refactor: 优化流式输出与状态管理

- 重构 DeepSeekAPI.js,实现稳健的 SSE 流式解析
- 将核心业务逻辑(生成、分析)移入 appStore.js
- 优化 WriterPanel 和 AnalysisPanel 组件,移除冗余逻辑
- 更新文档,补充架构演进说明
This commit is contained in:
empty
2026-01-08 10:54:48 +08:00
parent 03e4007149
commit 3d0d16a3e5
5 changed files with 217 additions and 192 deletions

View File

@@ -29,6 +29,21 @@ src/
└── main.js # 入口文件
```
## 架构演进与优化
### 1. 流式输出增强 (Stream Optimization)
项目在 `DeepSeekAPI.js` 中实现了稳健的 SSE (Server-Sent Events) 解析器:
- **逐行扫描**:放弃简单的 `split` 分割,改用逐字符缓冲区扫描,确保处理被网络分包截断的 JSON 数据。
- **协议兼容**:完美支持标准 SSE 协议,兼容 data: 后有无空格的各种情况。
- **状态管理**:通过 `appStore` 统一管理生成状态,实现组件间的数据实时同步,保证 UI 渲染无延迟。
### 2. 状态管理重构
- **逻辑收拢**:将所有 API 调用和业务逻辑(生成、分析)移入 Pinia Store (`appStore.js`)。
- **组件纯粹化**`WriterPanel``AnalysisPanel` 仅负责 UI 渲染和用户交互,不再包含核心业务逻辑。
### 3. 深度集成
- **应用到写作**:实现了从"范式分析"到"写作工坊"的深度数据迁移,包括原文引用、风格约束注入和范式模板自动匹配。
## 功能特性
### 1. AI 写作工坊

View File

@@ -1,79 +1,115 @@
import axios from 'axios'
class DeepSeekAPI {
constructor(config) {
this.baseURL = config.url
this.apiKey = config.key
this.client = axios.create({
baseURL: this.baseURL,
timeout: 60000,
headers: {
'Content-Type': 'application/json'
console.log('DeepSeekAPI 已初始化:', { baseURL: this.baseURL })
}
async _streamRequest(messages, options = {}, onContent) {
const authHeader = this.apiKey.trim().startsWith('Bearer ')
? this.apiKey.trim()
: `Bearer ${this.apiKey.trim()}`;
console.log('DeepSeekAPI: Starting stream request...', { messagesLength: messages.length })
try {
const response = await fetch(this.baseURL, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': authHeader
},
body: JSON.stringify({
model: 'deepseek-chat',
messages,
stream: true,
...options
})
})
if (!response.ok) {
const errorText = await response.text()
console.error('DeepSeekAPI: HTTP Error', response.status, errorText)
throw new Error(`API请求失败: ${response.status} ${errorText}`)
}
})
console.log('DeepSeekAPI: Stream connection established.')
const reader = response.body.getReader()
const decoder = new TextDecoder()
let buffer = ''
while (true) {
const { done, value } = await reader.read()
if (done) {
console.log('DeepSeekAPI: Stream finished by server.')
break
}
buffer += decoder.decode(value, { stream: true })
// Split by newline
let lines = buffer.split('\n')
// Keep the last part (potential incomplete line) in buffer
buffer = lines.pop() || ''
for (const line of lines) {
const trimmedLine = line.trim()
if (!trimmedLine) continue
if (trimmedLine.startsWith('data:')) {
const data = trimmedLine.substring(5).trim() // Remove 'data:' prefix
if (data === '[DONE]') {
console.log('DeepSeekAPI: Received [DONE] signal.')
return
}
try {
const parsed = JSON.parse(data)
const content = parsed.choices?.[0]?.delta?.content || ''
if (content) {
// console.log('DeepSeekAPI: Received content chunk:', content.length) // Too verbose for large text
if (onContent) {
onContent(content)
}
}
} catch (e) {
console.warn('DeepSeekAPI: JSON parse error for line:', trimmedLine, e)
}
}
}
}
} catch (err) {
console.error('DeepSeekAPI: Stream processing error:', err);
throw err;
}
}
async generateContent(prompt, options = {}) {
const response = await fetch(this.baseURL, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': `Bearer ${this.apiKey}`
async generateContent(prompt, onContent, options = {}) {
return this._streamRequest([
{
role: 'system',
content: '你是一个资深的专业写作助手,请严格按照用户的要求进行创作。'
},
body: JSON.stringify({
model: 'deepseek-chat',
messages: [
{
role: 'system',
content: '你是一个资深的专业写作助手,请严格按照用户的要求进行创作。'
},
{
role: 'user',
content: prompt
}
],
stream: true,
temperature: 0.7,
...options
})
})
if (!response.ok) {
throw new Error(`API请求失败: ${response.status}`)
}
return response
{
role: 'user',
content: prompt
}
], { temperature: 0.7, ...options }, onContent)
}
async analyzeContent(text) {
const response = await fetch(this.baseURL, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': `Bearer ${this.apiKey}`
async analyzeContent(text, onContent) {
return this._streamRequest([
{
role: 'system',
content: '你是一个专业的写作分析师擅长分析文章的写作范式、结构和特点。请从以下几个方面分析文章1. 主要写作范式类型 2. 文章结构特点 3. 语言风格特征 4. 目标读者群体 5. 写作技巧总结。请用简洁明了的语言回答使用Markdown格式。'
},
body: JSON.stringify({
model: 'deepseek-chat',
messages: [
{
role: 'system',
content: '你是一个专业的写作分析师擅长分析文章的写作范式、结构和特点。请从以下几个方面分析文章1. 主要写作范式类型 2. 文章结构特点 3. 语言风格特征 4. 目标读者群体 5. 写作技巧总结。请用简洁明了的语言回答使用Markdown格式。'
},
{
role: 'user',
content: `请分析以下文章的写作范式:\n\n${text}`
}
],
stream: true,
temperature: 0.3
})
})
if (!response.ok) {
throw new Error(`API请求失败: ${response.status}`)
}
return response
{
role: 'user',
content: `请分析以下文章的写作范式:\n\n${text}`
}
], { temperature: 0.3 }, onContent)
}
}

View File

@@ -203,81 +203,12 @@ const applyParadigm = (paradigm) => {
// 分析文章
const analyzeArticle = async () => {
if (!analysisText.value.trim()) {
alert('请输入要分析的文章内容')
return
}
isAnalyzing.value = true
appStore.analysisResult = {
paradigm: '分析中...',
paradigmType: null,
analysis: '',
timestamp: new Date()
}
try {
const api = new DeepSeekAPI({
url: appStore.apiUrl,
key: appStore.apiKey
})
const response = await api.analyzeContent(analysisText.value)
const reader = response.body.getReader()
const decoder = new TextDecoder()
let fullContent = ''
while (true) {
const { done, value } = await reader.read()
if (done) break
const chunk = decoder.decode(value)
const lines = chunk.split('\n').filter(line => line.trim())
for (const line of lines) {
if (line.startsWith('data: ')) {
const data = line.slice(6)
if (data === '[DONE]') continue
try {
const parsed = JSON.parse(data)
const content = parsed.choices?.[0]?.delta?.content || ''
if (content) {
fullContent += content
// 实时更新分析结果
appStore.analysisResult = {
paradigm: '分析中...',
paradigmType: null,
analysis: fullContent,
timestamp: new Date()
}
}
} catch (e) {
// 忽略解析错误
}
}
}
}
// 分析完成后检测范式类型
const detectedParadigm = detectParadigm(fullContent)
appStore.analysisResult = {
paradigm: detectedParadigm.name,
paradigmType: detectedParadigm.type,
analysis: fullContent,
timestamp: new Date()
}
const result = await appStore.analyzeArticleAction(analysisText.value, detectParadigm)
// 添加到历史记录
addToHistory(detectedParadigm.name, analysisText.value, fullContent)
addToHistory(result.paradigm, analysisText.value, result.content)
} catch (error) {
appStore.analysisResult = {
error: true,
message: error.message
}
} finally {
isAnalyzing.value = false
alert(error.message)
}
}

View File

@@ -130,9 +130,6 @@
<script setup>
import { storeToRefs } from 'pinia'
import { useAppStore } from '../stores/app'
import { buildPrompt } from '../utils/promptBuilder.js'
import DeepSeekAPI from '../api/deepseek.js'
import { marked } from 'marked'
const appStore = useAppStore()
const {
@@ -144,7 +141,6 @@ const {
selectedTags,
customConstraint,
isGenerating,
generatedContent,
showPromptDebug,
apiUrl,
apiKey
@@ -157,10 +153,7 @@ const presetTags = ['Markdown格式', '总分总结构', '数据支撑', '语气
// 添加参考案例
const addReference = () => {
if (!newRefTitle.value || !newRefContent.value) return
references.value.push({
title: newRefTitle.value,
content: newRefContent.value
})
appStore.addReferenceFromAnalysis(newRefTitle.value, newRefContent.value)
newRefTitle.value = ''
newRefContent.value = ''
showRefInput.value = false
@@ -182,54 +175,10 @@ const toggleTag = (tag) => {
// 生成内容
const generateContent = async () => {
if (!apiUrl.value || !apiKey.value || apiKey.value === 'YOUR_KEY') {
alert('请先配置 API 地址和 API Key')
return
}
isGenerating.value = true
generatedContent.value = ''
try {
const api = new DeepSeekAPI({ url: apiUrl.value, key: apiKey.value })
const prompt = buildPrompt(inputTask.value, [...selectedTags.value, customConstraint.value].filter(Boolean), references.value)
const response = await api.generateContent(prompt)
const reader = response.body.getReader()
const decoder = new TextDecoder()
let buffer = ''
while (true) {
const { done, value } = await reader.read()
if (done) break
buffer += decoder.decode(value, { stream: true })
const lines = buffer.split('\n')
buffer = lines.pop() // 最后一行可能不完整,存入 buffer
for (const line of lines) {
const trimmedLine = line.trim()
if (!trimmedLine || !trimmedLine.startsWith('data: ')) continue
const data = trimmedLine.slice(6)
if (data === '[DONE]') break
try {
const parsed = JSON.parse(data)
const content = parsed.choices?.[0]?.delta?.content || ''
if (content) {
generatedContent.value += content
}
} catch (e) {
console.warn('解析流数据失败:', e, data)
}
}
}
await appStore.generateContentAction()
} catch (error) {
console.error('生成内容失败:', error)
generatedContent.value = `## 错误\n\n${error.message}\n\n请检查 API 配置是否正确。`
} finally {
isGenerating.value = false
alert(error.message)
}
}
</script>

View File

@@ -1,6 +1,8 @@
import { defineStore } from 'pinia'
import { ref } from 'vue'
import { config } from '../utils/config.js'
import DeepSeekAPI from '../api/deepseek.js'
import { buildPrompt } from '../utils/promptBuilder.js'
export const useAppStore = defineStore('app', () => {
// 页面状态
@@ -42,6 +44,96 @@ export const useAppStore = defineStore('app', () => {
})
}
// 生成内容动作
const generateContentAction = async () => {
if (!apiUrl.value || !apiKey.value || apiKey.value === 'YOUR_KEY') {
throw new Error('请先配置 API 地址和 API Key')
}
console.log('Store: generateContentAction 启动')
isGenerating.value = true
generatedContent.value = ''
try {
const api = new DeepSeekAPI({ url: apiUrl.value, key: apiKey.value })
const prompt = buildPrompt(
inputTask.value,
[...selectedTags.value, customConstraint.value].filter(Boolean),
references.value
)
console.log('Store: 调用 API 生成内容...')
await api.generateContent(prompt, (content) => {
// console.log('Store: 收到生成内容块', content.length) // Verbose
generatedContent.value += content
})
console.log('Store: 生成内容完成')
} catch (error) {
console.error('Store: 生成内容失败:', error)
throw error
} finally {
isGenerating.value = false
console.log('Store: isGenerating 重置为 false')
}
}
// 分析文章动作
const analyzeArticleAction = async (text, detectParadigmFn) => {
if (!text.trim()) {
throw new Error('请输入要分析的文章内容')
}
console.log('Store: analyzeArticleAction 启动')
isAnalyzing.value = true
analysisResult.value = {
paradigm: '分析中...',
paradigmType: null,
analysis: '',
timestamp: new Date()
}
try {
const api = new DeepSeekAPI({ url: apiUrl.value, key: apiKey.value })
let fullContent = ''
console.log('Store: 调用 API 分析文章...')
await api.analyzeContent(text, (content) => {
fullContent += content
// 实时更新分析结果,用于 UI 展示流式效果
analysisResult.value = {
paradigm: '分析中...',
paradigmType: null,
analysis: fullContent,
timestamp: new Date()
}
})
console.log('Store: 分析文章流式接收完成')
// 分析完成后检测范式类型
const detectedParadigm = detectParadigmFn(fullContent)
console.log('Store: 检测到范式类型:', detectedParadigm.name)
analysisResult.value = {
paradigm: detectedParadigm.name,
paradigmType: detectedParadigm.type,
analysis: fullContent,
timestamp: new Date()
}
return { paradigm: detectedParadigm.name, content: fullContent }
} catch (error) {
console.error('Store: 分析失败:', error)
analysisResult.value = {
error: true,
message: error.message
}
throw error
} finally {
isAnalyzing.value = false
console.log('Store: isAnalyzing 重置为 false')
}
}
// 切换页面
const switchPage = (page) => {
currentPage.value = page
@@ -69,6 +161,8 @@ export const useAppStore = defineStore('app', () => {
// 方法
switchPage,
addReferenceFromAnalysis
addReferenceFromAnalysis,
generateContentAction,
analyzeArticleAction
}
})