|
| 1 | +// Shared helpers for git hooks — API-key allowlist + ANSI colors + |
| 2 | +// content scanners. Imported by .git-hooks/{commit-msg,pre-commit, |
| 3 | +// pre-push}.mts. No third-party deps; uses only Node built-ins. |
| 4 | +// |
| 5 | +// Requires Node 25+ for stable .mts type-stripping (no flag needed). |
| 6 | +// Earlier Node versions either lacked --experimental-strip-types or |
| 7 | +// shipped it under a flag, both unacceptable for hook ergonomics. |
| 8 | + |
| 9 | +import { spawnSync } from 'node:child_process' |
| 10 | +import { existsSync, readFileSync, statSync } from 'node:fs' |
| 11 | + |
| 12 | +// Hard-fail if Node is below 25. This runs at module load — every |
| 13 | +// hook invocation imports _helpers.mts before doing anything, so the |
| 14 | +// version check is the first thing that happens. |
| 15 | +const NODE_MIN_MAJOR = 25 |
| 16 | +const nodeMajor = Number.parseInt( |
| 17 | + process.versions.node.split('.')[0] || '0', |
| 18 | + 10, |
| 19 | +) |
| 20 | +if (nodeMajor < NODE_MIN_MAJOR) { |
| 21 | + process.stderr.write( |
| 22 | + `\x1b[0;31m✗ Hook requires Node >= ${NODE_MIN_MAJOR}.0.0 (have v${process.versions.node})\x1b[0m\n`, |
| 23 | + ) |
| 24 | + process.stderr.write( |
| 25 | + 'Install Node 25+ — these hooks rely on stable .mts type stripping.\n', |
| 26 | + ) |
| 27 | + process.exit(1) |
| 28 | +} |
| 29 | + |
| 30 | +// ── Allowlist constants ──────────────────────────────────────────── |
| 31 | +// These exempt known-safe matches from the API-key scanner. Each |
| 32 | +// allowlist entry is a substring; if the matched line contains it, |
| 33 | +// the line is dropped from the findings. |
| 34 | + |
| 35 | +// Real public API key shipped in socket-lib test fixtures. Safe to |
| 36 | +// appear anywhere in the fleet. |
| 37 | +export const ALLOWED_PUBLIC_KEY = |
| 38 | + 'sktsec_t_--RAN5U4ivauy4w37-6aoKyYPDt5ZbaT5JBVMqiwKo_api' |
| 39 | + |
| 40 | +// Substring marker used in test fixtures (see |
| 41 | +// socket-lib/test/unit/utils/fake-tokens.ts). Lines containing this |
| 42 | +// are treated as test fixtures. |
| 43 | +export const FAKE_TOKEN_MARKER = 'socket-test-fake-token' |
| 44 | + |
| 45 | +// Legacy lib-scoped marker — accepted during the rename from |
| 46 | +// `socket-lib-test-fake-token` to `socket-test-fake-token`. Drop when |
| 47 | +// lib's rename PR lands. |
| 48 | +export const FAKE_TOKEN_LEGACY = 'socket-lib-test-fake-token' |
| 49 | + |
| 50 | +// Name of the env var used in shell examples; not a token value. |
| 51 | +export const SOCKET_SECURITY_ENV = 'SOCKET_SECURITY_API_KEY=' |
| 52 | + |
| 53 | +// ── ANSI colors ──────────────────────────────────────────────────── |
| 54 | + |
| 55 | +export const RED = '\x1b[0;31m' |
| 56 | +export const GREEN = '\x1b[0;32m' |
| 57 | +export const YELLOW = '\x1b[1;33m' |
| 58 | +export const NC = '\x1b[0m' |
| 59 | + |
| 60 | +// ── Output helpers ───────────────────────────────────────────────── |
| 61 | + |
| 62 | +export const out = (msg: string): void => { |
| 63 | + process.stdout.write(msg + '\n') |
| 64 | +} |
| 65 | + |
| 66 | +export const err = (msg: string): void => { |
| 67 | + process.stderr.write(msg + '\n') |
| 68 | +} |
| 69 | + |
| 70 | +export const red = (msg: string): string => `${RED}${msg}${NC}` |
| 71 | +export const green = (msg: string): string => `${GREEN}${msg}${NC}` |
| 72 | +export const yellow = (msg: string): string => `${YELLOW}${msg}${NC}` |
| 73 | + |
| 74 | +// ── API-key allowlist filter ─────────────────────────────────────── |
| 75 | + |
| 76 | +// Drops any line that matches an allowlist entry. |
| 77 | +export const filterAllowedApiKeys = (lines: readonly string[]): string[] => { |
| 78 | + return lines.filter( |
| 79 | + line => |
| 80 | + !line.includes(ALLOWED_PUBLIC_KEY) && |
| 81 | + !line.includes(FAKE_TOKEN_MARKER) && |
| 82 | + !line.includes(FAKE_TOKEN_LEGACY) && |
| 83 | + !line.includes(SOCKET_SECURITY_ENV) && |
| 84 | + !line.includes('.example'), |
| 85 | + ) |
| 86 | +} |
| 87 | + |
| 88 | +// ── Personal-path scanner ────────────────────────────────────────── |
| 89 | + |
| 90 | +// Real personal paths to flag: /Users/foo/, /home/foo/, C:\Users\foo\. |
| 91 | +const PERSONAL_PATH_RE = |
| 92 | + /(\/Users\/[^/\s]+\/|\/home\/[^/\s]+\/|C:\\Users\\[^\\]+\\)/ |
| 93 | + |
| 94 | +// Placeholders we ALLOW (documentation, not real leaks): any path |
| 95 | +// component wrapped in <...> or starting with $VAR / ${VAR}. |
| 96 | +const PERSONAL_PATH_PLACEHOLDER_RE = |
| 97 | + /(\/Users\/<[^>]*>\/|\/home\/<[^>]*>\/|C:\\Users\\<[^>]*>\\|\/Users\/\$\{?[A-Z_]+\}?\/|\/home\/\$\{?[A-Z_]+\}?\/)/ |
| 98 | + |
| 99 | +export type LineHit = { lineNumber: number; line: string } |
| 100 | + |
| 101 | +// Returns lines that contain a real personal path (excludes lines |
| 102 | +// that are pure placeholders). Caller decides what to do with hits. |
| 103 | +export const scanPersonalPaths = (text: string): LineHit[] => { |
| 104 | + const hits: LineHit[] = [] |
| 105 | + const lines = text.split('\n') |
| 106 | + for (let i = 0; i < lines.length; i++) { |
| 107 | + const line = lines[i]! |
| 108 | + if (!PERSONAL_PATH_RE.test(line)) { |
| 109 | + continue |
| 110 | + } |
| 111 | + if (PERSONAL_PATH_PLACEHOLDER_RE.test(line)) { |
| 112 | + // Has placeholder — but might also have a real path on the |
| 113 | + // same line. Strip placeholder forms and re-test. |
| 114 | + const stripped = line.replace( |
| 115 | + new RegExp(PERSONAL_PATH_PLACEHOLDER_RE, 'g'), |
| 116 | + '', |
| 117 | + ) |
| 118 | + if (!PERSONAL_PATH_RE.test(stripped)) { |
| 119 | + continue |
| 120 | + } |
| 121 | + } |
| 122 | + hits.push({ lineNumber: i + 1, line }) |
| 123 | + } |
| 124 | + return hits |
| 125 | +} |
| 126 | + |
| 127 | +// ── Secret scanners ──────────────────────────────────────────────── |
| 128 | + |
| 129 | +const SOCKET_API_KEY_RE = /sktsec_[a-zA-Z0-9_-]+/ |
| 130 | +const AWS_KEY_RE = /(aws_access_key|aws_secret|\bAKIA[0-9A-Z]{16}\b)/i |
| 131 | +const GITHUB_TOKEN_RE = /gh[ps]_[a-zA-Z0-9]{36}/ |
| 132 | +const PRIVATE_KEY_RE = /-----BEGIN (RSA |EC |DSA )?PRIVATE KEY-----/ |
| 133 | + |
| 134 | +export const scanSocketApiKeys = (text: string): LineHit[] => { |
| 135 | + const hits: LineHit[] = [] |
| 136 | + const lines = text.split('\n') |
| 137 | + for (let i = 0; i < lines.length; i++) { |
| 138 | + const line = lines[i]! |
| 139 | + if (SOCKET_API_KEY_RE.test(line)) { |
| 140 | + hits.push({ lineNumber: i + 1, line }) |
| 141 | + } |
| 142 | + } |
| 143 | + // Filter the LineHit objects directly so duplicate-content lines |
| 144 | + // at different line numbers keep their correct numbers. |
| 145 | + const allowedSet = new Set(filterAllowedApiKeys(hits.map(h => h.line))) |
| 146 | + return hits.filter(h => allowedSet.has(h.line)) |
| 147 | +} |
| 148 | + |
| 149 | +export const scanAwsKeys = (text: string): LineHit[] => { |
| 150 | + const hits: LineHit[] = [] |
| 151 | + const lines = text.split('\n') |
| 152 | + for (let i = 0; i < lines.length; i++) { |
| 153 | + const line = lines[i]! |
| 154 | + if (AWS_KEY_RE.test(line)) { |
| 155 | + hits.push({ lineNumber: i + 1, line }) |
| 156 | + } |
| 157 | + } |
| 158 | + return hits |
| 159 | +} |
| 160 | + |
| 161 | +export const scanGitHubTokens = (text: string): LineHit[] => { |
| 162 | + const hits: LineHit[] = [] |
| 163 | + const lines = text.split('\n') |
| 164 | + for (let i = 0; i < lines.length; i++) { |
| 165 | + const line = lines[i]! |
| 166 | + if (GITHUB_TOKEN_RE.test(line)) { |
| 167 | + hits.push({ lineNumber: i + 1, line }) |
| 168 | + } |
| 169 | + } |
| 170 | + return hits |
| 171 | +} |
| 172 | + |
| 173 | +export const scanPrivateKeys = (text: string): LineHit[] => { |
| 174 | + const hits: LineHit[] = [] |
| 175 | + const lines = text.split('\n') |
| 176 | + for (let i = 0; i < lines.length; i++) { |
| 177 | + const line = lines[i]! |
| 178 | + if (PRIVATE_KEY_RE.test(line)) { |
| 179 | + hits.push({ lineNumber: i + 1, line }) |
| 180 | + } |
| 181 | + } |
| 182 | + return hits |
| 183 | +} |
| 184 | + |
| 185 | +// ── npx/dlx scanner ──────────────────────────────────────────────── |
| 186 | + |
| 187 | +const NPX_DLX_RE = /\b(npx|pnpm dlx|yarn dlx)\b/ |
| 188 | + |
| 189 | +export const scanNpxDlx = (text: string): LineHit[] => { |
| 190 | + const hits: LineHit[] = [] |
| 191 | + const lines = text.split('\n') |
| 192 | + for (let i = 0; i < lines.length; i++) { |
| 193 | + const line = lines[i]! |
| 194 | + if (NPX_DLX_RE.test(line) && !line.includes('# zizmor:')) { |
| 195 | + hits.push({ lineNumber: i + 1, line }) |
| 196 | + } |
| 197 | + } |
| 198 | + return hits |
| 199 | +} |
| 200 | + |
| 201 | +// ── Linear issue reference scanner ───────────────────────────────── |
| 202 | +// CLAUDE.md "ABSOLUTE RULES": NEVER reference Linear issues in commits. |
| 203 | +// Team keys enumerated from the Socket workspace. PATCH listed before |
| 204 | +// PAT so the alternation matches the longer prefix first. |
| 205 | + |
| 206 | +const LINEAR_TEAM_KEYS = |
| 207 | + 'ASK|AUTO|BOT|CE|CORE|DAT|DES|DEV|ENG|INFRA|LAB|MAR|MET|OPS|PAR|PATCH|PAT|PLAT|REA|SALES|SBOM|SEC|SMO|SUP|TES|TI|WEB' |
| 208 | + |
| 209 | +const LINEAR_ISSUE_RE = new RegExp( |
| 210 | + `(?:^|[^A-Za-z0-9_])((?:${LINEAR_TEAM_KEYS})-[0-9]+)(?:$|[^A-Za-z0-9_])`, |
| 211 | + 'gm', |
| 212 | +) |
| 213 | + |
| 214 | +const LINEAR_URL_RE = /linear\.app\/[A-Za-z0-9/_-]+/g |
| 215 | + |
| 216 | +export const scanLinearReferences = (commitMsg: string): string[] => { |
| 217 | + const hits: string[] = [] |
| 218 | + const lines = commitMsg.split('\n').filter(l => !l.startsWith('#')) |
| 219 | + const body = lines.join('\n') |
| 220 | + for (const m of body.matchAll(LINEAR_ISSUE_RE)) { |
| 221 | + hits.push(m[1]!) |
| 222 | + } |
| 223 | + for (const m of body.matchAll(LINEAR_URL_RE)) { |
| 224 | + hits.push(m[0]!) |
| 225 | + } |
| 226 | + return hits.slice(0, 5) |
| 227 | +} |
| 228 | + |
| 229 | +// ── AI attribution scanner ───────────────────────────────────────── |
| 230 | + |
| 231 | +const AI_ATTRIBUTION_RE = |
| 232 | + /(Generated with.*(Claude|AI)|Co-Authored-By: Claude|Co-Authored-By: AI|🤖 Generated|AI generated|@anthropic\.com|Assistant:|Generated by Claude|Machine generated|Claude Code)/i |
| 233 | + |
| 234 | +export const containsAiAttribution = (text: string): boolean => |
| 235 | + AI_ATTRIBUTION_RE.test(text) |
| 236 | + |
| 237 | +export const stripAiAttribution = ( |
| 238 | + text: string, |
| 239 | +): { cleaned: string; removed: number } => { |
| 240 | + const lines = text.split('\n') |
| 241 | + const kept: string[] = [] |
| 242 | + let removed = 0 |
| 243 | + for (const line of lines) { |
| 244 | + if (AI_ATTRIBUTION_RE.test(line)) { |
| 245 | + removed++ |
| 246 | + } else { |
| 247 | + kept.push(line) |
| 248 | + } |
| 249 | + } |
| 250 | + return { cleaned: kept.join('\n'), removed } |
| 251 | +} |
| 252 | + |
| 253 | +// ── File classification ──────────────────────────────────────────── |
| 254 | + |
| 255 | +// Files we never scan: hooks themselves, husky shims, test fixtures. |
| 256 | +const SKIP_FILE_RE = |
| 257 | + /\.(test|spec)\.(m?[jt]s|tsx?|cts|mts)$|\.example$|\/test\/|\/tests\/|fixtures\/|\.git-hooks\/|\.husky\/|node_modules\/|pnpm-lock\.yaml/ |
| 258 | + |
| 259 | +export const shouldSkipFile = (filePath: string): boolean => |
| 260 | + SKIP_FILE_RE.test(filePath) |
| 261 | + |
| 262 | +// Returns file content as a string. For binaries, runs `strings` to |
| 263 | +// extract printable byte sequences (catches paths embedded in WASM |
| 264 | +// or other compiled artifacts). |
| 265 | +export const readFileForScan = (filePath: string): string => { |
| 266 | + if (!existsSync(filePath)) { |
| 267 | + return '' |
| 268 | + } |
| 269 | + try { |
| 270 | + if (statSync(filePath).isDirectory()) { |
| 271 | + return '' |
| 272 | + } |
| 273 | + } catch { |
| 274 | + return '' |
| 275 | + } |
| 276 | + // Detect binary via grep -I (matches text-only); if grep says |
| 277 | + // binary, fall back to `strings`. |
| 278 | + const grepResult = spawnSync('grep', ['-qI', '', filePath]) |
| 279 | + if (grepResult.status === 0) { |
| 280 | + // Text file. |
| 281 | + try { |
| 282 | + return readFileSync(filePath, 'utf8') |
| 283 | + } catch { |
| 284 | + return '' |
| 285 | + } |
| 286 | + } |
| 287 | + // Binary — extract strings. |
| 288 | + const stringsResult = spawnSync('strings', [filePath], { |
| 289 | + encoding: 'utf8', |
| 290 | + }) |
| 291 | + return stringsResult.stdout || '' |
| 292 | +} |
| 293 | + |
| 294 | +// ── Git wrappers ─────────────────────────────────────────────────── |
| 295 | + |
| 296 | +export const git = (...args: string[]): string => { |
| 297 | + const result = spawnSync('git', args, { encoding: 'utf8' }) |
| 298 | + return result.stdout.trim() |
| 299 | +} |
| 300 | + |
| 301 | +export const gitLines = (...args: string[]): string[] => { |
| 302 | + const out = git(...args) |
| 303 | + return out ? out.split('\n') : [] |
| 304 | +} |
0 commit comments