Compare commits

..

1 Commits

Author SHA1 Message Date
9665ac448e fix: answer the page when a background handler throws (closes #280)
All checks were successful
check / check (push) Successful in 46s
handleRpc(...).then(sendResponse) had no .catch(), and sendResponse is the only
thing that settles the dApp's window.ethereum.request() promise. Any throw
inside handleRpc therefore sent nothing back: the content script posted nothing,
and the page's promise stayed pending forever with no error and no timeout,
indistinguishable from a slow wallet. handleRpc does real work -- state loads,
provider calls, transaction population, approval plumbing -- so "it does not
throw today" was not a property anyone was maintaining.

A rejected handleRpc now answers { code: -32603, message }. -32603 is the
JSON-RPC internal error EIP-1474 defines and EIP-1193 defers to for RPC-layer
failures; no EIP-1193 4xxx code describes "the wallet broke" and none was
invented for it. The cause is not put in the message: the page gets a stable
sentence, the background console gets the method and the throw, so the failure
is visible rather than swallowed.

The two async IIFEs behind AUTISTMASK_TX_RESPONSE and AUTISTMASK_SIGN_RESPONSE
are the same shape one level down. Every statement is inside a try, but a throw
from one of the catch blocks escapes as an unhandled rejection and neither the
popup nor the page is answered. Each gets a last-resort .catch() that settles
the approval through settleApproval() -- the existing chokepoint, with no new
delete or resolve -- and answers the popup. The transaction one reports the
broadcast stage, because it cannot tell whether the transaction reached the
network and that is the wording that does not invite a second send. Every other
message handler on the path is synchronous and cannot leave a promise pending.

Each of the three is driven by a real failure rather than a hook in the handler:
a rejecting extension-storage read, which getState() awaits unguarded, and a
failure classifier that throws while classifying a genuine verification failure.
All three were demonstrated failing against the unfixed code, the RPC one with
sendResponse at zero calls, which is precisely the page-side hang.
2026-08-14 04:06:17 +00:00
14 changed files with 417 additions and 993 deletions

View File

@@ -1,49 +0,0 @@
name: e2e
on: [push]
# The browser end-to-end suites, one job per browser, deliberately kept out
# of the check workflow: REPO_POLICIES.md caps make test at 20 seconds and
# script/cibuild is a plain `docker build .` whose Dockerfile runs
# make check, so folding a browser suite into either would blow that cap
# and slow the local fast path. Before this workflow every browser-level
# guarantee in this repo held only when a human remembered to run it.
#
# One job per browser rather than two steps in one job, so a Chrome failure
# does not hide the Firefox result.
#
# Each job is one script and nothing else. Both scripts need docker and
# nothing else — they deliver the repo to the daemon as a build context and
# build the extension inside the pinned image — which is what makes them
# runnable here at all: the runner executes the job in a container against
# the host's docker socket, so a `-v "$PWD:/work"` source path is resolved
# by the host daemon and mounts an empty directory, and the runner image's
# node is too old to install this repo's dependencies.
#
# These jobs REPORT, they do not gate. Whether a check blocks a merge is
# Gitea branch protection, which this repo does not configure, so a failure
# here is a red mark a reviewer has to account for rather than a hard
# block. Making e2e-chrome a required check is blocked on the measured
# flake in the dApp signing wait -- two of six runs of unmutated code on a
# loaded machine -- tracked as
# https://git.eeqj.de/sneak/AutistMask/issues/287. A gate that fails at
# random teaches people to merge past red.
#
# Nothing here may pass vacuously. There is no continue-on-error and no
# `|| true`. Both scripts exit non-zero when docker is missing, when the
# image build fails, and when the browser fails to start; the Chrome
# harness aborts the suite outright if its network interception is not in
# effect.
jobs:
e2e-chrome:
runs-on: ubuntu-latest
steps:
# actions/checkout v4.2.2, 2026-02-22
- uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- run: script/test-e2e
e2e-firefox:
runs-on: ubuntu-latest
steps:
# actions/checkout v4.2.2, 2026-02-22
- uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683
- run: script/test-e2e-firefox

View File

@@ -83,11 +83,10 @@ provide:
git pre-commit hook git pre-commit hook
- `script/projectname` — print the project name (used for the Docker image tag) - `script/projectname` — print the project name (used for the Docker image tag)
- `script/test` — run the test suite (jest) - `script/test` — run the test suite (jest)
- `script/test-e2e` — run the Chrome browser end-to-end suite (docker is the - `script/test-e2e` — run the Chrome browser end-to-end suite (docker required;
only prerequisite: it builds a pinned image that carries the repo and a fresh see [End-to-End Tests](#end-to-end-tests))
extension build, see [End-to-End Tests](#end-to-end-tests)) - `script/test-e2e-firefox` — run the Firefox browser end-to-end suite (docker
- `script/test-e2e-firefox` — run the Firefox browser end-to-end suite (same, required; builds its own pinned image, see
against an image with a pinned Firefox and geckodriver, see
[End-to-End Tests](#end-to-end-tests)) [End-to-End Tests](#end-to-end-tests))
- `script/lint` — run the linter - `script/lint` — run the linter
- `script/fmt` — format all files (writes) - `script/fmt` — format all files (writes)
@@ -137,12 +136,11 @@ are outside `make check`.
`make test-e2e` builds `dist/chrome/` and drives the **real popup in a real `make test-e2e` builds `dist/chrome/` and drives the **real popup in a real
Chrome**, loaded as an unpacked MV3 extension inside a pinned Chrome**, loaded as an unpacked MV3 extension inside a pinned
`mcr.microsoft.com/playwright` container (pinned by digest in `mcr.microsoft.com/playwright` container (pinned by digest in `script/test-e2e`;
`tests/e2e/Dockerfile`, which is also where the extension is built; docker is docker is required and the suite fails loudly rather than skipping if it is
required and the suite fails loudly rather than skipping if it is unavailable). unavailable). The suite lives in `tests/e2e/` and is driven by
The suite lives in `tests/e2e/` and is driven by `playwright-core`, whose `playwright-core`, whose version must stay matched to the container's Playwright
version must stay matched to the container's Playwright version — the browsers version — the browsers ship inside the image.
ship inside the image.
It covers popup load, WebAssembly compilation under the shipped CSP (see It covers popup load, WebAssembly compilation under the shipped CSP (see
[Content Security Policy](#content-security-policy)), wallet creation through [Content Security Policy](#content-security-policy)), wallet creation through
@@ -322,46 +320,9 @@ Two limits are worth knowing, both real differences from the Chrome suite:
Neither `make test-e2e` nor `make test-e2e-firefox` is part of `make check` or Neither `make test-e2e` nor `make test-e2e-firefox` is part of `make check` or
`make test`. `REPO_POLICIES.md` caps `make test` at 20 seconds and a browser `make test`. `REPO_POLICIES.md` caps `make test` at 20 seconds and a browser
suite does not fit; nothing in `tests/e2e/` is named `*.test.js`, so jest cannot suite does not fit; nothing in `tests/e2e/` is named `*.test.js`, so jest cannot
pick it up either. Run them locally before changing anything under pick it up either. Neither is wired into the Gitea workflow yet —
`src/popup/views/`. docker-in-docker in CI is a separate question. Run them locally before changing
anything under `src/popup/views/`.
### In CI
`.gitea/workflows/e2e.yml` runs both suites on every push, as two jobs —
`e2e-chrome` and `e2e-firefox` — separate from the `check` workflow, so the
20-second `make test` cap and the local fast path are untouched. Each job is a
checkout and the matching `script/` entrypoint, nothing else.
Docker is the only thing either job needs from the runner, and that is not an
accident. The runner executes a job inside a container against the **host's**
docker daemon, so a `docker run -v "$PWD:/work"` source path is resolved by the
host and mounts an empty directory, and the runner image's node is too old to
install this repo's dependencies. Both suites therefore ship the repo to the
daemon as a build context and build the extension inside the image, which works
identically on a laptop.
The jobs **report, they do not gate.** A failure is a red mark against the
commit that a reviewer has to account for, not a hard block: whether a check
blocks a merge is Gitea branch protection, which this repo does not configure.
That is not only a statement about configuration. The Chrome suite is
**measurably flaky under load** — two of six runs of unmutated code on a busy
machine lost the approval popup out from under the dApp signing wait, always in
the `#183` section, tracked as
[#287](https://git.eeqj.de/sneak/AutistMask/issues/287). So a red `e2e-chrome`
has to be read before it is believed, and that flake is the blocker to ever
making this a required check. Do not answer it with a retry wrapper: a suite
that reruns until it is green stops being evidence.
Nothing in either job can pass vacuously. There is no `continue-on-error` and no
`|| true`; both scripts exit non-zero when docker is missing, when the image
build fails, and when the browser fails to start; the Chrome harness aborts the
suite outright if its network interception is not in effect.
Measured on this repo's runner: `e2e-chrome` about 1m55s cold, almost all of it
the one-time pull of the pinned ~800MB Playwright layer, and well under a minute
once that layer is cached. `e2e-firefox` about 1m05s cold, and it caches its
Firefox and geckodriver downloads the same way.
## Rationale ## Rationale
@@ -677,21 +638,6 @@ ExportPrivKey and ShowRecoveryPhrase — are deliberately absent from that list,
so the popup can never reopen onto one of them with no password prompt in front so the popup can never reopen onto one of them with no password prompt in front
of it. of it.
A reopened popup renders the wallet list and the one screen it restores onto,
and nothing else, so every screen on the stack behind that one is still the
blank template from `index.html`. "Back" therefore renders its target rather
than only unhiding it, through the same dispatch and data guards as the restore
(`src/popup/viewRouter.js`), and falls back to Home when the state the target
would render is gone.
It renders only a screen this page load has not rendered yet. Forward navigation
renders as it goes, and `viewRouter.js` records every screen that reaches
`showView()`, so "Back" onto a screen already on the page unhides it and nothing
more — rendering it a second time would re-fetch and overwrite what it holds,
such as an edit typed into Settings and not yet saved. Home is the one screen
"Back" always re-renders, so the wallet list reflects anything that changed
while the user was away from it.
Every screen that holds secret material in the page registers a cleanup with Every screen that holds secret material in the page registers a cleanup with
`onViewLeave()` (`src/popup/views/helpers.js`), which `showView()` runs on every `onViewLeave()` (`src/popup/views/helpers.js`), which `showView()` runs on every
exit from that screen rather than only on its "Back" button, so nothing secret exit from that screen rather than only on its "Back" button, so nothing secret

61
TODO.md
View File

@@ -33,8 +33,7 @@ The backlog lives on the
authoritative; this file does not duplicate it. Full policy file set present. authoritative; this file does not duplicate it. Full policy file set present.
Real-browser end-to-end suites (`make test-e2e` for Chrome, Real-browser end-to-end suites (`make test-e2e` for Chrome,
`make test-e2e-firefox` for Firefox) now sit alongside `make check`, which `make test-e2e-firefox` for Firefox) now sit alongside `make check`, which
cannot see a runtime `ReferenceError` in a popup view, and cannot see a runtime `ReferenceError` in a popup view.
`.gitea/workflows/e2e.yml` runs both of them on every push.
# Next Step # Next Step
@@ -46,24 +45,25 @@ undefined identifiers, which is how
# Completed Steps # Completed Steps
- 2026-08-14: CI runs the browser end-to-end suites. `.gitea/workflows/e2e.yml` - 2026-08-14: A background message handler that throws now rejects the page
runs `script/test-e2e` and `script/test-e2e-firefox` as two jobs on every instead of hanging it. `handleRpc(...).then(sendResponse)` had no `.catch()`,
push, separate from `check`, so `make check` and its 20-second `make test` cap and `sendResponse` is the only thing that settles the dApp's
are untouched. Every browser-level guarantee in this repo — the WASM-under-CSP `window.ethereum.request()` promise — so any throw inside `handleRpc` left
check, the recovery-phrase and private-key DOM wipes, the ConfirmTx spend that promise pending forever, with no error and no timeout, indistinguishable
gate, the dApp approval round trips — was enforced only when a human from a slow wallet. It now answers `{ code: -32603, message }` (the JSON-RPC
remembered to run it by hand. The suites could not run on the runner as they internal error EIP-1474 defines and EIP-1193 defers to; no EIP-1193 4xxx code
stood: the runner executes a job in a container against the host's docker describes "the wallet broke" and none was invented) and logs the method and
daemon, so `docker run -v "$PWD:/work"` mounts an empty directory (measured), the throw to the background console rather than swallowing them. The two async
and the runner image's node cannot install this repo's dependencies. Both IIFEs behind `AUTISTMASK_TX_RESPONSE` and `AUTISTMASK_SIGN_RESPONSE` were the
suites now ship the repo to the daemon as a build context and build the same shape one level down — every statement inside a `try`, but a throw out of
extension inside the pinned image, so docker is the only prerequisite on a a `catch` block escaping unhandled — and each got a last-resort `.catch()`
runner or a laptop, and both run the image by ID rather than by tag so settling the approval through `settleApproval()` and answering the popup; the
concurrent clones cannot swap it. The jobs report rather than gate — this repo transaction one reports the broadcast stage, because it cannot tell whether
configures no branch protection, and the Chrome suite is measurably flaky the transaction reached the network. Every other handler on the path is
under load, filed as [#287](https://git.eeqj.de/sneak/AutistMask/issues/287) synchronous. All three are driven by real failures — a rejecting storage read,
rather than papered over and a failure classifier that throws while classifying a genuine verification
([#259](https://git.eeqj.de/sneak/AutistMask/issues/259)). failure — and were demonstrated failing first, the RPC one with `sendResponse`
at zero calls ([#280](https://git.eeqj.de/sneak/AutistMask/issues/280)).
- 2026-08-12: EIP-1193 error codes now reach the page. `src/content/inpage.js` - 2026-08-12: EIP-1193 error codes now reach the page. `src/content/inpage.js`
rebuilt every failure as `new Error(error.message)`, so the code the rebuilt every failure as `new Error(error.message)`, so the code the
background produced and the content script relayed intact was dropped in the background produced and the content script relayed intact was dropped in the
@@ -78,23 +78,6 @@ undefined identifiers, which is how
`tests/inpageErrors.test.js`, and the e2e probe that printed the missing code `tests/inpageErrors.test.js`, and the e2e probe that printed the missing code
now requires it on the page's Error as well as on the wire, for all four now requires it on the page's Error as well as on the wire, for all four
rejected flows ([#274](https://git.eeqj.de/sneak/AutistMask/issues/274)). rejected flows ([#274](https://git.eeqj.de/sneak/AutistMask/issues/274)).
- 2026-08-12: "Back" now renders the screen it lands on instead of only unhiding
it. A reopened popup renders the wallet list and the one screen it restores
onto, so every screen further down the stack was still the blank template from
`index.html`, and Back walked straight onto it — an empty address, no
balances, no QR code. The Back path now goes through the same per-view
dispatch and data guards as the restore (`src/popup/viewRouter.js`, shared
with `restoreView()`), falling back to Home when the state the target would
render is gone. It renders only a view this page load has not rendered yet:
`viewRouter.js` records every view that reaches `showView()`, which is where
forward navigation and the restore both end, so Back onto a view already on
the page unhides it and nothing more. That is what keeps a second render from
re-fetching and overwriting what the view holds — an unsaved edit in Settings,
a transaction list already loaded. Home is the exception and is always
re-rendered, as it was before. Covered by unit tests on the real `goBack()`
and by three end-to-end cases against the real popup, each demonstrated
failing on the unfixed build
([#268](https://git.eeqj.de/sneak/AutistMask/issues/268)).
- 2026-08-12: `KNOWN_SYMBOLS` now maps a symbol to the set of contract addresses - 2026-08-12: `KNOWN_SYMBOLS` now maps a symbol to the set of contract addresses
that bear it, not to one of them. A ticker is not unique: seven of the 512 that bear it, not to one of them. A ticker is not unique: seven of the 512
bundled tokens — `FRAX`, `REUSD`, `TON`, `EURE`, `MSUSD`, `MUSD` and `JPYC` bundled tokens — `FRAX`, `REUSD`, `TON`, `EURE`, `MSUSD`, `MUSD` and `JPYC`
@@ -387,5 +370,9 @@ tracker.
- Pre-1.0 security review of the extension (key handling, DEBUG mode policy, RPC - Pre-1.0 security review of the extension (key handling, DEBUG mode policy, RPC
input validation) before any 1.0rc tag. Individual filed issues are parts of input validation) before any 1.0rc tag. Individual filed issues are parts of
it, but the review is broader than any of them. it, but the review is broader than any of them.
- Decide whether docker-in-docker makes `make test-e2e` and
`make test-e2e-firefox` runnable in the Gitea workflow. Extending the Chrome
suite itself is tracked as
[#183](https://git.eeqj.de/sneak/AutistMask/issues/183).
- Cut 1.0.0 once the milestone is empty, then continue tagging as milestones - Cut 1.0.0 once the milestone is empty, then continue tagging as milestones
land. land.

View File

@@ -7,29 +7,17 @@
# caps make test at 20 seconds and a browser suite does not fit. Run it # caps make test at 20 seconds and a browser suite does not fit. Run it
# yourself before touching popup views; it is the only check that can see # yourself before touching popup views; it is the only check that can see
# a used-but-not-imported identifier blow up at runtime. # a used-but-not-imported identifier blow up at runtime.
# .gitea/workflows/e2e.yml also runs it on every push, in a job separate
# from check so that cap and the local fast path both stay intact.
#
# Docker is the only prerequisite. The repo reaches the container as a
# build context and the extension is built inside it (see
# tests/e2e/Dockerfile), so nothing here depends on the node, yarn or make
# on the machine that starts the run. That is not a convenience: a bind
# mount cannot work under Gitea Actions, and the runner image's node is too
# old to install this repo's dependencies.
set -eu set -eu
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)" ROOT="$(cd "$(dirname "$0")/.." && pwd -P)"
ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
IMAGE="$("$SCRIPT_DIR/projectname")-e2e-chrome" # mcr.microsoft.com/playwright:v1.56.0-noble, 2026-08-09
#
IIDFILE="" # The playwright-core devDependency is pinned to the matching Playwright
# version (1.56.0) and the two must be bumped together: the browsers ship
cleanup() { # inside this image, and playwright-core looks for the exact browser
if [ -n "$IIDFILE" ]; then # revision its own version expects. A mismatch fails at launch.
rm -f "$IIDFILE" IMAGE="mcr.microsoft.com/playwright@sha256:35246d87a7c88ea9b771c65d33171b2611b02a8253b4b12ce6f94376c55f99f2"
fi
}
main() { main() {
cd "$ROOT" cd "$ROOT"
@@ -39,23 +27,14 @@ main() {
exit 1 exit 1
fi fi
IIDFILE="$(mktemp)" echo "Building extension for e2e..."
trap cleanup EXIT yarn run build 2>&1
trap 'cleanup; exit 130' INT TERM
echo "Building the Chrome e2e image (extension included)..."
docker build --iidfile "$IIDFILE" -t "$IMAGE" -f tests/e2e/Dockerfile .
echo "Running e2e suite in the pinned Playwright container..." echo "Running e2e suite in the pinned Playwright container..."
# The image is run by ID, not by tag: where two clones of this repo run
# the suite at once, the other build can move the tag between this
# build and this run, and the suite would then silently test the other
# checkout.
#
# --ipc=host: Chromium's shared-memory needs more than the default # --ipc=host: Chromium's shared-memory needs more than the default
# 64MB /dev/shm or renderers crash. # 64MB /dev/shm or renderers crash.
# HOME=/tmp: the image's root home is not a reliable place for the # --user: keep files the suite touches owned by the caller, not root.
# browser profile. # HOME=/tmp: the mapped uid has no home directory in the image.
# PW_EXPERIMENTAL_SERVICE_WORKER_NETWORK_EVENTS=1: without it, # PW_EXPERIMENTAL_SERVICE_WORKER_NETWORK_EVENTS=1: without it,
# ctx.route() intercepts page requests only, and every fetch made by # ctx.route() intercepts page requests only, and every fetch made by
# the MV3 background service worker — including the phishing # the MV3 background service worker — including the phishing
@@ -72,10 +51,13 @@ main() {
# on a deliberate bump. # on a deliberate bump.
docker run --rm \ docker run --rm \
--ipc=host \ --ipc=host \
--user "$(id -u):$(id -g)" \
-e HOME=/tmp \ -e HOME=/tmp \
-e PW_EXPERIMENTAL_SERVICE_WORKER_NETWORK_EVENTS=1 \ -e PW_EXPERIMENTAL_SERVICE_WORKER_NETWORK_EVENTS=1 \
-e "E2E_TRACE_NETWORK=${E2E_TRACE_NETWORK:-0}" \ -e "E2E_TRACE_NETWORK=${E2E_TRACE_NETWORK:-0}" \
"$(cat "$IIDFILE")" \ -v "$ROOT:/work" \
-w /work \
"$IMAGE" \
node tests/e2e/run.js node tests/e2e/run.js
} }

View File

@@ -5,17 +5,12 @@
# #
# Deliberately NOT called by script/check or script/test, for the same # Deliberately NOT called by script/check or script/test, for the same
# reason as the Chrome suite: REPO_POLICIES.md caps make test at 20 seconds # reason as the Chrome suite: REPO_POLICIES.md caps make test at 20 seconds
# and a browser suite does not fit. .gitea/workflows/e2e.yml also runs it # and a browser suite does not fit.
# on every push, in a job separate from check.
# #
# Unlike script/test-e2e this builds its base image locally, because no # Unlike script/test-e2e this builds its image locally, because no
# published image carries both a pinned Firefox and a matching geckodriver. # published image carries both a pinned Firefox and a matching geckodriver.
# All three external artifacts are pinned by digest inside the Dockerfile; # All three external artifacts are pinned by digest inside the Dockerfile;
# see tests/e2e/firefox/Dockerfile, which also explains why the repo and # see tests/e2e/firefox/Dockerfile.
# the extension build are baked into the image rather than mounted.
#
# Docker is the only prerequisite: nothing here depends on the node, yarn
# or make on the machine that starts the run.
set -eu set -eu
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)" SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd -P)"
@@ -23,14 +18,6 @@ ROOT="$(cd "$SCRIPT_DIR/.." && pwd -P)"
IMAGE="$("$SCRIPT_DIR/projectname")-e2e-firefox" IMAGE="$("$SCRIPT_DIR/projectname")-e2e-firefox"
IIDFILE=""
cleanup() {
if [ -n "$IIDFILE" ]; then
rm -f "$IIDFILE"
fi
}
main() { main() {
cd "$ROOT" cd "$ROOT"
@@ -39,20 +26,16 @@ main() {
exit 1 exit 1
fi fi
IIDFILE="$(mktemp)" echo "Building extension for e2e..."
trap cleanup EXIT yarn run build 2>&1
trap 'cleanup; exit 130' INT TERM
echo "Building the pinned Firefox e2e image (extension included)..." # The build context is tests/e2e/firefox/ and holds nothing but the
docker build --iidfile "$IIDFILE" -t "$IMAGE" \ # Dockerfile: the harness itself arrives over the bind mount below, so
-f tests/e2e/firefox/Dockerfile . # editing it never invalidates an image layer.
echo "Building the pinned Firefox e2e image..."
docker build -t "$IMAGE" "$ROOT/tests/e2e/firefox"
echo "Running the Firefox e2e suite..." echo "Running the Firefox e2e suite..."
# The image is run by ID, not by tag: where two clones of this repo run
# the suite at once, the other build can move the tag between this
# build and this run, and the suite would then silently test the other
# checkout.
#
# --shm-size=1g: Firefox needs more than the default 64MB /dev/shm. # --shm-size=1g: Firefox needs more than the default 64MB /dev/shm.
# --network none: the suite stubs nothing, so this is what keeps the # --network none: the suite stubs nothing, so this is what keeps the
# run offline and deterministic. The extension swallows its own # run offline and deterministic. The extension swallows its own
@@ -60,8 +43,8 @@ main() {
# network note in README.md. Weaker than the Chrome suite's # network note in README.md. Weaker than the Chrome suite's
# fixture interception, and honestly so — it proves no request # fixture interception, and honestly so — it proves no request
# escaped, but it cannot report which ones were attempted. # escaped, but it cannot report which ones were attempted.
# HOME=/tmp: the image's root home is not a reliable place for the # --user: keep files the suite touches owned by the caller, not root.
# browser profile. # HOME=/tmp: the mapped uid has no home directory in the image.
# #
# No --privileged. Firefox's sandbox logs # No --privileged. Firefox's sandbox logs
# "CanCreateUserNamespace() clone() failure: EPERM" on startup here; # "CanCreateUserNamespace() clone() failure: EPERM" on startup here;
@@ -69,8 +52,11 @@ main() {
docker run --rm \ docker run --rm \
--shm-size=1g \ --shm-size=1g \
--network none \ --network none \
--user "$(id -u):$(id -g)" \
-e HOME=/tmp \ -e HOME=/tmp \
"$(cat "$IIDFILE")" \ -v "$ROOT:/work" \
-w /work \
"$IMAGE" \
node tests/e2e/firefox/run.js dist/firefox node tests/e2e/firefox/run.js dist/firefox
} }

View File

@@ -57,6 +57,16 @@ const connectedSites = {};
// Pending approval requests: { id: { origin, hostname, resolve } } // Pending approval requests: { id: { origin, hostname, resolve } }
const pendingApprovals = {}; const pendingApprovals = {};
// What the page is told when a request failed in a way the wallet has no
// specific answer for. -32603 is the JSON-RPC internal error EIP-1474 defines
// and EIP-1193 defers to for RPC-layer failures; no EIP-1193 4xxx code
// describes "the wallet broke", and one is not invented here. The cause is
// logged rather than put in the message: the page gets a stable sentence, the
// background console gets the throw.
const INTERNAL_ERROR_CODE = -32603;
const INTERNAL_ERROR_MESSAGE =
"AutistMask could not complete this request because of an internal error.";
async function getState() { async function getState() {
const result = await storageApi.get("autistmask"); const result = await storageApi.get("autistmask");
return ( return (
@@ -865,9 +875,26 @@ runtime.onMessage.addListener((msg, sender, sendResponse) => {
// keep fallback // keep fallback
} }
} }
handleRpc(msg.method, msg.params, trustedOrigin).then((response) => { handleRpc(msg.method, msg.params, trustedOrigin)
sendResponse(response); .then((response) => {
}); sendResponse(response);
})
.catch((err) => {
// Without this the page's window.ethereum.request() promise
// stays pending forever: no response is sent, the content
// script posts nothing back, and the dApp cannot tell the
// failure from a slow wallet. handleRpc does real work —
// state loads, provider calls, transaction population — so
// "it does not throw today" is not a property anyone is
// maintaining.
log.errorf("RPC request failed:", msg.method, err);
sendResponse({
error: {
code: INTERNAL_ERROR_CODE,
message: INTERNAL_ERROR_MESSAGE,
},
});
});
return true; return true;
} }
@@ -1051,7 +1078,31 @@ runtime.onMessage.addListener((msg, sender, sendResponse) => {
stage: TX_STAGE_BROADCAST, stage: TX_STAGE_BROADCAST,
}); });
} }
})(); })().catch((e) => {
// Every statement above is inside a try, but a throw from one of
// the catch blocks escapes as an unhandled rejection and neither
// the popup nor the page is ever answered. Settle both, through
// the same chokepoint as every other retirement. The stage is
// broadcast because this cannot tell whether the transaction
// reached the network, and that is the wording that does not
// invite a second send.
log.errorf("transaction approval response failed:", e);
settleApproval(
msg.id,
{
error: {
code: INTERNAL_ERROR_CODE,
message: INTERNAL_ERROR_MESSAGE,
},
},
{ holdsClaim: true },
);
sendResponse({
error: INTERNAL_ERROR_MESSAGE,
retryable: false,
stage: TX_STAGE_BROADCAST,
});
});
return true; return true;
} }
@@ -1135,7 +1186,25 @@ runtime.onMessage.addListener((msg, sender, sendResponse) => {
} }
sendResponse({ error: errMsg, retryable }); sendResponse({ error: errMsg, retryable });
} }
})(); })().catch((e) => {
// Same shape as the transaction path: a throw out of the catch
// block above would leave the popup and the page both waiting.
log.errorf("sign approval response failed:", e);
settleApproval(
msg.id,
{
error: {
code: INTERNAL_ERROR_CODE,
message: INTERNAL_ERROR_MESSAGE,
},
},
{ holdsClaim: true },
);
sendResponse({
error: INTERNAL_ERROR_MESSAGE,
retryable: false,
});
});
return true; return true;
} }

View File

@@ -9,17 +9,16 @@ const {
$, $,
showView, showView,
updateDebugBanner, updateDebugBanner,
setBackRenderer, setRenderMain,
pushCurrentView, pushCurrentView,
goBack, goBack,
clearViewStack, clearViewStack,
} = require("./views/helpers"); } = require("./views/helpers");
const { applyTheme } = require("./theme"); const { applyTheme } = require("./theme");
// Renders a view the popup lands on without having navigated to it forward: // Views that can be fully re-rendered from persisted state. All others fall
// on restore here, and on Back. Only the views that can be fully re-rendered // back to the nearest restorable parent; see the module for why the
// from persisted state (RESTORABLE_VIEWS, src/popup/restorableViews.js) go // secret-bearing views are absent.
// through it; anything else falls back to the nearest restorable parent. const { RESTORABLE_VIEWS } = require("./restorableViews");
const { renderView, makeBackRenderer } = require("./viewRouter");
const home = require("./views/home"); const home = require("./views/home");
const welcome = require("./views/welcome"); const welcome = require("./views/welcome");
@@ -109,22 +108,91 @@ const ctx = {
}, },
}; };
// The view modules the router renders through, keyed as it expects them. function needsAddress(view) {
const viewModules = { return (
main: { show: () => fallbackView() }, view === "address" ||
addressDetail, view === "address-token" ||
addressToken, view === "receive" ||
receive, view === "transaction"
settings, );
settingsAddToken, }
confirmTx,
transactionDetail, function hasValidAddress() {
txStatus, return (
}; state.selectedWallet !== null &&
state.selectedAddress !== null &&
state.wallets[state.selectedWallet] &&
state.wallets[state.selectedWallet].addresses[state.selectedAddress]
);
}
function restoreView() { function restoreView() {
if (!renderView(state.currentView, state, viewModules)) { const view = state.currentView;
fallbackView(); if (!view || !RESTORABLE_VIEWS.has(view)) {
return fallbackView();
}
if (needsAddress(view) && !hasValidAddress()) {
return fallbackView();
}
if (view === "address-token" && !state.selectedToken) {
return fallbackView();
}
switch (view) {
case "address":
addressDetail.show();
break;
case "address-token":
addressToken.show();
break;
case "receive":
receive.show();
break;
case "settings":
settings.show();
break;
case "settings-addtoken":
settingsAddToken.show();
break;
case "confirm-tx":
if (state.viewData && state.viewData.pendingTx) {
confirmTx.restore();
} else {
fallbackView();
}
break;
case "transaction":
if (state.viewData && state.viewData.tx) {
transactionDetail.render();
} else {
fallbackView();
}
break;
case "wait-tx":
// Resumes the receipt poll from the persisted broadcast time.
if (!txStatus.restoreWait()) {
fallbackView();
}
break;
case "success-tx":
if (state.viewData && state.viewData.hash) {
txStatus.renderSuccess();
} else {
fallbackView();
}
break;
case "error-tx":
if (state.viewData && state.viewData.message) {
txStatus.renderError();
} else {
fallbackView();
}
break;
default:
fallbackView();
break;
} }
} }
@@ -179,7 +247,7 @@ async function init() {
settings.show(); settings.show();
}); });
setBackRenderer(makeBackRenderer(state, viewModules)); setRenderMain(renderWalletList);
welcome.init(ctx); welcome.init(ctx);
addWallet.init(ctx); addWallet.init(ctx);

View File

@@ -1,167 +0,0 @@
// Rendering a view the popup lands on without having navigated to it
// forward: on restore, and on Back. In both cases the view may never have
// been rendered in this page load — a reopened popup renders only the
// wallet list and the view it restores onto, so every other view is still
// the blank static template from index.html — so unhiding it is not enough.
//
// Forward navigation renders as it goes and must NOT come through here:
// rendering a second time would re-fetch and clobber whatever the view has
// in flight.
//
// The view modules are injected and nothing here touches the DOM, so the
// dispatch and its data guards can be tested directly; src/popup/index.js
// cannot be required outside a browser.
const { RESTORABLE_VIEWS } = require("./restorableViews");
// The views this page load has rendered.
//
// The Back path cannot otherwise tell its two cases apart. A view the popup
// never rendered is still the blank template from index.html and has to be
// rendered; a view already on the page must NOT be rendered again, because
// a second render re-fetches and overwrites whatever the user has typed
// into it and not yet saved.
//
// Registration is showView() in views/helpers.js, which is the last thing
// every render path runs — restoreView()'s, the Back path's, and every
// forward show(). That is the point of putting it there rather than in the
// individual views: a view added later registers itself with no one having
// to remember it, so this cannot decay.
//
// Module scope is page-load scope: the popup loads this module once per
// page load, and a reopened popup gets a fresh, empty set — which is
// exactly the state that makes the Back path render.
const renderedViews = new Set();
function markViewRendered(view) {
if (view) renderedViews.add(view);
}
// Begin a fresh page-load scope. The popup gets one by being loaded; the
// unit tests, which simulate several page loads against one module
// instance, ask for one.
function resetRenderedViews() {
renderedViews.clear();
}
// Home is the exception: Back re-renders it every time, which is what the
// popup did before this router existed (index.js registered
// renderWalletList() as setRenderMain(), and goBack() called it on every
// Back onto "main"). It must stay that way — the wallet list has to reflect
// what changed while the user was away from it, such as a wallet renamed or
// an address removed in Settings — and Home holds no unsaved input to lose.
const ALWAYS_RENDER_ON_BACK = new Set(["main"]);
// Views that render an address the user picked and cannot be rendered
// without one.
const ADDRESS_VIEWS = new Set([
"address",
"address-token",
"receive",
"transaction",
]);
function needsAddress(view) {
return ADDRESS_VIEWS.has(view);
}
function hasValidAddress(state) {
return Boolean(
state.selectedWallet !== null &&
state.selectedAddress !== null &&
state.wallets[state.selectedWallet] &&
state.wallets[state.selectedWallet].addresses[state.selectedAddress],
);
}
// Render `view` from persisted state. Each view module shows itself, so a
// true return means the view is both rendered and on screen.
//
// Returns false when the view is not one the popup renders from state, or
// when the state it would render is gone — a token no longer selected, a
// transaction no longer persisted. The caller falls back rather than
// putting an empty template on screen.
function renderView(view, state, views) {
if (!view || !RESTORABLE_VIEWS.has(view)) return false;
if (needsAddress(view) && !hasValidAddress(state)) return false;
if (view === "address-token" && !state.selectedToken) return false;
const data = state.viewData || {};
switch (view) {
case "main":
views.main.show();
return true;
case "address":
views.addressDetail.show();
return true;
case "address-token":
views.addressToken.show();
return true;
case "receive":
views.receive.show();
return true;
case "settings":
views.settings.show();
return true;
case "settings-addtoken":
views.settingsAddToken.show();
return true;
case "confirm-tx":
if (!data.pendingTx) return false;
views.confirmTx.restore();
return true;
case "transaction":
if (!data.tx) return false;
views.transactionDetail.render();
return true;
case "wait-tx":
// Resumes the receipt poll from the persisted broadcast time,
// and answers false when there is nothing resumable left.
return Boolean(views.txStatus.restoreWait());
case "success-tx":
if (!data.hash) return false;
views.txStatus.renderSuccess();
return true;
case "error-tx":
if (!data.message) return false;
views.txStatus.renderError();
return true;
default:
return false;
}
}
// The Back-path renderer, registered with setBackRenderer() in
// views/helpers.js.
//
// Returns false — leaving goBack() to unhide the view, as it always did —
// in the two cases where the view is known to be on the page already:
//
// - It is not one the popup renders from persisted state. The restored
// stack is filtered against RESTORABLE_VIEWS, so such a view can only
// be on the stack from this page load, where forward navigation
// rendered it on the way in.
// - This page load has rendered it. Re-rendering would re-fetch and
// clobber what it holds; Home is rendered anyway, see above.
//
// What is left is the case the router exists for: a view on the stack that
// this page load has never rendered, whose template is still blank.
function makeBackRenderer(state, views) {
return function renderBack(view) {
if (!RESTORABLE_VIEWS.has(view)) return false;
if (renderedViews.has(view) && !ALWAYS_RENDER_ON_BACK.has(view)) {
return false;
}
if (!renderView(view, state, views)) {
views.main.show();
}
return true;
};
}
module.exports = {
renderView,
makeBackRenderer,
markViewRendered,
resetRenderedViews,
};

View File

@@ -7,7 +7,6 @@ const {
getAddressValueUsd, getAddressValueUsd,
} = require("../../shared/prices"); } = require("../../shared/prices");
const { state, saveState, currentNetwork } = require("../../shared/state"); const { state, saveState, currentNetwork } = require("../../shared/state");
const { markViewRendered } = require("../viewRouter");
// When views are added, removed, or transitions between them change, // When views are added, removed, or transitions between them change,
// update the view-navigation documentation in README.md to match. // update the view-navigation documentation in README.md to match.
@@ -77,10 +76,6 @@ function showView(name) {
} }
clearFlash(); clearFlash();
state.currentView = name; state.currentView = name;
// A view's show() ends here, so this is where the Back path learns the
// view is no longer the blank template from index.html and must not be
// rendered a second time. See viewRouter.js.
markViewRendered(name);
saveState(); saveState();
updateDebugBanner(name); updateDebugBanner(name);
} }
@@ -116,19 +111,12 @@ function updateDebugBanner(viewName) {
} }
} }
// Callback that renders a view being navigated BACK onto. Set once by // Callback to re-render the main/home view when navigating back to it.
// index.js via setBackRenderer(), which routes the view through the same // Set once by index.js via setRenderMain().
// per-view render and data guards restoreView() uses. let _renderMain = null;
//
// It answers true when it took the navigation — the view is rendered and
// shown, or its backing data was gone and it fell back — and false for a
// view the popup does not render from persisted state. Those can only be
// on the stack from this page load, because the stack is filtered on load,
// so they have already been rendered and only need unhiding.
let _renderBack = null;
function setBackRenderer(fn) { function setRenderMain(fn) {
_renderBack = fn; _renderMain = fn;
} }
// Push the current view onto the navigation stack so goBack() can // Push the current view onto the navigation stack so goBack() can
@@ -148,11 +136,9 @@ function goBack() {
} else { } else {
target = "main"; target = "main";
} }
// A popped view is landed on, not navigated to. If the popup has been if (target === "main" && _renderMain) {
// closed and reopened since the view was pushed, nothing has ever _renderMain();
// rendered it in this page load and its template is still blank, so it }
// has to be rendered here rather than merely unhidden.
if (_renderBack && _renderBack(target)) return;
showView(target); showView(target);
} }
@@ -484,7 +470,7 @@ module.exports = {
showView, showView,
onViewLeave, onViewLeave,
updateDebugBanner, updateDebugBanner,
setBackRenderer, setRenderMain,
pushCurrentView, pushCurrentView,
goBack, goBack,
clearViewStack, clearViewStack,

View File

@@ -1,329 +0,0 @@
// Back after reopening the popup (#268).
//
// A reopened popup renders the wallet list and the one view it restores
// onto; every other view is still the blank static template from
// index.html. goBack() used to only unhide its target, so Back landed on
// that blank template for any view the popup had not rendered in this page
// load. These tests drive the real goBack() with the real router wired to
// recording view modules, so what is asserted is which view render ran —
// the thing that was missing.
//
// The rendering itself is asserted against the real popup in a real
// browser by tests/e2e/run.js; here the DOM is a stub, because goBack()
// only needs showView() to work.
const els = new Map();
function fakeEl() {
return {
textContent: "",
innerHTML: "",
classList: {
toggle() {},
add() {},
remove() {},
contains: () => false,
},
remove() {},
};
}
globalThis.document = {
getElementById(id) {
if (!els.has(id)) els.set(id, fakeEl());
return els.get(id);
},
};
// helpers.js pulls in state.js, which reads chrome.storage.local at load.
globalThis.chrome = {
storage: { local: { get: async () => ({}), set: async () => {} } },
};
const {
showView,
goBack,
setBackRenderer,
pushCurrentView,
} = require("../src/popup/views/helpers");
const {
makeBackRenderer,
markViewRendered,
resetRenderedViews,
} = require("../src/popup/viewRouter");
const { state } = require("../src/shared/state");
const ADDRESS = "0x1111111111111111111111111111111111111111";
const TOKEN = "0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48";
let calls;
// Stand-ins for the view modules. Each records itself and then shows its
// view, which is what every real view render ends with — so the assertions
// can tell "rendered and shown" apart from "merely unhidden".
function recorder(name, view) {
return () => {
calls.push(name);
showView(view);
};
}
function makeViews() {
return {
main: { show: recorder("main", "main") },
addressDetail: { show: recorder("addressDetail", "address") },
addressToken: { show: recorder("addressToken", "address-token") },
receive: { show: recorder("receive", "receive") },
settings: { show: recorder("settings", "settings") },
settingsAddToken: {
show: recorder("settingsAddToken", "settings-addtoken"),
},
confirmTx: { restore: recorder("confirmTx", "confirm-tx") },
transactionDetail: {
render: recorder("transactionDetail", "transaction"),
},
txStatus: {
restoreWait: () => {
calls.push("waitTx");
showView("wait-tx");
return true;
},
renderSuccess: recorder("successTx", "success-tx"),
renderError: recorder("errorTx", "error-tx"),
},
};
}
// The popup as it stands just after a reopen: one wallet with one address,
// the view the popup restored onto, and the stack behind it.
//
// A reopen is a fresh page load, so the record of what has been rendered
// starts empty — that emptiness is what makes the Back path render at all.
// Returns the view modules so a test can drive forward navigation through
// the same recorders the router renders through.
function reopenedOn(view, stack, extra) {
calls = [];
resetRenderedViews();
state.wallets = [
{
name: "Wallet 1",
addresses: [{ address: ADDRESS, balance: "0", tokenBalances: [] }],
},
];
state.selectedWallet = 0;
state.selectedAddress = 0;
state.selectedToken = null;
state.viewData = null;
state.currentView = view;
state.viewStack = stack.slice();
Object.assign(state, extra || {});
// Restoring onto a view renders it, so the reopened popup has that one
// view on the page and nothing else.
markViewRendered(view);
const views = makeViews();
setBackRenderer(makeBackRenderer(state, views));
return views;
}
// The reproduction from the issue, step for step.
describe("Back onto a view the reopened popup never rendered", () => {
test("Back from settings renders the address detail underneath", () => {
reopenedOn("settings", ["main", "address"]);
goBack();
expect(calls).toEqual(["addressDetail"]);
expect(state.currentView).toBe("address");
expect(state.viewStack).toEqual(["main"]);
});
test("Back onto the token detail renders it", () => {
reopenedOn("settings", ["main", "address", "address-token"], {
selectedToken: TOKEN,
});
goBack();
expect(calls).toEqual(["addressToken"]);
expect(state.currentView).toBe("address-token");
});
test("Back onto Receive renders it", () => {
reopenedOn("settings", ["main", "address", "receive"]);
goBack();
expect(calls).toEqual(["receive"]);
expect(state.currentView).toBe("receive");
});
test("Back onto the transaction detail renders it", () => {
reopenedOn("settings", ["main", "transaction"], {
viewData: { tx: { hash: "0xdead" } },
});
goBack();
expect(calls).toEqual(["transactionDetail"]);
expect(state.currentView).toBe("transaction");
});
test("Back onto the transaction confirmation restores it", () => {
reopenedOn("settings", ["main", "confirm-tx"], {
viewData: { pendingTx: { to: ADDRESS, amount: "1" } },
});
goBack();
expect(calls).toEqual(["confirmTx"]);
expect(state.currentView).toBe("confirm-tx");
});
test("Back onto the success screen renders it", () => {
reopenedOn("settings", ["main", "success-tx"], {
viewData: { hash: "0xdead" },
});
goBack();
expect(calls).toEqual(["successTx"]);
expect(state.currentView).toBe("success-tx");
});
test("Back onto the failure screen renders it", () => {
reopenedOn("settings", ["main", "error-tx"], {
viewData: { message: "execution reverted" },
});
goBack();
expect(calls).toEqual(["errorTx"]);
expect(state.currentView).toBe("error-tx");
});
test("Back onto Home renders the wallet list", () => {
reopenedOn("settings", ["main"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("Back with an empty stack renders Home", () => {
reopenedOn("settings", []);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
});
// The guards are restoreView()'s, so a popped view whose backing data is
// gone lands on Home rather than on an empty template.
describe("Back onto a view whose backing data is gone", () => {
test("the token detail with no token selected falls back to Home", () => {
reopenedOn("settings", ["main", "address-token"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("the transaction detail with no transaction falls back to Home", () => {
reopenedOn("settings", ["main", "transaction"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("the confirmation with no pending transaction falls back to Home", () => {
reopenedOn("settings", ["main", "confirm-tx"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("an address view with no address selected falls back to Home", () => {
reopenedOn("settings", ["main", "receive"], {
selectedAddress: null,
});
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("the success screen with no transaction hash falls back to Home", () => {
reopenedOn("settings", ["main", "success-tx"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("the failure screen with no message falls back to Home", () => {
reopenedOn("settings", ["main", "error-tx"]);
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
test("a wait that can no longer be resumed falls back to Home", () => {
reopenedOn("settings", ["main", "wait-tx"]);
const views = makeViews();
views.txStatus.restoreWait = () => false;
setBackRenderer(makeBackRenderer(state, views));
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
});
// Forward navigation renders as it goes, and a second render would re-fetch
// and clobber whatever the view holds — an unsaved edit, a request in
// flight. So the Back path renders only a view this page load has never
// rendered, and merely unhides every other one: the views it does not
// render from persisted state, and the views already on the page.
describe("what the Back path leaves alone", () => {
test("forward navigation renders nothing by itself", () => {
reopenedOn("address", ["main"]);
pushCurrentView();
showView("send");
expect(calls).toEqual([]);
expect(state.viewStack).toEqual(["main", "address"]);
});
test("Back onto a live-session view only unhides it", () => {
reopenedOn("confirm-tx", ["main", "address", "send"]);
goBack();
expect(calls).toEqual([]);
expect(state.currentView).toBe("send");
});
test("Back renders its target exactly once", () => {
reopenedOn("settings", ["main", "address"]);
goBack();
expect(calls.filter((c) => c === "addressDetail")).toHaveLength(1);
});
test("Back onto a view this page load already rendered only unhides it", () => {
const views = reopenedOn("main", []);
pushCurrentView();
views.addressDetail.show();
pushCurrentView();
views.settings.show();
calls = [];
goBack();
expect(calls).toEqual([]);
expect(state.currentView).toBe("address");
});
// The unit mirror of the regression the browser suite pins: Settings
// reassigns its fields from persisted state on every render, so a
// re-render on the way back discards an edit the user has not saved.
test("Back onto Settings visited earlier in this page load does not re-render it", () => {
const views = reopenedOn("main", []);
pushCurrentView();
views.settings.show();
pushCurrentView();
views.settingsAddToken.show();
calls = [];
goBack();
expect(calls).toEqual([]);
expect(state.currentView).toBe("settings");
});
// Home is the deliberate exception, unchanged from the popup's
// behaviour before the router existed: it re-renders on every Back so
// the wallet list reflects what changed while the user was away.
test("Back onto Home renders it again even when it is already on the page", () => {
const views = reopenedOn("main", []);
pushCurrentView();
views.addressDetail.show();
calls = [];
goBack();
expect(calls).toEqual(["main"]);
expect(state.currentView).toBe("main");
});
});

View File

@@ -133,6 +133,14 @@ function loadBackground(options) {
ensureRecurringAlarms: jest.fn(async () => {}), ensureRecurringAlarms: jest.fn(async () => {}),
registerAlarmHandlers: jest.fn(), registerAlarmHandlers: jest.fn(),
})); }));
// The real verification module, except where a test replaces one export
// with a throw to drive the handler's own error handling into failing.
if (opts.approvalVerify) {
jest.doMock("../src/shared/approvalVerify", () => ({
...jest.requireActual("../src/shared/approvalVerify"),
...opts.approvalVerify,
}));
}
const persisted = { const persisted = {
wallets: [ wallets: [
@@ -152,7 +160,10 @@ function loadBackground(options) {
global.chrome = { global.chrome = {
storage: { storage: {
local: { local: {
get: jest.fn(async () => ({ autistmask: persisted })), get: jest.fn(
opts.storageGet ||
(async () => ({ autistmask: persisted })),
),
set: jest.fn(async () => {}), set: jest.fn(async () => {}),
}, },
}, },
@@ -280,6 +291,25 @@ async function settle() {
for (let i = 0; i < 50; i++) await Promise.resolve(); for (let i = 0; i < 50; i++) await Promise.resolve();
} }
// Node aborts the worker process on an unhandled rejection; an extension
// service worker does not — the promise is simply never settled, nothing is
// sent back, and the page's window.ethereum.request() waits forever. Recording
// them instead of dying on them keeps that difference visible: the assertion
// that the page WAS answered is what reports the failure, and the recording is
// asserted empty alongside it.
const unhandledRejections = [];
process.on("unhandledRejection", (reason) => {
unhandledRejections.push(reason);
});
// Node reports an unhandled rejection on the macrotask turn after the promise
// was left unhandled, which is past everything settle() waits for.
async function settleIncludingRejections() {
await settle();
await new Promise((resolve) => setImmediate(resolve));
await new Promise((resolve) => setImmediate(resolve));
}
afterEach(() => { afterEach(() => {
delete global.chrome; delete global.chrome;
jest.resetModules(); jest.resetModules();
@@ -1034,6 +1064,147 @@ describe("a claimed approval outlives every other retirement path", () => {
}); });
}); });
// A handler that throws must still answer. `sendResponse` is the only thing
// that settles the page's window.ethereum.request() promise, so a throw that
// escapes a handler leaves that promise pending forever — no error, no
// timeout, indistinguishable from a slow wallet. Each case below drives a real
// throw out of a handler rather than asserting the catch block exists.
describe("a handler that throws still settles the page", () => {
const INTERNAL_ERROR = {
code: -32603,
message:
"AutistMask could not complete this request because of an internal error.",
};
let errorLog;
beforeEach(() => {
errorLog = jest.spyOn(console, "error").mockImplementation(() => {});
unhandledRejections.length = 0;
});
afterEach(() => {
errorLog.mockRestore();
});
// getState() awaits extension storage unguarded, and every read path in
// handleRpc goes through it. A storage read that rejects is the whole
// failure — no hook in the handler itself.
test("a rejected handleRpc rejects the page instead of hanging it", async () => {
const bg = loadBackground({
storageGet: async () => {
throw new Error("storage unavailable");
},
});
const answer = bg.send(
{ type: "AUTISTMASK_RPC", method: "eth_accounts", params: [] },
{ origin: ORIGIN },
);
await settleIncludingRejections();
// The channel is held open for the async answer, and the answer
// arrives.
expect(answer.kept).toBe(true);
expect(answer.sendResponse).toHaveBeenCalledWith({
error: INTERNAL_ERROR,
});
expect(unhandledRejections).toEqual([]);
// Not swallowed: the throw is on the background console, which is how
// this class gets caught in future.
expect(errorLog).toHaveBeenCalledWith(
"[AutistMask]",
"RPC request failed:",
"eth_accounts",
expect.objectContaining({ message: "storage unavailable" }),
);
});
// The transaction response handler wraps every statement in a try, so what
// escapes it is a throw from inside one of its catch blocks. Here the
// failure classifier itself throws while classifying a real verification
// failure — the approval is left claimed, so nothing else can settle it.
test("a throw while handling a failed transaction settles both the page and the popup", async () => {
const bg = loadBackground({
approvalVerify: {
describeTxFailure: () => {
throw new Error("classifier broke");
},
},
});
const pending = bg.requestTx();
await settle();
const id = pending.id();
// A real verification failure: the artifact is signed at a nonce the
// approval never displayed.
const answer = bg.send(
{
type: "AUTISTMASK_TX_RESPONSE",
id,
approved: true,
rawSignedTx: await signedAtNonce(NONCE + 1),
},
{ url: bg.fromPopup.url },
);
await settleIncludingRejections();
expect(bg.broadcastTransaction).not.toHaveBeenCalled();
expect(pending.result()).toEqual({ error: INTERNAL_ERROR });
expect(answer.sendResponse).toHaveBeenCalledWith({
error: INTERNAL_ERROR.message,
retryable: false,
// The handler cannot tell whether the transaction reached the
// network, so the popup must not say "start again from the site".
stage: "broadcast",
});
expect(unhandledRejections).toEqual([]);
expect(errorLog).toHaveBeenCalledWith(
"[AutistMask]",
"transaction approval response failed:",
expect.objectContaining({ message: "classifier broke" }),
);
});
test("a throw while handling a failed signature settles both the page and the popup", async () => {
const bg = loadBackground({
approvalVerify: {
failureIsRetryable: () => {
throw new Error("classifier broke");
},
},
});
const pending = bg.requestSign();
await settle();
// A real verification failure: the active address moved after the
// approval was raised.
bg.setActiveAddress(other.address);
const answer = bg.send(
{
type: "AUTISTMASK_SIGN_RESPONSE",
id: pending.id(),
approved: true,
signature: await signer.signMessage(
Buffer.from(MESSAGE.slice(2), "hex"),
),
},
{ url: bg.fromPopup.url },
);
await settleIncludingRejections();
expect(pending.result()).toEqual({ error: INTERNAL_ERROR });
expect(answer.sendResponse).toHaveBeenCalledWith({
error: INTERNAL_ERROR.message,
retryable: false,
});
expect(unhandledRejections).toEqual([]);
expect(errorLog).toHaveBeenCalledWith(
"[AutistMask]",
"sign approval response failed:",
expect.objectContaining({ message: "classifier broke" }),
);
});
});
describe("popup-only messages", () => { describe("popup-only messages", () => {
test("a page sender cannot answer an approval", async () => { test("a page sender cannot answer an approval", async () => {
const bg = loadBackground(); const bg = loadBackground();

View File

@@ -1,34 +0,0 @@
# Chrome end-to-end image: the pinned Playwright image with this repo and a
# freshly built extension inside it, built by script/test-e2e. The suite is
# still started with `docker run`, so every runtime flag the harness needs
# (--ipc=host in particular) applies as before.
#
# The repo is baked in rather than bind-mounted because a bind mount does
# not resolve under Gitea Actions: the runner runs the job in a container
# against the HOST's docker socket, so the source side of a -v is resolved
# by the host daemon while the job's checkout lives on a docker volume that
# is not a host path -- the mount silently succeeds and /work is empty. A
# build context is streamed to the daemon and so works from anywhere.
# Building the extension here too means the machine starting a run needs
# docker and nothing else.
# mcr.microsoft.com/playwright:v1.56.0-noble, 2026-08-09
#
# The playwright-core devDependency is pinned to the matching Playwright
# version (1.56.0) and the two must be bumped together: the browsers ship
# inside this image, and playwright-core looks for the exact browser
# revision its own version expects. A mismatch fails at launch.
FROM mcr.microsoft.com/playwright@sha256:35246d87a7c88ea9b771c65d33171b2611b02a8253b4b12ce6f94376c55f99f2
WORKDIR /work
# Same layering as the root Dockerfile: script/bootstrap installs the
# prerequisites and the dependencies, and the manifests are copied first so
# that layer is cached until they change.
COPY script/ script/
COPY package.json yarn.lock ./
RUN script/bootstrap
COPY . .
RUN make build

View File

@@ -1,24 +1,10 @@
# Firefox end-to-end image: stock Firefox plus geckodriver on a node base, # Firefox end-to-end image: stock Firefox plus geckodriver on a node base,
# with this repo and a freshly built extension inside it, built by # built by script/test-e2e-firefox. The repo is bind-mounted at /work; the
# script/test-e2e-firefox. The harness itself has no dependencies, so # harness itself has no dependencies, so nothing is installed for it.
# nothing is installed for it.
# #
# The build context is the repo root. The repo is baked in rather than # All three external artifacts are pinned by digest. The Firefox version in
# bind-mounted because a bind mount does not resolve under Gitea Actions: # particular must not float: -remote-allow-system-access is mandatory on 153
# the runner runs the job in a container against the HOST's docker socket, # and was not on 142, so the flag the harness passes is version-coupled.
# so the source side of a -v is resolved by the host daemon while the job's
# checkout lives on a docker volume that is not a host path -- the mount
# silently succeeds and /work is empty. Baking the build in is also the
# only way this suite can have both a built extension and the
# `--network none` it runs under, since a container with no network cannot
# install anything.
#
# All three external artifacts are pinned by digest, and are fetched in
# layers above the repo copy, so editing the harness or any source file
# re-runs only the two cheap layers at the bottom. The Firefox version in
# particular must not float: -remote-allow-system-access is mandatory on
# 153 and was not on 142, so the flag the harness passes is
# version-coupled.
# node:22-bookworm-slim, 2026-08-12 # node:22-bookworm-slim, 2026-08-12
FROM node@sha256:d649c27dae7ba0137b3cef5dd75baa422c08dc3d9e3fc0c23dfb172dc3cc6436 FROM node@sha256:d649c27dae7ba0137b3cef5dd75baa422c08dc3d9e3fc0c23dfb172dc3cc6436
@@ -62,16 +48,4 @@ ENV FIREFOX_BIN=/opt/firefox/firefox
ENV GECKODRIVER=/usr/local/bin/geckodriver ENV GECKODRIVER=/usr/local/bin/geckodriver
WORKDIR /work WORKDIR /work
# Same layering as the root Dockerfile: script/bootstrap installs the
# prerequisites and the dependencies, and the manifests are copied first so
# that layer is cached until they change.
COPY script/ script/
COPY package.json yarn.lock ./
RUN script/bootstrap
COPY . .
RUN make build
CMD ["node", "tests/e2e/firefox/run.js", "dist/firefox"] CMD ["node", "tests/e2e/firefox/run.js", "dist/firefox"]

View File

@@ -419,172 +419,6 @@ test("reopening the popup never lands on the phrase screen (#161)", async (env)
assertWiped(st, env.phrase, "after reopening the popup"); assertWiped(st, env.phrase, "after reopening the popup");
}); });
// ------------------------------- Back after reopening the popup (#268)
// A reopened popup renders the wallet list and the view it restores onto,
// and nothing else: every other screen is still the blank static template
// from index.html. Back used to only unhide its target, which is why these
// have to run against the real popup — the template is present and
// well-formed, so only its emptiness distinguishes the defect, and only a
// real reopen produces it.
// Everything the address screen must have on it, read out of the DOM.
function addressScreenState(page) {
return page.evaluate(() => {
const line = document.getElementById("address-line");
const balances = document.getElementById("address-balances");
return {
hidden: document
.getElementById("view-address")
.classList.contains("hidden"),
line: line ? line.innerText.trim() : "",
balances: balances ? balances.innerText.trim() : "",
};
});
}
// Close and reopen the page rather than reload it: that is what the toolbar
// popup does, and it is the only thing that produces the unrendered views.
async function reopenPopup(env, restoredView) {
await env.page.close();
env.page = await openPopup(env.ctx, env.popupUrl);
await visible(env.page, restoredView);
}
// The reproduction from the issue, step for step.
test("Back after reopening the popup renders the address screen (#268)", async (env) => {
await openAddressDetail(env.page);
const before = await addressScreenState(env.page);
assert(
before.line.length > 0,
"the address screen was blank to begin with",
);
await env.page.click("#btn-settings");
await visible(env.page, "#view-settings");
await reopenPopup(env, "#view-settings");
await env.page.click("#btn-settings-back");
await visible(env.page, "#view-address");
const after = await addressScreenState(env.page);
assert(
after.line === before.line,
"the address line reads " +
JSON.stringify(after.line) +
", expected " +
JSON.stringify(before.line),
);
assert(
after.balances.includes("ETH"),
"the balances read " + JSON.stringify(after.balances),
);
});
// The same defect one screen further in. Receive holds the address twice
// over — as text and as the QR code the sender scans — and a blank one is
// worse than a missing screen.
// Everything the Receive screen must have on it. The QR code is read as
// pixels, not as an element: the blank template carries the canvas too, a
// default 300x150 one with nothing drawn on it and every pixel fully
// transparent. A drawn QR paints an opaque background across the whole
// canvas, so a single opaque pixel is the whole question.
function receiveScreenState(page) {
return page.evaluate(() => {
const block = document.getElementById("receive-address-block");
const canvas = document.getElementById("receive-qr");
const px = canvas
.getContext("2d")
.getImageData(0, 0, canvas.width, canvas.height).data;
let opaque = 0;
for (let i = 3; i < px.length; i += 4) {
if (px[i] > 0) opaque += 1;
}
return {
address: block.dataset.full || "",
text: block.innerText.trim(),
qrOpaquePixels: opaque,
};
});
}
test("Back after reopening the popup renders the Receive screen (#268)", async (env) => {
await openAddressDetail(env.page);
await env.page.click("#btn-receive");
await visible(env.page, "#view-receive");
const before = await receiveScreenState(env.page);
assert(
/^0x[0-9a-fA-F]{40}$/.test(before.address),
"Receive showed no address to begin with: " +
JSON.stringify(before.address),
);
await env.page.click("#btn-settings");
await visible(env.page, "#view-settings");
await reopenPopup(env, "#view-settings");
await env.page.click("#btn-settings-back");
await visible(env.page, "#view-receive");
const shown = await receiveScreenState(env.page);
assert(
shown.address === before.address,
"Receive shows " +
JSON.stringify(shown.address) +
", expected " +
JSON.stringify(before.address),
);
assert(
shown.text.includes(before.address),
"the Receive address is not on screen: " + JSON.stringify(shown.text),
);
assert(shown.qrOpaquePixels > 0, "Receive shows an unpainted QR code");
// Leave the suite where it found it.
await env.page.click("#btn-receive-back");
await visible(env.page, "#view-address");
await env.page.click("#btn-address-back");
await visible(env.page, "#view-main");
});
// The other half of the requirement: Back renders a screen this page load
// never rendered, and must NOT re-render one it already has on screen.
// settings.show() reassigns #settings-rpc from persisted state, so
// re-rendering Settings on the way back would silently revert whatever the
// user typed and had not saved yet — and they could then press Save and
// store the value they believed they had replaced. No reopen here: this is
// an ordinary in-session forward-and-back, which is exactly why the render
// must not happen.
test("Back onto Settings keeps unsaved input (#268)", async (env) => {
await visible(env.page, "#view-main");
await env.page.click("#btn-settings");
await visible(env.page, "#view-settings");
const typed = "https://rpc.example.invalid/unsaved";
await env.page.fill("#settings-rpc", typed);
await env.page.click("#btn-settings-add-token");
await visible(env.page, "#view-settings-addtoken");
await env.page.click("#btn-settings-addtoken-back");
await visible(env.page, "#view-settings");
const kept = await env.page.inputValue("#settings-rpc");
assert(
kept === typed,
"the unsaved RPC URL reads " +
JSON.stringify(kept) +
", expected " +
JSON.stringify(typed),
);
// Leave the suite where it found it. The typed value was never saved,
// and Settings reloads the field from state next time it renders.
await env.page.click("#btn-settings-back");
await visible(env.page, "#view-main");
});
// -------------------------------------------- address removal (#162) // -------------------------------------------- address removal (#162)
// Number of address rows across every wallet in the list, counted in the DOM // Number of address rows across every wallet in the list, counted in the DOM