scottpan/msh-system
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(ai-chat): 新增豆包API + AI模型配置项支持动态切换

Browse Source 
- 后端新增豆包（火山引擎Ark）API集成：DoubaoController、ToolDoubaoServiceImpl，
  使用OkHttp3 SSE流式对话，兼容OpenAI Chat Completions格式
- 新增DoubaoConfig配置类，读取doubao.api.*配置
- 在eb_system_config表新增ai_chat_model配置项，支持doubao/coze/gemini三种模型切换
- 新增GET /api/front/doubao/ai-model-config接口供前端读取当前模型配置
- 前端ai-nutritionist.vue的sendToAI按系统配置分发到_sendViaDoubao/_sendViaCoze/_sendViaGemini
- 前端models-api.js新增doubaoChatStream/doubaoChat/getAiModelConfig函数
- 附带豆包API测试脚本和数据库初始化SQL

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
msh-agent
2026-04-11 18:03:21 +08:00
parent 58ea76498f
commit b164d8ba11
14 changed files with 1369 additions and 119 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
msh_crmeb_22/crmeb-common/src/main/java/com/zbkj/common/config/DoubaoConfig.java Normal file
View File

@@ -0,0 +1,43 @@
package com.zbkj.common.config;
import lombok.Data;
import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.context.annotation.Configuration;
/**
* 豆包（火山引擎 Ark）API 配置类
* 使用 OpenAI 兼容的 Chat Completions 接口
* +----------------------------------------------------------------------
* | Author:ScottPan
* +----------------------------------------------------------------------
*/
@Data
@Configuration
@ConfigurationProperties(prefix = "doubao.api")
public class DoubaoConfig {
/**
* API 基础 URL
*/
private String baseUrl = "https://ark.cn-beijing.volces.com/api/v3";
/**
* API Key（火山引擎控制台获取）
*/
private String apiKey;
/**
* 默认模型名称
*/
private String model = "doubao-seed-2-0-pro-260215";
/**
* 连接超时时间（毫秒）
*/
private Integer connectTimeout = 30000;
/**
* 读取超时时间（毫秒）
*/
private Integer readTimeout = 120000;
}

3
msh_crmeb_22/crmeb-common/src/main/java/com/zbkj/common/constants/SysConfigConstants.java
View File

@@ -202,4 +202,7 @@ public class SysConfigConstants {
/** 京东云存储端点 */ /** 京东云存储端点 */
public static final String CONFIG_JD_CLOUD_ENDPOINT = "jdEndpoint"; public static final String CONFIG_JD_CLOUD_ENDPOINT = "jdEndpoint";
// ====== AI 模型配置 ======
/** AI 营养师对话模型选择: doubao / coze / gemini */
public static final String CONFIG_KEY_AI_CHAT_MODEL = "ai_chat_model";
} }

58
msh_crmeb_22/crmeb-common/src/main/java/com/zbkj/common/request/doubao/DoubaoChatRequest.java Normal file
View File

@@ -0,0 +1,58 @@
package com.zbkj.common.request.doubao;
import io.swagger.annotations.ApiModel;
import io.swagger.annotations.ApiModelProperty;
import lombok.Data;
import lombok.NoArgsConstructor;
import javax.validation.Valid;
import javax.validation.constraints.NotEmpty;
import javax.validation.constraints.NotNull;
import java.io.Serializable;
import java.util.List;
/**
* 豆包（火山引擎 Ark）Chat Completions 请求 DTO
* 兼容 OpenAI Chat Completions 格式
* +----------------------------------------------------------------------
* | Author:ScottPan
* +----------------------------------------------------------------------
*/
@Data
@NoArgsConstructor
@ApiModel(value = "DoubaoChatRequest", description = "豆包 Chat Completions 请求")
public class DoubaoChatRequest implements Serializable {
private static final long serialVersionUID = 1L;
@Valid
@NotEmpty(message = "messages不能为空")
@ApiModelProperty(value = "消息列表", required = true)
private List<Message> messages;
@ApiModelProperty(value = "是否流式返回", example = "false")
private Boolean stream;
@ApiModelProperty(value = "模型名称（可选，不传则使用服务端默认配置）")
private String model;
@ApiModelProperty(value = "温度参数 (0-2)", example = "0.7")
private Double temperature;
@ApiModelProperty(value = "最大输出 token 数")
private Integer maxTokens;
@Data
@NoArgsConstructor
@ApiModel(value = "Message", description = "单条消息")
public static class Message implements Serializable {
private static final long serialVersionUID = 1L;
@ApiModelProperty(value = "角色: system, user, assistant")
private String role;
@NotNull(message = "content不能为空")
@ApiModelProperty(value = "内容：字符串或多模态数组", required = true)
private Object content;
}
}

74
msh_crmeb_22/crmeb-front/src/main/java/com/zbkj/front/controller/DoubaoController.java Normal file
View File

@@ -0,0 +1,74 @@
package com.zbkj.front.controller;
import com.zbkj.common.constants.SysConfigConstants;
import com.zbkj.common.request.doubao.DoubaoChatRequest;
import com.zbkj.common.result.CommonResult;
import com.zbkj.service.service.SystemConfigService;
import com.zbkj.service.service.tool.ToolDoubaoService;
import io.swagger.annotations.Api;
import io.swagger.annotations.ApiOperation;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.validation.annotation.Validated;
import org.springframework.web.bind.annotation.*;
import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
import javax.servlet.http.HttpServletResponse;
import java.util.HashMap;
import java.util.Map;
/**
* 豆包（火山引擎 Ark）对话接口 + AI 模型配置
* +----------------------------------------------------------------------
* | Author:ScottPan
* +----------------------------------------------------------------------
*/
@RestController
@RequestMapping("/api/front/doubao")
@Api(tags = "豆包 AI 对话")
public class DoubaoController {
@Autowired
private ToolDoubaoService toolDoubaoService;
@Autowired
private SystemConfigService systemConfigService;
/**
* 获取当前 AI 对话模型配置
* 返回 { model: "doubao" | "coze" | "gemini" }
*/
@ApiOperation(value = "获取AI模型配置", notes = "获取当前系统配置的AI对话模型")
@GetMapping("/ai-model-config")
public CommonResult<Map<String, String>> getAiModelConfig() {
String model = systemConfigService.getValueByKey(SysConfigConstants.CONFIG_KEY_AI_CHAT_MODEL);
if (model == null || model.isEmpty()) {
model = "doubao"; // 默认使用豆包
}
Map<String, String> result = new HashMap<>();
result.put("model", model.trim().toLowerCase());
return CommonResult.success(result);
}
/**
* 非流式对话
*/
@ApiOperation(value = "对话", notes = "与豆包模型进行对话")
@PostMapping("/chat")
public Map<String, Object> chat(@RequestBody @Validated DoubaoChatRequest request) {
request.setStream(false);
return toolDoubaoService.chat(request);
}
/**
* 流式对话（SSE）
*/
@ApiOperation(value = "流式对话", notes = "与豆包模型进行流式对话，使用 SSE 实时推送响应")
@PostMapping(value = "/chat/stream", produces = "text/event-stream")
public SseEmitter chatStream(@RequestBody @Validated DoubaoChatRequest request,
HttpServletResponse response) {
response.setHeader("X-Accel-Buffering", "no");
response.setHeader("Cache-Control", "no-cache");
request.setStream(true);
return toolDoubaoService.chatStream(request);
}
}

11
msh_crmeb_22/crmeb-front/src/main/resources/application-sophia.yml
View File

@@ -41,7 +41,7 @@ spring:
min-idle: 0 # 连接池中的最小空闲连接 min-idle: 0 # 连接池中的最小空闲连接
time-between-eviction-runs: -1 #逐出扫描的时间间隔(毫秒) 如果为负数,则不运行逐出线程, 默认-1 time-between-eviction-runs: -1 #逐出扫描的时间间隔(毫秒) 如果为负数,则不运行逐出线程, 默认-1
second: second:
database: 23 # 微信accessToken存储库 database: 3 # 微信accessToken存储库
debug: true debug: true
@@ -87,6 +87,15 @@ coze:
nutrition-analysis-workflow-id: 1180790412263 #饮食打卡记录ai分析 nutrition-analysis-workflow-id: 1180790412263 #饮食打卡记录ai分析
diet-analysis-id: 1180790412263 diet-analysis-id: 1180790412263
# 豆包（火山引擎 Ark）API 配置
doubao:
api:
base-url: https://ark.cn-beijing.volces.com/api/v3
api-key: 18480c26-ebcd-4263-8a6f-48359b8bd65d
model: doubao-seed-2-0-pro-260215
connect-timeout: 30000
read-timeout: 120000
# 腾讯云语音识别配置 # 腾讯云语音识别配置
tencent-asr: tencent-asr:
# API密钥ID # API密钥ID

1
msh_crmeb_22/crmeb-front/src/main/resources/application.yml
View File

@@ -21,6 +21,7 @@ crmeb:
- api/front/upload/imageOuter - api/front/upload/imageOuter
- api/front/coze/** - api/front/coze/**
- api/front/kieai/** - api/front/kieai/**
- api/front/doubao/**
# 配置端口 # 配置端口
server: server:

171
msh_crmeb_22/crmeb-service/src/main/java/com/zbkj/service/service/impl/tool/ToolDoubaoServiceImpl.java Normal file
View File

@@ -0,0 +1,171 @@
package com.zbkj.service.service.impl.tool;
import com.alibaba.fastjson.JSON;
import com.alibaba.fastjson.JSONObject;
import com.zbkj.common.config.DoubaoConfig;
import com.zbkj.common.request.doubao.DoubaoChatRequest;
import com.zbkj.common.utils.SseEmitterUtil;
import com.zbkj.service.service.tool.ToolDoubaoService;
import okhttp3.*;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.http.client.SimpleClientHttpRequestFactory;
import org.springframework.stereotype.Service;
import org.springframework.web.client.RestTemplate;
import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.nio.charset.StandardCharsets;
import java.util.*;
import java.util.concurrent.TimeUnit;
/**
* 豆包（火山引擎 Ark）服务实现类
* 使用 OpenAI 兼容的 Chat Completions 接口 + OkHttp SSE 流式
* +----------------------------------------------------------------------
* | Author:ScottPan
* +----------------------------------------------------------------------
*/
@Service
public class ToolDoubaoServiceImpl implements ToolDoubaoService {
private static final Logger logger = LoggerFactory.getLogger(ToolDoubaoServiceImpl.class);
private static final String CHAT_COMPLETIONS_PATH = "/chat/completions";
@Autowired
private DoubaoConfig config;
/**
* 构建发送给豆包 API 的请求体
*/
private Map<String, Object> buildRequestBody(DoubaoChatRequest request, boolean stream) {
Map<String, Object> body = new LinkedHashMap<>();
// 模型：优先使用请求指定的，否则使用配置默认的
String model = (request.getModel() != null && !request.getModel().isEmpty())
? request.getModel() : config.getModel();
body.put("model", model);
// 消息列表 — 用 FastJSON 解析保留原始结构（与 KieAI 实现一致）
List<Object> messages = JSON.parseArray(JSON.toJSONString(request.getMessages()));
body.put("messages", messages);
body.put("stream", stream);
if (request.getTemperature() != null) {
body.put("temperature", request.getTemperature());
}
if (request.getMaxTokens() != null) {
body.put("max_tokens", request.getMaxTokens());
}
return body;
}
@Override
public Map<String, Object> chat(DoubaoChatRequest request) {
if (config.getApiKey() == null || config.getApiKey().isEmpty()) {
throw new RuntimeException("豆包 API Key 未配置");
}
Map<String, Object> body = buildRequestBody(request, false);
String url = config.getBaseUrl() + CHAT_COMPLETIONS_PATH;
SimpleClientHttpRequestFactory factory = new SimpleClientHttpRequestFactory();
factory.setReadTimeout(config.getReadTimeout());
factory.setConnectTimeout(config.getConnectTimeout());
RestTemplate rt = new RestTemplate(factory);
org.springframework.http.HttpHeaders headers = new org.springframework.http.HttpHeaders();
headers.set("Authorization", "Bearer " + config.getApiKey());
headers.set("Content-Type", "application/json");
org.springframework.http.HttpEntity<String> entity =
new org.springframework.http.HttpEntity<>(JSON.toJSONString(body), headers);
try {
org.springframework.http.ResponseEntity<String> resp =
rt.postForEntity(url, entity, String.class);
if (resp.getBody() != null) {
return JSON.parseObject(resp.getBody(), Map.class);
}
return Collections.emptyMap();
} catch (Exception e) {
logger.error("Doubao chat error", e);
throw new RuntimeException("豆包对话失败: " + e.getMessage());
}
}
@Override
public SseEmitter chatStream(DoubaoChatRequest request) {
if (config.getApiKey() == null || config.getApiKey().isEmpty()) {
throw new RuntimeException("豆包 API Key 未配置");
}
SseEmitter emitter = new SseEmitter(120000L);
Map<String, Object> body = buildRequestBody(request, true);
String url = config.getBaseUrl() + CHAT_COMPLETIONS_PATH;
OkHttpClient client = new OkHttpClient.Builder()
.connectTimeout(config.getConnectTimeout(), TimeUnit.MILLISECONDS)
.readTimeout(config.getReadTimeout(), TimeUnit.MILLISECONDS)
.writeTimeout(30, TimeUnit.SECONDS)
.build();
RequestBody requestBody = RequestBody.create(
MediaType.parse("application/json; charset=utf-8"),
JSON.toJSONString(body));
Request okRequest = new Request.Builder()
.url(url)
.post(requestBody)
.addHeader("Authorization", "Bearer " + config.getApiKey())
.addHeader("Content-Type", "application/json")
.addHeader("Accept", "text/event-stream")
.build();
try {
client.newCall(okRequest).enqueue(new Callback() {
@Override
public void onFailure(Call call, IOException e) {
logger.error("Doubao chat stream request failed", e);
SseEmitterUtil.completeWithError(emitter, e);
}
@Override
public void onResponse(Call call, Response response) throws IOException {
if (!response.isSuccessful() || response.body() == null) {
String errBody = response.body() != null ? response.body().string() : "";
logger.error("Doubao stream failed: {} {}", response.code(), errBody);
SseEmitterUtil.completeWithError(emitter,
new RuntimeException("豆包流式请求失败: " + response.code()));
return;
}
try (BufferedReader reader = new BufferedReader(
new InputStreamReader(response.body().byteStream(), StandardCharsets.UTF_8))) {
String line;
while ((line = reader.readLine()) != null) {
if (line.startsWith("data: ")) {
String json = line.substring(6).trim();
if ("[DONE]".equals(json) || json.isEmpty()) {
break;
}
SseEmitterUtil.send(emitter, json);
}
}
SseEmitterUtil.complete(emitter);
} catch (Exception e) {
logger.error("Doubao chat stream read error", e);
SseEmitterUtil.completeWithError(emitter, e);
}
}
});
} catch (Exception e) {
logger.error("Doubao chat stream error", e);
SseEmitterUtil.completeWithError(emitter, e);
}
return emitter;
}
}

31
msh_crmeb_22/crmeb-service/src/main/java/com/zbkj/service/service/tool/ToolDoubaoService.java Normal file
View File

@@ -0,0 +1,31 @@
package com.zbkj.service.service.tool;
import com.zbkj.common.request.doubao.DoubaoChatRequest;
import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
import java.util.Map;
/**
* 豆包（火山引擎 Ark）服务接口
* +----------------------------------------------------------------------
* | Author:ScottPan
* +----------------------------------------------------------------------
*/
public interface ToolDoubaoService {
/**
* 豆包对话（非流式）
*
* @param request 对话请求参数
* @return 对话响应
*/
Map<String, Object> chat(DoubaoChatRequest request);
/**
* 豆包对话（流式 SSE）
*
* @param request 对话请求参数
* @return SSE 事件流
*/
SseEmitter chatStream(DoubaoChatRequest request);
}

164
msh_single_uniapp/api/models-api.js
View File

@@ -812,6 +812,163 @@ function cozeTextToSpeech(data) {
}) })
} }
/**
* 获取 AI 对话模型配置
* @returns {Promise} { data: { model: "doubao" | "coze" | "gemini" } }
*/
function getAiModelConfig() {
return request('/api/front/doubao/ai-model-config')
}
/**
* 豆包（火山引擎 Ark）- 非流式对话
* @param {object} data 请求参数 { messages: [{role, content}], model?, temperature?, maxTokens? }
* @returns {Promise} 对话响应（OpenAI Chat Completions 格式）
*/
function doubaoChat(data) {
return request('/api/front/doubao/chat', {
method: 'POST',
data: data
})
}
/**
* 豆包（火山引擎 Ark）- 流式对话 (SSE + enableChunked)
* 返回 OpenAI 兼容的 SSE 事件流：choices[0].delta.content
* @param {object} data 请求参数 { messages: [{role, content}] }
* @returns {object} 控制器 { onMessage(deltaText), onError, onComplete, abort }
*/
function doubaoChatStream(data) {
let _onMessage = () => {}
let _onError = () => {}
let _onComplete = () => {}
let _buffer = ''
let _task = null
let _gotChunks = false
const controller = {
onMessage(fn) { _onMessage = fn; return controller },
onError(fn) { _onError = fn; return controller },
onComplete(fn) { _onComplete = fn; return controller },
abort() { if (_task) _task.abort() },
getTask() { return _task }
}
/** 从 SSE data JSON 中提取增量文本（OpenAI 兼容格式） */
const extractDeltaText = (evt) => {
if (evt && Array.isArray(evt.choices) && evt.choices[0]) {
const delta = evt.choices[0].delta
if (delta && typeof delta.content === 'string') return delta.content
const msg = evt.choices[0].message
if (msg && typeof msg.content === 'string') return msg.content
}
return ''
}
const parseSseLines = (text) => {
_buffer += text
const lines = _buffer.split('\n')
_buffer = lines.pop() || ''
for (const line of lines) {
const trimmed = line.trim()
if (!trimmed || trimmed.startsWith(':')) continue
if (trimmed === 'data: [DONE]') continue
if (trimmed.startsWith('data:')) {
const jsonStr = trimmed.slice(5).trim()
if (!jsonStr) continue
try {
const evt = JSON.parse(jsonStr)
const delta = extractDeltaText(evt)
if (delta) _onMessage(delta)
} catch (e) {
console.warn('[doubaoChatStream] JSON parse failed:', jsonStr.slice(0, 200))
}
}
}
}
const parseSseResponseBody = (body) => {
if (!body || typeof body !== 'string') return
const lines = body.split('\n')
for (const line of lines) {
const trimmed = line.trim()
if (!trimmed || trimmed.startsWith(':')) continue
if (trimmed === 'data: [DONE]') continue
if (trimmed.startsWith('data:')) {
const jsonStr = trimmed.slice(5).trim()
if (!jsonStr) continue
try {
const evt = JSON.parse(jsonStr)
const delta = extractDeltaText(evt)
if (delta) _onMessage(delta)
} catch (e) {
console.warn('[doubaoChatStream] JSON parse failed in body fallback:', jsonStr.slice(0, 200))
}
}
}
}
const token = store.state && store.state.app && store.state.app.token
_task = uni.request({
url: `${API_BASE_URL}/api/front/doubao/chat/stream`,
method: 'POST',
data: data,
header: {
'Content-Type': 'application/json',
'Accept': 'text/event-stream',
...(token ? { [TOKENNAME]: token } : {})
},
enableChunked: true,
responseType: 'text',
success: (res) => {
console.log('[doubaoChatStream] success: statusCode=', res.statusCode, '_gotChunks=', _gotChunks)
if (res.statusCode !== 200) {
let errMsg = '请求失败: ' + res.statusCode
try {
const body = typeof res.data === 'string' ? JSON.parse(res.data) : res.data
if (body && body.message) errMsg = body.message
else if (body && body.error && body.error.message) errMsg = body.error.message
} catch (e) { /* keep default message */ }
_onError(new Error(errMsg))
return
}
// 处理 buffer 中残余内容
if (_buffer.trim()) {
parseSseLines('\n')
}
// 不支持 chunked 的环境降级：从完整 response body 解析
if (!_gotChunks && res && res.data) {
const body = typeof res.data === 'string' ? res.data : JSON.stringify(res.data)
parseSseResponseBody(body)
}
_onComplete()
},
fail: (err) => {
console.error('[doubaoChatStream] request fail:', err)
_onError(err)
}
})
if (_task && _task.onChunkReceived) {
_task.onChunkReceived((res) => {
_gotChunks = true
try {
const bytes = new Uint8Array(res.data)
let text = ''
for (let i = 0; i < bytes.length; i++) {
text += String.fromCharCode(bytes[i])
}
text = decodeURIComponent(escape(text))
parseSseLines(text)
} catch (e) {
console.warn('[doubaoChatStream] chunk decode error:', e)
}
})
}
return controller
}
export default { export default {
request, request,
getArticleById, getArticleById,
@@ -835,5 +992,10 @@ export default {
cozeWorkflowStream, cozeWorkflowStream,
cozeWorkflowResume, cozeWorkflowResume,
cozeUploadFile, cozeUploadFile,
cozeTextToSpeech cozeTextToSpeech,
// 豆包 API
doubaoChat,
doubaoChatStream,
// AI 模型配置
getAiModelConfig
} }

341
msh_single_uniapp/pages/tool/ai-nutritionist.vue
View File

@@ -224,6 +224,7 @@ export default {
}, },
data() { data() {
return { return {
aiModel: 'doubao', // 当前AI模型: doubao / coze / gemini（从系统配置读取）
botId: '7591133240535449654', botId: '7591133240535449654',
conversationId: '', conversationId: '',
scrollTop: 0, scrollTop: 0,
@@ -261,6 +262,7 @@ export default {
}); });
this.initRecorder(); this.initRecorder();
this.initAudioContext(); this.initAudioContext();
this.loadAiModelConfig();
}, },
onUnload() { onUnload() {
this.stopRecordTimer(); this.stopRecordTimer();
@@ -277,6 +279,19 @@ export default {
} }
}, },
methods: { methods: {
/** 加载 AI 模型配置 */
loadAiModelConfig() {
api.getAiModelConfig().then(res => {
const data = res && res.data
if (data && data.model) {
this.aiModel = data.model.trim().toLowerCase()
console.log('[ai-nutritionist] AI模型配置:', this.aiModel)
}
}).catch(err => {
console.warn('[ai-nutritionist] 获取AI模型配置失败，使用默认 doubao:', err)
})
},
// 初始化录音管理器 // 初始化录音管理器
initRecorder() { initRecorder() {
// #ifdef MP-WEIXIN || APP-PLUS // #ifdef MP-WEIXIN || APP-PLUS
@@ -744,87 +759,65 @@ export default {
return ''; return '';
}, },
/** 从 Gemini(KieAI) 非流式响应中提取回复文本 */
getGeminiReplyFromResponse(response) {
if (!response || typeof response !== 'object') return ''
let data = response.data
if (typeof data === 'string') {
try { data = JSON.parse(data) } catch (e) { return '' }
}
const payload = this.getGeminiPayload(data || response)
if (!payload) return ''
// OpenAI 形态: choices[0].message.content
if (Array.isArray(payload.choices) && payload.choices[0]) {
const msg = payload.choices[0].message
if (msg) return this.extractReplyContent(msg.content)
const delta = payload.choices[0].delta
if (delta) return this.extractReplyContent(delta.content)
}
// Gemini 形态: candidates[0].content
if (Array.isArray(payload.candidates) && payload.candidates[0]) {
return this.extractReplyContent(payload.candidates[0].content)
}
return ''
},
/** 工具方法：sleep ms 毫秒 */ /** 工具方法：sleep ms 毫秒 */
sleep(ms) { sleep(ms) {
return new Promise(resolve => setTimeout(resolve, ms)); return new Promise(resolve => setTimeout(resolve, ms));
}, },
buildGeminiMessages(content, type) {
if (type === 'text') {
return [{ role: 'user', content: typeof content === 'string' ? content : String(content) }];
}
const parts = Array.isArray(content) ? content : [{ type: 'text', text: String(content) }];
return [{ role: 'user', content: parts }];
},
/** /**
* BUG-005：主路径为 response.data.choices[0].message.content（models-api 已尽量规范 data 形态）。 * 构建 OpenAI 兼容格式的消息列表（用于豆包 API）
* 兼容上游 candidates、或根级即 completion 等变体；仅使用接口返回字段， * @param {string|Array} content 消息内容
* 成功路径不使用 api/tool.js 的 getAIResponse 或本地关键词话术。 * @param {string} type 消息类型 text / image / multimodal
* @returns {Array} messages 数组 [{role, content}]
*/ */
getGeminiReplyFromResponse(response) { buildChatMessages(content, type) {
if (!response || typeof response !== 'object') return ''; if (type === 'text') {
const looksLikeCompletion = (o) => return [{ role: 'user', content: typeof content === 'string' ? content : String(content) }]
o && typeof o === 'object' &&
(Array.isArray(o.choices) || Array.isArray(o.candidates));
let data = response.data;
if (typeof data === 'string') {
try {
data = JSON.parse(data);
} catch (e) {
return '';
} }
if (type === 'image') {
// 图片：尝试构建多模态消息
let fileInfo = content
if (typeof fileInfo === 'string') {
try { fileInfo = JSON.parse(fileInfo) } catch (e) { /* 非JSON */ }
} }
// 主路径：OpenAI 形态 choices[0].message.content（与需求 data.choices[0].message.content 一致） const imageUrl = (fileInfo && fileInfo.url) || (fileInfo && fileInfo.path) || ''
if (data && typeof data === 'object' && Array.isArray(data.choices) && data.choices.length > 0) { if (imageUrl) {
const choice0 = data.choices[0]; return [{
const msg = choice0 && choice0.message; role: 'user',
if (msg && typeof msg === 'object') { content: [
const fromMsg = this.extractReplyContent(msg.content); { type: 'image_url', image_url: { url: imageUrl } },
if (fromMsg.trim()) return fromMsg; { type: 'text', text: '请分析这张图片' }
]
}]
} }
const delta = choice0 && choice0.delta; return [{ role: 'user', content: '我发送了一张图片，请帮我分析' }]
if (delta && typeof delta === 'object') {
const fromDelta = this.extractReplyContent(delta.content);
if (fromDelta.trim()) return fromDelta;
} }
if (choice0 && typeof choice0.text === 'string' && choice0.text.trim()) { // multimodal：直接传多模态 parts
return choice0.text; const parts = Array.isArray(content) ? content : [{ type: 'text', text: String(content) }]
} return [{ role: 'user', content: parts }]
}
if (!looksLikeCompletion(data) && looksLikeCompletion(response)) {
data = response;
}
if (!data || typeof data !== 'object') return '';
const payload = this.getGeminiPayload(data);
if (!payload || typeof payload !== 'object') return '';
const choices = payload.choices;
if (Array.isArray(choices) && choices.length > 0) {
const choice0 = choices[0];
const msg = choice0 && choice0.message;
if (msg && typeof msg === 'object') {
const fromMsg = this.extractReplyContent(msg.content);
if (fromMsg.trim()) return fromMsg;
}
const delta = choice0 && choice0.delta;
if (delta && typeof delta === 'object') {
const fromDelta = this.extractReplyContent(delta.content);
if (fromDelta.trim()) return fromDelta;
}
if (choice0 && typeof choice0.text === 'string' && choice0.text.trim()) {
return choice0.text;
}
}
const cands = payload.candidates;
if (Array.isArray(cands) && cands.length > 0) {
const c0 = cands[0];
if (c0 && typeof c0 === 'object') {
const fromCand = this.extractReplyContent(c0.content);
if (fromCand.trim()) return fromCand;
if (typeof c0.text === 'string' && c0.text.trim()) return c0.text;
}
}
return '';
}, },
async sendToAI(content, type) { async sendToAI(content, type) {
@@ -833,54 +826,15 @@ export default {
this.messageList.push(aiMsg); this.messageList.push(aiMsg);
this.scrollToBottom(); this.scrollToBottom();
const messages = this.buildGeminiMessages(content, type);
// 优先尝试流式输出以改善响应速度感知
try { try {
await new Promise((resolve, reject) => { // 根据系统配置的 aiModel 分发到不同的 API
const ctrl = api.kieaiGeminiChatStream({ messages }) if (this.aiModel === 'coze') {
this._streamCtrl = ctrl await this._sendViaCoze(content, type, aiMsg)
} else if (this.aiModel === 'gemini') {
// 收到第一个 chunk 后切换为 streaming 状态 await this._sendViaGemini(content, type, aiMsg)
ctrl.onMessage((deltaText) => { } else {
if (aiMsg.loading) { // 默认：豆包
aiMsg.loading = false await this._sendViaDoubao(content, type, aiMsg)
aiMsg.streaming = true
}
aiMsg.content += deltaText
this.messageList = [...this.messageList]
this.scrollToBottom()
})
ctrl.onError((err) => {
console.warn('[sendToAI] 流式请求失败，降级为非流式:', err)
this._streamCtrl = null
reject(err)
})
ctrl.onComplete(() => {
this._streamCtrl = null
resolve()
})
})
// 流式完成，检查内容
if (!aiMsg.content.trim()) {
aiMsg.content = '模型未返回有效内容，请稍后重试。'
}
} catch (streamError) {
// 流式失败，降级为非流式请求
try {
const response = await api.kieaiGeminiChat({
messages,
stream: false
});
const reply = this.getGeminiReplyFromResponse(response);
aiMsg.content = reply.trim() ? reply : '模型未返回有效内容，请稍后重试。';
} catch (error) {
console.error('KieAI Gemini 对话失败:', error);
const errText = error && error.message ? String(error.message) : '';
aiMsg.content = errText || '抱歉，处理您的请求时出现错误，请稍后再试。';
} }
} finally { } finally {
aiMsg.loading = false; aiMsg.loading = false;
@@ -897,6 +851,159 @@ export default {
} }
}, },
// ========== 豆包 API（OpenAI 兼容，默认） ==========
async _sendViaDoubao(content, type, aiMsg) {
const messages = this.buildChatMessages(content, type)
try {
await new Promise((resolve, reject) => {
const ctrl = api.doubaoChatStream({ messages })
this._streamCtrl = ctrl
ctrl.onMessage((deltaText) => {
if (aiMsg.loading) { aiMsg.loading = false; aiMsg.streaming = true }
aiMsg.content += deltaText
this.messageList = [...this.messageList]
this.scrollToBottom()
})
ctrl.onError((err) => { this._streamCtrl = null; reject(err) })
ctrl.onComplete(() => { this._streamCtrl = null; resolve() })
})
if (!aiMsg.content.trim()) aiMsg.content = '模型未返回有效内容，请稍后重试。'
} catch (streamError) {
console.warn('[sendToAI:doubao] 流式失败，降级非流式:', streamError)
try {
const response = await api.doubaoChat({ messages, stream: false })
let reply = ''
if (response && response.choices && response.choices[0]) {
const msg = response.choices[0].message
if (msg && msg.content) reply = msg.content
}
aiMsg.content = reply.trim() ? reply : '模型未返回有效内容，请稍后重试。'
} catch (error) {
console.error('豆包对话失败:', error)
aiMsg.content = (error && error.message) || '抱歉，处理您的请求时出现错误，请稍后再试。'
}
}
},
// ========== Coze API ==========
async _sendViaCoze(content, type, aiMsg) {
const userId = this.uid || (uni.getStorageSync('userInfo') || {}).id || 'default_user'
const cozeMessages = this.buildChatMessages(content, type).map(m => ({
...m, content_type: 'text'
}))
const reqData = {
botId: this.botId,
userId: String(userId),
additionalMessages: cozeMessages
}
if (this.conversationId) reqData.conversationId = this.conversationId
try {
await new Promise((resolve, reject) => {
const ctrl = api.cozeChatStream(reqData)
this._streamCtrl = ctrl
ctrl.onMessage((evt) => {
if (evt.conversation_id && !this.conversationId) {
this.conversationId = evt.conversation_id
}
if (evt.event === 'conversation.message.delta' && evt.type === 'answer' && evt.content) {
if (aiMsg.loading) { aiMsg.loading = false; aiMsg.streaming = true }
aiMsg.content += evt.content
this.messageList = [...this.messageList]
this.scrollToBottom()
}
if (evt.event === 'conversation.chat.completed' && evt.conversation_id) {
this.conversationId = evt.conversation_id
}
})
ctrl.onError((err) => { this._streamCtrl = null; reject(err) })
ctrl.onComplete(() => { this._streamCtrl = null; resolve() })
})
if (!aiMsg.content.trim()) aiMsg.content = '模型未返回有效内容，请稍后重试。'
} catch (streamError) {
console.warn('[sendToAI:coze] 流式失败，降级非流式:', streamError)
try {
reqData.stream = false
const response = await api.cozeChat(reqData)
if (response && response.data && response.data.chat) {
const { conversation_id, id: chat_id } = response.data.chat
if (conversation_id) this.conversationId = conversation_id
await this.pollCozeResult(conversation_id, chat_id, aiMsg)
} else {
aiMsg.content = '模型未返回有效内容，请稍后重试。'
}
} catch (error) {
console.error('Coze 对话失败:', error)
aiMsg.content = (error && error.message) || '抱歉，处理您的请求时出现错误，请稍后再试。'
}
}
},
// ========== Gemini API (KieAI) ==========
async _sendViaGemini(content, type, aiMsg) {
const messages = this.buildChatMessages(content, type)
try {
await new Promise((resolve, reject) => {
const ctrl = api.kieaiGeminiChatStream({ messages })
this._streamCtrl = ctrl
ctrl.onMessage((deltaText) => {
if (aiMsg.loading) { aiMsg.loading = false; aiMsg.streaming = true }
aiMsg.content += deltaText
this.messageList = [...this.messageList]
this.scrollToBottom()
})
ctrl.onError((err) => { this._streamCtrl = null; reject(err) })
ctrl.onComplete(() => { this._streamCtrl = null; resolve() })
})
if (!aiMsg.content.trim()) aiMsg.content = '模型未返回有效内容，请稍后重试。'
} catch (streamError) {
console.warn('[sendToAI:gemini] 流式失败，降级非流式:', streamError)
try {
const response = await api.kieaiGeminiChat({ messages, stream: false })
const reply = this.getGeminiReplyFromResponse(response)
aiMsg.content = reply.trim() ? reply : '模型未返回有效内容，请稍后重试。'
} catch (error) {
console.error('Gemini 对话失败:', error)
aiMsg.content = (error && error.message) || '抱歉，处理您的请求时出现错误，请稍后再试。'
}
}
},
/**
* 非流式 Coze 降级：轮询对话状态并获取回复
*/
async pollCozeResult(conversationId, chatId, aiMsg) {
const maxAttempts = 30
const interval = 2000
for (let i = 0; i < maxAttempts; i++) {
await this.sleep(interval)
try {
const res = await api.cozeRetrieveChat({ conversationId, chatId })
const status = res && res.data && res.data.status
if (status === 'completed') {
const msgRes = await api.cozeMessageList({ conversationId, chatId })
if (msgRes && msgRes.data && Array.isArray(msgRes.data)) {
const answer = msgRes.data.find(m => m.role === 'assistant' && m.type === 'answer')
if (answer && answer.content) {
aiMsg.content = answer.content
this.messageList = [...this.messageList]
this.scrollToBottom()
return
}
}
aiMsg.content = '模型未返回有效内容，请稍后重试。'
return
} else if (status === 'failed') {
aiMsg.content = '对话处理失败，请稍后重试。'
return
}
} catch (e) {
console.warn('[pollCozeResult] 轮询出错:', e)
}
}
aiMsg.content = '等待超时，请稍后重试。'
},
// ---------- TTS 方法 ---------- // ---------- TTS 方法 ----------
initAudioContext() { initAudioContext() {

21
sql/add_ai_chat_model_config.sql Normal file
View File

@@ -0,0 +1,21 @@
-- ============================================================
-- AI 营养师对话模型配置项
-- 在 eb_system_config 表中新增模型选择配置
-- 可选值: doubao / coze / gemini
-- ============================================================
-- 插入配置项（如果不存在）
INSERT INTO eb_system_config (`name`, `title`, `value`, `form_id`, `status`, `create_time`, `update_time`)
SELECT 'ai_chat_model', 'AI对话模型选择(doubao/coze/gemini)', 'doubao', 0, 0, NOW(), NOW()
FROM DUAL
WHERE NOT EXISTS (
SELECT 1 FROM eb_system_config WHERE `name` = 'ai_chat_model'
);
-- 如果需要修改模型，可以直接更新：
-- UPDATE eb_system_config SET `value` = 'coze', `update_time` = NOW() WHERE `name` = 'ai_chat_model';
-- UPDATE eb_system_config SET `value` = 'gemini', `update_time` = NOW() WHERE `name` = 'ai_chat_model';
-- UPDATE eb_system_config SET `value` = 'doubao', `update_time` = NOW() WHERE `name` = 'ai_chat_model';
-- 验证：
-- SELECT * FROM eb_system_config WHERE `name` = 'ai_chat_model';

104
test-doubao-api.py Normal file
View File

@@ -0,0 +1,104 @@
#!/usr/bin/env python3
"""
豆包（火山引擎 Ark）API 快速测试
用法: python3 test-doubao-api.py
"""
import json, time, sys
try:
import requests
except ImportError:
print("需要安装 requests: pip install requests")
sys.exit(1)
API_KEY = "18480c26-ebcd-4263-8a6f-48359b8bd65d"
BASE_URL = "https://ark.cn-beijing.volces.com/api/v3"
MODEL = "doubao-seed-2-0-pro-260215"
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {API_KEY}"
}
print("=" * 50)
print(" 豆包 API 测试")
print("=" * 50)
# ---- 测试 1: 非流式 ----
print("\n【测试 1】非流式调用")
payload = {
"model": MODEL,
"messages": [{"role": "user", "content": "你好，请用一句话介绍你自己"}],
"stream": False
}
start = time.time()
try:
resp = requests.post(f"{BASE_URL}/chat/completions", headers=headers, json=payload, timeout=30)
elapsed = (time.time() - start) * 1000
print(f"HTTP 状态码: {resp.status_code}")
print(f"耗时: {elapsed:.0f} ms")
if resp.status_code == 200:
data = resp.json()
choices = data.get("choices", [])
if choices:
print(f"回复: {choices[0]['message']['content']}")
usage = data.get("usage", {})
print(f"Token: prompt={usage.get('prompt_tokens',0)}, completion={usage.get('completion_tokens',0)}, total={usage.get('total_tokens',0)}")
print(f"模型: {data.get('model', '?')}")
else:
print(f"错误: {resp.text[:500]}")
except Exception as e:
elapsed = (time.time() - start) * 1000
print(f"请求失败 ({elapsed:.0f}ms): {e}")
# ---- 测试 2: 流式 ----
print("\n【测试 2】流式调用 (SSE)")
payload_stream = {
"model": MODEL,
"messages": [{"role": "user", "content": "简单说一下健康饮食的三个要点，每个要点一句话"}],
"stream": True
}
start2 = time.time()
first_token_time = None
full_content = ""
try:
resp2 = requests.post(f"{BASE_URL}/chat/completions", headers=headers, json=payload_stream, stream=True, timeout=30)
print(f"HTTP 状态码: {resp2.status_code}")
if resp2.status_code == 200:
print("流式输出: ", end="", flush=True)
for line in resp2.iter_lines(decode_unicode=True):
if not line:
continue
if line.startswith("data: "):
json_str = line[6:].strip()
if json_str == "[DONE]":
break
try:
evt = json.loads(json_str)
delta = evt.get("choices", [{}])[0].get("delta", {}).get("content", "")
if delta:
if first_token_time is None:
first_token_time = time.time()
full_content += delta
print(delta, end="", flush=True)
except json.JSONDecodeError:
pass
total_time = (time.time() - start2) * 1000
ttft = (first_token_time - start2) * 1000 if first_token_time else 0
print(f"\n---")
print(f"首字时间 (TTFT): {ttft:.0f} ms")
print(f"总耗时: {total_time:.0f} ms")
print(f"回复长度: {len(full_content)} 字符")
else:
print(f"错误: {resp2.text[:500]}")
except Exception as e:
elapsed2 = (time.time() - start2) * 1000
print(f"\n请求失败 ({elapsed2:.0f}ms): {e}")
print("\n" + "=" * 50)
print(" 测试完成")
print("=" * 50)

144
test-doubao-api.sh Normal file
View File

@@ -0,0 +1,144 @@
#!/bin/bash
# 豆包 API 连通性测试脚本
# 用法: bash test-doubao-api.sh
API_KEY="18480c26-ebcd-4263-8a6f-48359b8bd65d"
BASE_URL="https://ark.cn-beijing.volces.com/api/v3"
MODEL="doubao-seed-2-0-pro-260215"
echo "============================================"
echo " 豆包（火山引擎 Ark）API 测试"
echo "============================================"
echo ""
# ---- 测试 1: 非流式调用 ----
echo "【测试 1】非流式调用"
echo "请求: 你好，请用一句话介绍你自己"
echo "---"
START=$(date +%s%3N 2>/dev/null || python3 -c "import time; print(int(time.time()*1000))")
RESPONSE=$(curl -s -w "\n---HTTP_CODE:%{http_code}---" \
-X POST "${BASE_URL}/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer ${API_KEY}" \
-d "{
\"model\": \"${MODEL}\",
\"messages\": [{\"role\": \"user\", \"content\": \"你好，请用一句话介绍你自己\"}],
\"stream\": false
}" 2>&1)
END=$(date +%s%3N 2>/dev/null || python3 -c "import time; print(int(time.time()*1000))")
ELAPSED=$((END - START))
HTTP_CODE=$(echo "$RESPONSE" | grep -oP '(?<=---HTTP_CODE:)\d+(?=---)')
BODY=$(echo "$RESPONSE" | sed 's/---HTTP_CODE:[0-9]*---//')
echo "HTTP 状态码: ${HTTP_CODE}"
echo "耗时: ${ELAPSED} ms"
if [ "$HTTP_CODE" = "200" ]; then
# 提取回复内容
CONTENT=$(echo "$BODY" | python3 -c "
import sys, json
data = json.load(sys.stdin)
choices = data.get('choices', [])
if choices:
print('回复:', choices[0].get('message', {}).get('content', '(empty)'))
usage = data.get('usage', {})
print(f'Token: prompt={usage.get(\"prompt_tokens\",0)}, completion={usage.get(\"completion_tokens\",0)}, total={usage.get(\"total_tokens\",0)}')
print(f'模型: {data.get(\"model\", \"?\")}')
" 2>&1)
echo "$CONTENT"
else
echo "错误响应: $BODY"
fi
echo ""
echo ""
# ---- 测试 2: 流式调用 ----
echo "【测试 2】流式调用 (SSE)"
echo "请求: 简单说一下健康饮食的三个要点"
echo "---"
START2=$(date +%s%3N 2>/dev/null || python3 -c "import time; print(int(time.time()*1000))")
# 流式请求，逐行输出
STREAM_OUTPUT=""
FIRST_CHUNK_TIME=""
curl -s -N \
-X POST "${BASE_URL}/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer ${API_KEY}" \
-d "{
\"model\": \"${MODEL}\",
\"messages\": [{\"role\": \"user\", \"content\": \"简单说一下健康饮食的三个要点，每个要点一句话\"}],
\"stream\": true
}" 2>/dev/null | while IFS= read -r line; do
if [[ "$line" == data:* ]]; then
JSON_DATA="${line#data: }"
if [ "$JSON_DATA" = "[DONE]" ]; then
break
fi
# 提取 delta content
DELTA=$(echo "$JSON_DATA" | python3 -c "
import sys, json
try:
d = json.load(sys.stdin)
c = d.get('choices',[{}])[0].get('delta',{}).get('content','')
if c: print(c, end='', flush=True)
except: pass
" 2>/dev/null)
if [ -z "$FIRST_CHUNK_TIME" ] && [ -n "$DELTA" ]; then
FIRST_CHUNK_TIME=$(date +%s%3N 2>/dev/null || python3 -c "import time; print(int(time.time()*1000))")
fi
fi
done
END2=$(date +%s%3N 2>/dev/null || python3 -c "import time; print(int(time.time()*1000))")
ELAPSED2=$((END2 - START2))
echo ""
echo "---"
echo "流式总耗时: ${ELAPSED2} ms"
echo ""
echo ""
# ---- 测试 3: 通过后端接口调用（需先启动后端服务） ----
echo "【测试 3】通过后端接口调用（需后端服务运行中）"
echo "请确认后端已启动，默认地址: http://localhost:8081"
echo "---"
BACKEND_URL="http://localhost:8081/api/front/doubao/chat"
RESP3=$(curl -s -w "\n---HTTP_CODE:%{http_code}---" \
-X POST "${BACKEND_URL}" \
-H "Content-Type: application/json" \
-d "{
\"messages\": [{\"role\": \"user\", \"content\": \"你好\"}]
}" 2>&1)
HTTP3=$(echo "$RESP3" | grep -oP '(?<=---HTTP_CODE:)\d+(?=---)')
BODY3=$(echo "$RESP3" | sed 's/---HTTP_CODE:[0-9]*---//')
if [ "$HTTP3" = "200" ]; then
echo "后端接口调用成功! HTTP $HTTP3"
echo "$BODY3" | python3 -c "
import sys, json
data = json.load(sys.stdin)
choices = data.get('choices', [])
if choices:
print('回复:', choices[0].get('message', {}).get('content', '(empty)')[:100])
" 2>&1
else
echo "后端接口调用失败: HTTP ${HTTP3:-连接失败}"
echo "（如果后端未启动，这是正常的）"
fi
echo ""
echo "============================================"
echo " 测试完成"
echo "============================================"

322
客户反馈分析报告.md Normal file
View File

@@ -0,0 +1,322 @@
# 测试问题与优化建议 — 修改解决方案
**分析日期：** 2026年4月11日
**项目：** msh_single_uniapp（UniApp 小程序）
**涉及页面：** calculator-result / ai-nutritionist / food-encyclopedia
---
## 一、总览：需处理项 vs 暂不处理项
| # | 问题 | 决策 | 涉及文件 |
|---|------|------|----------|
| 1 | 食物份数 → 克数 | **采纳** | `pages/tool/calculator-result.vue` |
| 2 | AI对话交互按钮 | **部分采纳** | `pages/tool/ai-nutritionist.vue` |
| 3 | AI回复速度慢 | **已优化（切换Coze流式API）** | `ai-nutritionist.vue` + `models-api.js` |
| 4 | 食物百科图片缺失 | **不处理**（后台可改） | — |
| 5 | 百科卡片误导点击 | **采纳**（跳转详情页） | `pages/tool/food-encyclopedia.vue` |
| 6 | 角色权限与分销 | **暂不处理**（后台人工配置） | — |
| 7 | 百科点击报错 Bug | **必须修复** | `pages/tool/food-encyclopedia.vue` |
---
## 二、需修改的问题及代码级解决方案
---
### 问题1：食物份数建议改为克数
**页面：** `pages/tool/calculator-result.vue`
**现状分析：**
当前"食物份数建议"卡片（约第94-113行）中，食物列表渲染逻辑为：
```html
<text class="food-portion">{{ item.portion }} 份</text>
```
数据结构 `foodList` 中每个 item 包含 `{ number, name, portion }`，`portion` 为份数值。
**修改方案：**
1. **后端接口改造**（推荐）：`getCalculatorResult` 接口返回数据中增加 `gram` 字段，或将 `portion` 的含义改为克数。
2. **前端模板修改**（`calculator-result.vue` 约第94-113行）：
```html
<!-- 修改前 -->
<text class="card-title">食物份数建议</text>
...
<text class="food-portion">{{ item.portion }} 份</text>
<!-- 修改后 -->
<text class="card-title">每日食物建议</text>
...
<text class="food-portion">{{ item.gram || item.portion }} 克</text>
```
3. **数据层适配**（`applyResult` 方法，约第320-350行）：解析接口返回时，确保 `foodList` 中的 item 携带 `gram` 字段。如后端暂未改造，可前端用换算公式临时过渡：
```javascript
// 在 applyResult 方法中
this.foodList = (res.data.foodList || []).map(item => ({
...item,
gram: item.gram || Math.round(item.portion * item.gramPerServing) || item.portion
}))
```
**工作量估计：** 前端 0.5天，后端（如需改接口）0.5天
---
### 问题2：AI营养师对话页面新增交互按钮
**页面：** `pages/tool/ai-nutritionist.vue`
**现状分析：**
当前每条 AI 消息仅有 **1个操作按钮**：TTS 语音朗读（约第84-90行）。消息数据结构为 `{ role, content, type, loading, streaming }`。AI 回复通过 `api.kieaiGeminiChat()` 调用（实际走 `/api/front/kieai/gemini/chat`，`stream: false` 非流式），整体响应一次性返回。
**部分采纳后的需求：** 新增"复制"、"重新生成"、"删除"按钮（语音朗读已有）。
**修改方案：**
在消息气泡下方的操作区域（约第84-90行 TTS 按钮位置），扩展为按钮组：
```html
<!-- 修改前：仅 TTS 按钮 -->
<view class="tts-play-btn" v-if="msg.role === 'ai' && !msg.loading && !msg.streaming && msg.content && msg.type !== 'image'" @click="...">
<text>{{ ttsPlayingIndex === index ? '⏹' : '▶' }}</text>
</view>
<!-- 修改后：操作按钮组 -->
<view class="msg-actions" v-if="msg.role === 'ai' && !msg.loading && !msg.streaming && msg.content && msg.type !== 'image'">
<!-- 复制 -->
<view class="action-btn" @click="copyMessage(index)">
<text class="action-icon">📋</text>
</view>
<!-- 重新生成（仅最后一条AI消息显示） -->
<view class="action-btn" v-if="isLastAiMessage(index)" @click="regenerateMessage(index)">
<text class="action-icon">🔄</text>
</view>
<!-- 语音朗读（已有） -->
<view class="action-btn" @click="ttsPlayingIndex === index ? stopTTS() : playTTS(index)">
<text class="action-icon">{{ ttsPlayingIndex === index ? '⏹' : '▶' }}</text>
</view>
<!-- 删除 -->
<view class="action-btn" @click="deleteMessage(index)">
<text class="action-icon">🗑️</text>
</view>
</view>
```
**新增 methods：**
```javascript
// 复制消息
copyMessage(index) {
const msg = this.messageList[index]
uni.setClipboardData({
data: msg.content,
success: () => uni.showToast({ title: '已复制', icon: 'success' })
})
},
// 删除单条消息
deleteMessage(index) {
uni.showModal({
title: '提示',
content: '确定删除这条消息吗？',
success: (res) => {
if (res.confirm) {
this.messageList.splice(index, 1)
}
}
})
},
// 重新生成（重发上一条用户消息）
regenerateMessage(index) {
// 找到该AI消息对应的上一条用户消息
let userMsgIndex = index - 1
while (userMsgIndex >= 0 && this.messageList[userMsgIndex].role !== 'user') {
userMsgIndex--
}
if (userMsgIndex < 0) return
// 移除当前AI回复
this.messageList.splice(index, 1)
// 重新发送
this.sendToAI()
},
// 判断是否为最后一条AI消息
isLastAiMessage(index) {
for (let i = this.messageList.length - 1; i >= 0; i--) {
if (this.messageList[i].role === 'ai') return i === index
}
return false
}
```
**新增样式：**
```scss
.msg-actions {
display: flex;
gap: 16rpx;
margin-top: 8rpx;
justify-content: flex-start;
}
.action-btn {
width: 56rpx;
height: 56rpx;
display: flex;
align-items: center;
justify-content: center;
border-radius: 50%;
background: #f4f5f7;
}
.action-icon {
font-size: 24rpx;
}
```
**工作量估计：** 1天
---
### 问题3：AI回复速度慢（已优化 — 切换至 Coze API）
**优化前调用链路：**
```
小程序 → POST /api/front/kieai/gemini/chat/stream
→ Spring KieAIController (SseEmitter)
→ OkHttp3 异步 POST https://api.kie.ai/gemini-2.5-flash/v1/chat/completions
→ KieAI 代理网关 → Gemini 2.5 Flash 模型
```
**优化后调用链路：**
```
小程序 → POST /api/front/coze/chat/stream
→ Spring CozeController (SseEmitter, 120s timeout, 15s heartbeat)
→ Coze 官方 SDK (client.chat().stream())
→ https://api.coze.cn → Coze Bot (7591133240535449654)
```
**已完成的优化：**
1. **前端切换至 Coze 流式 API：** `sendToAI()` 改为调用 `api.cozeChatStream()`，直接使用 Coze 的流式事件（`conversation.message.delta`），逐字渲染 AI 回复。首字可见时间大幅缩短。
2. **Coze 事件解析：** 正确解析 Coze SSE 事件格式 `{ event, conversation_id, chat_id, content, role, type }`，仅累积 `event=conversation.message.delta` 且 `type=answer` 的文本增量。
3. **多轮对话支持：** 自动保存 `conversation_id`，后续消息在同一会话中进行，Coze Bot 可获取完整上下文。
4. **消息格式适配：** 新增 `buildCozeMessages()` / `buildCozeRequest()` 方法，将用户输入（文本/图片/多模态）转换为 Coze 格式的 `additionalMessages`。
5. **降级策略：** 流式失败时自动降级为非流式 `api.cozeChat()`，并通过 `pollCozeResult()` 轮询获取最终回复。
**Coze 相比 KieAI 代理的优势：**
- 去掉一层 KieAI 代理网关，减少一跳网络延迟
- Coze 后端已配置 `X-Accel-Buffering: no` + `Cache-Control: no-cache`，SSE 不会被 nginx 缓冲
- Coze SDK 内置连接池管理，不再每次请求 new OkHttpClient
- 内置 15s heartbeat 防止连接超时断开
- 支持通过 `conversation_id` 进行多轮上下文对话
---
### 问题4：食物百科图片缺失
**决策：不处理。** 图片数据通过商城 PC 管理后台维护，属运营侧工作。
---
### 问题5：百科卡片点击 → 跳转食物详情页
**页面：** `pages/tool/food-encyclopedia.vue`
**现状分析：**
当前卡片已绑定了 `@click="goToFoodDetail(item)"`（约第103行），`goToFoodDetail` 方法（约第1012-1030行）会尝试提取 item 的 id 并跳转到 `/pages/tool/food-detail`。项目中已存在 `pages/tool/food-detail.vue` 页面。
**问题：** 跳转逻辑本身已实现，但存在 Bug（见问题7），导致点击报错而非正常跳转，用户看到的效果就是"点击无反应"或"点击后闪退"。
**修改方案：** 修复问题7的 Bug 后，卡片点击跳转详情页的功能即可正常工作。需确认：
1. `food-detail.vue` 详情页内容是否完整可用
2. `pages.json` 中是否已注册 `/pages/tool/food-detail` 路由
**工作量估计：** 与问题7合并处理，0.5天
---
### 问题6：角色权限与分销系统
**决策：暂不处理。** 临时方案为运营人员在后台手动为指定用户账号配置角色标签（医生/护士/普通用户）。
---
### 问题7（Bug）：食物百科点击报错 TypeError
**页面：** `pages/tool/food-encyclopedia.vue`
**错误信息：**
```
TypeError: Cannot read property 'id' of undefined
at pickNumericIdStr (food-encyclopedia.vue:1015)
at VueComponent.goToFoodDetail (food-encyclopedia.vue:1022)
```
**根因分析：**
`goToFoodDetail(item)` 方法（第1012行）中，内部闭包 `pickNumericIdStr` 直接访问 `item.id`（第1015行），但未做空值校验。当 `filteredFoodList` 中的某个 item 为 `undefined` 或 `null` 时（可能因 API 返回脏数据、`normalizeFoodItem` 边界情况），点击即触发 TypeError。
**修复方案：**
在 `goToFoodDetail` 方法开头增加防御性校验（约第1012行）：
```javascript
goToFoodDetail(item) {
// 防御性校验：避免 item 为空时报错
if (!item || typeof item !== 'object') {
console.warn('[food-encyclopedia] goToFoodDetail: item 为空，跳过跳转')
return
}
const pickNumericIdStr = () => {
const cands = [item.id, item.foodId, item.food_id, item.v2FoodId, item.v2_food_id, item.foodID]
// ... 原有逻辑不变
}
// ... 后续逻辑不变
}
```
同时建议加固 `filteredFoodList` 计算属性（约第261行），确保过滤掉无效项：
```javascript
filteredFoodList() {
const list = (this.foodList || []).filter(item => item != null && typeof item === 'object' && item.name)
// ... 后续过滤逻辑不变
}
```
**工作量估计：** 0.5天（含测试验证）
---
## 三、执行优先级排序
| 优先级 | 问题 | 预估工时 | 负责方 |
|--------|------|----------|--------|
| **P0** | #7 百科点击报错 Bug | 0.5天 | 前端 |
| **P1** | #1 份数→克数 | 0.5-1天 | 前端+后端 |
| **P1** | #5 百科卡片跳转详情页 | 与#7合并 | 前端 |
| **P2** | #3 AI响应速度（已切换Coze流式API） | ✅ 已完成 | 前端 |
| **P2** | #2 AI对话新增操作按钮 | 1天 | 前端 |
| **—** | #4 百科图片 | 不处理 | 运营 |
| **—** | #6 角色权限 | 暂不处理 | 运营（后台手动） |
**总预估：** 前端约 3-4天，后端约 2-3天
---
*基于 msh_single_uniapp 项目代码分析生成，所有代码引用均已核对源文件。*