Source code

Revision control

Copy as Markdown

Other Tools

/* Any copyright is dedicated to the Public Domain.
/* eslint no-unused-vars: [2, {"vars": "local"}] */
/* import-globals-from ../../inspector/test/shared-head.js */
"use strict";
// This shared-head.js file is used by most mochitests
// and we start using it in xpcshell tests as well.
// It contains various common helper functions.
const isMochitest = "gTestPath" in this;
const isXpcshell = !isMochitest;
if (isXpcshell) {
// gTestPath isn't exposed to xpcshell tests
// _TEST_FILE is an array for a unique string
/* global _TEST_FILE */
this.gTestPath = _TEST_FILE[0];
}
const { Constructor: CC } = Components;
// Print allocation count if DEBUG_DEVTOOLS_ALLOCATIONS is set to "normal",
// and allocation sites if DEBUG_DEVTOOLS_ALLOCATIONS is set to "verbose".
const DEBUG_ALLOCATIONS = Services.env.get("DEBUG_DEVTOOLS_ALLOCATIONS");
if (DEBUG_ALLOCATIONS) {
// Use a custom loader with `invisibleToDebugger` flag for the allocation tracker
// as it instantiates custom Debugger API instances and has to be running in a distinct
// compartments from DevTools and system scopes (JSMs, XPCOM,...)
const {
useDistinctSystemPrincipalLoader,
releaseDistinctSystemPrincipalLoader,
} = ChromeUtils.importESModule(
);
const requester = {};
const loader = useDistinctSystemPrincipalLoader(requester);
registerCleanupFunction(() =>
releaseDistinctSystemPrincipalLoader(requester)
);
const { allocationTracker } = loader.require(
);
const tracker = allocationTracker({ watchAllGlobals: true });
registerCleanupFunction(() => {
if (DEBUG_ALLOCATIONS == "normal") {
tracker.logCount();
} else if (DEBUG_ALLOCATIONS == "verbose") {
tracker.logAllocationSites();
}
tracker.stop();
});
}
// When DEBUG_STEP environment variable is set,
// automatically start a tracer which will log all line being executed
// in the running test (and nothing else) and also pause its execution
// for the given amount of milliseconds.
//
// Be careful that these pause have significant side effect.
// This will pause the test script event loop and allow running the other
// tasks queued in the parent process's main thread event loop queue.
//
// Passing any non-number value, like `DEBUG_STEP=true` will still
// log the executed lines without any pause, and without this side effect.
//
// For now, the tracer can only work once per thread.
// So when using this feature you will not be able to use the JS tracer
// in any other way on parent process's main thread.
const DEBUG_STEP = Services.env.get("DEBUG_STEP");
if (DEBUG_STEP) {
// Use a custom loader with `invisibleToDebugger` flag for the allocation tracker
// as it instantiates custom Debugger API instances and has to be running in a distinct
// compartments from DevTools and system scopes (JSMs, XPCOM,...)
const {
useDistinctSystemPrincipalLoader,
releaseDistinctSystemPrincipalLoader,
} = ChromeUtils.importESModule(
);
const requester = {};
const loader = useDistinctSystemPrincipalLoader(requester);
const stepper = loader.require(
);
stepper.start(globalThis, gTestPath, DEBUG_STEP);
registerCleanupFunction(() => {
stepper.stop();
releaseDistinctSystemPrincipalLoader(requester);
});
}
const DEBUG_TRACE_LINE = Services.env.get("DEBUG_TRACE_LINE");
if (DEBUG_TRACE_LINE) {
// Use a custom loader with `invisibleToDebugger` flag for the allocation tracker
// as it instantiates custom Debugger API instances and has to be running in a distinct
// compartments from DevTools and system scopes (ESMs, XPCOM,...)
const {
useDistinctSystemPrincipalLoader,
releaseDistinctSystemPrincipalLoader,
} = ChromeUtils.importESModule(
);
const requester = {};
const loader = useDistinctSystemPrincipalLoader(requester);
const lineTracer = loader.require(
);
lineTracer.start(globalThis, gTestPath, DEBUG_TRACE_LINE);
registerCleanupFunction(() => {
lineTracer.stop();
releaseDistinctSystemPrincipalLoader(requester);
});
}
const { loader, require } = ChromeUtils.importESModule(
);
const { sinon } = ChromeUtils.importESModule(
);
// When loaded from xpcshell test, this file is loaded via xpcshell.ini's head property
// and so it loaded first before anything else and isn't having access to Services global.
// Whereas many head.js files from mochitest import this file via loadSubScript
// and already expose Services as a global.
const {
gDevTools,
const {
CommandsFactory,
const DevToolsUtils = require("resource://devtools/shared/DevToolsUtils.js");
const KeyShortcuts = require("resource://devtools/client/shared/key-shortcuts.js");
const { LocalizationHelper } = require("resource://devtools/shared/l10n.js");
loader.lazyRequireGetter(
this,
"ResponsiveUIManager",
);
loader.lazyRequireGetter(
this,
"localTypes",
);
loader.lazyRequireGetter(
this,
"ResponsiveMessageHelper",
);
loader.lazyRequireGetter(
this,
"FluentReact",
);
const TEST_DIR = gTestPath.substr(0, gTestPath.lastIndexOf("/"));
const CHROME_URL_ROOT = TEST_DIR + "/";
const URL_ROOT = CHROME_URL_ROOT.replace(
);
const URL_ROOT_SSL = CHROME_URL_ROOT.replace(
);
// Add aliases which make it more explicit that URL_ROOT uses a com TLD.
const URL_ROOT_COM = URL_ROOT;
const URL_ROOT_COM_SSL = URL_ROOT_SSL;
// test Fission scenarios easily.
// Note: example.net is not available for https.
const URL_ROOT_ORG = CHROME_URL_ROOT.replace(
);
const URL_ROOT_ORG_SSL = CHROME_URL_ROOT.replace(
);
const URL_ROOT_NET = CHROME_URL_ROOT.replace(
);
const URL_ROOT_NET_SSL = CHROME_URL_ROOT.replace(
);
// mochi.test:8888 is the actual primary location where files are served.
const URL_ROOT_MOCHI_8888 = CHROME_URL_ROOT.replace(
);
try {
if (isMochitest) {
Services.scriptloader.loadSubScript(
this
);
}
} catch (e) {
ok(
false,
"MISSING DEPENDENCY ON telemetry-test-helpers.js\n" +
"Please add the following line in browser.ini:\n" +
" !/devtools/client/shared/test/telemetry-test-helpers.js\n"
);
throw e;
}
// Force devtools to be initialized so menu items and keyboard shortcuts get installed
// All tests are asynchronous
if (isMochitest) {
waitForExplicitFinish();
}
var EXPECTED_DTU_ASSERT_FAILURE_COUNT = 0;
registerCleanupFunction(function () {
if (
DevToolsUtils.assertionFailureCount !== EXPECTED_DTU_ASSERT_FAILURE_COUNT
) {
ok(
false,
"Should have had the expected number of DevToolsUtils.assert() failures." +
" Expected " +
EXPECTED_DTU_ASSERT_FAILURE_COUNT +
", got " +
DevToolsUtils.assertionFailureCount
);
}
});
// Uncomment this pref to dump all devtools emitted events to the console.
// Services.prefs.setBoolPref("devtools.dump.emit", true);
/**
* Watch console messages for failed propType definitions in React components.
*/
function onConsoleMessage(subject) {
const message = subject.wrappedJSObject.arguments[0];
if (message && /Failed propType/.test(message.toString())) {
ok(false, message);
}
}
const ConsoleAPIStorage = Cc["@mozilla.org/consoleAPI-storage;1"].getService(
Ci.nsIConsoleAPIStorage
);
ConsoleAPIStorage.addLogEventListener(
onConsoleMessage,
Cc["@mozilla.org/systemprincipal;1"].createInstance(Ci.nsIPrincipal)
);
registerCleanupFunction(() => {
ConsoleAPIStorage.removeLogEventListener(onConsoleMessage);
});
Services.prefs.setBoolPref("devtools.inspector.three-pane-enabled", true);
// Disable this preference to reduce exceptions related to pending `listWorkers`
// requests occuring after a process is created/destroyed. See Bug 1620983.
Services.prefs.setBoolPref("dom.ipc.processPrelaunch.enabled", false);
// Disable this preference to capture async stacks across all locations during
// DevTools mochitests. Async stacks provide very valuable information to debug
// intermittents, but come with a performance overhead, which is why they are
// only captured in Debuggees by default.
Services.prefs.setBoolPref(
"javascript.options.asyncstack_capture_debuggee_only",
false
);
// On some Linux platforms, prefers-reduced-motion is enabled, which would
// trigger the notification to be displayed in the toolbox. Dismiss the message
// by default.
Services.prefs.setBoolPref(
"devtools.inspector.simple-highlighters.message-dismissed",
true
);
registerCleanupFunction(() => {
Services.prefs.clearUserPref("devtools.dump.emit");
Services.prefs.clearUserPref("devtools.inspector.three-pane-enabled");
Services.prefs.clearUserPref("dom.ipc.processPrelaunch.enabled");
Services.prefs.clearUserPref("devtools.toolbox.host");
Services.prefs.clearUserPref("devtools.toolbox.previousHost");
Services.prefs.clearUserPref("devtools.toolbox.splitconsole.open");
Services.prefs.clearUserPref("devtools.toolbox.splitconsoleHeight");
Services.prefs.clearUserPref(
"javascript.options.asyncstack_capture_debuggee_only"
);
Services.prefs.clearUserPref(
"devtools.inspector.simple-highlighters.message-dismissed"
);
});
var {
BrowserConsoleManager,
registerCleanupFunction(async function cleanup() {
// Closing the browser console if there's one
const browserConsole = BrowserConsoleManager.getBrowserConsole();
if (browserConsole) {
await safeCloseBrowserConsole({ clearOutput: true });
}
// Close any tab opened by the test.
// There should be only one tab opened by default when firefox starts the test.
while (isMochitest && gBrowser.tabs.length > 1) {
await closeTabAndToolbox(gBrowser.selectedTab);
}
// Note that this will run before cleanup functions registered by tests or other head.js files.
// So all connections must be cleaned up by the test when the test ends,
// before the harness starts invoking the cleanup functions
await waitForTick();
// All connections must be cleaned up by the test when the test ends.
const {
DevToolsServer,
ok(
!DevToolsServer.hasConnection(),
"The main process DevToolsServer has no pending connection when the test ends"
);
// If there is still open connection, close all of them so that following tests
// could pass.
if (DevToolsServer.hasConnection()) {
for (const conn of Object.values(DevToolsServer._connections)) {
conn.close();
}
}
});
async function safeCloseBrowserConsole({ clearOutput = false } = {}) {
const hud = BrowserConsoleManager.getBrowserConsole();
if (!hud) {
return;
}
if (clearOutput) {
info("Clear the browser console output");
const { ui } = hud;
const promises = [ui.once("messages-cleared")];
// If there's an object inspector, we need to wait for the actors to be released.
if (ui.outputNode.querySelector(".object-inspector")) {
promises.push(ui.once("fronts-released"));
}
await ui.clearOutput(true);
await Promise.all(promises);
info("Browser console cleared");
}
info("Wait for all Browser Console targets to be attached");
// It might happen that waitForAllTargetsToBeAttached does not resolve, so we set a
// timeout of 1s before closing
await Promise.race([
waitForAllTargetsToBeAttached(hud.commands.targetCommand),
wait(1000),
]);
info("Close the Browser Console");
await BrowserConsoleManager.closeBrowserConsole();
info("Browser Console closed");
}
/**
* Observer code to register the test actor in every DevTools server which
* starts registering its own actors.
*
* We require immediately the highlighter test actor file, because it will force to load and
* register the front and the spec for HighlighterTestActor. Normally specs and fronts are
* in separate files registered in specs/index.js. But here to simplify the
* setup everything is in the same file and we force to load it here.
*
* DevToolsServer will emit "devtools-server-initialized" after finishing its
* initialization. We watch this observable to add our custom actor.
*
* As a single test may create several DevTools servers, we keep the observer
* alive until the test ends.
*
* To avoid leaks, the observer needs to be removed at the end of each test.
* The test cleanup will send the async message "remove-devtools-highlightertestactor-observer",
* we listen to this message to cleanup the observer.
*/
function highlighterTestActorBootstrap() {
/* eslint-env mozilla/process-script */
const HIGHLIGHTER_TEST_ACTOR_URL =
const { require: _require } = ChromeUtils.importESModule(
);
_require(HIGHLIGHTER_TEST_ACTOR_URL);
const actorRegistryObserver = subject => {
const actorRegistry = subject.wrappedJSObject;
actorRegistry.registerModule(HIGHLIGHTER_TEST_ACTOR_URL, {
prefix: "highlighterTest",
constructor: "HighlighterTestActor",
type: { target: true },
});
};
Services.obs.addObserver(
actorRegistryObserver,
"devtools-server-initialized"
);
const unloadListener = () => {
Services.cpmm.removeMessageListener(
"remove-devtools-testactor-observer",
unloadListener
);
Services.obs.removeObserver(
actorRegistryObserver,
"devtools-server-initialized"
);
};
Services.cpmm.addMessageListener(
"remove-devtools-testactor-observer",
unloadListener
);
}
if (isMochitest) {
const highlighterTestActorBootstrapScript =
"data:,(" + highlighterTestActorBootstrap + ")()";
Services.ppmm.loadProcessScript(
highlighterTestActorBootstrapScript,
// Load this script in all processes (created or to be created)
true
);
registerCleanupFunction(() => {
Services.ppmm.broadcastAsyncMessage("remove-devtools-testactor-observer");
Services.ppmm.removeDelayedProcessScript(
highlighterTestActorBootstrapScript
);
});
}
/**
* Spawn an instance of the highlighter test actor for the given toolbox
*
* @param {Toolbox} toolbox
* @param {Object} options
* @param {Function} options.target: Optional target to get the highlighterTestFront for.
* If not provided, the top level target will be used.
* @returns {HighlighterTestFront}
*/
async function getHighlighterTestFront(toolbox, { target } = {}) {
// Loading the Inspector panel in order to overwrite the TestActor getter for the
// highlighter instance with a method that points to the currently visible
// Box Model Highlighter managed by the Inspector panel.
const inspector = await toolbox.loadTool("inspector");
const highlighterTestFront = await (target || toolbox.target).getFront(
"highlighterTest"
);
// Override the highligher getter with a method to return the active box model
// highlighter. Adaptation for multi-process scenarios where there can be multiple
// highlighters, one per process.
highlighterTestFront.highlighter = () => {
return inspector.highlighters.getActiveHighlighter(
inspector.highlighters.TYPES.BOXMODEL
);
};
return highlighterTestFront;
}
/**
* Spawn an instance of the highlighter test actor for the given tab, when we need the
* highlighter test front before opening or without a toolbox.
*
* @param {Tab} tab
* @returns {HighlighterTestFront}
*/
async function getHighlighterTestFrontWithoutToolbox(tab) {
const commands = await CommandsFactory.forTab(tab);
// Initialize the TargetCommands which require some async stuff to be done
// before being fully ready. This will define the `targetCommand.targetFront` attribute.
await commands.targetCommand.startListening();
const targetFront = commands.targetCommand.targetFront;
return targetFront.getFront("highlighterTest");
}
/**
* Returns a Promise that resolves when all the targets are fully attached.
*
* @param {TargetCommand} targetCommand
*/
function waitForAllTargetsToBeAttached(targetCommand) {
return Promise.allSettled(
targetCommand
.getAllTargets(targetCommand.ALL_TYPES)
.map(target => target.initialized)
);
}
/**
* Add a new test tab in the browser and load the given url.
* @param {String} url The url to be loaded in the new tab
* @param {Object} options Object with various optional fields:
* - {Boolean} background If true, open the tab in background
* - {ChromeWindow} window Firefox top level window we should use to open the tab
* - {Number} userContextId The userContextId of the tab.
* - {String} preferredRemoteType
* - {Boolean} waitForLoad Wait for the page in the new tab to load. (Defaults to true.)
* @return a promise that resolves to the tab object when the url is loaded
*/
async function addTab(url, options = {}) {
info("Adding a new tab with URL: " + url);
const {
background = false,
userContextId,
preferredRemoteType,
waitForLoad = true,
} = options;
const { gBrowser } = options.window ? options.window : window;
const tab = BrowserTestUtils.addTab(gBrowser, url, {
userContextId,
preferredRemoteType,
});
if (!background) {
gBrowser.selectedTab = tab;
}
if (waitForLoad) {
await BrowserTestUtils.browserLoaded(tab.linkedBrowser);
// Waiting for presShell helps with test timeouts in webrender platforms.
await waitForPresShell(tab.linkedBrowser);
info("Tab added and finished loading");
} else {
info("Tab added");
}
return tab;
}
/**
* Remove the given tab.
* @param {Object} tab The tab to be removed.
* @return Promise<undefined> resolved when the tab is successfully removed.
*/
async function removeTab(tab) {
info("Removing tab.");
const { gBrowser } = tab.ownerDocument.defaultView;
const onClose = once(gBrowser.tabContainer, "TabClose");
gBrowser.removeTab(tab);
await onClose;
info("Tab removed and finished closing");
}
/**
* Alias for navigateTo which will reuse the current URI of the provided browser
* to trigger a navigation.
*/
async function reloadBrowser({
browser = gBrowser.selectedBrowser,
isErrorPage = false,
waitForLoad = true,
} = {}) {
return navigateTo(browser.currentURI.spec, {
browser,
isErrorPage,
waitForLoad,
});
}
/**
* Navigate the currently selected tab to a new URL and wait for it to load.
* Also wait for the toolbox to attach to the new target, if we navigated
* to a new process.
*
* @param {String} url The url to be loaded in the current tab.
* @param {JSON} options Optional dictionary object with the following keys:
* - {XULBrowser} browser
* The browser element which should navigate. Defaults to the selected
* browser.
* - {Boolean} isErrorPage
* You may pass `true` if the URL is an error page. Otherwise
* BrowserTestUtils.browserLoaded will wait for 'load' event, which
* never fires for error pages.
* - {Boolean} waitForLoad
* You may pass `false` if the page load is expected to be blocked by
* a script or a breakpoint.
*
* @return a promise that resolves when the page has fully loaded.
*/
async function navigateTo(
uri,
{
browser = gBrowser.selectedBrowser,
isErrorPage = false,
waitForLoad = true,
} = {}
) {
const waitForDevToolsReload = await watchForDevToolsReload(browser, {
isErrorPage,
waitForLoad,
});
uri = uri.replaceAll("\n", "");
info(`Navigating to "${uri}"`);
const onBrowserLoaded = BrowserTestUtils.browserLoaded(
browser,
// includeSubFrames
false,
// resolve on this specific page to load (if null, it would be any page load)
loadedUrl => {
// loadedUrl is encoded, while uri might not be.
return loadedUrl === uri || decodeURI(loadedUrl) === uri;
},
isErrorPage
);
// if we're navigating to the same page we're already on, use reloadTab instead as the
// behavior slightly differs from loadURI (e.g. scroll position isn't keps with the latter).
if (uri === browser.currentURI.spec) {
gBrowser.reloadTab(gBrowser.getTabForBrowser(browser));
} else {
BrowserTestUtils.startLoadingURIString(browser, uri);
}
if (waitForLoad) {
info(`Waiting for page to be loaded…`);
await onBrowserLoaded;
info(`→ page loaded`);
}
await waitForDevToolsReload();
}
/**
* This method should be used to watch for completion of any browser navigation
* performed with a DevTools UI.
*
* It should watch for:
* - Toolbox reload
* - Toolbox commands reload
* - RDM reload
* - RDM commands reload
*
* And it should work both for target switching or old-style navigations.
*
* This method, similarly to all the other watch* navigation methods in this file,
* is async but returns another method which should be called after the navigation
* is done. Browser navigation might be monitored differently depending on the
* situation, so it's up to the caller to handle it as needed.
*
* Typically, this would be used as follows:
* ```
* async function someNavigationHelper(browser) {
* const waitForDevToolsFn = await watchForDevToolsReload(browser);
*
* // This step should wait for the load to be completed from the browser's
* // point of view, so that waitForDevToolsFn can compare pIds, browsing
* // contexts etc... and check if we should expect a target switch
* await performBrowserNavigation(browser);
*
* await waitForDevToolsFn();
* }
* ```
*/
async function watchForDevToolsReload(
browser,
{ isErrorPage = false, waitForLoad = true } = {}
) {
const waitForToolboxReload = await _watchForToolboxReload(browser, {
isErrorPage,
waitForLoad,
});
const waitForResponsiveReload = await _watchForResponsiveReload(browser, {
isErrorPage,
waitForLoad,
});
return async function () {
info("Wait for the toolbox to reload");
await waitForToolboxReload();
info("Wait for Responsive UI to reload");
await waitForResponsiveReload();
};
}
/**
* Start watching for the toolbox reload to be completed:
* - watch for the toolbox's commands to be fully reloaded
* - watch for the toolbox's current panel to be reloaded
*/
async function _watchForToolboxReload(
browser,
{ isErrorPage, waitForLoad } = {}
) {
const tab = gBrowser.getTabForBrowser(browser);
const toolbox = gDevTools.getToolboxForTab(tab);
if (!toolbox) {
// No toolbox to wait for
return function () {};
}
const waitForCurrentPanelReload = watchForCurrentPanelReload(toolbox);
const waitForToolboxCommandsReload = await watchForCommandsReload(
toolbox.commands,
{ isErrorPage, waitForLoad }
);
const checkTargetSwitching = await watchForTargetSwitching(
toolbox.commands,
browser
);
return async function () {
const isTargetSwitching = checkTargetSwitching();
info(`Waiting for toolbox commands to be reloaded…`);
await waitForToolboxCommandsReload(isTargetSwitching);
// TODO: We should wait for all loaded panels to reload here, because some
// of them might still perform background updates.
if (waitForCurrentPanelReload) {
info(`Waiting for ${toolbox.currentToolId} to be reloaded…`);
await waitForCurrentPanelReload();
info(`→ panel reloaded`);
}
};
}
/**
* Start watching for Responsive UI (RDM) reload to be completed:
* - watch for the Responsive UI's commands to be fully reloaded
* - watch for the Responsive UI's target switch to be done
*/
async function _watchForResponsiveReload(
browser,
{ isErrorPage, waitForLoad } = {}
) {
const tab = gBrowser.getTabForBrowser(browser);
const ui = ResponsiveUIManager.getResponsiveUIForTab(tab);
if (!ui) {
// No responsive UI to wait for
return function () {};
}
const onResponsiveTargetSwitch = ui.once("responsive-ui-target-switch-done");
const waitForResponsiveCommandsReload = await watchForCommandsReload(
ui.commands,
{ isErrorPage, waitForLoad }
);
const checkTargetSwitching = await watchForTargetSwitching(
ui.commands,
browser
);
return async function () {
const isTargetSwitching = checkTargetSwitching();
info(`Waiting for responsive ui commands to be reloaded…`);
await waitForResponsiveCommandsReload(isTargetSwitching);
if (isTargetSwitching) {
await onResponsiveTargetSwitch;
}
};
}
/**
* Watch for the current panel selected in the provided toolbox to be reloaded.
* Some panels implement custom events that should be expected for every reload.
*
* Note about returning a method instead of a promise:
* In general this pattern is useful so that we can check if a target switch
* occurred or not, and decide which events to listen for. So far no panel is
* behaving differently whether there was a target switch or not. But to remain
* consistent with other watch* methods we still return a function here.
*
* @param {Toolbox}
* The Toolbox instance which is going to experience a reload
* @return {function} An async method to be called and awaited after the reload
* started. Will return `null` for panels which don't implement any
* specific reload event.
*/
function watchForCurrentPanelReload(toolbox) {
return _watchForPanelReload(toolbox, toolbox.currentToolId);
}
/**
* Watch for all the panels loaded in the provided toolbox to be reloaded.
* Some panels implement custom events that should be expected for every reload.
*
* Note about returning a method instead of a promise:
* See comment for watchForCurrentPanelReload
*
* @param {Toolbox}
* The Toolbox instance which is going to experience a reload
* @return {function} An async method to be called and awaited after the reload
* started.
*/
function watchForLoadedPanelsReload(toolbox) {
const waitForPanels = [];
for (const [id] of toolbox.getToolPanels()) {
// Store a watcher method for each panel already loaded.
waitForPanels.push(_watchForPanelReload(toolbox, id));
}
return function () {
return Promise.all(
waitForPanels.map(async watchPanel => {
// Wait for all panels to be reloaded.
if (watchPanel) {
await watchPanel();
}
})
);
};
}
function _watchForPanelReload(toolbox, toolId) {
const panel = toolbox.getPanel(toolId);
if (toolId == "inspector") {
const markuploaded = panel.once("markuploaded");
const onNewRoot = panel.once("new-root");
const onUpdated = panel.once("inspector-updated");
const onReloaded = panel.once("reloaded");
return async function () {
info("Waiting for markup view to load after navigation.");
await markuploaded;
info("Waiting for new root.");
await onNewRoot;
info("Waiting for inspector to update after new-root event.");
await onUpdated;
info("Waiting for inspector updates after page reload");
await onReloaded;
};
} else if (
["netmonitor", "accessibility", "webconsole", "jsdebugger"].includes(toolId)
) {
const onReloaded = panel.once("reloaded");
return async function () {
info(`Waiting for ${toolId} updates after page reload`);
await onReloaded;
};
}
return null;
}
/**
* Watch for a Commands instance to be reloaded after a navigation.
*
* As for other navigation watch* methods, this should be called before the
* navigation starts, and the function it returns should be called after the
* navigation is done from a Browser point of view.
*
* !!! The wait function expects a `isTargetSwitching` argument to be provided,
* which needs to be monitored using watchForTargetSwitching !!!
*/
async function watchForCommandsReload(
commands,
{ isErrorPage = false, waitForLoad = true } = {}
) {
// If we're switching origins, we need to wait for the 'switched-target'
// event to make sure everything is ready.
// Navigating from/to pages loaded in the parent process, like about:robots,
// also spawn new targets.
// (If target switching is disabled, the toolbox will reboot)
const onTargetSwitched = commands.targetCommand.once("switched-target");
// Wait until we received a page load resource:
// - dom-complete if we can wait for a full page load
// - dom-loading otherwise
// This allows to wait for page load for consumers calling directly
// waitForDevTools instead of navigateTo/reloadBrowser.
// This is also useful as an alternative to target switching, when no target
// switch is supposed to happen.
const waitForCompleteLoad = waitForLoad && !isErrorPage;
const documentEventName = waitForCompleteLoad
? "dom-complete"
: "dom-loading";
const { onResource: onTopLevelDomEvent } =
await commands.resourceCommand.waitForNextResource(
commands.resourceCommand.TYPES.DOCUMENT_EVENT,
{
ignoreExistingResources: true,
predicate: resource =>
resource.targetFront.isTopLevel &&
resource.name === documentEventName,
}
);
return async function (isTargetSwitching) {
if (typeof isTargetSwitching === "undefined") {
throw new Error("isTargetSwitching was not provided to the wait method");
}
if (isTargetSwitching) {
info(`Waiting for target switch…`);
await onTargetSwitched;
info(`→ switched-target emitted`);
}
info(`Waiting for '${documentEventName}' resource…`);
await onTopLevelDomEvent;
info(`→ '${documentEventName}' resource emitted`);
return isTargetSwitching;
};
}
/**
* Watch if an upcoming navigation will trigger a target switching, for the
* provided Commands instance and the provided Browser.
*
* As for other navigation watch* methods, this should be called before the
* navigation starts, and the function it returns should be called after the
* navigation is done from a Browser point of view.
*/
async function watchForTargetSwitching(commands, browser) {
browser = browser || gBrowser.selectedBrowser;
const currentPID = browser.browsingContext.currentWindowGlobal.osPid;
const currentBrowsingContextID = browser.browsingContext.id;
// If the current top-level target follows the window global lifecycle, a
// target switch will occur regardless of process changes.
const targetFollowsWindowLifecycle =
commands.targetCommand.targetFront.targetForm.followWindowGlobalLifeCycle;
return function () {
// Compare the PIDs (and not the toolbox's targets) as PIDs are updated also immediately,
// while target may be updated slightly later.
const switchedProcess =
currentPID !== browser.browsingContext.currentWindowGlobal.osPid;
const switchedBrowsingContext =
currentBrowsingContextID !== browser.browsingContext.id;
return (
targetFollowsWindowLifecycle || switchedProcess || switchedBrowsingContext
);
};
}
/**
* Create a Target for the provided tab and attach to it before resolving.
* This should only be used for tests which don't involve the frontend or a
* toolbox. Typically, retrieving the target and attaching to it should be
* handled at framework level when a Toolbox is used.
*
* @param {XULTab} tab
* The tab for which a target should be created.
* @return {WindowGlobalTargetFront} The attached target front.
*/
async function createAndAttachTargetForTab(tab) {
info("Creating and attaching to a local tab target");
const commands = await CommandsFactory.forTab(tab);
// Initialize the TargetCommands which require some async stuff to be done
// before being fully ready. This will define the `targetCommand.targetFront` attribute.
await commands.targetCommand.startListening();
const target = commands.targetCommand.targetFront;
return target;
}
function isFissionEnabled() {
return SpecialPowers.useRemoteSubframes;
}
function isEveryFrameTargetEnabled() {
return Services.prefs.getBoolPref(
"devtools.every-frame-target.enabled",
false
);
}
/**
* Open the inspector in a tab with given URL.
* @param {string} url The URL to open.
* @param {String} hostType Optional hostType, as defined in Toolbox.HostType
* @return A promise that is resolved once the tab and inspector have loaded
* with an object: { tab, toolbox, inspector, highlighterTestFront }.
*/
async function openInspectorForURL(url, hostType) {
const tab = await addTab(url);
const { inspector, toolbox, highlighterTestFront } = await openInspector(
hostType
);
return { tab, inspector, toolbox, highlighterTestFront };
}
function getActiveInspector() {
const toolbox = gDevTools.getToolboxForTab(gBrowser.selectedTab);
return toolbox.getPanel("inspector");
}
/**
* Simulate a key event from an electron key shortcut string:
*
* @param {String} key
* @param {DOMWindow} target
* Optional window where to fire the key event
*/
function synthesizeKeyShortcut(key, target) {
// parseElectronKey requires any window, just to access `KeyboardEvent`
const window = Services.appShell.hiddenDOMWindow;
const shortcut = KeyShortcuts.parseElectronKey(window, key);
const keyEvent = {
altKey: shortcut.alt,
ctrlKey: shortcut.ctrl,
metaKey: shortcut.meta,
shiftKey: shortcut.shift,
};
if (shortcut.keyCode) {
keyEvent.keyCode = shortcut.keyCode;
}
info("Synthesizing key shortcut: " + key);
EventUtils.synthesizeKey(shortcut.key || "", keyEvent, target);
}
var waitForTime = DevToolsUtils.waitForTime;
/**
* Wait for a tick.
* @return {Promise}
*/
function waitForTick() {
return new Promise(resolve => DevToolsUtils.executeSoon(resolve));
}
/**
* This shouldn't be used in the tests, but is useful when writing new tests or
* debugging existing tests in order to introduce delays in the test steps
*
* @param {Number} ms
* The time to wait
* @return A promise that resolves when the time is passed
*/
function wait(ms) {
return new Promise(resolve => {
setTimeout(resolve, ms);
info("Waiting " + ms / 1000 + " seconds.");
});
}
/**
* Wait for a predicate to return a result.
*
* @param function condition
* Invoked once in a while until it returns a truthy value. This should be an
* idempotent function, since we have to run it a second time after it returns
* true in order to return the value.
* @param string message [optional]
* A message to output if the condition fails.
* @param number interval [optional]
* How often the predicate is invoked, in milliseconds.
* Can be set globally for a test via `waitFor.overrideIntervalForTestFile = someNumber;`.
* @param number maxTries [optional]
* How many times the predicate is invoked before timing out.
* Can be set globally for a test via `waitFor.overrideMaxTriesForTestFile = someNumber;`.
* @return object
* A promise that is resolved with the result of the condition.
*/
async function waitFor(condition, message = "", interval = 10, maxTries = 500) {
// Update interval & maxTries if overrides are defined on the waitFor object.
interval =
typeof waitFor.overrideIntervalForTestFile !== "undefined"
? waitFor.overrideIntervalForTestFile
: interval;
maxTries =
typeof waitFor.overrideMaxTriesForTestFile !== "undefined"
? waitFor.overrideMaxTriesForTestFile
: maxTries;
try {
const value = await BrowserTestUtils.waitForCondition(
condition,
message,
interval,
maxTries
);
return value;
} catch (e) {
const errorMessage = `Failed waitFor(): ${message} \nFailed condition: ${condition} \nException Message: ${e}`;
throw new Error(errorMessage);
}
}
/**
* Wait for eventName on target to be delivered a number of times.
*
* @param {Object} target
* An observable object that either supports on/off or
* addEventListener/removeEventListener
* @param {String} eventName
* @param {Number} numTimes
* Number of deliveries to wait for.
* @param {Boolean} useCapture
* Optional, for addEventListener/removeEventListener
* @return A promise that resolves when the event has been handled
*/
function waitForNEvents(target, eventName, numTimes, useCapture = false) {
info("Waiting for event: '" + eventName + "' on " + target + ".");
let count = 0;
return new Promise(resolve => {
for (const [add, remove] of [
["on", "off"],
["addEventListener", "removeEventListener"],
["addListener", "removeListener"],
["addMessageListener", "removeMessageListener"],
]) {
if (add in target && remove in target) {
target[add](
eventName,
function onEvent(...args) {
if (typeof info === "function") {
info("Got event: '" + eventName + "' on " + target + ".");
}
if (++count == numTimes) {
target[remove](eventName, onEvent, useCapture);
resolve(...args);
}
},
useCapture
);
break;
}
}
});
}
/**
* Wait for DOM change on target.
*
* @param {Object} target
* The Node on which to observe DOM mutations.
* @param {String} selector
* Given a selector to watch whether the expected element is changed
* on target.
* @param {Number} expectedLength
* Optional, default set to 1
* There may be more than one element match an array match the selector,
* give an expected length to wait for more elements.
* @return A promise that resolves when the event has been handled
*/
function waitForDOM(target, selector, expectedLength = 1) {
return new Promise(resolve => {
const observer = new MutationObserver(mutations => {
mutations.forEach(mutation => {
const elements = mutation.target.querySelectorAll(selector);
if (elements.length === expectedLength) {
observer.disconnect();
resolve(elements);
}
});
});
observer.observe(target, {
attributes: true,
childList: true,
subtree: true,
});
});
}
/**
* Wait for eventName on target.
*
* @param {Object} target
* An observable object that either supports on/off or
* addEventListener/removeEventListener
* @param {String} eventName
* @param {Boolean} useCapture
* Optional, for addEventListener/removeEventListener
* @return A promise that resolves when the event has been handled
*/
function once(target, eventName, useCapture = false) {
return waitForNEvents(target, eventName, 1, useCapture);
}
/**
* Some tests may need to import one or more of the test helper scripts.
* A test helper script is simply a js file that contains common test code that
* is either not common-enough to be in head.js, or that is located in a
* separate directory.
* The script will be loaded synchronously and in the test's scope.
* @param {String} filePath The file path, relative to the current directory.
* Examples:
* - "helper_attributes_test_runner.js"
*/
function loadHelperScript(filePath) {
const testDir = gTestPath.substr(0, gTestPath.lastIndexOf("/"));
Services.scriptloader.loadSubScript(testDir + "/" + filePath, this);
}
/**
* Open the toolbox in a given tab.
* @param {XULNode} tab The tab the toolbox should be opened in.
* @param {String} toolId Optional. The ID of the tool to be selected.
* @param {String} hostType Optional. The type of toolbox host to be used.
* @return {Promise} Resolves with the toolbox, when it has been opened.
*/
async function openToolboxForTab(tab, toolId, hostType) {
info("Opening the toolbox");
// Check if the toolbox is already loaded.
let toolbox = gDevTools.getToolboxForTab(tab);
if (toolbox) {
if (!toolId || (toolId && toolbox.getPanel(toolId))) {
info("Toolbox is already opened");
return toolbox;
}
}
// If not, load it now.
toolbox = await gDevTools.showToolboxForTab(tab, { toolId, hostType });
// Make sure that the toolbox frame is focused.
await new Promise(resolve => waitForFocus(resolve, toolbox.win));
info("Toolbox opened and focused");
return toolbox;
}
/**
* Add a new tab and open the toolbox in it.
* @param {String} url The URL for the tab to be opened.
* @param {String} toolId Optional. The ID of the tool to be selected.
* @param {String} hostType Optional. The type of toolbox host to be used.
* @return {Promise} Resolves when the tab has been added, loaded and the
* toolbox has been opened. Resolves to the toolbox.
*/
async function openNewTabAndToolbox(url, toolId, hostType) {
const tab = await addTab(url);
return openToolboxForTab(tab, toolId, hostType);
}
/**
* Close a tab and if necessary, the toolbox that belongs to it
* @param {Tab} tab The tab to close.
* @return {Promise} Resolves when the toolbox and tab have been destroyed and
* closed.
*/
async function closeTabAndToolbox(tab = gBrowser.selectedTab) {
if (gDevTools.hasToolboxForTab(tab)) {
await gDevTools.closeToolboxForTab(tab);
}
await removeTab(tab);
await new Promise(resolve => setTimeout(resolve, 0));
}
/**
* Close a toolbox and the current tab.
* @param {Toolbox} toolbox The toolbox to close.
* @return {Promise} Resolves when the toolbox and tab have been destroyed and
* closed.
*/
async function closeToolboxAndTab(toolbox) {
await toolbox.destroy();
await removeTab(gBrowser.selectedTab);
}
/**
* Waits until a predicate returns true.
*
* @param function predicate
* Invoked once in a while until it returns true.
* @param number interval [optional]
* How often the predicate is invoked, in milliseconds.
*/
function waitUntil(predicate, interval = 10) {
if (predicate()) {
return Promise.resolve(true);
}
return new Promise(resolve => {
setTimeout(function () {
waitUntil(predicate, interval).then(() => resolve(true));
}, interval);
});
}
/**
* Variant of waitUntil that accepts a predicate returning a promise.
*/
async function asyncWaitUntil(predicate, interval = 10) {
let success = await predicate();
while (!success) {
// Wait for X milliseconds.
await new Promise(resolve => setTimeout(resolve, interval));
// Test the predicate again.
success = await predicate();
}
}
/**
* Wait for a context menu popup to open.
*
* @param Element popup
* The XUL popup you expect to open.
* @param Element button
* The button/element that receives the contextmenu event. This is
* expected to open the popup.
* @param function onShown
* Function to invoke on popupshown event.
* @param function onHidden
* Function to invoke on popuphidden event.
* @return object
* A Promise object that is resolved after the popuphidden event
* callback is invoked.
*/
function waitForContextMenu(popup, button, onShown, onHidden) {
return new Promise(resolve => {
function onPopupShown() {
info("onPopupShown");
popup.removeEventListener("popupshown", onPopupShown);
onShown && onShown();
// Use executeSoon() to get out of the popupshown event.
popup.addEventListener("popuphidden", onPopupHidden);
DevToolsUtils.executeSoon(() => popup.hidePopup());
}
function onPopupHidden() {
info("onPopupHidden");
popup.removeEventListener("popuphidden", onPopupHidden);
onHidden && onHidden();
resolve(popup);
}
popup.addEventListener("popupshown", onPopupShown);
info("wait for the context menu to open");
synthesizeContextMenuEvent(button);
});
}
function synthesizeContextMenuEvent(el) {
el.scrollIntoView();
const eventDetails = { type: "contextmenu", button: 2 };
EventUtils.synthesizeMouse(
el,
5,
2,
eventDetails,
el.ownerDocument.defaultView
);
}
/**
* Promise wrapper around SimpleTest.waitForClipboard
*/
function waitForClipboardPromise(setup, expected) {
return new Promise((resolve, reject) => {
SimpleTest.waitForClipboard(expected, setup, resolve, reject);
});
}
/**
* Simple helper to push a temporary preference. Wrapper on SpecialPowers
* pushPrefEnv that returns a promise resolving when the preferences have been
* updated.
*
* @param {String} preferenceName
* The name of the preference to updated
* @param {} value
* The preference value, type can vary
* @return {Promise} resolves when the preferences have been updated
*/
function pushPref(preferenceName, value) {
const options = { set: [[preferenceName, value]] };
return SpecialPowers.pushPrefEnv(options);
}
async function closeToolbox() {
await gDevTools.closeToolboxForTab(gBrowser.selectedTab);
}
/**
* Clean the logical clipboard content. This method only clears the OS clipboard on
* Windows (see Bug 666254).
*/
function emptyClipboard() {
const clipboard = Services.clipboard;
clipboard.emptyClipboard(clipboard.kGlobalClipboard);
}
/**
* Check if the current operating system is Windows.
*/
function isWindows() {
return Services.appinfo.OS === "WINNT";
}
/**
* Create an HTTP server that can be used to simulate custom requests within
* a test. It is automatically cleaned up when the test ends, so no need to
* call `destroy`.
*
* for more information about how to register handlers.
*
* The server can be accessed like:
*
* const server = createTestHTTPServer();
* let url = "http://localhost: " + server.identity.primaryPort + "/path";
* @returns {HttpServer}
*/
function createTestHTTPServer() {
const { HttpServer } = ChromeUtils.importESModule(
);
const server = new HttpServer();
registerCleanupFunction(async function cleanup() {
await new Promise(resolve => server.stop(resolve));
});
server.start(-1);
return server;
}
/*
* Register an actor in the content process of the current tab.
*
* Calling ActorRegistry.registerModule only registers the actor in the current process.
* As all test scripts are ran in the parent process, it is only registered here.
* This function helps register them in the content process used for the current tab.
*
* @param {string} url
* Actor module URL or absolute require path
* @param {json} options
* Arguments to be passed to DevToolsServer.registerModule
*/
async function registerActorInContentProcess(url, options) {
function convertChromeToFile(uri) {
return Cc["@mozilla.org/chrome/chrome-registry;1"]
.getService(Ci.nsIChromeRegistry)
.convertChromeURL(Services.io.newURI(uri)).spec;
}
// chrome://mochitests URI is registered only in the parent process, so convert these
// URLs to file:// one in order to work in the content processes
url = url.startsWith("chrome://mochitests") ? convertChromeToFile(url) : url;
return SpecialPowers.spawn(
gBrowser.selectedBrowser,
[{ url, options }],
args => {
// eslint-disable-next-line no-shadow
const { require } = ChromeUtils.importESModule(
);
const {
ActorRegistry,
ActorRegistry.registerModule(args.url, args.options);
}
);
}
/**
* Move the provided Window to the provided left, top coordinates and wait for
* the window position to be updated.
*/
async function moveWindowTo(win, left, top) {
// Check that the expected coordinates are within the window available area.
left = Math.max(win.screen.availLeft, left);
left = Math.min(win.screen.width, left);
top = Math.max(win.screen.availTop, top);
top = Math.min(win.screen.height, top);
info(`Moving window to {${left}, ${top}}`);
win.moveTo(left, top);
// Bug 1600809: window move/resize can be async on Linux sometimes.
// Wait so that the anchor's position is correctly measured.
return waitUntil(() => {
info(
`Wait for window screenLeft and screenTop to be updated: (${win.screenLeft}, ${win.screenTop})`
);
return win.screenLeft === left && win.screenTop === top;
});
}
function getCurrentTestFilePath() {
return gTestPath.replace("chrome://mochitests/content/browser/", "");
}
/**
* Unregister all registered service workers.
*
* @param {DevToolsClient} client
*/
async function unregisterAllServiceWorkers(client) {
info("Wait until all workers have a valid registrationFront");
let workers;
await asyncWaitUntil(async function () {
workers = await client.mainRoot.listAllWorkers();
const allWorkersRegistered = workers.service.every(
worker => !!worker.registrationFront
);
return allWorkersRegistered;
});
info("Unregister all service workers");
const promises = [];
for (const worker of workers.service) {
promises.push(worker.registrationFront.unregister());
}
await Promise.all(promises);
}
/**********************
* Screenshot helpers *
**********************/
/**
* Returns an object containing the r,g and b colors of the provided image at
* the passed position
*
* @param {Image} image
* @param {Int} x
* @param {Int} y
* @returns Object with the following properties:
* - {Int} r: The red component of the pixel
* - {Int} g: The green component of the pixel
* - {Int} b: The blue component of the pixel
*/
function colorAt(image, x, y) {
// Create a test canvas element.
const HTML_NS = "http://www.w3.org/1999/xhtml";
const canvas = document.createElementNS(HTML_NS, "canvas");
canvas.width = image.width;
canvas.height = image.height;
// Draw the image in the canvas
const context = canvas.getContext("2d");
context.drawImage(image, 0, 0, image.width, image.height);
// Return the color found at the provided x,y coordinates as a "r, g, b" string.
const [r, g, b] = context.getImageData(x, y, 1, 1).data;
return { r, g, b };
}
let allDownloads = [];
/**
* Returns a Promise that resolves when a new screenshot is available in the download folder.
*
* @param {Object} [options]
* @param {Boolean} options.isWindowPrivate: Set to true if the window from which the screenshot
* is taken is a private window. This will ensure that we check that the
* screenshot appears in the private window, not the non-private one (See Bug 1783373)
*/
async function waitUntilScreenshot({ isWindowPrivate = false } = {}) {
const { Downloads } = ChromeUtils.importESModule(
);
const list = await Downloads.getList(Downloads.ALL);
return new Promise(function (resolve) {
const view = {
onDownloadAdded: async download => {
await download.whenSucceeded();
if (allDownloads.includes(download)) {
return;
}
is(
!!download.source.isPrivate,
isWindowPrivate,
`The download occured in the expected${
isWindowPrivate ? " private" : ""
} window`
);
allDownloads.push(download);
resolve(download.target.path);
list.removeView(view);
},
};
list.addView(view);
});
}
/**
* Clear all the download references.
*/
async function resetDownloads() {
info("Reset downloads");
const { Downloads } = ChromeUtils.importESModule(
);
const downloadList = await Downloads.getList(Downloads.ALL);
const downloads = await downloadList.getAll();
for (const download of downloads) {
downloadList.remove(download);
await download.finalize(true);
}
allDownloads = [];
}
/**
* Return a screenshot of the currently selected node in the inspector (using the internal
* Inspector#screenshotNode method).
*
* @param {Inspector} inspector
* @returns {Image}
*/
async function takeNodeScreenshot(inspector) {
// Cleanup all downloads at the end of the test.
registerCleanupFunction(resetDownloads);
info(
"Call screenshotNode() and wait until the screenshot is found in the Downloads"
);
const whenScreenshotSucceeded = waitUntilScreenshot();
inspector.screenshotNode();
const filePath = await whenScreenshotSucceeded;
info("Create an image using the downloaded fileas source");
const image = new Image();
const onImageLoad = once(image, "load");
image.src = PathUtils.toFileURI(filePath);
await onImageLoad;
info("Remove the downloaded screenshot file");
await IOUtils.remove(filePath);
// See intermittent Bug 1508435. Even after removing the file, tests still manage to
// reuse files from the previous test if they have the same name. Since our file name
// is based on a timestamp that has "second" precision, wait for one second to make sure
// screenshots will have different names.
info(
"Wait for one second to make sure future screenshots will use a different name"
);
await new Promise(r => setTimeout(r, 1000));
return image;
}
/**
* Check that the provided image has the expected width, height, and color.
* NOTE: This test assumes that the image is only made of a single color and will only
* check one pixel.
*/
async function assertSingleColorScreenshotImage(
image,
width,
height,
{ r, g, b }
) {
info(`Assert ${image.src} content`);
const ratio = await SpecialPowers.spawn(
gBrowser.selectedBrowser,
[],
() => content.wrappedJSObject.devicePixelRatio
);
is(
image.width,
ratio * width,
`node screenshot has the expected width (dpr = ${ratio})`
);
is(
image.height,
height * ratio,
`node screenshot has the expected height (dpr = ${ratio})`
);
const color = colorAt(image, 0, 0);
is(color.r, r, "node screenshot has the expected red component");
is(color.g, g, "node screenshot has the expected green component");
is(color.b, b, "node screenshot has the expected blue component");
}
/**
* Check that the provided image has the expected color at a given position
*/
function checkImageColorAt({ image, x = 0, y, expectedColor, label }) {
const color = colorAt(image, x, y);
is(`rgb(${Object.values(color).join(", ")})`, expectedColor, label);
}
/**
* Wait until the store has reached a state that matches the predicate.
* @param Store store
* The Redux store being used.
* @param function predicate
* A function that returns true when the store has reached the expected
* state.
* @return Promise
* Resolved once the store reaches the expected state.
*/
function waitUntilState(store, predicate) {
return new Promise(resolve => {
const unsubscribe = store.subscribe(check);
info(`Waiting for state predicate "${predicate}"`);
function check() {
if (predicate(store.getState())) {
info(`Found state predicate "${predicate}"`);
unsubscribe();
resolve();
}
}
// Fire the check immediately in case the action has already occurred
check();
});
}
/**
* Wait for a specific action type to be dispatched.
*
* If the action is async and defines a `status` property, this helper will wait
* for the status to reach either "error" or "done".
*
* @param {Object} store
* Redux store where the action should be dispatched.
* @param {String} actionType
* The actionType to wait for.
* @param {Number} repeat
* Optional, number of time the action is expected to be dispatched.
* Defaults to 1
* @return {Promise}
*/
function waitForDispatch(store, actionType, repeat = 1) {
let count = 0;
return new Promise(resolve => {
store.dispatch({
type: "@@service/waitUntil",
predicate: action => {
const isDone =
!action.status ||
action.status === "done" ||
action.status === "error";
if (action.type === actionType && isDone && ++count == repeat) {
return true;
}
return false;
},
run: (dispatch, getState, action) => {
resolve(action);
},
});
});
}
/**
* Retrieve a browsing context in nested frames.
*
* @param {BrowsingContext|XULBrowser} browsingContext
* The topmost browsing context under which we should search for the
* browsing context.
* @param {Array<String>} selectors
* Array of CSS selectors that form a path to a specific nested frame.
* @return {BrowsingContext} The nested browsing context.
*/
async function getBrowsingContextInFrames(browsingContext, selectors) {
let context = browsingContext;
if (!Array.isArray(selectors)) {
throw new Error(
"getBrowsingContextInFrames called with an invalid selectors argument"
);
}
if (selectors.length === 0) {
throw new Error(
"getBrowsingContextInFrames called with an empty selectors array"
);
}
const clonedSelectors = [...selectors];
while (clonedSelectors.length) {
const selector = clonedSelectors.shift();
context = await SpecialPowers.spawn(context, [selector], _selector => {
return content.document.querySelector(_selector).browsingContext;
});
}
return context;
}
/**
* Synthesize a mouse event on an element, after ensuring that it is visible
* in the viewport.
*
* @param {String|Array} selector: The node selector to get the node target for the event.
* To target an element in a specific iframe, pass an array of CSS selectors
* (e.g. ["iframe", ".el-in-iframe"])
* @param {number} x
* @param {number} y
* @param {object} options: Options that will be passed to BrowserTestUtils.synthesizeMouse
*/
async function safeSynthesizeMouseEventInContentPage(
selector,
x,
y,
options = {}
) {
let context = gBrowser.selectedBrowser.browsingContext;
// If an array of selector is passed, we need to retrieve the context in which the node
// lives in.
if (Array.isArray(selector)) {
if (selector.length === 1) {
selector = selector[0];
} else {
context = await getBrowsingContextInFrames(
context,
// only pass the iframe path
selector.slice(0, -1)
);
// retrieve the last item of the selector, which should be the one for the node we want.
selector = selector.at(-1);
}
}
await scrollContentPageNodeIntoView(context, selector);
BrowserTestUtils.synthesizeMouse(selector, x, y, options, context);
}
/**
* Synthesize a mouse event at the center of an element, after ensuring that it is visible
* in the viewport.
*
* @param {String|Array} selector: The node selector to get the node target for the event.
* To target an element in a specific iframe, pass an array of CSS selectors
* (e.g. ["iframe", ".el-in-iframe"])
* @param {object} options: Options that will be passed to BrowserTestUtils.synthesizeMouse
*/
async function safeSynthesizeMouseEventAtCenterInContentPage(
selector,
options = {}
) {
let context = gBrowser.selectedBrowser.browsingContext;
// If an array of selector is passed, we need to retrieve the context in which the node
// lives in.
if (Array.isArray(selector)) {
if (selector.length === 1) {
selector = selector[0];
} else {
context = await getBrowsingContextInFrames(
context,
// only pass the iframe path
selector.slice(0, -1)
);
// retrieve the last item of the selector, which should be the one for the node we want.
selector = selector.at(-1);
}
}
await scrollContentPageNodeIntoView(context, selector);
BrowserTestUtils.synthesizeMouseAtCenter(selector, options, context);
}
/**
* Scroll into view an element in the content page matching the passed selector
*
* @param {BrowsingContext} browsingContext: The browsing context the element lives in.
* @param {String} selector: The node selector to get the node to scroll into view
* @returns {Promise}
*/
function scrollContentPageNodeIntoView(browsingContext, selector) {
return SpecialPowers.spawn(
browsingContext,
[selector],
function (innerSelector) {
const node =
content.wrappedJSObject.document.querySelector(innerSelector);
node.scrollIntoView();
}
);
}
/**
* Change the zoom level of the selected page.
*
* @param {Number} zoomLevel
*/
function setContentPageZoomLevel(zoomLevel) {
gBrowser.selectedBrowser.fullZoom = zoomLevel;
}
/**
* Wait for the next DOCUMENT_EVENT dom-complete resource on a top-level target
*
* @param {Object} commands
* @return {Promise<Object>}
* Return a promise which resolves once we fully settle the resource listener.
* You should await for its resolution before doing the action which may fire
* your resource.
* This promise will resolve with an object containing a `onDomCompleteResource` property,
* which is also a promise, that will resolve once a "top-level" DOCUMENT_EVENT dom-complete
* is received.
*/
async function waitForNextTopLevelDomCompleteResource(commands) {
const { onResource: onDomCompleteResource } =
await commands.resourceCommand.waitForNextResource(
commands.resourceCommand.TYPES.DOCUMENT_EVENT,
{
ignoreExistingResources: true,
predicate: resource =>
resource.name === "dom-complete" && resource.targetFront.isTopLevel,
}
);
return { onDomCompleteResource };
}
/**
* Wait for the provided context to have a valid presShell. This can be useful
* for tests which try to create popup panels or interact with the document very
* early.
*
* @param {BrowsingContext} context
**/
function waitForPresShell(context) {
return SpecialPowers.spawn(context, [], async () => {
const winUtils = SpecialPowers.getDOMWindowUtils(content);
await ContentTaskUtils.waitForCondition(() => {
try {
return !!winUtils.getPresShellId();
} catch (e) {
return false;
}
}, "Waiting for a valid presShell");
});
}
/**
* In tests using Fluent localization, it is preferable to match DOM elements using
* a message ID rather than the raw string as:
*
* 1. It allows testing infrastructure to be multilingual if needed.
* 2. It isolates the tests from localization changes.
*
* @param {Array<string>} resourceIds A list of .ftl files to load.
* @returns {(id: string, args?: Record<string, FluentVariable>) => string}
*/
async function getFluentStringHelper(resourceIds) {
const locales = Services.locale.appLocalesAsBCP47;
const generator = L10nRegistry.getInstance().generateBundles(
locales,
resourceIds
);
const bundles = [];
for await (const bundle of generator) {
bundles.push(bundle);
}
const reactLocalization = new FluentReact.ReactLocalization(bundles);
/**
* Get the string from a message id. It throws when the message is not found.
*
* @param {string} id
* @param {string} attributeName: attribute name if you need to access a specific attribute
* defined in the fluent string, e.g. setting "title" for this param
* will retrieve the `title` string in
* compatibility-issue-browsers-list =
* .title = This is the title
* @param {Record<string, FluentVariable>} [args] optional
* @returns {string}
*/
return (id, attributeName, args) => {
let string;
if (!attributeName) {
string = reactLocalization.getString(id, args);
} else {
for (const bundle of reactLocalization.bundles) {
const msg = bundle.getMessage(id);
if (msg?.attributes[attributeName]) {
string = bundle.formatPattern(
msg.attributes[attributeName],
args,
[]
);
break;
}
}
}
if (!string) {
throw new Error(
`Could not find a string for "${id}"${
attributeName ? ` and attribute "${attributeName}")` : ""
}. Was the correct resource bundle loaded?`
);
}
return string;
};
}
/**
* Open responsive design mode for the given tab.
*/
async function openRDM(tab, { waitForDeviceList = true } = {}) {
info("Opening responsive design mode");
const manager = ResponsiveUIManager;
const ui = await manager.openIfNeeded(tab.ownerGlobal, tab, {
trigger: "test",
});
info("Responsive design mode opened");
await ResponsiveMessageHelper.wait(ui.toolWindow, "post-init");
info("Responsive design initialized");
await waitForRDMLoaded(ui, { waitForDeviceList });
return { ui, manager };
}
async function waitForRDMLoaded(ui, { waitForDeviceList = true } = {}) {
// Always wait for the viewport to be added.
const { store } = ui.toolWindow;
await waitUntilState(store, state => state.viewports.length == 1);
if (waitForDeviceList) {
// Wait until the device list has been loaded.
await waitUntilState(
store,
state => state.devices.listState == localTypes.loadableState.LOADED
);
}
}
/**
* Close responsive design mode for the given tab.
*/
async function closeRDM(tab, options) {
info("Closing responsive design mode");
const manager = ResponsiveUIManager;
await manager.closeIfNeeded(tab.ownerGlobal, tab, options);
info("Responsive design mode closed");
}
function getInputStream(data) {
const BufferStream = Components.Constructor(
"@mozilla.org/io/arraybuffer-input-stream;1",
"nsIArrayBufferInputStream",
"setData"
);
const buffer = new TextEncoder().encode(data).buffer;
return new BufferStream(buffer, 0, buffer.byteLength);
}
/**
* Wait for a specific target to have been fully processed by targetCommand.
*
* @param {Commands} commands
* The commands instance
* @param {Function} isExpectedTargetFn
* Predicate which will be called with a target front argument. Should
* return true if the target front is the expected one, false otherwise.
* @return {Promise}
* Promise which resolves when a target matching `isExpectedTargetFn`
* has been processed by targetCommand.
*/
function waitForTargetProcessed(commands, isExpectedTargetFn) {
return new Promise(resolve => {
const onProcessed = targetFront => {
try {
if (isExpectedTargetFn(targetFront)) {
commands.targetCommand.off("processed-available-target", onProcessed);
resolve();
}
} catch {
// Ignore errors from isExpectedTargetFn.
}
};
commands.targetCommand.on("processed-available-target", onProcessed);
});
}
/**
* Instantiate a HTTP Server that serves files from a given test folder.
* The test folder should be made of multiple sub folder named: v1, v2, v3,...
* We will serve the content from one of these sub folder
* and switch to the next one, each time `httpServer.switchToNextVersion()`
* is called.
*
* @return Object Test server with two functions:
* - urlFor(path)
* Returns the absolute url for a given file.
* - switchToNextVersion()
* Start serving files from the next available sub folder.
* - backToFirstVersion()
* When running more than one test, helps restart from the first folder.
*/
function createVersionizedHttpTestServer(testFolderName) {
const httpServer = createTestHTTPServer();
let currentVersion = 1;
httpServer.registerPrefixHandler("/", async (request, response) => {
response.processAsync();
response.setStatusLine(request.httpVersion, 200, "OK");
if (request.path.endsWith(".js")) {
response.setHeader("Content-Type", "application/javascript");
} else if (request.path.endsWith(".js.map")) {
response.setHeader("Content-Type", "application/json");
}
if (request.path == "/" || request.path.endsWith(".html")) {
response.setHeader("Content-Type", "text/html");
}
// If a query string is passed, lookup with a matching file, if available
// The '?' is replaced by '.'
let fetchResponse;
if (request.queryString) {
const url = `${URL_ROOT_SSL}${testFolderName}/v${currentVersion}${request.path}.${request.queryString}`;
try {
fetchResponse = await fetch(url);
// Log this only if the request succeed
info(`[test-http-server] serving: ${url}`);
} catch (e) {
// Ignore any error and proceed without the query string
fetchResponse = null;
}
}
if (!fetchResponse) {
const url = `${URL_ROOT_SSL}${testFolderName}/v${currentVersion}${request.path}`;
info(`[test-http-server] serving: ${url}`);
fetchResponse = await fetch(url);
}
// Ensure forwarding the response headers generated by the other http server
// (this can be especially useful when query .sjs files)
for (const [name, value] of fetchResponse.headers.entries()) {
response.setHeader(name, value);
}
// Override cache settings so that versionized requests are never cached
// and we get brand new content for any request.
response.setHeader("Cache-Control", "no-store");
const text = await fetchResponse.text();
response.write(text);
response.finish();
});
return {
switchToNextVersion() {
currentVersion++;
},
backToFirstVersion() {
currentVersion = 1;
},
urlFor(path) {
const port = httpServer.identity.primaryPort;
return `http://localhost:${port}/${path}`;
},
};
}
/**
* Fake clicking a link and return the URL we would have navigated to.
* This function should be used to check external links since we can't access
* network in tests.
* This can also be used to test that a click will not be fired.
*
* @param ElementNode element
* The <a> element we want to simulate click on.
* @returns Promise
* A Promise that is resolved when the link click simulation occured or
* when the click is not dispatched.
* The promise resolves with an object that holds the following properties
* - link: url of the link or null(if event not fired)
* - where: "tab" if tab is active or "tabshifted" if tab is inactive
* or null(if event not fired)
*/
function simulateLinkClick(element) {
const browserWindow = Services.wm.getMostRecentWindow(
gDevTools.chromeWindowType
);
const onOpenLink = new Promise(resolve => {
const openLinkIn = (link, where) => resolve({ link, where });
sinon.replace(browserWindow, "openTrustedLinkIn", openLinkIn);
sinon.replace(browserWindow, "openWebLinkIn", openLinkIn);
});
element.click();
// Declare a timeout Promise that we can use to make sure spied methods were not called.
const onTimeout = new Promise(function (resolve) {
setTimeout(() => {
resolve({ link: null, where: null });
}, 1000);
});
const raceResult = Promise.race([onOpenLink, onTimeout]);
sinon.restore();
return raceResult;
}
/**
* Since the MDN data is updated frequently, it might happen that the properties used in
* this test are not in the dataset anymore/now have URLs.
* This function will return properties in the dataset that don't have MDN url so you
* can easily find a replacement.
*/
function logCssCompatDataPropertiesWithoutMDNUrl() {
const cssPropertiesCompatData = require("resource://devtools/shared/compatibility/dataset/css-properties.json");
function walk(node) {
for (const propertyName in node) {
const property = node[propertyName];
if (property.__compat) {
if (!property.__compat.mdn_url) {
dump(
`"${propertyName}" - MDN URL: ${
property.__compat.mdn_url || "❌"
} - Spec URL: ${property.__compat.spec_url || "❌"}\n`
);
}
} else if (typeof property == "object") {
walk(property);
}
}
}
walk(cssPropertiesCompatData);
}
/**
* Craft a CssProperties instance without involving RDP for tests
* manually spawning OutputParser, CssCompleter, Editor...
*
* Otherwise this should instead be fetched from CssPropertiesFront.
*
* @return {CssProperties}
*/
function getClientCssProperties() {
const {
generateCssProperties,
const {
CssProperties,
normalizeCssData,
return new CssProperties(
normalizeCssData({ properties: generateCssProperties(document) })
);
}
/**
* Helper method to stop a Service Worker promptly.
*
* @param {String} workerUrl
* Absolute Worker URL to stop.
*/
async function stopServiceWorker(workerUrl) {
info(`Stop Service Worker: ${workerUrl}\n`);
// Help the SW to be immediately destroyed after unregistering it.
Services.prefs.setIntPref("dom.serviceWorkers.idle_timeout", 0);
const swm = Cc["@mozilla.org/serviceworkers/manager;1"].getService(
Ci.nsIServiceWorkerManager
);
// Unfortunately we can't use swm.getRegistrationByPrincipal, as it requires a "scope", which doesn't seem to be the worker URL.
// So let's use getAllRegistrations to find the nsIServiceWorkerInfo in order to:
// - retrieve its active worker,
// - call attach+detachDebugger,
// - reset the idle timeout.
// This way, the unregister instruction is immediate, thanks to the 0 dom.serviceWorkers.idle_timeout we set at the beginning of the function
const registrations = swm.getAllRegistrations();
let matchedInfo;
for (let i = 0; i < registrations.length; i++) {
const info = registrations.queryElementAt(
i,
Ci.nsIServiceWorkerRegistrationInfo
);
// Lookup for an exact URL match.
if (info.scriptSpec === workerUrl) {
matchedInfo = info;
break;
}
}
ok(!!matchedInfo, "Found the service worker info");
info("Wait for the worker to be active");
await waitFor(() => matchedInfo.activeWorker, "Wait for the SW to be active");
// We need to attach+detach the debugger in order to reset the idle timeout.
// Otherwise the worker would still be waiting for a previously registered timeout
// which would be the 0ms one we set by tweaking the preference.
function resetWorkerTimeout(worker) {
worker.attachDebugger();
worker.detachDebugger();
}
resetWorkerTimeout(matchedInfo.activeWorker);
// Also reset all the other possible worker instances
if (matchedInfo.evaluatingWorker) {
resetWorkerTimeout(matchedInfo.evaluatingWorker);
}
if (matchedInfo.installingWorker) {
resetWorkerTimeout(matchedInfo.installingWorker);
}
if (matchedInfo.waitingWorker) {
resetWorkerTimeout(matchedInfo.waitingWorker);
}
// Reset this preference in order to ensure other SW are not immediately destroyed.
Services.prefs.clearUserPref("dom.serviceWorkers.idle_timeout");
// Spin the event loop to ensure the worker had time to really be shut down.
await wait(0);
return matchedInfo;
}
/**
* Helper method to stop and unregister a Service Worker promptly.
*
* @param {String} workerUrl
* Absolute Worker URL to unregister.
*/
async function unregisterServiceWorker(workerUrl) {
const swInfo = await stopServiceWorker(workerUrl);
info(`Unregister Service Worker: ${workerUrl}\n`);
// Now call unregister on that worker so that it can be destroyed immediately
const swm = Cc["@mozilla.org/serviceworkers/manager;1"].getService(
Ci.nsIServiceWorkerManager
);
const unregisterSuccess = await new Promise(resolve => {
swm.unregister(
swInfo.principal,
{
unregisterSucceeded(success) {
resolve(success);
},
},
swInfo.scope
);
});
ok(unregisterSuccess, "Service worker successfully unregistered");
}