eta_crawler/services/commodity_coal_mine.go

package services

import (
	"fmt"
	"github.com/mozillazg/go-pinyin"
	"github.com/tealeg/xlsx"
	"hongze/hongze_data_crawler/models"
	"hongze/hongze_data_crawler/utils"
	"strconv"
	"strings"
	"time"
)

func FileCoalJsm() () {
	var err error
	defer func() {
		if err != nil {
			fmt.Println("RefreshDataFromCoalJsm Err:" + err.Error())
			go utils.SendEmail(utils.APPNAME+"【"+utils.RunMode+"】"+"失败提醒", "RefreshDataFromCoalJsm ErrMsg:"+err.Error(), utils.EmailSendToUsers)
		}
	}()
	//path := "static/file/442家晋陕蒙煤矿周度产量数据-20220328.xlsx"
	path := "/home/code/python/coal_mail/emailFile/442家晋陕蒙煤矿周度产量数据-%s.xlsx"
	path = fmt.Sprintf(path,time.Now().Format(utils.FormatDateUnSpace))

	//exist,err := PathExists(path)
	//if err != nil {
	//	fmt.Println(err)
	//	return
	//}
	xlFile, err := xlsx.OpenFile(path)
	if err != nil {
		return
	}
	//var xlFile *xlsx.File
	//if exist {
	//	xlFile, err = xlsx.OpenFile(path)
	//	if err != nil {
	//		return
	//	}
	//}else {
	//	return
	//}


	var province string
	var description string
	var exchange string
	var city string
	var companyName string
	var group string

	var dateMap = make(map[int]string)
	var codeMap = make(map[string]string)
	var indexMap = make(map[string]string)
	var codeCompanyMap = make(map[string]string)
	var indexCompanyMap = make(map[string]string)
	var items []*models.BaseFromCoalmineMapping
	var itemsCompany []*models.BaseFromCoalmineMapping
	var itemsIndex []*models.BaseFromCoalmineJsmIndex
	var itemsCompanyIndex []*models.BaseFromCoalmineCompanyIndex
	codeList, err := models.GetBaseFromCoalmineMapping()
	if err != nil && err.Error() != utils.ErrNoRow() {
		utils.FileLog.Info("获取煤炭指标失败：", err)
		return
	}

	if len(codeList) > 0 {
		for _, v := range codeList {
			codeMap[v.IndexName] = v.IndexCode
		}
	}

	codeCompanyList, err := models.GetBaseFromCoalmineMapping()
	if err != nil && err.Error() != utils.ErrNoRow() {
		utils.FileLog.Info("获取煤炭公司指标失败：", err)
		return
	}
	if len(codeCompanyList) > 0 {
		for _, v := range codeCompanyList {
			codeCompanyMap[v.IndexName] = v.IndexCode
		}
	}

	indexList, err := models.GetBaseFromCoalmineIndex()
	if err != nil && err.Error() != utils.ErrNoRow() {
		utils.FileLog.Info("获取煤炭公司指标失败：", err)
		return
	}
	if len(indexList) > 0 {
		for _, v := range indexList {
			indexMap[v.IndexName+v.DataTime] = v.DealValue
		}
	}

	indexCompanyList, err := models.GetBaseFromCoalmineCompanyIndex()
	if err != nil && err.Error() != utils.ErrNoRow() {
		utils.FileLog.Info("获取煤炭公司指标失败：", err)
		return
	}
	if len(indexCompanyList) > 0 {
		for _, v := range indexCompanyList {
			indexCompanyMap[v.IndexName+v.DataTime] = v.DealValue
		}
	}

	for _, sheet := range xlFile.Sheets {
		//遍历行读取
		maxRow := sheet.MaxRow
		for i := 0; i < maxRow; i++ {
			//获取样本情况
			if i > 3 && i < 16 {
				row := sheet.Row(i)
				cells := row.Cells
			Loop2:
				for k, cell := range cells {
					text := cell.String()
					if text != "" {
						if k == 1 {
							province = text
						}
						if k == 3 {
							description = text
						}
						if k == 4 {
							item := new(models.BaseFromCoalmineMapping)
							exchange = text

							exists := ContainsSpecialName(province)
							var strResult string
							if exists {
								abbr := TrimProvinceName(province)
								//取处理后公司名首字母缩写
								a := pinyin.NewArgs()
								rows := pinyin.Pinyin(exchange, a)
								for i := 0; i < len(rows); i++ {
									if len(rows[i]) != 0 {
										str := rows[i][0]
										pi := str[0:1]
										strResult += pi
									}
								}
								item.IndexCode = abbr + strResult + "jsm"
							} else {
								//取处理后公司名首字母缩写
								a := pinyin.NewArgs()
								rows := pinyin.Pinyin(province+exchange, a)
								for i := 0; i < len(rows); i++ {
									if len(rows[i]) != 0 {
										str := rows[i][0]
										pi := str[0:1]
										strResult += pi
									}
								}
								item.IndexCode = strResult + "jsm"
							}

							item.IndexName = province + exchange


							item.CreateTime = time.Now()
							items = append(items, item)
							break Loop2
						}
					}
				}
			}
			//获取日期
			if i == 3 {
				row := sheet.Row(i)
				cells := row.Cells
				for k, cell := range cells {
					text := cell.String()
					stamp, _ := time.ParseInLocation("01-02-06", text, time.Now().Location())
					if k > 4 {
						dateMap[k] = stamp.Format(utils.FormatDate)
					}
				}
			}
			//获取具体产量
			if i > 3 && i < 16 {
				row := sheet.Row(i)
				cells := row.Cells
				for k, cell := range cells {
					text := cell.String()
					if k > 4 {
						item := new(models.BaseFromCoalmineJsmIndex)
						item.IndexName = province + exchange
						item.IndexCode = codeMap[item.IndexName]
						item.Exchange = exchange
						item.DealValue = text
						item.DataTime = dateMap[k]
						item.Exchange = exchange
						item.Description = description
						item.Province = province
						item.Source = "三省周度"
						if exchange == "产量"{
							item.Unit = "万吨"
						}else {
							item.Unit = "百分比"
						}
						item.Frequency = "周度"
						item.ModifyTime = time.Now()
						item.CreateTime = time.Now()
						itemsIndex = append(itemsIndex, item)
					}
				}
			}

			//获取公司指标名称
			if i > 17 {
				row := sheet.Row(i)
				cells := row.Cells
			Loop3:
				for k, cell := range cells {
					text := cell.String()
					if text != "" {
						if k == 1 {
							province = text
						}
						if k == 2 {
							city = text
						}
						if k == 3 {
							companyName = text
						}
						if k == 4 {
							item := new(models.BaseFromCoalmineMapping)
							group = text
							trimName := TrimCompanyName(companyName)
							item.IndexName = trimName
							//取处理后公司名首字母缩写
							strResult := ""
							a := pinyin.NewArgs()
							rows := pinyin.Pinyin(trimName, a)
							for i := 0; i < len(rows); i++ {
								if len(rows[i]) != 0 {
									str := rows[i][0]
									pi := str[0:1]
									strResult += pi
								}
							}
							item.IndexCode = strResult + "company"
							item.CreateTime = time.Now()
							itemsCompany = append(itemsCompany, item)
							break Loop3
						}
					}
				}
			}

			//获取公司具体产量
			if i > 18 {
				row := sheet.Row(i)
				cells := row.Cells
				for k, cell := range cells {
					text := cell.String()
					if k > 4 {
						item := new(models.BaseFromCoalmineCompanyIndex)
						companyName = TrimCompanyName(companyName)
						item.IndexName = companyName
						item.IndexCode = codeCompanyMap[item.IndexName]
						item.DealValue = text
						item.DataTime = dateMap[k]
						item.Province = province
						item.City = city

						//处理无类别名时的情况
						if group == "无" {
							item.GroupName = companyName
						} else {
							item.GroupName = group
						}

						item.Source = "三省企业"
						item.Unit = "万吨"
						item.Frequency = "周度"
						item.ModifyTime = time.Now()
						item.CreateTime = time.Now()
						//fmt.Println(item)
						itemsCompanyIndex = append(itemsCompanyIndex, item)
					}
				}
			}
		}
	}

	//添加数据到数据库
	for _, v := range items {
		if codeMap[v.IndexName] == "" {
			codeMap[v.IndexName] = v.IndexCode
			newId, err := models.AddBaseFromCoalmineMapping(v)
			if err != nil {
				fmt.Println("添加指标名称错误")
			} else {
				fmt.Println("添加指标名称成功", newId)
			}
		}
	}
	fmt.Println("指标操作完成")

	for _, v := range itemsIndex {
		v.IndexCode = codeMap[v.IndexName]
		if indexMap[v.IndexName+v.DataTime] == "" && v.DealValue != "" {
			newId, err := models.AddBaseFromCoalmineIndex(v)
			if err != nil {
				fmt.Println("添加数据错误", err)
			} else {
				fmt.Println("新增成功", newId)
			}
		} else {
			if indexMap[v.IndexName+v.DataTime] != v.DealValue && v.DealValue != "" {
				err = models.UpdateBaseFromCoalmineIndex(v)
				if err != nil {
					fmt.Println("修改数据错误错误", err)
					return
				}
			}
		}
	}
	fmt.Println("数据操作完成")

	for _, v := range itemsCompany {
		if codeCompanyMap[v.IndexName] == "" {
			codeCompanyMap[v.IndexName] = v.IndexCode
			newId, err := models.AddBaseFromCoalmineMapping(v)
			if err != nil {
				for i := 0; i<10; i++ {
					v.IndexCode = v.IndexCode + strconv.Itoa(i)
					codeCompanyMap[v.IndexName] = v.IndexCode
					newId, err := models.AddBaseFromCoalmineMapping(v)
					if err != nil {
						fmt.Println("再次添加公司指标名称错误", err)
						fmt.Println(v.IndexName, v.IndexCode)
						continue
					} else {
						fmt.Println("新增公司成功", newId)
						break
					}
				}
			} else {
				fmt.Println("新增公司成功", newId)
			}
		}
	}
	fmt.Println("公司指标操作完成")

	for _, v := range itemsCompanyIndex {
		v.IndexCode = codeCompanyMap[v.IndexName]
		if indexCompanyMap[v.IndexName+v.DataTime] == "" && v.DealValue != "" {
			newId, err := models.AddBaseFromCoalmineCompanyIndex(v)
			if err != nil {
				fmt.Println("添加公司数据错误", err)
			} else {
				fmt.Println("新增公司数据成功", newId)
			}
		} else {
			if indexCompanyMap[v.IndexName+v.DataTime] != v.DealValue && v.DealValue != "" {
				err = models.UpdateBaseFromCoalmineCompanyIndex(v)
				if err != nil {
					fmt.Println("修改数据错误错误", err)
				}
			}
		}

	}
	return
}

func TrimCompanyName(name string) string {
	name = strings.Replace(name, "有限", "", -1)
	name = strings.Replace(name, "股份", "", -1)
	name = strings.Replace(name, "责任", "", -1)
	name = strings.Replace(name, "公司", "", -1)
	name = strings.Replace(name, "（", "", -1)
	name = strings.Replace(name, "）", "", -1)
	return name
}

func ContainsSpecialName(name string) bool {
	return strings.Contains(name, "陕西") ||
		strings.Contains(name, "海南") ||
		strings.Contains(name, "河南") ||
		strings.Contains(name, "河北") ||
		strings.Contains(name, "澳门") ||
		strings.Contains(name, "内蒙古") ||
		strings.Contains(name, "黑龙江")
}

func TrimProvinceName(name string) string {
	name = strings.Replace(name, "陕西省", "sns", -1)
	name = strings.Replace(name, "陕西", "sn", -1)
	name = strings.Replace(name, "海南市", "hi", -1)
	name = strings.Replace(name, "海南", "hi", -1)
	name = strings.Replace(name, "河南省", "has", -1)
	name = strings.Replace(name, "河南", "ha", -1)
	name = strings.Replace(name, "河北省", "hes", -1)
	name = strings.Replace(name, "河北", "he", -1)
	name = strings.Replace(name, "澳门", "mo", -1)
	name = strings.Replace(name, "内蒙古自治区", "nm", -1)
	name = strings.Replace(name, "内蒙古", "nm", -1)
	name = strings.Replace(name, "黑龙江", "hl", -1)
	name = strings.Replace(name, "（", "", -1)
	name = strings.Replace(name, "）", "", -1)
	name = strings.Replace(name, "+", "", -1)
	return name
}