lib/web_ui/dev/test_runner.dart - mirrors/engine - Git at Google

 // Copyright 2013 The Flutter Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.

 import 'dart:async';
 import 'dart:io' as io;

 import 'package:args/command_runner.dart';
 import 'package:path/path.dart' as path;

 import 'package:watcher/src/watch_event.dart';

 import 'pipeline.dart';
 import 'steps/compile_tests_step.dart';
 import 'steps/run_tests_step.dart';
 import 'utils.dart';

 /// Runs tests.
 class TestCommand extends Command<bool> with ArgUtils<bool> {
   TestCommand() {
     argParser
       ..addFlag(
         'debug',
         help: 'Pauses the browser before running a test, giving you an '
             'opportunity to add breakpoints or inspect loaded code before '
             'running the code.',
       )
       ..addFlag(
         'watch',
         abbr: 'w',
         help: 'Run in watch mode so the tests re-run whenever a change is '
             'made.',
       )
       ..addFlag(
         'use-system-flutter',
         help: 'integration tests are using flutter repository for various tasks'
             ', such as flutter drive, flutter pub get. If this flag is set, felt '
             'will use flutter command without cloning the repository. This flag '
             'can save internet bandwidth. However use with caution. Note that '
             'since flutter repo is always synced to youngest commit older than '
             'the engine commit for the tests running in CI, the tests results '
             "won't be consistent with CIs when this flag is set. flutter "
             'command should be set in the PATH for this flag to be useful.'
             'This flag can also be used to test local Flutter changes.')
       ..addFlag(
         'require-skia-gold',
         help:
             'Whether we require Skia Gold to be available or not. When this '
             'flag is true, the tests will fail if Skia Gold is not available.',
       )
       ..addFlag(
         'update-screenshot-goldens',
         help:
             'When running screenshot tests writes them to the file system into '
             '.dart_tool/goldens. Use this option to bulk-update all screenshots, '
             'for example, when a new browser version affects pixels.',
       )
       ..addOption(
         'browser',
         defaultsTo: 'chrome',
         help: 'An option to choose a browser to run the tests. By default '
               'tests run in Chrome.',
       )
       ..addFlag(
         'fail-early',
         help: 'If set, causes the test runner to exit upon the first test '
               'failure. If not set, the test runner will continue running '
               'test despite failures and will report them after all tests '
               'finish.',
       )
       ..addOption(
         'canvaskit-path',
         help: 'Optional. The path to a local build of CanvasKit to use in '
               'tests. If omitted, the test runner uses the default CanvasKit '
               'build.',
       )
       ..addFlag(
         'wasm',
         help: 'Whether the test we are running are compiled to webassembly.'
       )
       ..addFlag(
         'use-local-canvaskit',
         help: 'Optional. Whether or not to use the locally built version of '
               'CanvasKit in the tests.',
       );
   }

   @override
   final String name = 'test';

   @override
   final String description = 'Run tests.';

   bool get isWatchMode => boolArg('watch');

   bool get failEarly => boolArg('fail-early');

   bool get isWasm => boolArg('wasm');

   /// Whether to start the browser in debug mode.
   ///
   /// In this mode the browser pauses before running the test to allow
   /// you set breakpoints or inspect the code.
   bool get isDebug => boolArg('debug');

   /// Paths to targets to run, e.g. a single test.
   List<String> get targets => argResults!.rest;

   /// The target test files to run.
   List<FilePath> get targetFiles => targets.map((String t) => FilePath.fromCwd(t)).toList();

   /// Whether all tests should run.
   bool get runAllTests => targets.isEmpty;

   /// The name of the browser to run tests in.
   String get browserName => stringArg('browser');

   /// When running screenshot tests, require Skia Gold to be available and
   /// reachable.
   bool get requireSkiaGold => boolArg('require-skia-gold');

   /// When running screenshot tests writes them to the file system into
   /// ".dart_tool/goldens".
   bool get doUpdateScreenshotGoldens => boolArg('update-screenshot-goldens');

   /// Path to a CanvasKit build. Overrides the default CanvasKit.
   String? get overridePathToCanvasKit => argResults!['canvaskit-path'] as String?;

   /// Whether or not to use the locally built version of CanvasKit.
   bool get useLocalCanvasKit => boolArg('use-local-canvaskit');

   @override
   Future<bool> run() async {
     final List<FilePath> testFiles = runAllTests
           ? findAllTests()
           : targetFiles;

     final Pipeline testPipeline = Pipeline(steps: <PipelineStep>[
       if (isWatchMode) ClearTerminalScreenStep(),
       CompileTestsStep(testFiles: testFiles, useLocalCanvasKit: useLocalCanvasKit, isWasm: isWasm),
       RunTestsStep(
         browserName: browserName,
         testFiles: testFiles,
         isDebug: isDebug,
         isWasm: isWasm,
         doUpdateScreenshotGoldens: doUpdateScreenshotGoldens,
         requireSkiaGold: requireSkiaGold,
         overridePathToCanvasKit: overridePathToCanvasKit,
       ),
     ]);

     try {
       await testPipeline.run();
       if (isWatchMode) {
         print('');
         print('Initial test succeeded!');
       }
     } catch(error, stackTrace) {
       if (isWatchMode) {
         // The error is printed but not rethrown in watch mode because
         // failures are expected. The idea is that the developer corrects the
         // error, saves the file, and the pipeline reruns.
         print('');
         print('Initial test failed!\n');
         print(error);
         print(stackTrace);
       } else {
         rethrow;
       }
     }

     if (isWatchMode) {
       final FilePath dir = FilePath.fromWebUi('');
       print('');
       print(
           'Watching ${dir.relativeToCwd}/lib and ${dir.relativeToCwd}/test to re-run tests');
       print('');
       await PipelineWatcher(
           dir: dir.absolute,
           pipeline: testPipeline,
           ignore: (WatchEvent event) {
             // Ignore font files that are copied whenever tests run.
             if (event.path.endsWith('.ttf')) {
               return true;
             }

             // React to changes in lib/ and test/ folders.
             final String relativePath =
                 path.relative(event.path, from: dir.absolute);
             if (path.isWithin('lib', relativePath) ||
                 path.isWithin('test', relativePath)) {
               return false;
             }

             // Ignore anything else.
             return true;
           }).start();
     }
     return true;
   }
 }

 /// Clears the terminal screen and places the cursor at the top left corner.
 ///
 /// This works on Linux and Mac. On Windows, it's a no-op.
 class ClearTerminalScreenStep implements PipelineStep {
   @override
   String get description => 'clearing terminal screen';

   @override
   bool get isSafeToInterrupt => false;

   @override
   Future<void> interrupt() async {}

   @override
   Future<void> run() async {
     if (!io.Platform.isWindows) {
       // See: https://en.wikipedia.org/wiki/ANSI_escape_code#CSI_sequences
       print('\x1B[2J\x1B[1;2H');
     }
   }
 }
	// Copyright 2013 The Flutter Authors. All rights reserved.
	// Use of this source code is governed by a BSD-style license that can be
	// found in the LICENSE file.

	import 'dart:async';
	import 'dart:io' as io;

	import 'package:args/command_runner.dart';
	import 'package:path/path.dart' as path;

	import 'package:watcher/src/watch_event.dart';

	import 'pipeline.dart';
	import 'steps/compile_tests_step.dart';
	import 'steps/run_tests_step.dart';
	import 'utils.dart';

	/// Runs tests.
	class TestCommand extends Command<bool> with ArgUtils<bool> {
	TestCommand() {
	argParser
	..addFlag(
	'debug',
	help: 'Pauses the browser before running a test, giving you an '
	'opportunity to add breakpoints or inspect loaded code before '
	'running the code.',
	)
	..addFlag(
	'watch',
	abbr: 'w',
	help: 'Run in watch mode so the tests re-run whenever a change is '
	'made.',
	)
	..addFlag(
	'use-system-flutter',
	help: 'integration tests are using flutter repository for various tasks'
	', such as flutter drive, flutter pub get. If this flag is set, felt '
	'will use flutter command without cloning the repository. This flag '
	'can save internet bandwidth. However use with caution. Note that '
	'since flutter repo is always synced to youngest commit older than '
	'the engine commit for the tests running in CI, the tests results '
	"won't be consistent with CIs when this flag is set. flutter "
	'command should be set in the PATH for this flag to be useful.'
	'This flag can also be used to test local Flutter changes.')
	..addFlag(
	'require-skia-gold',
	help:
	'Whether we require Skia Gold to be available or not. When this '
	'flag is true, the tests will fail if Skia Gold is not available.',
	)
	..addFlag(
	'update-screenshot-goldens',
	help:
	'When running screenshot tests writes them to the file system into '
	'.dart_tool/goldens. Use this option to bulk-update all screenshots, '
	'for example, when a new browser version affects pixels.',
	)
	..addOption(
	'browser',
	defaultsTo: 'chrome',
	help: 'An option to choose a browser to run the tests. By default '
	'tests run in Chrome.',
	)
	..addFlag(
	'fail-early',
	help: 'If set, causes the test runner to exit upon the first test '
	'failure. If not set, the test runner will continue running '
	'test despite failures and will report them after all tests '
	'finish.',
	)
	..addOption(
	'canvaskit-path',
	help: 'Optional. The path to a local build of CanvasKit to use in '
	'tests. If omitted, the test runner uses the default CanvasKit '
	'build.',
	)
	..addFlag(
	'wasm',
	help: 'Whether the test we are running are compiled to webassembly.'
	)
	..addFlag(
	'use-local-canvaskit',
	help: 'Optional. Whether or not to use the locally built version of '
	'CanvasKit in the tests.',
	);
	}

	@override
	final String name = 'test';

	@override
	final String description = 'Run tests.';

	bool get isWatchMode => boolArg('watch');

	bool get failEarly => boolArg('fail-early');

	bool get isWasm => boolArg('wasm');

	/// Whether to start the browser in debug mode.
	///
	/// In this mode the browser pauses before running the test to allow
	/// you set breakpoints or inspect the code.
	bool get isDebug => boolArg('debug');

	/// Paths to targets to run, e.g. a single test.
	List<String> get targets => argResults!.rest;

	/// The target test files to run.
	List<FilePath> get targetFiles => targets.map((String t) => FilePath.fromCwd(t)).toList();

	/// Whether all tests should run.
	bool get runAllTests => targets.isEmpty;

	/// The name of the browser to run tests in.
	String get browserName => stringArg('browser');

	/// When running screenshot tests, require Skia Gold to be available and
	/// reachable.
	bool get requireSkiaGold => boolArg('require-skia-gold');

	/// When running screenshot tests writes them to the file system into
	/// ".dart_tool/goldens".
	bool get doUpdateScreenshotGoldens => boolArg('update-screenshot-goldens');

	/// Path to a CanvasKit build. Overrides the default CanvasKit.
	String? get overridePathToCanvasKit => argResults!['canvaskit-path'] as String?;

	/// Whether or not to use the locally built version of CanvasKit.
	bool get useLocalCanvasKit => boolArg('use-local-canvaskit');

	@override
	Future<bool> run() async {
	final List<FilePath> testFiles = runAllTests
	? findAllTests()
	: targetFiles;

	final Pipeline testPipeline = Pipeline(steps: <PipelineStep>[
	if (isWatchMode) ClearTerminalScreenStep(),
	CompileTestsStep(testFiles: testFiles, useLocalCanvasKit: useLocalCanvasKit, isWasm: isWasm),
	RunTestsStep(
	browserName: browserName,
	testFiles: testFiles,
	isDebug: isDebug,
	isWasm: isWasm,
	doUpdateScreenshotGoldens: doUpdateScreenshotGoldens,
	requireSkiaGold: requireSkiaGold,
	overridePathToCanvasKit: overridePathToCanvasKit,
	),
	]);

	try {
	await testPipeline.run();
	if (isWatchMode) {
	print('');
	print('Initial test succeeded!');
	}
	} catch(error, stackTrace) {
	if (isWatchMode) {
	// The error is printed but not rethrown in watch mode because
	// failures are expected. The idea is that the developer corrects the
	// error, saves the file, and the pipeline reruns.
	print('');
	print('Initial test failed!\n');
	print(error);
	print(stackTrace);
	} else {
	rethrow;
	}
	}

	if (isWatchMode) {
	final FilePath dir = FilePath.fromWebUi('');
	print('');
	print(
	'Watching ${dir.relativeToCwd}/lib and ${dir.relativeToCwd}/test to re-run tests');
	print('');
	await PipelineWatcher(
	dir: dir.absolute,
	pipeline: testPipeline,
	ignore: (WatchEvent event) {
	// Ignore font files that are copied whenever tests run.
	if (event.path.endsWith('.ttf')) {
	return true;
	}

	// React to changes in lib/ and test/ folders.
	final String relativePath =
	path.relative(event.path, from: dir.absolute);
	if (path.isWithin('lib', relativePath) \|\|
	path.isWithin('test', relativePath)) {
	return false;
	}

	// Ignore anything else.
	return true;
	}).start();
	}
	return true;
	}
	}

	/// Clears the terminal screen and places the cursor at the top left corner.
	///
	/// This works on Linux and Mac. On Windows, it's a no-op.
	class ClearTerminalScreenStep implements PipelineStep {
	@override
	String get description => 'clearing terminal screen';

	@override
	bool get isSafeToInterrupt => false;

	@override
	Future<void> interrupt() async {}

	@override
	Future<void> run() async {
	if (!io.Platform.isWindows) {
	// See: https://en.wikipedia.org/wiki/ANSI_escape_code#CSI_sequences
	print('\x1B[2J\x1B[1;2H');
	}
	}
	}