jy-data-extract/fullproject/src_v1/init.go

package main

import (
	"log"
	"math"
	mu "mfw/util"
	"qfw/util"
	"regexp"
	"sort"
	"strings"
	"sync"

	"go.mongodb.org/mongo-driver/bson/primitive"
)

const (
	ProjectCache = "info" //存放每条项目信息，key为项目ID
)

var (
	Sysconfig                                      map[string]interface{} //读取配置文件
	MongoTool                                      *MongodbSim            //mongodb连接
	ExtractColl, ProjectColl, BackupColl, SiteColl string                 //抽取表、项目表、项目快照表、站点表
	UpdateColl                                     string                 // 金额修改数据表
	Thread                                         int                    //配置项线程数
	//NextNode                 []interface{}
	BlackList    								   []interface{}
	BlaskListMap								   map[string]bool
)

var (
	//判断是日期
	_datereg   = regexp.MustCompile("20[0-2][0-9][年-][0-9]{1,2}[月-][0-9]{1,2}[日-]([0-9]{1,2}时[0-9]{0,2})?")
	_numreg1   = regexp.MustCompile("^[0-9-]{1,8}$")
	_zimureg1  = regexp.MustCompile("^[a-zA-Z-]{1,7}$")
	_nzreg     = regexp.MustCompile("^[0-9a-zA-Z-]+$")
	_hanreg    = regexp.MustCompile(`^[\p{Han}:：【】\\[\\]（）()－-、]+$`)
	replaceStr = regexp.MustCompile("(工程|采购|项目|[?！、【】（）—()－-]|栏标价|中标候选人|招标代理)")
	//判断带有分包、等特定词的
	pStr = regexp.MustCompile("(勘察|监理|施工|设计|验收|标段|分包|子包|[0-9A-Z]包|[一二三四五六七八九十0-9]批)")
	//判断包含数值
	nreg1 = regexp.MustCompile("[0-9]{2,}")
	//判断包含字母
	zreg1 = regexp.MustCompile("[a-zA-Z]{1,}")
	//判断包含汉字
	hreg1 = regexp.MustCompile(`[\p{Han}]+`)
	//判断项目编号是在10以内的纯数字结构
	numCheckPc = regexp.MustCompile("^[0-9-]{1,10}$")
	//仅初始化使用
	compareNoPass = map[string]bool{}
	compareAB     = map[string]bool{}
	compareAB2D   = map[string]bool{}
	compareABD    = map[string]bool{}
	compareAB2CD  = map[string]bool{}
	compareABCD   = map[string]bool{}
)

func init() {
	util.ReadConfig(&Sysconfig)
	MongoTool = &MongodbSim{
		MongodbAddr: Sysconfig["mongodbServers"].(string),
		Size:        util.IntAll(Sysconfig["mongodbPoolSize"]),
		DbName:      Sysconfig["mongodbName"].(string),
	}
	MongoTool.InitPool()

	ExtractColl = Sysconfig["extractColl"].(string)
	ProjectColl = Sysconfig["projectColl"].(string)
	UpdateColl = Sysconfig["updateColl"].(string)
	BackupColl = Sysconfig["projectColl"].(string) + "_back"
	SiteColl = Sysconfig["siteColl"].(string)
	Thread = util.IntAll(Sysconfig["thread"])
	//NextNode = Sysconfig["nextNode"].([]interface{})
	udpport, _ := Sysconfig["udpport"].(string)
	udpclient = mu.UdpClient{Local: udpport, BufSize: 1024}
	udpclient.Listen(processUdpMsg)
	log.Println("Udp服务监听", udpport)

	BlackList = Sysconfig["blacklist"].([]interface{})
	BlaskListMap = make(map[string]bool)
	for _, v := range BlackList {
		BlaskListMap[util.ObjToString(v)] = true
	}

	//加载项目数据
	//---不能通过
	vm := []string{"C", "D"}
	for i := 0; i < 2; i++ {
		for j := 0; j < 2; j++ {
			for k := 0; k < 2; k++ {
				key := vm[i] + vm[j] + vm[k]
				compareNoPass[key] = true
				//fmt.Println(key)
			}
		}
	}
	//fmt.Println("-------------------")

	//三个元素一致 [AB][AB][AB]，分值最高
	vm = []string{"A", "B"}
	for i := 0; i < 2; i++ {
		for j := 0; j < 2; j++ {
			for k := 0; k < 2; k++ {
				key := vm[i] + vm[j] + vm[k]
				compareAB[key] = true
				//fmt.Println(key)
			}
		}
	}
	//fmt.Println("-------------------", len(compareAB))
	//---至少两个一致，其他可能不存在
	//[AB][AB][ABD]
	//[AB][ABD][AB]
	vm = []string{"A", "B"}
	vm2 := []string{"A", "B", "D"}
	for i := 0; i < 2; i++ {
		for j := 0; j < 2; j++ {
			for k := 0; k < 3; k++ {
				key := vm[i] + vm[j] + vm2[k]
				if !compareAB[key] {
					compareAB2D[key] = true
					//fmt.Println(key)

				}
			}
		}
	}
	for i := 0; i < 2; i++ {
		for j := 0; j < 3; j++ {
			for k := 0; k < 2; k++ {
				key := vm[i] + vm2[j] + vm[k]
				if !compareAB[key] {
					compareAB2D[key] = true
					//fmt.Println(key)

				}
			}
		}
	}
	//fmt.Println("-------------------", len(compareAB2D))
	//---至少一个一致，其他可能不存在
	//[ABD][ABD][ABD] //已经删除DDD
	vm = []string{"A", "B", "D"}
	for i := 0; i < 3; i++ {
		for j := 0; j < 3; j++ {
			for k := 0; k < 3; k++ {
				key := vm[i] + vm[j] + vm[k]
				if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] {
					compareABD[key] = true
					//fmt.Println(key)
				}
			}
		}
	}
	//fmt.Println("-------------------", len(compareABD))

	//[AB][ABCD][AB]
	//[AB][AB][ABCD]
	vm = []string{"A", "B"}
	vm2 = []string{"A", "B", "C", "D"}
	for i := 0; i < 2; i++ {
		for j := 0; j < 4; j++ {
			for k := 0; k < 2; k++ {
				key := vm[i] + vm2[j] + vm[k]
				if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
					compareAB2CD[key] = true
					//fmt.Println(key)
				}
			}
		}
	}
	for i := 0; i < 2; i++ {
		for j := 0; j < 2; j++ {
			for k := 0; k < 4; k++ {
				key := vm[i] + vm[j] + vm2[k]
				if !compareAB[key] && !compareAB2D[key] && !compareNoPass[key] && !compareABD[key] {
					compareAB2CD[key] = true
					//fmt.Println(key)
				}
			}
		}
	}
	//fmt.Println("-------------------", len(compareAB2CD))
	//[ABECD][ABECD][ABECD]  //已经删除[CD][CD][CD]   //这个要重点讨论
	vm = []string{"A", "B", "C", "D"}
	for i := 0; i < 4; i++ {
		for j := 0; j < 4; j++ {
			for k := 0; k < 4; k++ {
				key := vm[i] + vm[j] + vm[k]
				if !compareAB[key] && !compareAB2D[key] && !compareABD[key] && !compareNoPass[key] && !compareAB2CD[key] {
					compareABCD[key] = true
					//fmt.Println(key)
				}
			}
		}
	}
}

func CheckHanAndNum(str string) (b bool) {
	return nreg1.MatchString(str) && hreg1.MatchString(str)
}
func CheckZimuAndNum(str string) (b bool) {
	return zreg1.MatchString(str) && nreg1.MatchString(str)
}

type KeyMap struct {
	Lock sync.Mutex
	Map  map[string]*Key
}

type ID struct {
	Id   string
	Lock sync.Mutex
	P    *ProjectInfo
}
type Key struct {
	Arr  []string
	Lock sync.Mutex
}
type IdAndLock struct {
	Id   string
	Lock sync.Mutex
}

func NewKeyMap() *KeyMap {
	return &KeyMap{
		Map:  map[string]*Key{},
		Lock: sync.Mutex{},
	}
}

//招标信息实体类
type Info struct {
	Id            string                   `json:"_id"`
	Href          string                   `json:"href"` //源地址
	Publishtime   int64                    `json:"publishtime"`
	Comeintime    int64                    `json:"comeintime"`
	Title         string                   `json:"title"`
	TopType       string                   `json:"toptype"`
	SubType       string                   `json:"subtype"`
	ProjectName   string                   `json:"projectname"`
	ProjectCode   string                   `json:"projectcode"`
	ProjectScope  string                   `json:"projectscope"`
	ContractCode  string                   `json:"contractcode"`
	Buyer         string                   `json:"buyer"`
	Buyerperson   string                   `json:"buyerperson"`
	Buyertel      string                   `json:"buyertel"`
	Agency        string                   `json:"agency"`
	Area          string                   `json:"area"`
	City          string                   `json:"city"`
	District      string                   `json:"district"`
	Infoformat    int                      `json:"infoformat"`
	ReviewExperts []string                 `json:"review_experts"`
	Purchasing    string                   `json:"purchasing"`
	WinnerOrder   []map[string]interface{} `json:"winnerorder"`
	HasPackage    bool                     // `json:"haspackage"`
	Package       map[string]interface{}   `json:"package"`
	//PNum          string                 `json:"pnum"`
	Topscopeclass []string `json:"topscopeclass"`
	Subscopeclass []string `json:"subscopeclass"`
	Buyerclass    string   `json:"buyerclass"`
	Bidopentime   int64    `json:"bidopentime"`
	Budget        float64  `json:"budget"`
	Bidamount     float64  `json:"bidamount"`
	Winners       []string
	dealtype      int
	PTC           string //从标题中抽的项目编号
	pnbval        int    //项目名称、编号、采购单位存在的个数
	LenPC         int    //项目编号长度
	LenPN         int    //项目名称长度
	LenPTC        int    //标题抽的项目编号长度
	//以下三个元素做对比，计算包含时候使用
	PNBH  int //0初始，+包含，-被包含
	PCBH  int
	PTCBH int
}

//项目实体类
type ProjectInfo struct {
	Id            primitive.ObjectID `json:"_id"`
	FirstTime     int64              `json:"firsttime,omitempty"` //项目的最早时间
	LastTime      int64              `json:"lasttime,omitempty"`  //项目的最后时间
	Ids           []string           `json:"ids,omitempty"`
	Topscopeclass []string           `json:"topscopeclass,omitempty"`
	Subscopeclass []string           `json:"subscopeclass,omitempty"` //子行业分类
	Winners       []string           `json:"s_winner,omitempty"`      //中标人
	ProjectName   string             `json:"projectname,omitempty"`   //项目名称
	ProjectCode   string             `json:"projectcode,omitempty"`   //项目代码唯一（纯数字的权重低）
	ContractCode  string             `json:"contractcode,omitempty"`  //项目编号
	Buyer         string             `json:"buyer,omitempty"`         //采购单位唯一
	MPN           []string           `json:"mpn,omitempty"`           //合并后多余的项目名称
	MPC           []string           `json:"mpc,omitempty"`           //合并后多余的项目编号
	Buyerperson   string             `json:"buyerperson"`             //采购联系人
	Buyertel      string             `json:"buyertel"`                //采购联系人电话
	Agency        string             `json:"agency"`                  //代理机构
	Area          string             `json:"area"`                    //地区
	City          string             `json:"city"`                    //地市
	District      string             `json:"district"`                //区县
	Bidstatus     string             `json:"bidstatus"`               //
	Bidtype       string             `json:"bidtype"`                 //
	ReviewExperts []string           `json:"review_experts"`          // 项目评审专家
	Purchasing    string             `json:"purchasing"`              // 标的物
	//HasPackage    bool                   `json:"haspackage"`              //是否有分包
	Package     map[string]interface{} `json:"package,omitempty"`     //分包的对比对象
	Buyerclass  string                 `json:"buyerclass"`            //采购单位分类
	Bidopentime int64                  `json:"bidopentime,omitempty"` //开标时间
	//	Zbtime        int64                  `json:"zbtime"`        //招标时间
	Jgtime        int64    `json:"jgtime"`              //结果中标时间
	Zbtime        int64    `json:"zbtime"`              //招标时间
	Bidamount     float64  `json:"bidamount,omitempty"` //中标金额
	Budget        float64  `json:"budget,omitempty"`    //预算
	Winnerorder   []string `json:"winnerorder"`         //中标候选人
	score         int
	comStr        string
	resVal, pjVal int
	InfoFiled     map[string]InfoField `json:"infofield"`    //逻辑处理需要的info字段
	Budgettag     int                  `json:"budgettag"`    //预算是否有效标记
	Bidamounttag  int                  `json:"bidamounttag"` //中标金额是否有效标记
}

//存储部分招标信息字段，业务逻辑处理需要
type InfoField struct {
	Budget       float64 `json:"budget"`
	Bidamount    float64 `json:"bidamount"`
	ContractCode string  `json:"contractcode"`
	ProjectName  string  `json:"projectname"`
	ProjectCode  string  `json:"projectcode"`
	Bidstatus    string  `json:"bidstatus"`
}

//站点信息
type Site struct {
	Id       string `json:"_id"`
	Site     string `json:"site"`     //站点名字
	Area     string `json:"area"`     //省
	City     string `json:"city"`     //市
	District string `json:"district"` //区、县
	Domain   string `json:"domain"`   //地址
	Status   int    `json:"status"`   //
}

//二分字符串查找
func BinarySearch(s []string, k string) int {
	sort.Strings(s)
	lo, hi := 0, len(s)-1
	for lo <= hi {
		m := (lo + hi) >> 1
		if s[m] < k {
			lo = m + 1
		} else if s[m] > k {
			hi = m - 1
		} else {
			return m
		}
	}
	return -1
}

//计算文本相似度
func CosineSimilar(srcWords1, dstWords1 string) float64 {
	srcWords, dstWords := strings.Split(srcWords1, ""), strings.Split(dstWords1, "")
	// get all words
	allWordsMap := make(map[string]int, 0)
	for _, word := range srcWords {
		if _, found := allWordsMap[word]; !found {
			allWordsMap[word] = 1
		} else {
			allWordsMap[word] += 1
		}
	}
	for _, word := range dstWords {
		if _, found := allWordsMap[word]; !found {
			allWordsMap[word] = 1
		} else {
			allWordsMap[word] += 1
		}
	}

	// stable the sort
	allWordsSlice := make([]string, 0)
	for word, _ := range allWordsMap {
		allWordsSlice = append(allWordsSlice, word)
	}

	// assemble vector
	srcVector := make([]int, len(allWordsSlice))
	dstVector := make([]int, len(allWordsSlice))
	for _, word := range srcWords {
		if index := BinarySearch(allWordsSlice, word); index != -1 {
			srcVector[index] += 1
		}
	}
	for _, word := range dstWords {
		if index := BinarySearch(allWordsSlice, word); index != -1 {
			dstVector[index] += 1
		}
	}

	// calc cos
	numerator := float64(0)
	srcSq := 0
	dstSq := 0
	for i, srcCount := range srcVector {
		dstCount := dstVector[i]
		numerator += float64(srcCount * dstCount)
		srcSq += srcCount * srcCount
		dstSq += dstCount * dstCount
	}
	denominator := math.Sqrt(float64(srcSq * dstSq))

	v1 := numerator / denominator
	//	if v1 > 0.6 {
	//		log.Println(v1, srcWords1, dstWords1)
	//	}
	return v1
}