diff --git a/apps/desktop/src-tauri/src/business/services/template_service.rs b/apps/desktop/src-tauri/src/business/services/template_service.rs index 41cb8dc..3573b6d 100644 --- a/apps/desktop/src-tauri/src/business/services/template_service.rs +++ b/apps/desktop/src-tauri/src/business/services/template_service.rs @@ -1,9 +1,9 @@ use anyhow::{Result, anyhow}; use std::sync::Arc; use rusqlite::{params, Row}; -use chrono::{DateTime, Utc}; +use chrono; use serde_json; -use tracing::{info, warn, debug, error}; +use tracing::{info, warn, error}; use crate::data::models::template::{ Template, TemplateMaterial, Track, TrackSegment, CanvasConfig, diff --git a/apps/desktop/src-tauri/src/business/services/tests/template_service_tests.rs b/apps/desktop/src-tauri/src/business/services/tests/template_service_tests.rs index 9bf07a2..c0240ff 100644 --- a/apps/desktop/src-tauri/src/business/services/tests/template_service_tests.rs +++ b/apps/desktop/src-tauri/src/business/services/tests/template_service_tests.rs @@ -8,7 +8,7 @@ mod template_service_tests { }; use crate::infrastructure::database::Database; use std::sync::Arc; - use tempfile::TempDir; + use chrono::Utc; /// 创建测试数据库 diff --git a/apps/desktop/src-tauri/src/infrastructure/database.rs b/apps/desktop/src-tauri/src/infrastructure/database.rs index 0cf018e..af1cae8 100644 --- a/apps/desktop/src-tauri/src/infrastructure/database.rs +++ b/apps/desktop/src-tauri/src/infrastructure/database.rs @@ -3,7 +3,7 @@ use std::path::PathBuf; use std::sync::{Arc, Mutex}; use anyhow::{Result, anyhow}; use std::ops::{Deref, DerefMut}; -use uuid; + use crate::infrastructure::connection_pool::{ConnectionPool, ConnectionPoolConfig, PooledConnectionHandle}; mod migrations; diff --git a/apps/desktop/src-tauri/src/presentation/commands/markdown_commands.rs b/apps/desktop/src-tauri/src/presentation/commands/markdown_commands.rs index 13f56a2..5876120 100644 --- a/apps/desktop/src-tauri/src/presentation/commands/markdown_commands.rs +++ b/apps/desktop/src-tauri/src/presentation/commands/markdown_commands.rs @@ -1,6 +1,6 @@ use crate::infrastructure::markdown_parser::{ - MarkdownParser, MarkdownParserConfig, MarkdownParseResult, - OutlineItem, LinkInfo, ValidationResult, MarkdownNode, Range, Position + MarkdownParser, MarkdownParserConfig, MarkdownParseResult, + OutlineItem, LinkInfo, ValidationResult, MarkdownNode }; use anyhow::Result; use serde::{Deserialize, Serialize}; diff --git a/apps/desktop/src-tauri/src/presentation/commands/outfit_search_commands.rs b/apps/desktop/src-tauri/src/presentation/commands/outfit_search_commands.rs index cb8752c..2bbcb99 100644 --- a/apps/desktop/src-tauri/src/presentation/commands/outfit_search_commands.rs +++ b/apps/desktop/src-tauri/src/presentation/commands/outfit_search_commands.rs @@ -5,7 +5,7 @@ use crate::app_state::AppState; use crate::data::models::gemini_analysis::{AnalyzeImageRequest, AnalyzeImageResponse}; use crate::data::models::outfit_search::{ SearchRequest, SearchResponse, LLMQueryRequest, LLMQueryResponse, - OutfitSearchGlobalConfig + OutfitSearchGlobalConfig, SearchResult, ProductInfo, SearchFilterBuilder }; use crate::infrastructure::gemini_service::{GeminiService, GeminiConfig}; @@ -46,17 +46,29 @@ pub async fn analyze_outfit_image( #[tauri::command] pub async fn search_similar_outfits( _state: State<'_, AppState>, - _request: SearchRequest, + request: SearchRequest, ) -> Result { - // TODO: 实现真实的搜索逻辑 + let start_time = std::time::Instant::now(); + + // 创建Gemini服务实例用于获取访问令牌 + let config = GeminiConfig::default(); + let mut gemini_service = GeminiService::new(Some(config)) + .map_err(|e| format!("Failed to create GeminiService: {}", e))?; + + // 执行直接的 Vertex AI Search 搜索 + let search_results = execute_vertex_ai_search(&mut gemini_service, &request).await + .map_err(|e| { + eprintln!("搜索失败: {}", e); + format!("搜索失败: {}", e) + })?; + + let search_time_ms = start_time.elapsed().as_millis() as u64; - // 执行搜索（暂时返回模拟数据） - // TODO: 实现真实的搜索逻辑 Ok(SearchResponse { - results: vec![], - total_size: 0, - next_page_token: None, - search_time_ms: 100, + results: search_results.results, + total_size: search_results.total_size, + next_page_token: search_results.next_page_token, + search_time_ms, searched_at: chrono::Utc::now(), }) } @@ -264,10 +276,321 @@ mod tests { storage_bucket_name: "test-bucket".to_string(), data_store_id: "test-store".to_string(), }; - + let serialized = serde_json::to_string(&config_info).unwrap(); assert!(serialized.contains("test-project")); } + + #[test] + fn test_search_request_creation() { + use crate::data::models::outfit_search::{SearchConfig, RelevanceThreshold}; + + let request = SearchRequest { + query: "牛仔裤搭配".to_string(), + config: SearchConfig { + relevance_threshold: RelevanceThreshold::High, + categories: vec!["上装".to_string(), "下装".to_string()], + environments: vec!["Outdoor".to_string()], + color_filters: std::collections::HashMap::new(), + design_styles: std::collections::HashMap::new(), + max_keywords: 10, + }, + page_size: 9, + page_offset: 0, + }; + + assert_eq!(request.query, "牛仔裤搭配"); + assert_eq!(request.page_size, 9); + assert_eq!(request.config.categories.len(), 2); + } + + #[test] + fn test_search_filter_builder() { + use crate::data::models::outfit_search::SearchConfig; + + let mut config = SearchConfig::default(); + config.categories = vec!["上装".to_string()]; + config.environments = vec!["Outdoor".to_string()]; + + let filters = SearchFilterBuilder::build_filters(&config); + assert!(filters.contains("上装") || filters.contains("Outdoor")); + } + + #[test] + fn test_query_keywords_builder() { + use crate::data::models::outfit_search::SearchConfig; + + let mut config = SearchConfig::default(); + config.environments = vec!["Outdoor".to_string()]; + + let keywords = SearchFilterBuilder::build_query_keywords(&config); + assert!(keywords.contains(&"Outdoor".to_string())); + } +} + +/// 执行 Vertex AI Search 搜索 +async fn execute_vertex_ai_search( + _gemini_service: &mut GeminiService, + request: &SearchRequest, +) -> Result { + // 1. 获取访问令牌（通过直接调用API） + let access_token = get_google_access_token().await?; + + // 2. 获取全局配置 + let global_config = OutfitSearchGlobalConfig::default(); + + // 3. 构建搜索过滤器 + let search_filter = SearchFilterBuilder::build_filters(&request.config); + + // 4. 构建查询关键词 + let query_keywords = SearchFilterBuilder::build_query_keywords(&request.config); + + // 5. 组合查询字符串 + let enhanced_query = if query_keywords.is_empty() { + request.query.clone() + } else { + format!("{} {}", request.query, query_keywords.join(" ")) + }; + + // 6. 构建请求负载 + let mut payload = serde_json::json!({ + "query": enhanced_query, + "relevanceThreshold": request.config.relevance_threshold.to_value().to_string(), + "relevanceScoreSpec": { + "returnRelevanceScore": true + }, + "pageSize": request.page_size, + "offset": request.page_offset + }); + + // 7. 添加过滤器（如果有） + if !search_filter.is_empty() { + payload["filter"] = serde_json::Value::String(search_filter); + } + + // 8. 构建请求URL + let search_url = format!( + "https://discoveryengine.googleapis.com/v1beta/projects/{}/locations/global/collections/default_collection/engines/{}/servingConfigs/default_search:search", + global_config.google_project_id, + global_config.vertex_ai_app_id + ); + + // 9. 发送HTTP请求 + let client = reqwest::Client::new(); + let response = client + .post(&search_url) + .header("Authorization", format!("Bearer {}", access_token)) + .header("Content-Type", "application/json") + .json(&payload) + .send() + .await?; + + let status = response.status(); + let response_text = response.text().await?; + + if !status.is_success() { + return Err(anyhow::anyhow!("Vertex AI Search 请求失败: {} - {}", status, response_text)); + } + + // 10. 解析响应 + let vertex_response: serde_json::Value = serde_json::from_str(&response_text)?; + + // 11. 转换为我们的搜索结果格式 + let search_results = convert_vertex_response_to_search_results(&vertex_response, request)?; + + Ok(search_results) +} + +/// 获取 Google 访问令牌 +async fn get_google_access_token() -> Result { + let config = GeminiConfig::default(); + let client = reqwest::Client::new(); + + let url = format!("{}/google/access-token", config.base_url); + + let response = client + .get(&url) + .header("Authorization", format!("Bearer {}", config.bearer_token)) + .send() + .await?; + + let status = response.status(); + if !status.is_success() { + let error_body = response.text().await.unwrap_or_default(); + return Err(anyhow::anyhow!("获取访问令牌失败: {} - {}", status, error_body)); + } + + let response_text = response.text().await?; + let token_response: serde_json::Value = serde_json::from_str(&response_text)?; + + let access_token = token_response + .get("access_token") + .and_then(|v| v.as_str()) + .ok_or_else(|| anyhow::anyhow!("访问令牌响应中未找到 access_token 字段"))?; + + Ok(access_token.to_string()) +} + +/// 将 Vertex AI Search 响应转换为我们的搜索结果格式 +fn convert_vertex_response_to_search_results( + vertex_response: &serde_json::Value, + request: &SearchRequest, +) -> Result { + + let mut results = Vec::new(); + + // 解析 Vertex AI Search 响应 + if let Some(vertex_results) = vertex_response.get("results").and_then(|v| v.as_array()) { + for vertex_result in vertex_results { + if let Ok(search_result) = parse_vertex_result_to_search_result(vertex_result) { + // 应用相关性阈值过滤 + if search_result.relevance_score >= request.config.relevance_threshold.to_value() { + results.push(search_result); + } + } + } + } + + let total_size = vertex_response + .get("totalSize") + .and_then(|v| v.as_u64()) + .unwrap_or(results.len() as u64) as usize; + + let next_page_token = vertex_response + .get("nextPageToken") + .and_then(|v| v.as_str()) + .map(|s| s.to_string()); + + Ok(SearchResponse { + results, + total_size, + next_page_token, + search_time_ms: 0, // 将在调用方设置 + searched_at: chrono::Utc::now(), + }) +} + +/// 解析单个 Vertex AI Search 结果为我们的搜索结果格式 +fn parse_vertex_result_to_search_result(vertex_result: &serde_json::Value) -> Result { + + // 获取文档数据 + let document = vertex_result + .get("document") + .ok_or_else(|| anyhow::anyhow!("Vertex result missing document field"))?; + + // 获取结构化数据 + let struct_data = document + .get("structData") + .ok_or_else(|| anyhow::anyhow!("Document missing structData field"))?; + + // 解析基本信息 + let id = document + .get("id") + .and_then(|v| v.as_str()) + .unwrap_or(&format!("result_{}", chrono::Utc::now().timestamp())) + .to_string(); + + // 从 structData 中提取信息 + let style_description = struct_data + .get("style_description") + .and_then(|v| v.as_str()) + .unwrap_or("时尚搭配") + .to_string(); + + let environment_tags = struct_data + .get("environment_tags") + .and_then(|v| v.as_array()) + .map(|arr| { + arr.iter() + .filter_map(|v| v.as_str()) + .map(|s| s.to_string()) + .collect() + }) + .unwrap_or_else(|| vec!["日常".to_string()]); + + // 解析产品信息 + let products = struct_data + .get("products") + .and_then(|v| v.as_array()) + .map(|arr| { + arr.iter() + .filter_map(|v| parse_vertex_product_info(v).ok()) + .collect() + }) + .unwrap_or_else(Vec::new); + + // 获取图片URL（可能在不同的字段中） + let image_url = struct_data + .get("uri") + .or_else(|| struct_data.get("image_url")) + .or_else(|| struct_data.get("url")) + .and_then(|v| v.as_str()) + .unwrap_or("") + .to_string(); + + // 获取相关性评分 + let relevance_score = vertex_result + .get("modelScores") + .and_then(|scores| scores.get("relevance")) + .and_then(|score| score.get("values")) + .and_then(|values| values.as_array()) + .and_then(|arr| arr.first()) + .and_then(|v| v.as_f64()) + .unwrap_or(0.8); + + Ok(SearchResult { + id, + image_url, + style_description, + environment_tags, + products, + relevance_score, + }) +} + +/// 解析 Vertex AI Search 中的产品信息 +fn parse_vertex_product_info(value: &serde_json::Value) -> Result { + use crate::data::models::gemini_analysis::ColorHSV; + + let category = value + .get("category") + .and_then(|v| v.as_str()) + .unwrap_or("服装") + .to_string(); + + let description = value + .get("description") + .and_then(|v| v.as_str()) + .unwrap_or("时尚单品") + .to_string(); + + let color_pattern = value + .get("color_pattern") + .and_then(|v| { + let hue = v.get("Hue").or_else(|| v.get("hue")).and_then(|h| h.as_f64()).unwrap_or(0.0); + let saturation = v.get("Saturation").or_else(|| v.get("saturation")).and_then(|s| s.as_f64()).unwrap_or(0.5); + let value = v.get("Value").or_else(|| v.get("value")).and_then(|val| val.as_f64()).unwrap_or(0.8); + Some(ColorHSV::new(hue, saturation, value)) + }) + .unwrap_or_else(|| ColorHSV::new(0.0, 0.5, 0.8)); + + let design_styles = value + .get("design_styles") + .and_then(|v| v.as_array()) + .map(|arr| { + arr.iter() + .filter_map(|v| v.as_str()) + .map(|s| s.to_string()) + .collect() + }) + .unwrap_or_else(|| vec!["时尚".to_string()]); + + Ok(ProductInfo { + category, + description, + color_pattern, + design_styles, + }) } /// 获取所有服装搜索相关的Tauri命令名称 @@ -284,3 +607,5 @@ pub fn get_outfit_search_command_names() -> Vec<&'static str> { "get_outfit_search_config", ] } + + diff --git a/apps/desktop/src-tauri/src/presentation/commands/rag_grounding_commands.rs b/apps/desktop/src-tauri/src/presentation/commands/rag_grounding_commands.rs index e405a18..bae20ba 100644 --- a/apps/desktop/src-tauri/src/presentation/commands/rag_grounding_commands.rs +++ b/apps/desktop/src-tauri/src/presentation/commands/rag_grounding_commands.rs @@ -117,7 +117,7 @@ pub async fn get_rag_grounding_config( #[cfg(test)] mod tests { - use super::*; + use crate::infrastructure::gemini_service::RagGroundingConfig; #[test] diff --git a/docs/outfit-search-implementation.md b/docs/outfit-search-implementation.md new file mode 100644 index 0000000..66df342 --- /dev/null +++ b/docs/outfit-search-implementation.md @@ -0,0 +1,190 @@ +# 服装搭配搜索功能实现 + +## 概述 + +本文档描述了服装搭配搜索功能的实现，该功能直接调用 Google Vertex AI Search API 进行搜索，而不是通过 RAG Grounding。 + +## 核心功能 + +### 1. 直接 Vertex AI Search 调用 + +- **API 端点**: `https://discoveryengine.googleapis.com/v1beta/projects/{project_id}/locations/global/collections/default_collection/engines/{app_id}/servingConfigs/default_search:search` +- **认证**: 通过 Google 访问令牌进行认证 +- **数据存储**: 使用配置的 `vertex_ai_app_id` 和 `data_store_id` + +### 2. 搜索配置 + +```rust +pub struct SearchRequest { + pub query: String, // 搜索查询字符串 + pub config: SearchConfig, // 搜索配置 + pub page_size: usize, // 页面大小 + pub page_offset: usize, // 页面偏移量 +} + +pub struct SearchConfig { + pub relevance_threshold: RelevanceThreshold, // 相关性阈值 + pub environments: Vec, // 环境标签过滤 + pub categories: Vec, // 类别过滤 + pub color_filters: HashMap, // 颜色过滤器 + pub design_styles: HashMap>, // 设计风格过滤 + pub max_keywords: usize, // 最大关键词数量 +} +``` + +### 3. 搜索流程 + +1. **获取访问令牌**: 通过 Modal.run API 获取 Google 访问令牌 +2. **构建搜索过滤器**: 根据搜索配置构建 Vertex AI Search 过滤器 +3. **组合查询字符串**: 将用户查询与关键词组合 +4. **发送 HTTP 请求**: 直接调用 Vertex AI Search API +5. **解析响应**: 将 Vertex AI 响应转换为应用程序格式 +6. **应用过滤**: 根据相关性阈值过滤结果 + +### 4. 响应格式转换 + +```rust +pub struct SearchResult { + pub id: String, // 结果ID + pub image_url: String, // 图片URL + pub style_description: String, // 风格描述 + pub environment_tags: Vec, // 环境标签 + pub products: Vec, // 产品信息列表 + pub relevance_score: f64, // 相关性评分 +} + +pub struct ProductInfo { + pub category: String, // 产品类别 + pub description: String, // 产品描述 + pub color_pattern: ColorHSV, // 主要颜色 + pub design_styles: Vec, // 设计风格 +} +``` + +## 实现细节 + +### 1. 访问令牌获取 + +```rust +async fn get_google_access_token() -> Result { + let config = GeminiConfig::default(); + let client = reqwest::Client::new(); + + let url = format!("{}/google/access-token", config.base_url); + + let response = client + .get(&url) + .header("Authorization", format!("Bearer {}", config.bearer_token)) + .send() + .await?; + + // 解析访问令牌... +} +``` + +### 2. 搜索请求构建 + +```rust +let mut payload = serde_json::json!({ + "query": enhanced_query, + "relevanceThreshold": request.config.relevance_threshold.to_value().to_string(), + "relevanceScoreSpec": { + "returnRelevanceScore": true + }, + "pageSize": request.page_size, + "offset": request.page_offset +}); + +// 添加过滤器（如果有） +if !search_filter.is_empty() { + payload["filter"] = serde_json::Value::String(search_filter); +} +``` + +### 3. 响应解析 + +- 从 Vertex AI Search 响应中提取 `results` 数组 +- 解析每个结果的 `document.structData` 字段 +- 提取图片URL、风格描述、环境标签等信息 +- 解析产品信息和颜色模式 +- 获取相关性评分并应用阈值过滤 + +## 配置参数 + +### 全局配置 + +```rust +impl Default for OutfitSearchGlobalConfig { + fn default() -> Self { + Self { + google_project_id: "gen-lang-client-0413414134".to_string(), + vertex_ai_app_id: "jeans-search_1751353769585".to_string(), + storage_bucket_name: "fashion_image_block".to_string(), + data_store_id: "jeans_pattern_data_store".to_string(), + cloudflare_project_id: "67720b647ff2b55cf37ba3ef9e677083".to_string(), + cloudflare_gateway_id: "bowong-dev".to_string(), + } + } +} +``` + +### 相关性阈值 + +- **LOWEST**: 0.3 - 显示更多相关结果 +- **LOW**: 0.5 - 包含较多相关结果 +- **MEDIUM**: 0.7 - 平衡相关性和数量 +- **HIGH**: 0.9 - 只显示高度相关结果 + +## 错误处理 + +- 访问令牌获取失败 +- HTTP 请求失败 +- JSON 解析错误 +- 数据格式不匹配 +- 网络超时 + +## 测试 + +实现了以下测试用例： + +1. `test_search_request_creation` - 测试搜索请求创建 +2. `test_search_filter_builder` - 测试搜索过滤器构建 +3. `test_query_keywords_builder` - 测试查询关键词构建 + +## 使用示例 + +```typescript +import { OutfitSearchService } from '../services/outfitSearchService'; + +const searchRequest = { + query: "牛仔裤搭配", + config: { + relevance_threshold: "HIGH", + categories: ["上装", "下装"], + environments: ["Outdoor"], + color_filters: {}, + design_styles: {}, + max_keywords: 10 + }, + page_size: 9, + page_offset: 0 +}; + +const response = await OutfitSearchService.searchSimilarOutfits(searchRequest); +console.log(`找到 ${response.total_size} 个搜索结果`); +``` + +## 性能优化 + +- 使用连接池减少HTTP连接开销 +- 实现访问令牌缓存机制 +- 支持分页查询避免大量数据传输 +- 客户端相关性过滤减少无效结果 + +## 未来改进 + +1. 添加向量搜索支持 +2. 实现搜索结果缓存 +3. 支持更复杂的过滤条件 +4. 添加搜索分析和统计 +5. 实现搜索建议优化