aere-quantum/crypto-inventory/lib/scan.mjs

// Parcurgerea dosarului si orchestrarea detectorilor. Nimic nu se sare in tacere: fiecare
// fisier vazut ajunge intr-o singura categorie numarata, si suma se verifica la sfarsit.
import { readdirSync, statSync, readFileSync, lstatSync } from 'node:fs';
import { join, relative, sep, basename, resolve, extname } from 'node:path';
import { facContext, aplicaSuprimarea, REZOLVARI_PE_FISIER } from './context.mjs';
import { detecteazaJs } from './detect-js.mjs';
import { detecteazaPy } from './detect-py.mjs';
import { detecteazaJava } from './detect-java.mjs';
import { detecteazaGo } from './detect-go.mjs';
import { detecteazaPem } from './pem.mjs';
import { citesteManifest, NUME_MANIFEST } from './manifeste.mjs';

export const VERSIUNE = '0.2.0';
export const DOSARE_EXCLUSE_IMPLICIT = ['.git', '.hg', '.svn', 'node_modules', '__pycache__', '.venv', 'venv', '.tox', '.gradle', '.idea', '.mypy_cache', '.pytest_cache'];

const LIMBAJ = {
  '.js': 'javascript', '.mjs': 'javascript', '.cjs': 'javascript', '.jsx': 'javascript',
  '.ts': 'javascript', '.tsx': 'javascript', '.mts': 'javascript', '.cts': 'javascript',
  '.py': 'python', '.pyw': 'python',
  '.java': 'java',
  '.go': 'go',
};
const DETECTOR = { javascript: detecteazaJs, python: detecteazaPy, java: detecteazaJava, go: detecteazaGo };
const EXEMPLE = 5;

export function limbajul(cale) {
  return LIMBAJ[extname(cale).toLowerCase()] || null;
}

// binar = contine un octet NUL in primii 8192 octeti
export function eBinar(buf) {
  const n = Math.min(buf.length, 8192);
  for (let i = 0; i < n; i++) if (buf[i] === 0) return true;
  return false;
}

export function scaneaza(radacina, optiuni = {}) {
  const o = {
    maxFileBytes: optiuni.maxFileBytes ?? 1048576,
    maxFiles: optiuni.maxFiles ?? 50000,
    excludeDirs: new Set([...(optiuni.noDefaultExcludes ? [] : DOSARE_EXCLUSE_IMPLICIT), ...(optiuni.excludeDirs || [])]),
  };
  const root = resolve(radacina);
  const st = statSync(root);
  if (!st.isDirectory()) throw new Error(`not a directory: ${radacina}`);
  const inceput = new Date();
  const stats = {
    filesSeen: 0,
    codeFiles: {},
    textFilesPemOnly: 0,
    manifests: 0,
    manifestErrors: [],
    notRead: { binary: 0, tooLarge: 0, fileLimit: 0, unreadable: 0 },
    notReadExamples: { binary: [], tooLarge: [], fileLimit: [], unreadable: [] },
    symlinksNotFollowed: 0,
    resolutionLimitFiles: 0,
    resolutionLimitExamples: [],
    dirsExcluded: {},
    dirsUnreadable: 0,
  };
  const gasiri = [];
  const biblioteci = [];
  let cititeSauIncercate = 0;
  const noteaza = (motiv, rel, extra) => {
    stats.notRead[motiv]++;
    if (stats.notReadExamples[motiv].length < EXEMPLE) stats.notReadExamples[motiv].push(extra ? `${rel} (${extra})` : rel);
  };

  const stiva = [root];
  while (stiva.length) {
    const dir = stiva.pop();
    let intrari;
    try { intrari = readdirSync(dir, { withFileTypes: true }); } catch { stats.dirsUnreadable++; continue; }
    intrari.sort((a, b) => (a.name < b.name ? -1 : a.name > b.name ? 1 : 0));
    const subdosare = [];
    for (const e of intrari) {
      const abs = join(dir, e.name);
      if (e.isSymbolicLink()) { stats.symlinksNotFollowed++; continue; }
      if (e.isDirectory()) {
        if (o.excludeDirs.has(e.name)) { stats.dirsExcluded[e.name] = (stats.dirsExcluded[e.name] || 0) + 1; continue; }
        subdosare.push(abs);
        continue;
      }
      if (!e.isFile()) continue;
      stats.filesSeen++;
      const rel = relative(root, abs).split(sep).join('/');
      if (cititeSauIncercate >= o.maxFiles) { noteaza('fileLimit', rel); continue; }
      cititeSauIncercate++;
      let marime;
      try { marime = lstatSync(abs).size; } catch (err) { noteaza('unreadable', rel, err.code); continue; }
      if (marime > o.maxFileBytes) { noteaza('tooLarge', rel, `${marime} bytes`); continue; }
      let buf;
      try { buf = readFileSync(abs); } catch (err) { noteaza('unreadable', rel, err.code); continue; }
      if (eBinar(buf)) { noteaza('binary', rel); continue; }
      const text = buf.toString('utf8');
      const limbaj = limbajul(abs);
      const ctx = facContext(text, limbaj || 'text', rel);
      if (limbaj) {
        stats.codeFiles[limbaj] = (stats.codeFiles[limbaj] || 0) + 1;
        DETECTOR[limbaj](ctx);
        if (ctx.rezolvariPestePlafon) {
          stats.resolutionLimitFiles++;
          if (stats.resolutionLimitExamples.length < EXEMPLE) stats.resolutionLimitExamples.push(rel);
        }
      } else {
        stats.textFilesPemOnly++;
      }
      detecteazaPem(text, ctx);
      gasiri.push(...aplicaSuprimarea(ctx.gasiri));
      if (NUME_MANIFEST.test(basename(abs))) {
        stats.manifests++;
        const r = citesteManifest(text, basename(abs), rel);
        if (r.eroare) stats.manifestErrors.push(`${rel}: ${r.eroare}`);
        biblioteci.push(...r.biblioteci);
      }
    }
    for (let i = subdosare.length - 1; i >= 0; i--) stiva.push(subdosare[i]);
  }

  // garda: fiecare fisier vazut e intr-o singura categorie
  const cod = Object.values(stats.codeFiles).reduce((a, b) => a + b, 0);
  const nr = stats.notRead;
  const suma = cod + stats.textFilesPemOnly + nr.binary + nr.tooLarge + nr.fileLimit + nr.unreadable;
  if (suma !== stats.filesSeen) throw new Error(`internal accounting error: ${stats.filesSeen} files seen, ${suma} accounted for`);
  stats.codeFilesTotal = cod;

  gasiri.sort((a, b) => (a.file < b.file ? -1 : a.file > b.file ? 1 : a.line - b.line || a.column - b.column));
  biblioteci.sort((a, b) => (a.file < b.file ? -1 : a.file > b.file ? 1 : a.line - b.line));
  return {
    tool: { name: 'aere-crypto-inventory', version: VERSIUNE },
    root,
    rootName: basename(root),
    options: { maxFileBytes: o.maxFileBytes, maxFiles: o.maxFiles, excludeDirs: [...o.excludeDirs].sort(), maxResolvedVariablesPerFile: REZOLVARI_PE_FISIER },
    startedAt: inceput.toISOString(),
    finishedAt: new Date().toISOString(),
    findings: gasiri,
    libraries: biblioteci,
    stats,
  };
}