Files
AutistMask/eslint.config.js
sneak 18ad93be45
All checks were successful
check / check (push) Successful in 54s
e2e / e2e-chrome (push) Successful in 2m0s
e2e / e2e-firefox (push) Successful in 54s
harden: make the background physically unable to read the shared state singleton (closes #324)
Five defects traced to one fact: src/background/index.js read and wrote the
module-level `state` singleton in src/shared/state.js, which the MV3 service
worker never populates and which answered an unpopulated read out of
DEFAULT_STATE in silence. Every previous fix added a loadState() before the
access, and that is what produced the fifth: a load detaches the objects an
in-flight handler is holding.

So the reachability goes rather than a sixth call site.

The background now has its own storage layer, src/background/state.js:
getState() is a detached, normalized per-call read, and updateState() is a
queued read-modify-write whose read is one storage round trip ahead of its
write. Nothing in the background holds an in-memory copy of the profile.

- Every handler takes one snapshot and answers from it, including the address
  it names: activeAddressOf(s) replaced a second, later storage read that
  could disagree with the first.
- wallet_switchEthereumChain applies applyChainSwitchFields() (split out of
  chainSwitch.js, which keeps the singleton path for the popup) inside
  updateState() instead of calling onChainSwitch() on the singleton.
- The remembered site decision is a read-modify-write, not a load-mutate-save
  around a prompt the user takes seconds to answer.
- backgroundRefresh() refreshes a private copy of the wallets and applies the
  balances that came back by address, so it never publishes an object other
  in-flight work holds, and a wallet added or deleted during the round trip
  survives its write.
- The transaction attempt takes its chain id and its endpoint from the same
  snapshot. They used to come from different moments, so a chain switch
  committed in between moved the endpoint under an artifact already verified
  against the old chain.

getProvider(rpcUrl, networkId) now REQUIRES the network id and validates it
against networks.js. That closes the cold-worker wrong-chain send at its shape
rather than at one call site: the hint used to default to currentNetwork() off
the unpopulated singleton, so the endpoint was the user's chain and ethers
fixed chainId at 0x1, and the wallet's own verifySignedTx then refused every
non-mainnet dApp send. refreshBalances(), lookupTokenInfo(), scanForAddresses()
and resolveEnsName() carry the id through; balances.js no longer requires
state.js at all.

The prohibition is enforced mechanically, not by review, and it is enforced by
the bundler rather than by a guess at what the bundler does. The table of
modules an entry point's bundle may not contain lives in
script/lib/forbiddenBundleInputs.js — one copy, read by both layers that act on
it — and build.js's assertNoForbiddenInputs() fails the build when esbuild's
metafile reports src/shared/state.js as an input of a background bundle, naming
the import chain from the metafile's own graph. That is the resolution the
shipped bundle was built from, so no specifier syntax, no hop and no resolution
rule can slip past it; Dockerfile:42 runs make build, so it holds in CI.

Both halves of the table are checked for rot, because either one turns the
prohibition into a pass that checks nothing: a key no bundled entry point
matched fails, and so does a listed module this build bundled nowhere. The
second is what makes a rename of src/shared/state.js loud instead of silently
disarming the check, and it is stronger than an existsSync() because it also
fails when the module is still there but has dropped out of every bundle. What
the assertion does NOT cover is a COPY of the singleton at another path: it is
keyed by path, so a copy builds and lints clean. That is stated where the table
lives, with the reason it is accepted — a copy carries the singleton's own
guard, so a background read of it throws rather than being served DEFAULT_STATE.

make check does not run make build, so the assertion is unit tested against
synthetic metafiles in tests/buildForbiddenInputs.test.js: build.js runs its
build() only as a program now and exports the checks. Executing a check in CI
is not testing it — without that file, inverting the condition leaves every
check in this repo green with the singleton back in the worker.

A custom ESLint rule walks the CommonJS require graph from every src/background/
file and reports the same thing in the editor, before a full bundle. It reads
the same table, and it matches specifiers textually, so it is best-effort fast
feedback and not the guarantee — two earlier revisions of it shipped holes (a
template literal, a dynamic import(), a comment inside the call, a directory
resolved through package.json main). Those are covered now and pinned by
tests/backgroundStateLintRule.test.js. Two shapes it does not report are
recorded in its header as known divergences the build catches, both measured: a
computed specifier (require("../shared/" + "state"), which esbuild
constant-folds into the bundle) and a symlink to the module (esbuild reports the
real path). Each is make lint exit 0 and make build exit 2.

Reading a persisted field of the singleton before any load now throws
StateNotLoadedError instead of serving DEFAULT_STATE.

Test stubs: chrome.storage.local is a serialization boundary, and eight files
stubbed it with an aliasing get, so the object a module held and the object
"storage" held were one object — an assertion could pass on a build that never
wrote anything. Every test that drives real persistence now goes through
tests/support/storageStub.js, which structured-clones in both directions.

closes #320
2026-08-23 15:04:59 +00:00

184 lines
6.8 KiB
JavaScript

// ESLint flat config. Static analysis for make check; formatting stays with
// prettier (script/fmt-check), so nothing here touches style.
//
// The sources are CommonJS and are bundled per entrypoint by build.js, so the
// globals differ by tree and are declared per tree below. Getting that wrong in
// either direction defeats the point: too few globals buries a real no-undef in
// false positives, too many hides the next unimported identifier.
const js = require("@eslint/js");
const globals = require("globals");
const backgroundState = require("./script/lib/eslint/noStateSingletonInBackground");
// The extension APIs. MV3 Chrome exposes `chrome`; Firefox exposes both, and
// the code feature-detects between them.
const extensionGlobals = {
chrome: "readonly",
browser: "readonly",
};
const commonjs = {
ecmaVersion: 2024,
sourceType: "commonjs",
};
module.exports = [
{
ignores: ["dist/", "node_modules/"],
},
js.configs.recommended,
{
rules: {
// The two rules this config exists for. Both are already
// error-level in the recommended set; restated so a future
// recommended-set change cannot silently downgrade them.
"no-undef": "error",
// `_`-prefixed arguments are the deliberate "present for the
// interface, unused here" marker: the popup views share one
// init(ctx) signature and three of the eight do not read ctx.
// An unused catch binding is written `catch {`, which the repo
// already does, so caught errors stay checked.
"no-unused-vars": ["error", { argsIgnorePattern: "^_" }],
// Off tree-wide: it requires every rethrow to carry `{ cause }`,
// at 3 sites today (src/shared/balances.js 207 and 215,
// tests/e2e/firefox/run.js 131). That is a change to what the
// wallet's error paths actually throw, and it is a decision of its
// own rather than a side effect of turning a linter on — so it is
// off everywhere, including for new code, until that decision is
// made. Unlike no-useless-assignment below, this is not an
// accommodation of particular sites and must not be scoped to
// them.
"preserve-caught-error": "off",
},
},
// no-useless-assignment stays on everywhere except the two files that
// wipe decrypted key material: the `password = null` and
// `decryptedSecret = null` assignments after use are dead by construction
// — that is what a best-effort wipe is — and the rule's fix is to delete
// the wipe. 9 sites: approval.js 582, 593, 618, 648, 692, 703, 728, 764
// and confirmTx.js 459. Everything else in the tree is still checked, so
// an ordinary dead store elsewhere is still an error.
{
files: ["src/popup/views/approval.js", "src/popup/views/confirmTx.js"],
rules: {
"no-useless-assignment": "off",
},
},
// Popup and content scripts: page/window context.
{
files: ["src/popup/**/*.js", "src/content/**/*.js"],
languageOptions: {
...commonjs,
globals: { ...globals.browser, ...extensionGlobals },
},
},
// MV3 background: a service worker, with no window and no document.
//
// It also may not reach src/shared/state.js. That module's `state` export
// is a per-bundle singleton loaded once and mutated in place, which is the
// popup's lifetime and not the worker's: the worker is killed when idle,
// nothing loads state at module scope, and an unpopulated read used to be
// served DEFAULT_STATE silently. Five defects came from background code
// reading or writing it (https://git.eeqj.de/sneak/AutistMask/issues/324),
// and each point fix added a loadState() that created the next one. The
// rule below checks reachability through the whole require graph, not just
// the direct require, because a re-export from any shared module the
// background already pulls in would put the singleton back in the bundle
// with no background file naming it.
//
// It is not the guarantee: build.js asserts the same prohibition against
// esbuild's own metafile, from the shared table in
// script/lib/forbiddenBundleInputs.js. This is the early report.
{
files: ["src/background/**/*.js"],
plugins: { background: backgroundState },
languageOptions: {
...commonjs,
globals: { ...globals.serviceworker, ...extensionGlobals },
},
rules: {
"background/no-state-singleton-in-background": "error",
},
},
// src/shared is bundled into both, so it may only use what both provide:
// the service worker globals are the intersection, plus the extension APIs.
{
files: ["src/shared/**/*.js"],
languageOptions: {
...commonjs,
globals: { ...globals.serviceworker, ...extensionGlobals },
},
},
// src/shared/ens.js is the documented exception to the line above: its own
// header says POPUP ONLY, it caches in localStorage, and only popup views
// require it. Linting it as a service worker would be wrong about the file.
{
files: ["src/shared/ens.js"],
languageOptions: {
...commonjs,
globals: { ...globals.browser, ...extensionGlobals },
},
},
// Unit tests, and the helpers they require: jest on node.
{
files: ["tests/**/*.test.js", "tests/support/**/*.js"],
languageOptions: {
...commonjs,
globals: { ...globals.node, ...globals.jest },
},
},
// The build script is a plain node program.
{
files: ["build.js"],
languageOptions: {
...commonjs,
globals: { ...globals.node },
},
},
// The helpers the script/ entrypoints call: plain node programs too, run
// from a shell script rather than from yarn, and never bundled.
{
files: ["script/lib/**/*.js"],
languageOptions: {
...commonjs,
globals: { ...globals.node },
},
},
// The e2e harnesses are node programs that also carry, inline, the
// callbacks they ship into the browser via page.evaluate — so both
// contexts really are present in the same file and both sets of globals
// are in scope somewhere in it.
{
files: ["tests/e2e/**/*.js"],
languageOptions: {
...commonjs,
globals: {
...globals.node,
...globals.browser,
...extensionGlobals,
},
},
},
// This config file itself.
{
files: ["eslint.config.js"],
languageOptions: {
...commonjs,
globals: { ...globals.node },
},
},
];