jy-data-extract/fullproject/src_dev1/comparePNCB.go

package main

import (
	"regexp"
	"strings"
)

func comparePNCB(info *Info, compareProject *ProjectInfo) (compareStr string, score int) {
	retv := 0
	if info.ProjectName != "" {
		pns := []string{}
		if compareProject.ProjectName != "" {
			pns = append(pns, compareProject.ProjectName)
		}
		if len(compareProject.MPN) > 0 {
			pns = append(pns, compareProject.MPN...)
		}
		ifind := 0
		for _, v := range pns {
			if info.ProjectName == v {
				ifind = 1
				break
			} else {
				//if strings.Contains(info.ProjectName, v) || strings.Contains(v, info.ProjectName) ||
				retv = CheckContain(info.ProjectName, v, info.Buyer)
				//if CheckContain(info.ProjectName, v, info.Buyer)
				if retv == 3 {
					ifind = 1
					break
				} else if retv == 1 {
					ifind = 2
				} else if ifind == 0 {
					ifind = 3
				}
			}
		}
		switch ifind {
		case 0:
			compareStr = "D"
		case 1:
			compareStr = "A"
			score += 4
			if len([]rune(info.ProjectName)) > 19 {
				score += 2
			}
		case 2:
			compareStr = "B"
			score += 2
			if len([]rune(info.ProjectName)) > 19 {
				score += 1
			}
		case 3:
			compareStr = "C"
		}
	} else {
		compareStr = "D"
	}
	PNStr := compareStr
	for _, pc := range []string{info.ProjectCode, info.PTC} {
		if pc != "" {
			pcs := []string{}
			if compareProject.ProjectCode != "" {
				pcs = append(pcs, compareProject.ProjectCode)
			}
			if len(compareProject.MPC) > 0 {
				pcs = append(pcs, compareProject.MPC...)
			}
			ifind := 0
			for _, v := range pcs {
				if pc == v {
					ifind = 1
					break
				} else {
					// math.Abs(float64(len([]rune(pc))-len([]rune(v)))) < 6
					//if !_numreg1.MatchString(pc) && !_zimureg1.MatchString(pc) && !_numreg1.MatchString(v) && !_zimureg1.MatchString(v)
					if (PNStr == "A" || PNStr == "B" || (!_nzreg.MatchString(pc) && !_nzreg.MatchString(v))) && (strings.Contains(pc, v) || strings.Contains(v, pc)) {
						t1 := pc
						t2 := v
						if len(v) > len(pc) {
							t1 = v
							t2 = pc
						}
						t3 := strings.Replace(t1, t2, "", -1)
						t3 = _datereg.ReplaceAllString(t3, "")
						if t3 == "" {
							ifind = 1
							break
						} else if _hanreg.MatchString(t3) || _numreg1.MatchString(t3) {
							ifind = 2
						} else {
							ifind = 3
						}
					} else if ifind == 0 {
						ifind = 3
					}
				}
			}
			switch ifind {
			case 0:
				compareStr += "D"
			case 1:
				compareStr += "A"
				score += 4
				if len([]rune(pc)) > 19 {
					score += 2
				}
			case 2:
				if compareStr == "C" && retv == 2 {
					compareStr += "C"
				} else {
					compareStr += "B"
					score += 2
					if len([]rune(pc)) > 19 {
						score += 1
					}
				}
			case 3:
				compareStr += "C"
			}

		} else {
			compareStr += "D"
		}
	}
	return
}

var _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})?")
var _numreg1 = regexp.MustCompile("^[0-9-]{1,8}$")
var _zimureg1 = regexp.MustCompile("^[a-zA-Z-]{1,7}$")
var _nzreg = regexp.MustCompile("^[0-9a-zA-Z-]+$")
var _hanreg = regexp.MustCompile(`^[\p{Han}:：【】\\[\\]（）()－-、]+$`)

var replaceStr = regexp.MustCompile("(工程|采购|项目|[?！、【】（）]|栏标价|中标候选人|招标代理)")
var 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)")

func CheckContain(b1, b2, infoBuyer string) (res int) {
	b1 = replaceStr.ReplaceAllString(b1, "")
	b2 = replaceStr.ReplaceAllString(b2, "")
	b1 = strings.Replace(b1, infoBuyer, "", 1)
	b2 = strings.Replace(b2, infoBuyer, "", 1)

	if b1 == b2 {
		res = 3
		return
	}
	bs1 := []rune(b1)
	bs2 := []rune(b2)
	tmp := ""
	for i := 0; i < len(bs1); i++ {
		for j := 0; j < len(bs2); j++ {
			if bs1[i] == bs2[j] {
				tmp += string(bs1[i])
			} else if tmp != "" {
				b1 = strings.Replace(b1, tmp, "", -1)
				b2 = strings.Replace(b2, tmp, "", -1)
				tmp = ""
			}
		}
	}
	if tmp != "" {
		b1 = strings.Replace(b1, tmp, "", -1)
		b2 = strings.Replace(b2, tmp, "", -1)
	}
	if b1 == b2 {
		res = 3
	} else if (b1 == "" && !pStr.MatchString(b2)) || (b2 == "" && !pStr.MatchString(b1)) {
		res = 1
	} else {
		res = 2
	}
	return
}