [...new Set(hits.filter((h) => h.isTest).map((h) => h.stratum))]
    .sort((a, b) => order.indexOf(a) - order.indexOf(b));
  if (strata.length === 0) return { etat: 'aucun_locus', strates: [] };
  if (!strata.includes(cfg.due)) return { etat: 'mal_domicilie', strates: strata };
  return { etat: strata.length === 1 ? 'domicilie' : 'renforce', strates: strata };
}

/** One sentence a reader understands without the doctrine — layer 1 wording. */
function sentence(id, verdict, due) {
  const above = verdict.strates.filter((s) => s !== due);
  if (verdict.etat === 'aucun_locus') return `${id} : aucune épreuve ne le nomme — aucun endroit constaté.`;
  if (verdict.etat === 'mal_domicilie')
    return `${id} : vérifié seulement hors de son étage (${verdict.strates.join(', ')}, jamais ${due}) — la garde n'est pas à sa place.`;
  if (verdict.etat === 'renforce')
    return `${id} : vérifié à son étage (${due}) et au-dessus (${above.join(', ')}) — défense en profondeur.`;
  return `${id} : vérifié à son étage (${due}).`;
}

/** Last recorded finding per invariant, and how many findings each has. */
function readPrevious(sortiePath) {
  const last = new Map();
  const counts = new Map();
  if (!existsSync(sortiePath)) return { last, counts };
  const items = parseYaml(readFileSync(sortiePath, 'utf8'));
  for (const it of Array.isArray(items) ? items : []) {
    if (typeof it?.invariant !== 'string') continue;
    counts.set(it.invariant, (counts.get(it.invariant) ?? 0) + 1);
    last.set(it.invariant, `${it.etat}|${(it.strates_constatees ?? []).join(',')}`);
  }
  return { last, counts };
}

function pieces(hits) {
  const kept = [];
  const omitted = { n: 0 };
  const perStratum = new Map();
  for (const h of hits.filter((x) => x.isTest)) {
    const n = (perStratum.get(h.stratum) ?? 0) + 1;
    perStratum.set(h.stratum, n);
    if (n <= MAX_PIECES_PER_STRATUM) kept.push({ chemin: h.path, ligne: h.line, strate: h.stratum });
    else omitted.n += 1;
  }
  return { kept, omitted: omitted.n };
}

function constatItem(cfg, inv, verdict, hits, seq, meta) {
  const { kept, omitted } = pieces(hits);
  const item = {
    id: `${cfg.juridiction}/MES-${inv.id}-${seq}`,
    classe: 'measure',
    invariant: inv.id,
    declare_par: inv.declaredBy,
    enonce: inv.statements.size > 0 ? [...inv.statements.keys()][0] : null,
    etat: verdict.etat,
    constat: sentence(inv.id, verdict, cfg.due),
    strates_constatees: verdict.strates,
    ...rangsConstates(cfg, verdict.strates),
    pieces: kept,
    ...(omitted > 0 ? { pieces_omises: omitted } : {}),
    citations_hors_epreuve: hits.filter((h) => !h.isTest).length,
    ...(inv.statements.size > 1
      ? { homonymie: [...inv.statements.entries()].map(([enonce, par]) => ({ enonce, declare_par: par })) }
      : {}),
    provenance: 'measured',
    confiance: 1.0,
    mesure_le: meta.now,
    revision: meta.revision,
    reglage_empreinte: meta.reglageHash,
  };
  return item;
}

/**
 * La correspondance vers une échelle tierce (p. ex. les rangs de Percursus),
 * si la juridiction la déclare dans son réglage. Émise SEULEMENT quand toutes
 * les strates constatées portent un rang : une correspondance partielle
 * laisserait le consommateur deviner — deviner n'est pas constater.
 */
function rangsConstates(cfg, strates) {
  const rang = new Map(cfg.strates.filter((s) => typeof s.rang === 'number').map((s) => [s.nom, s.rang]));
  if (strates.length === 0 || !strates.every((n) => rang.has(n))) return {};
  return { rangs_constates: strates.map((n) => rang.get(n)) };
}

function gitRevision(repo) {
  try {
    return execFileSync('git', ['-C', repo, 'rev-parse', 'HEAD'], { encoding: 'utf8', stdio: ['ignore', 'pipe', 'ignore'] }).trim();
  } catch {
    return null;
  }
}

const FILE_HEADER = `# Constats de l'instrument — en ajout seul, un constat ne se réécrit jamais.
# Un état qui change s'écrit comme un constat nouveau (temoin/ADR-018).
`;

function main() {
  const args = parseArgs(process.argv.slice(2));
  const cfg = loadConfig(args.repo, args.config);
  const { invariants, journeys } = collectInvariants(args.repo, cfg.parcours);
  const files = classifyFiles(args.repo, cfg);
  const hits = scanCitations(args.repo, files, [...invariants.keys()]);
  const meta = {
    now: new Date().toISOString(),
    revision: gitRevision(args.repo),
    reglageHash: `sha256:${createHash('sha256').update(cfg.raw).digest('hex')}`,
  };
  const sortiePath = join(args.repo, cfg.sortie);
  const { last, counts } = readPrevious(sortiePath);
  const tally = { domicilie: 0, renforce: 0, mal_domicilie: 0, aucun_locus: 0 };
  const fresh = [];
  for (const inv of [...invariants.values()].sort((a, b) => a.id.localeCompare(b.id))) {
    const verdict = judge(hits.get(inv.id), cfg);
    tally[verdict.etat] += 1;
    const key = `${verdict.etat}|${verdict.strates.join(',')}`;
    if (last.get(inv.id) === key) continue;
    fresh.push(constatItem(cfg, inv, verdict, hits.get(inv.id), (counts.get(inv.id) ?? 0) + 1, meta));
  }
  emit(args, cfg, { journeys, invariants: invariants.size, tally, fresh, meta, sortiePath });
}

function emit(args, cfg, r) {
  const journal = {
    juridiction: cfg.juridiction,
    executee_le: r.meta.now,
    revision: r.meta.revision,
    ...(typeof cfg.cadence_heures === 'number' ? { cadence_heures: cfg.cadence_heures } : {}),
    parcours_lus: r.journeys,
    invariants: r.invariants,
    ...(cfg.strates.some((s) => typeof s.rang === 'number')
      ? { correspondance_rangs: Object.fromEntries(cfg.strates.filter((s) => typeof s.rang === 'number').map((s) => [s.nom, s.rang])) }
      : {}),
    etats: r.tally,
    constats_nouveaux: r.fresh.length,
    reglage_empreinte: r.meta.reglageHash,
  };
  if (args.dryRun) {
    if (r.fresh.length > 0) process.stdout.write(stringifyYaml(r.fresh));
    console.log('--- (--dry-run : rien n\'est écrit)');
  } else {
    if (r.fresh.length > 0) {
      mkdirSync(dirname(r.sortiePath), { recursive: true });
      const prefix = existsSync(r.sortiePath) ? '' : FILE_HEADER;
      appendFileSync(r.sortiePath, prefix + stringifyYaml(r.fresh));
    }
    mkdirSync(dirname(join(args.repo, cfg.journal_execution)), { recursive: true });
    writeFileSync(join(args.repo, cfg.journal_execution), stringifyYaml(journal));
  }
  console.log(`Instrument — ${cfg.juridiction}${r.meta.revision === null ? '' : ` @ ${r.meta.revision.slice(0, 10)}`}`);
  console.log(`  ${r.invariants} invariant(s) déclaré(s) par ${r.journeys} parcours`);
  console.log(`  à son étage : ${r.tally.domicilie} · renforcé : ${r.tally.renforce} · hors de son étage : ${r.tally.mal_domicilie} · jamais éprouvé : ${r.tally.aucun_locus}`);
  console.log(`  constat(s) nouveau(x) : ${r.fresh.length}${args.dryRun ? '' : ` → ${cfg.sortie}`}`);
}

main();