feat: check if file is binary when collecting

Author: Hweinstock

packages/core/src/shared/utilities/workspaceUtils.ts

@@ -20,6 +20,7 @@ import fs from '../fs/fs'
 import { ChildProcess } from './processUtils'
 import { isWin } from '../vscode/env'
 import { maxRepoSizeBytes } from '../../amazonqFeatureDev/constants'
+import { ZipConstants } from '../../codewhisperer/util/zipUtil'
 
 type GitIgnoreRelativeAcceptor = {
     folderPath: string
@@ -335,6 +336,7 @@ export type CollectFilesResultItem = {
     fileUri: vscode.Uri
     fileContent: string
     fileSizeBytes: number
+    isText: boolean
     zipFilePath: string
 }
 export type CollectFilesFilter = (relativePath: string) => boolean // returns true if file should be filtered out
@@ -421,12 +423,13 @@ export async function collectFiles(
                 continue
             }
 
-            const result = {
+            const result: Omit<CollectFilesResultItem, 'fileContent'> = {
                 workspaceFolder: relativePath.workspaceFolder,
                 relativeFilePath: relativePath.relativePath,
                 fileUri: file,
                 fileSizeBytes: fileStat.size,
                 zipFilePath: prefixWithFolderPrefix(relativePath.workspaceFolder, relativePath.relativePath),
+                isText: !ZipConstants.knownBinaryFileExts.includes(path.extname(file.fsPath)),
             }
             if (includeContent) {
                 const content = await readFile(file)

packages/core/src/testInteg/shared/utilities/workspaceUtils.test.ts

@@ -267,48 +267,56 @@ describe('workspaceUtils', () => {
                         relativeFilePath: '.gitignore',
                         fileContent: gitignoreContent,
                         fileSizeBytes: 162,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: 'file1',
                         fileContent: 'test content',
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: 'file3',
                         fileContent: 'test content',
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: 'range_file9',
                         fileContent: 'test content',
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: path.join('src', '.gitignore'),
                         fileContent: gitignore2,
                         fileSizeBytes: 8,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: path.join('src', 'folder2', 'a.js'),
                         fileContent: fileContent,
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: path.join('src', 'folder3', '.gitignore'),
                         fileContent: gitignore3,
                         fileSizeBytes: 42,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: path.join('src', 'folder3', 'negate_test1'),
                         fileContent: fileContent,
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                 ] satisfies typeof result,
                 result
@@ -517,11 +525,13 @@ describe('workspaceUtils', () => {
                         workspaceFolder,
                         relativeFilePath: path.join('src', 'folder2', 'a.js'),
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: path.join('src', 'folder2', 'b.cs'),
                         fileSizeBytes: 12,
+                        isText: true,
                     },
                 ] satisfies typeof result,
                 result
@@ -548,11 +558,13 @@ describe('workspaceUtils', () => {
                         workspaceFolder,
                         relativeFilePath: 'a.js',
                         fileSizeBytes: 14,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: 'b.java',
                         fileSizeBytes: 14,
+                        isText: true,
                     },
                 ] satisfies typeof result,
                 result
@@ -579,11 +591,13 @@ describe('workspaceUtils', () => {
                         workspaceFolder,
                         relativeFilePath: 'top.java',
                         fileSizeBytes: 14,
+                        isText: true,
                     },
                     {
                         workspaceFolder,
                         relativeFilePath: 'top.js',
                         fileSizeBytes: 14,
+                        isText: true,
                     },
                 ] satisfies typeof result,
                 result