Add output code language setting

src/main/backend/__tests__/deepseekClient.test.ts

@@ -15,6 +15,7 @@ function mockSettings(provider = 'deepseek', model?: string, apiKey = 'test-key'
     provider,
     model: model ?? (provider === 'kimi' ? 'kimi-k2.6' : 'deepseek-v4-flash'),
     language: 'zh-CN',
+    selectedCodeLanguage: 'Python',
   })
 }
 

src/main/backend/__tests__/paperAnalyzerFlow.test.ts

@@ -70,6 +70,7 @@ describe('analyzePaper flow', () => {
       provider: 'deepseek',
       model: 'deepseek-v4-flash',
       language: 'zh-CN',
+      selectedCodeLanguage: 'Python',
     })
     mockedParsePDF.mockResolvedValue({ text: 'paper text', pageCount: 3 })
   })
@@ -95,6 +96,24 @@ describe('analyzePaper flow', () => {
     expect(progress).toEqual(expect.arrayContaining(['parsing', 'summarizing', 'generating_code', 'done']))
   })
 
+  it('passes the selected output code language into the combined analysis prompt', async () => {
+    mockedGetActiveSettings.mockReturnValueOnce({
+      apiKey: 'key',
+      provider: 'deepseek',
+      model: 'deepseek-v4-flash',
+      language: 'en-US',
+      selectedCodeLanguage: 'MATLAB',
+    })
+    mockLlmOutput(outputWithoutCode('MATLAB summary'))
+
+    await analyzePaper('paper.pdf', () => {})
+
+    const messages = mockedCallDeepSeek.mock.calls[0][0]
+    expect(messages[0].content).toContain('Demo Code and core code files MUST be generated in MATLAB')
+    expect(messages[1].content).toContain('The selected output code language is MATLAB')
+    expect(messages[1].content).toContain('core_code/descriptive_file_name.m')
+  })
+
   it('caches generated core code and returns hasCoreCode for valid code output', async () => {
     mockLlmOutput(outputWithCode())
 

src/main/backend/__tests__/promptBuilder.test.ts

@@ -2,6 +2,22 @@ import { describe, expect, it } from 'vitest'
 import { buildCodePrompt, buildCombinedAnalysisPrompt, buildSummaryPrompt } from '../promptBuilder'
 
 describe('promptBuilder', () => {
+  it.each([
+    ['Python', 'py'],
+    ['C', 'c'],
+    ['C++', 'cpp'],
+    ['Java', 'java'],
+    ['Go', 'go'],
+    ['Rust', 'rs'],
+    ['MATLAB', 'm'],
+    ['R', 'R'],
+  ])('uses the expected source extension for %s code output', (codeLanguage, extension) => {
+    const prompt = buildCombinedAnalysisPrompt('paper body', 'en-US', codeLanguage)
+
+    expect(prompt.system).toContain(`generated in ${codeLanguage}`)
+    expect(prompt.user).toContain(`core_code/descriptive_file_name.${extension}`)
+  })
+
   it('builds Chinese summary prompts by default with README and math rules', () => {
     const prompt = buildSummaryPrompt('paper body')
 
@@ -24,14 +40,16 @@ describe('promptBuilder', () => {
   })
 
   it('builds code prompts with strict JSON schema and summary context', () => {
-    const prompt = buildCodePrompt('paper text', 'summary text', 'en-US')
+    const prompt = buildCodePrompt('paper text', 'summary text', 'en-US', 'Rust')
 
     expect(prompt.system).toContain('Use English')
     expect(prompt.user).toContain('paper text')
     expect(prompt.user).toContain('summary text')
     expect(prompt.user).toContain('Return ONLY strict JSON')
     expect(prompt.user).toContain('"files"')
     expect(prompt.user).toContain('"notApplicableReason"')
+    expect(prompt.user).toContain('Generate all demo/core code in Rust')
+    expect(prompt.user).toContain('core_code/descriptive_file_name.rs')
     expect(prompt.user).toContain('Every file path MUST be relative')
   })
 
@@ -47,13 +65,15 @@ describe('promptBuilder', () => {
   })
 
   it('builds combined prompts with blueprint and exact code bundle constraints', () => {
-    const prompt = buildCombinedAnalysisPrompt('combined paper', 'en-US')
+    const prompt = buildCombinedAnalysisPrompt('combined paper', 'en-US', 'C++')
 
     expect(prompt.system).toContain('Use English')
+    expect(prompt.system).toContain('Demo Code and core code files MUST be generated in C++')
     expect(prompt.user).toContain('# Paper Title')
     expect(prompt.user).toContain('<P2CC_CODE_BLUEPRINT>')
     expect(prompt.user).toContain('<P2CC_CODE_BUNDLE>')
-    expect(prompt.user).toContain('<P2CC_FILE path="core_code/descriptive_file_name.py">')
+    expect(prompt.user).toContain('The selected output code language is C++')
+    expect(prompt.user).toContain('<P2CC_FILE path="core_code/descriptive_file_name.cpp">')
     expect(prompt.user).toContain('no more and no fewer')
     expect(prompt.user).toContain('Avoid generic files')
   })

src/main/backend/__tests__/settingsStore.test.ts

@@ -43,6 +43,7 @@ describe('settingsStore', () => {
       provider: 'deepseek',
       model: PROVIDER_SETTINGS.deepseek.defaultModel,
       language: 'zh-CN',
+      selectedCodeLanguage: 'Python',
     })
   })
 
@@ -54,6 +55,7 @@ describe('settingsStore', () => {
       provider: 'deepseek',
       model: PROVIDER_SETTINGS.deepseek.defaultModel,
       language: 'zh-CN',
+      selectedCodeLanguage: 'Python',
     })
   })
 
@@ -76,6 +78,25 @@ describe('settingsStore', () => {
       provider: 'kimi',
       model: 'kimi-k2.5',
       language: 'en-US',
+      selectedCodeLanguage: 'Python',
+    })
+  })
+
+  it('normalizes and preserves the selected output code language', () => {
+    writeConfig({
+      provider: 'deepseek',
+      selectedCodeLanguage: 'Rust',
+      providers: {
+        deepseek: { apiKey: 'deep-key', model: 'deepseek-v4-pro' },
+      },
+    })
+
+    expect(getActiveSettings()).toEqual({
+      apiKey: 'deep-key',
+      provider: 'deepseek',
+      model: 'deepseek-v4-pro',
+      language: 'zh-CN',
+      selectedCodeLanguage: 'Rust',
     })
   })
 
@@ -93,6 +114,21 @@ describe('settingsStore', () => {
       provider: 'deepseek',
       model: PROVIDER_SETTINGS.deepseek.defaultModel,
       language: 'zh-CN',
+      selectedCodeLanguage: 'Python',
+    })
+  })
+
+  it('falls back invalid stored output code languages to Python', () => {
+    writeConfig({
+      provider: 'deepseek',
+      selectedCodeLanguage: 'TypeScript',
+      providers: {
+        deepseek: { apiKey: 'deep-key', model: 'deepseek-v4-flash' },
+      },
+    })
+
+    expect(getActiveSettings()).toMatchObject({
+      selectedCodeLanguage: 'Python',
     })
   })
 
@@ -111,6 +147,7 @@ describe('settingsStore', () => {
       provider: 'kimi',
       model: PROVIDER_SETTINGS.kimi.defaultModel,
       language: 'en-US',
+      selectedCodeLanguage: 'Python',
     })
   })
 
@@ -127,36 +164,53 @@ describe('settingsStore', () => {
       provider: 'glm',
       model: 'glm-5-turbo',
       language: 'en-US',
+      selectedCodeLanguage: 'Python',
     })
   })
 
   it('saves active provider settings and preserves provider-specific keys', () => {
-    expect(saveSettingsPatch({ apiKey: 'deep-key', language: 'en-US' })).toEqual({
+    expect(saveSettingsPatch({ apiKey: 'deep-key', language: 'en-US', selectedCodeLanguage: 'Go' })).toEqual({
       apiKey: 'deep-key',
       provider: 'deepseek',
       model: PROVIDER_SETTINGS.deepseek.defaultModel,
       language: 'en-US',
+      selectedCodeLanguage: 'Go',
     })
 
     expect(saveSettingsPatch({ provider: 'kimi' })).toEqual({
       apiKey: '',
       provider: 'kimi',
       model: PROVIDER_SETTINGS.kimi.defaultModel,
       language: 'en-US',
+      selectedCodeLanguage: 'Go',
     })
 
     expect(saveSettingsPatch({ apiKey: 'kimi-key', model: 'kimi-k2.5' })).toEqual({
       apiKey: 'kimi-key',
       provider: 'kimi',
       model: 'kimi-k2.5',
       language: 'en-US',
+      selectedCodeLanguage: 'Go',
     })
 
     expect(saveSettingsPatch({ provider: 'deepseek' })).toEqual({
       apiKey: 'deep-key',
       provider: 'deepseek',
       model: PROVIDER_SETTINGS.deepseek.defaultModel,
       language: 'en-US',
+      selectedCodeLanguage: 'Go',
+    })
+  })
+
+  it('saves selected output code language patches and normalizes unsupported values', () => {
+    expect(saveSettingsPatch({ selectedCodeLanguage: 'MATLAB' })).toMatchObject({
+      selectedCodeLanguage: 'MATLAB',
+    })
+
+    expect(readStoredConfig().selectedCodeLanguage).toBe('MATLAB')
+
+    expect(saveSettingsPatch({ selectedCodeLanguage: 'CUDA' })).toMatchObject({
+      selectedCodeLanguage: 'Python',
     })
   })
 

src/main/backend/paperAnalyzer.ts

@@ -94,8 +94,11 @@ export async function analyzePaper(
   signal?: AbortSignal
 ): Promise<AnalysisResult> {
   let language = 'zh-CN'
+  let selectedCodeLanguage = 'Python'
   try {
-    language = getActiveSettings().language || 'zh-CN'
+    const settings = getActiveSettings()
+    language = settings.language || 'zh-CN'
+    selectedCodeLanguage = settings.selectedCodeLanguage || 'Python'
   } catch {}
 
   const msg = (zh: string, en: string) => language === 'en-US' ? en : zh
@@ -116,7 +119,7 @@ export async function analyzePaper(
     onProgress({ stage: 'parsing', message: msg(`PDF 解析完成 (${pageCount} 页)`, `PDF parsed (${pageCount} pages)`) })
 
     onProgress({ stage: 'summarizing', message: msg('正在分析论文结构并生成总结...', 'Analyzing paper structure and generating summary...') })
-    const analysisPrompt = buildCombinedAnalysisPrompt(text, language)
+    const analysisPrompt = buildCombinedAnalysisPrompt(text, language, selectedCodeLanguage)
     let rawOutput = ''
     let streamedSummary = ''
     let summaryClosed = false

src/main/backend/promptBuilder.ts

@@ -1,3 +1,24 @@
+function getCodeExtension(selectedCodeLanguage: string): string {
+  switch (selectedCodeLanguage) {
+    case 'C':
+      return 'c'
+    case 'C++':
+      return 'cpp'
+    case 'Java':
+      return 'java'
+    case 'Go':
+      return 'go'
+    case 'Rust':
+      return 'rs'
+    case 'MATLAB':
+      return 'm'
+    case 'R':
+      return 'R'
+    default:
+      return 'py'
+  }
+}
+
 export function buildSummaryPrompt(text: string, language: string = 'zh-CN'): { system: string; user: string } {
   const isEn = language === 'en-US'
 
@@ -71,8 +92,9 @@ $$`
   }
 }
 
-export function buildCodePrompt(text: string, summaryContext: string, language: string = 'zh-CN'): { system: string; user: string } {
+export function buildCodePrompt(text: string, summaryContext: string, language: string = 'zh-CN', selectedCodeLanguage: string = 'Python'): { system: string; user: string } {
   const isEn = language === 'en-US'
+  const codeExtension = getCodeExtension(selectedCodeLanguage)
 
   const system = `You are an expert AI research paper analyst. Your task is to analyze academic papers and produce structured summaries and, when applicable, extract core algorithmic implementations.
 
@@ -97,9 +119,9 @@ ${summaryContext}
 Your task:
 1. Identify the smallest implementable core method from the paper.
 2. Implement it as multiple focused files inside a folder-style project structure.
-3. Prefer Python for ML/algorithm papers.
+3. Generate all demo/core code in ${selectedCodeLanguage}. Do NOT switch to Python unless the selected output code language is Python.
 4. Keep files componentized and minimal: configuration, model/algorithm, loss/metrics if needed, training or inference, and an example entry point.
-5. Include a requirements.txt file when dependencies are needed.
+5. Use file extensions and dependency notes appropriate for ${selectedCodeLanguage}.
 
 Output format:
 Return ONLY strict JSON. Do not wrap it in Markdown fences. Do not add prose before or after the JSON.
@@ -108,35 +130,7 @@ The JSON schema MUST be:
 {
   "files": [
     {
-      "path": "requirements.txt",
-      "content": "..."
-    },
-    {
-      "path": "core_code/__init__.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/config.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/model.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/losses.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/train.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/inference.py",
-      "content": "..."
-    },
-    {
-      "path": "core_code/example.py",
+      "path": "core_code/descriptive_file_name.${codeExtension}",
       "content": "..."
     }
   ]
@@ -157,6 +151,7 @@ Rules:
 - If the paper uses pseudo-code, translate it to real code.
 - If the paper describes an architecture, implement the forward pass.
 - If the paper has a training procedure, implement the training loop.
+- Demo Code MUST use ${selectedCodeLanguage}.
 - Keep code APIs, filenames, classes, and variables in English even when the UI language is Chinese.`
 
   return {
@@ -165,15 +160,17 @@ Rules:
   }
 }
 
-export function buildCombinedAnalysisPrompt(text: string, language: string = 'zh-CN'): { system: string; user: string } {
+export function buildCombinedAnalysisPrompt(text: string, language: string = 'zh-CN', selectedCodeLanguage: string = 'Python'): { system: string; user: string } {
   const isEn = language === 'en-US'
+  const codeExtension = getCodeExtension(selectedCodeLanguage)
 
   const system = `You are an expert AI research paper analyst. Analyze the paper, write a README.md-style summary, decide whether core component code is needed, and generate code only when applicable.
 
 Rules:
 - Be precise and factual. Do not fabricate details not present in the paper.
 - Use ${isEn ? 'English' : 'Chinese'} for paper summary and decision reason.
 - Keep generated code APIs, filenames, classes, and variables in English.
+- Demo Code and core code files MUST be generated in ${selectedCodeLanguage}.
 - Format mathematical expressions in the summary with Markdown LaTeX delimiters: inline math as $...$ and block math as $$...$$.
 - Follow the exact tagged output protocol. Do not add prose before, between, or after the required tags.`
 
@@ -251,7 +248,7 @@ Use "needed": true only if the paper clearly describes at least one implementabl
   "minimalImplementationBoundary": "exactly what the generated code should implement, and what it should not implement",
   "files": [
     {
-      "path": "core_code/descriptive_file_name.py",
+      "path": "core_code/descriptive_file_name.${codeExtension}",
       "purpose": "why this file is necessary for the minimal core contribution",
       "mainSymbols": ["function_or_class_name"],
       "mustInclude": ["specific method elements that must appear in this file"],
@@ -277,6 +274,9 @@ Use "needed": true only if the paper clearly describes at least one implementabl
 </P2CC_CODE_BLUEPRINT>
 
 Blueprint rules:
+- The selected output code language is ${selectedCodeLanguage}. Every generated source file MUST use ${selectedCodeLanguage}.
+- Do NOT switch to Python unless the selected output code language is Python.
+- Choose file extensions appropriate for ${selectedCodeLanguage}.
 - Do NOT assume the paper is about AI, machine learning, or software engineering. Infer the domain from the paper.
 - Do NOT use a fixed project template. The file list must be designed from the paper's smallest implementable computational contribution.
 - Each file must be justified by the core contribution. If a file is not necessary for that contribution, omit it.
@@ -288,7 +288,7 @@ Blueprint rules:
 
 5. Then output code files inside these exact tags. The code bundle MUST contain exactly the files listed in P2CC_CODE_BLUEPRINT.files, no more and no fewer:
 <P2CC_CODE_BUNDLE>
-<P2CC_FILE path="core_code/descriptive_file_name.py">
+<P2CC_FILE path="core_code/descriptive_file_name.${codeExtension}">
 ...
 </P2CC_FILE>
 </P2CC_CODE_BUNDLE>
@@ -301,7 +301,7 @@ Code generation rules:
 - Do NOT encode file contents as JSON strings. Write raw file content directly inside the file block.
 - Use the exact paths declared in P2CC_CODE_BLUEPRINT.files.
 - Do not generate extra helper files outside the blueprint.
-- Prefer Python for algorithmic papers, but choose simple, dependency-light code that best fits the paper's method.
+- Demo Code MUST use ${selectedCodeLanguage}; choose simple, dependency-light code that best fits the paper's method within that language.
 - Keep the implementation reusable as a core component, not as a full experiment reproduction project.
 - Use comments to explain key implementation decisions only where helpful.`