fix(input files): support emojis in file names (#2430)

Support emojis and other encoded characters in file names when using `git ls-files`.

e2e/test/mocha-javascript/package.json

@@ -7,6 +7,7 @@
   "scripts": {
     "pretest": "rimraf \"reports\"",
     "test": "stryker run",
+    "test:unit": "mocha",
     "posttest": "mocha --no-config --require ../../tasks/ts-node-register.js verify/*.ts"
   },
   "author": "",

e2e/test/mocha-javascript/src/🐱‍👓ninja.cat.js

@@ -0,0 +1,4 @@
+
+module.exports = function ninjaCatSays(toSay) {
+  return `🐱‍👓 ${toSay}`;
+}

e2e/test/mocha-javascript/test/unit/ninjaCat.spec.js

@@ -0,0 +1,10 @@
+const ninjaCatSays = require('../../src/🐱‍👓ninja.cat');
+const { expect } = require('chai');
+
+describe('🐱‍👓', () => {
+  describe('ninjaCatSays', () => {
+    it('should speak', () => {
+      expect(ninjaCatSays('ATTACK!')).eq('🐱‍👓 ATTACK!')
+    });
+  });
+});

e2e/test/mocha-javascript/verify/verify.ts

@@ -10,16 +10,17 @@ describe('Verify stryker has ran correctly', () => {
 
   it('should report correct score', async () => {
     await expectMetricsResult({
+
       metrics: produceMetrics({
-        killed: 16,
-        mutationScore: 64,
-        mutationScoreBasedOnCoveredCode: 64,
+        killed: 18,
+        mutationScore: 66.67,
+        mutationScoreBasedOnCoveredCode: 66.67,
         survived: 9,
-        totalCovered: 25,
-        totalDetected: 16,
-        totalMutants: 25,
+        totalCovered: 27,
+        totalDetected: 18,
+        totalMutants: 27,
         totalUndetected: 9,
-        totalValid: 25
+        totalValid: 27
       })
     });
   });

packages/core/src/input/InputFileResolver.ts

@@ -1,6 +1,8 @@
 import * as path from 'path';
 import { promises as fs } from 'fs';
 
+import { StringDecoder } from 'string_decoder';
+
 import { File, StrykerOptions } from '@stryker-mutator/api/core';
 import { Logger } from '@stryker-mutator/api/logging';
 import { commonTokens, tokens } from '@stryker-mutator/api/plugin';
@@ -23,6 +25,29 @@ function toReportSourceFile(file: File): SourceFile {
 
 const IGNORE_PATTERN_CHARACTER = '!';
 
+/**
+ *  When characters are represented as the octal values of its utf8 encoding
+ *  e.g. å becomes \303\245 in git.exe output
+ */
+function decodeGitLsOutput(line: string) {
+  if (line.startsWith('"') && line.endsWith('"')) {
+    return line
+      .substr(1, line.length - 2)
+      .replace(/\\\\/g, '\\')
+      .replace(/(?:\\\d+)*/g, (octalEscapeSequence) =>
+        new StringDecoder('utf-8').write(
+          Buffer.from(
+            octalEscapeSequence
+              .split('\\')
+              .filter(Boolean)
+              .map((octal) => parseInt(octal, 8))
+          )
+        )
+      );
+  }
+  return line;
+}
+
 export default class InputFileResolver {
   private readonly mutatePatterns: readonly string[];
   private readonly filePatterns: readonly string[] | undefined;
@@ -103,12 +128,14 @@ export default class InputFileResolver {
       const { stdout } = await childProcessAsPromised.exec(`git ls-files --others --exclude-standard --cached --exclude /${this.tempDirName}/*`, {
         maxBuffer: 10 * 1000 * 1024,
       });
-      const fileNames = stdout
+      const relativeFileNames = stdout
         .toString()
         .split('\n')
         .map((line) => line.trim())
-        .filter((line) => line) // remove empty lines
-        .map((relativeFileName) => path.resolve(relativeFileName));
+        .filter(Boolean) // remove empty lines
+        .map(decodeGitLsOutput);
+
+      const fileNames = relativeFileNames.map((relativeFileName) => path.resolve(relativeFileName));
       return fileNames;
     } catch (error) {
       throw new StrykerError(
@@ -135,21 +162,19 @@ export default class InputFileResolver {
     return files.filter(notEmpty);
   }
 
-  private readFile(fileName: string): Promise<File | null> {
-    return fs
-      .readFile(fileName)
-      .then((content: Buffer) => new File(fileName, content))
-      .then((file: File) => {
-        this.reportSourceFilesRead(file);
-        return file;
-      })
-      .catch((error) => {
-        if ((isErrnoException(error) && error.code === 'ENOENT') || error.code === 'EISDIR') {
-          return null; // file is deleted or a directory. This can be a valid result of the git command
-        } else {
-          // Rethrow
-          throw error;
-        }
-      });
+  private async readFile(fileName: string): Promise<File | null> {
+    try {
+      const content = await fs.readFile(fileName);
+      const file = new File(fileName, content);
+      this.reportSourceFilesRead(file);
+      return file;
+    } catch (error) {
+      if ((isErrnoException(error) && error.code === 'ENOENT') || error.code === 'EISDIR') {
+        return null; // file is deleted or a directory. This can be a valid result of the git command
+      } else {
+        // Rethrow
+        throw error;
+      }
+    }
   }
 }

packages/core/test/unit/input/InputFileResolver.spec.ts

@@ -4,7 +4,7 @@ import * as fs from 'fs';
 
 import { File } from '@stryker-mutator/api/core';
 import { SourceFile } from '@stryker-mutator/api/report';
-import { testInjector, factory } from '@stryker-mutator/test-helpers';
+import { testInjector, factory, assertions } from '@stryker-mutator/test-helpers';
 import { createIsDirError, fileNotFoundError } from '@stryker-mutator/test-helpers/src/factory';
 import { childProcessAsPromised, errorToString } from '@stryker-mutator/util';
 import { expect } from 'chai';
@@ -33,8 +33,6 @@ describe(InputFileResolver.name, () => {
       .stub(fs.promises, 'readFile')
       .withArgs(sinon.match.string)
       .resolves(Buffer.from('')) // fallback
-      .withArgs(sinon.match.string)
-      .resolves(Buffer.from('')) // fallback
       .withArgs(sinon.match('file1'))
       .resolves(Buffer.from('file 1 content'))
       .withArgs(sinon.match('file2'))
@@ -134,6 +132,26 @@ describe(InputFileResolver.name, () => {
     expect(result.files).lengthOf(0);
   });
 
+  it('should decode encoded file names', async () => {
+    sut = createSut();
+    childProcessExecStub.resolves({
+      // \303\245 = å
+      // \360\237\220\261\342\200\215\360\237\221\223 = 🐱‍👓
+      // On linux, files are allowed to contain `\`, which is also escaped in git output
+      stdout: Buffer.from(`
+      "\\303\\245.js"
+      "src/\\360\\237\\220\\261\\342\\200\\215\\360\\237\\221\\223ninja.cat.js"
+      "a\\\\test\\\\file.js"
+    `),
+    });
+    const files = await sut.resolve();
+    assertions.expectTextFilesEqual(files.files, [
+      new File(path.resolve('å.js'), ''),
+      new File(path.resolve('src/🐱‍👓ninja.cat.js'), ''),
+      new File(path.resolve('a\\test\\file.js'), ''),
+    ]);
+  });
+
   describe('with mutate file expressions', () => {
     it('should result in the expected mutate files', async () => {
       testInjector.options.mutate = ['mute*'];

packages/test-helpers/src/assertions.ts

@@ -49,7 +49,7 @@ export function expectTextFileEqual(actual: File, expected: File) {
   expect(fileToJson(actual)).deep.eq(fileToJson(expected));
 }
 
-export function expectTextFilesEqual(actual: File[], expected: File[]) {
+export function expectTextFilesEqual(actual: readonly File[], expected: readonly File[]) {
   expect(actual.map(fileToJson)).deep.eq(expected.map(fileToJson));
 }