fix(amazonq): decide project language by file count aws#4862

Problem
Project scans fail server side validation if the active file is an unsupported codewhisperer language, even if there are other supported files in the project.

Solution
Get the most-frequent language (based on file extension) when preparing the zip file and use that as the language.

Author: ctlai95

packages/core/src/codewhisperer/commands/startSecurityScan.ts

@@ -131,6 +131,9 @@ export async function startSecurityScan(
         codeScanTelemetryEntry.codewhispererCodeScanBuildPayloadBytes = zipMetadata.buildPayloadSizeInBytes
         codeScanTelemetryEntry.codewhispererCodeScanSrcZipFileBytes = zipMetadata.zipFileSizeInBytes
         codeScanTelemetryEntry.codewhispererCodeScanLines = zipMetadata.lines
+        if (zipMetadata.language) {
+            codeScanTelemetryEntry.codewhispererLanguage = zipMetadata.language
+        }
 
         /**
          * Step 2: Get presigned Url, upload and clean up

packages/core/src/codewhisperer/util/runtimeLanguageContext.ts

@@ -180,6 +180,17 @@ export class RuntimeLanguageContext {
         return { language: this.normalizeLanguage(languageId) ?? 'plaintext' }
     }
 
+    /**
+     *
+     * @param fileExtension File extension i.e. py, js, java
+     * @returns The corresponding {@link CodewhispererLanguage} of the file extension
+     */
+    public getLanguageFromFileExtension(fileExtension: string): CodewhispererLanguage | undefined {
+        return [...this.supportedLanguageExtensionMap.entries()].find(
+            ([, extension]) => extension === fileExtension
+        )?.[0]
+    }
+
     /**
      * Mapping the field ProgrammingLanguage of codewhisperer generateRecommendationRequest | listRecommendationRequest to
      * its Codewhisperer runtime language e.g. jsx -> typescript, typescript -> typescript

packages/core/src/codewhisperer/util/zipUtil.ts

@@ -12,6 +12,8 @@ import { ToolkitError } from '../../shared/errors'
 import { fsCommon } from '../../srcShared/fs'
 import { collectFiles } from '../../amazonqFeatureDev/util/files'
 import { getLoggerForScope } from '../service/securityScanHandler'
+import { runtimeLanguageContext } from './runtimeLanguageContext'
+import { CodewhispererLanguage } from '../../shared/telemetry/telemetry.gen'
 
 export interface ZipMetadata {
     rootDir: string
@@ -21,6 +23,7 @@ export interface ZipMetadata {
     buildPayloadSizeInBytes: number
     zipFileSizeInBytes: number
     lines: number
+    language: CodewhispererLanguage | undefined
 }
 
 export const ZipConstants = {
@@ -38,6 +41,7 @@ export class ZipUtil {
     protected _zipDir: string = ''
     protected _totalLines: number = 0
     protected _fetchedDirs: Set<string> = new Set<string>()
+    protected _language: CodewhispererLanguage | undefined
 
     constructor() {}
 
@@ -119,6 +123,7 @@ export class ZipUtil {
         }
 
         const files = await collectFiles([projectPath], [workspaceFolder])
+        const languageCount = new Map<CodewhispererLanguage, number>()
         for (const file of files) {
             const isFileOpenAndDirty = this.isFileOpenAndDirty(file.fileUri)
             const fileContent = isFileOpenAndDirty ? await this.getTextContent(file.fileUri) : file.fileContent
@@ -137,6 +142,13 @@ export class ZipUtil {
                 this._pickedSourceFiles.add(file.fileUri.fsPath)
                 this._totalSize += fileSize
                 this._totalLines += fileContent.split(ZipConstants.newlineRegex).length
+
+                const language = runtimeLanguageContext.getLanguageFromFileExtension(
+                    path.extname(file.fileUri.fsPath).slice(1)
+                )
+                if (language) {
+                    languageCount.set(language, (languageCount.get(language) || 0) + 1)
+                }
             }
 
             if (isFileOpenAndDirty) {
@@ -146,6 +158,10 @@ export class ZipUtil {
             }
         }
 
+        if (languageCount.size === 0) {
+            throw new ToolkitError('Project does not contain valid files to scan')
+        }
+        this._language = [...languageCount.entries()].reduce((a, b) => (b[1] > a[1] ? b : a))[0]
         const zipFilePath = this.getZipDirPath() + CodeWhispererConstants.codeScanZipExt
         zip.writeZip(zipFilePath)
         return zipFilePath
@@ -193,6 +209,7 @@ export class ZipUtil {
                 zipFileSizeInBytes: zipFileSize,
                 buildPayloadSizeInBytes: this._totalBuildSize,
                 lines: this._totalLines,
+                language: this._language,
             }
         } catch (error) {
             getLogger().error('Zip error caused by:', error)

packages/core/src/test/codewhisperer/commands/startSecurityScan.test.ts

@@ -284,7 +284,7 @@ describe('startSecurityScan', function () {
             CodeAnalysisScope.PROJECT
         )
         assertTelemetry('codewhisperer_securityScan', {
-            codewhispererLanguage: 'python',
+            codewhispererLanguage: 'yaml',
             codewhispererCodeScanTotalIssues: 1,
             codewhispererCodeScanIssuesWithFixes: 0,
         } as CodewhispererSecurityScan)