billai/server/handler/upload.go

package handler

import (
	"fmt"
	"io"
	"net/http"
	"os"
	"path/filepath"
	"strings"
	"time"

	"github.com/gin-gonic/gin"

	"billai-server/config"
	"billai-server/model"
	"billai-server/service"
)

// Upload 处理账单上传和清理请求
func Upload(c *gin.Context) {
	// 1. 获取上传的文件
	file, header, err := c.Request.FormFile("file")
	if err != nil {
		c.JSON(http.StatusBadRequest, model.UploadResponse{
			Result:  false,
			Message: "请上传账单文件 (参数名: file)",
		})
		return
	}
	defer file.Close()

	// 2. 解析请求参数
	var req model.UploadRequest
	c.ShouldBind(&req)
	if req.Format == "" {
		req.Format = "csv"
	}

	// 3. 保存上传的文件
	timestamp := time.Now().Format("20060102_150405")
	inputFileName := fmt.Sprintf("%s_%s", timestamp, header.Filename)
	uploadDirAbs := config.ResolvePath(config.Global.UploadDir)
	inputPath := filepath.Join(uploadDirAbs, inputFileName)

	dst, err := os.Create(inputPath)
	if err != nil {
		c.JSON(http.StatusInternalServerError, model.UploadResponse{
			Result:  false,
			Message: "保存文件失败: " + err.Error(),
		})
		return
	}
	defer dst.Close()
	io.Copy(dst, file)

	// 4. 对原始数据进行去重检查
	fmt.Printf("📋 开始去重检查...\n")
	dedupResult, dedupErr := service.DeduplicateRawFile(inputPath, timestamp)
	if dedupErr != nil {
		c.JSON(http.StatusInternalServerError, model.UploadResponse{
			Result:  false,
			Message: "去重检查失败: " + dedupErr.Error(),
		})
		return
	}

	// 账单类型从去重结果获取
	billType := dedupResult.BillType

	fmt.Printf("   原始记录: %d 条\n", dedupResult.OriginalCount)
	if dedupResult.DuplicateCount > 0 {
		fmt.Printf("   重复记录: %d 条（已跳过）\n", dedupResult.DuplicateCount)
	}
	fmt.Printf("   新增记录: %d 条\n", dedupResult.NewCount)

	// 如果全部重复，返回提示
	if dedupResult.NewCount == 0 {
		c.JSON(http.StatusOK, model.UploadResponse{
			Result:  true,
			Message: fmt.Sprintf("文件中的 %d 条记录全部已存在，无需重复导入", dedupResult.OriginalCount),
			Data: &model.UploadData{
				BillType:       billType,
				RawCount:       0,
				CleanedCount:   0,
				DuplicateCount: dedupResult.DuplicateCount,
			},
		})
		return
	}

	// 使用去重后的文件路径进行后续处理
	processFilePath := dedupResult.DedupFilePath

	// 5. 构建输出文件路径
	baseName := strings.TrimSuffix(header.Filename, filepath.Ext(header.Filename))
	outputExt := ".csv"
	if req.Format == "json" {
		outputExt = ".json"
	}
	outputFileName := fmt.Sprintf("%s_%s_cleaned%s", timestamp, baseName, outputExt)
	outputDirAbs := config.ResolvePath(config.Global.OutputDir)
	outputPath := filepath.Join(outputDirAbs, outputFileName)

	// 6. 执行 Python 清洗脚本
	cleanOpts := &service.CleanOptions{
		Year:   req.Year,
		Month:  req.Month,
		Start:  req.Start,
		End:    req.End,
		Format: req.Format,
	}
	cleanResult, cleanErr := service.RunCleanScript(processFilePath, outputPath, cleanOpts)
	if cleanErr != nil {
		c.JSON(http.StatusInternalServerError, model.UploadResponse{
			Result:  false,
			Message: cleanErr.Error(),
		})
		return
	}

	// 7. 如果去重检测没有识别出类型，从 Python 输出中检测
	if billType == "" {
		billType = cleanResult.BillType
	}

	// 8. 将去重后的原始数据存入 MongoDB（原始数据集合）
	rawCount, rawErr := service.SaveRawBillsFromFile(processFilePath, billType, header.Filename, timestamp)
	if rawErr != nil {
		fmt.Printf("⚠️  存储原始数据到 MongoDB 失败: %v\n", rawErr)
	} else {
		fmt.Printf("✅ 已存储 %d 条原始账单记录到 MongoDB\n", rawCount)
	}

	// 9. 将清洗后的数据存入 MongoDB（清洗后数据集合）
	cleanedCount, _, cleanedErr := service.SaveCleanedBillsFromFile(outputPath, req.Format, billType, header.Filename, timestamp)
	if cleanedErr != nil {
		fmt.Printf("⚠️  存储清洗后数据到 MongoDB 失败: %v\n", cleanedErr)
	} else {
		fmt.Printf("✅ 已存储 %d 条清洗后账单记录到 MongoDB\n", cleanedCount)
	}

	// 10. 清理临时的去重文件（如果生成了的话）
	if dedupResult.DedupFilePath != inputPath && dedupResult.DedupFilePath != "" {
		os.Remove(dedupResult.DedupFilePath)
	}

	// 11. 返回成功响应
	message := fmt.Sprintf("处理成功，新增 %d 条记录", cleanedCount)
	if dedupResult.DuplicateCount > 0 {
		message = fmt.Sprintf("处理成功，新增 %d 条，跳过 %d 条重复记录", cleanedCount, dedupResult.DuplicateCount)
	}

	c.JSON(http.StatusOK, model.UploadResponse{
		Result:  true,
		Message: message,
		Data: &model.UploadData{
			BillType:       billType,
			FileURL:        fmt.Sprintf("/download/%s", outputFileName),
			FileName:       outputFileName,
			RawCount:       rawCount,
			CleanedCount:   cleanedCount,
			DuplicateCount: dedupResult.DuplicateCount,
		},
	})
}