dataIdentify/service/rule.go

package service

import (
	"github.com/gogf/gf/v2/container/gvar"
	"github.com/gogf/gf/v2/frame/g"
	"github.com/gogf/gf/v2/os/gctx"
	"github.com/gogf/gf/v2/util/gconv"
	"log"
	"regexp"
	"strings"
)

var (
	clearPatterns                            = g.Config().MustGet(gctx.New(), "clearPatterns").Strings()
	bidCommonwealth_blacklistPatterns        = g.Config().MustGet(gctx.New(), "bidCommonwealth.blacklistPatterns").Strings()
	bidCommonwealth_whitelistPatterns        = g.Config().MustGet(gctx.New(), "bidCommonwealth.whitelistPatterns").Strings()
	bidCommonwealth_modelPatterns            = g.Config().MustGet(gctx.New(), "bidCommonwealth.modelPatterns").Strings()
	bidCommonwealth_firstWinnerOrder         = g.Config().MustGet(gctx.New(), "bidCommonwealth.firstWinnerOrder").String()
	bidCommonwealth_firstWinnerOrderClearReg = regexp.MustCompile(g.Config().MustGet(gctx.New(), "bidCommonwealth.firstWinnerOrderClear").String())
	quoteModeRules                           = g.Config().MustGet(gctx.New(), "quoteMode.rules").Maps()
	quoteModeTableKReg                       = g.Config().MustGet(gctx.New(), "quoteMode.tableK").String()
	showOnlyOnce                             = [][]string{}
	moreWinnerReg                            = regexp.MustCompile("[,，、]")
)

func init() {
	for _, v := range g.Config().MustGet(gctx.New(), "bidCommonwealth.showOnlyOnce").Array() {
		showOnlyOnce = append(showOnlyOnce, gconv.Strings(v))
	}
}

type Rule struct{}

func (r *Rule) Execute(b *BidInfo) (bool, string, bool, int, bool) {
	var bidCommonwealth int
	var quoteMode string
	var quoteMode_bigModel, bidCommonwealth_bigModel bool
	var bigModel map[string]bool
	if b.Type == 0 {
		bidCommonwealth = r.bidCommonwealth(b)
		quoteMode, bigModel = r.quoteMode(b)
		if bidCommonwealth == -2 && len(bigModel) > 0 {
			_, quoteMode, quoteMode_bigModel, bidCommonwealth, bidCommonwealth_bigModel = (&Model{}).Do(b, 2)
		} else if bidCommonwealth == -2 {
			_, _, _, bidCommonwealth, bidCommonwealth_bigModel = (&Model{}).Do(b, 3)
		} else if len(bigModel) > 0 {
			_, quoteMode, quoteMode_bigModel, _, _ = (&Model{}).Do(b, 1)
			if !bigModel[quoteMode] {
				quoteMode = ""
			}
		}
	} else if b.Type == 1 {
		quoteMode, bigModel = r.quoteMode(b)
		if len(bigModel) > 0 {
			_, quoteMode, quoteMode_bigModel, _, _ = (&Model{}).Do(b, 1)
			if !bigModel[quoteMode] {
				quoteMode = ""
			}
		}
	} else if b.Type == 2 {
		bidCommonwealth = r.bidCommonwealth(b)
		if bidCommonwealth == -2 {
			_, _, _, bidCommonwealth, bidCommonwealth_bigModel = (&Model{}).Do(b, 3)
		}
	}
	if b.Type == 0 || b.Type == 1 {
		if quoteMode == "" && (b.Bidamount == 0 || b.Bidamount > g.Config().MustGet(gctx.New(), "quoteMode.minBidamount").Float64()) {
			quoteMode = QuoteMode_Whole
		}
	}
	log.Println(b.Id, "规则", "报价模式", quoteMode, "中标联合体", bidCommonwealth)
	return true, quoteMode, quoteMode_bigModel, bidCommonwealth, bidCommonwealth_bigModel
}

// 识别报价模式
func (r *Rule) quoteMode(b *BidInfo) (string, map[string]bool) {
	for _, v := range quoteModeRules {
		vv := gvar.New(v).MapStrVar()
		modeName := vv["mode"].String()
		for _, row := range b.TableKv {
			for k, v := range row {
				if tableV := vv["tableV"].String(); tableV != "" {
					matchedK, _ := regexp.MatchString(quoteModeTableKReg, k)
					if !matchedK {
						if tableK := vv["tableK"].String(); tableK != "" {
							matchedK, _ = regexp.MatchString(tableK, k)
						}
					}
					matchedV, _ := regexp.MatchString(tableV, v)
					if matchedK && matchedV {
						log.Println(b.Id, "tableKv", k, v)
						return modeName, nil
					}
				}
			}
		}
	}
	bigModel := map[string]bool{}
	for _, v := range quoteModeRules {
		vv := gvar.New(v).MapStrVar()
		modeName := vv["mode"].String()
		var lineUnderstandPatterns []string
		if lineUnderstand := vv["lineUnderstand"].MapStrVar(); lineUnderstand != nil {
			lineUnderstandPatterns = lineUnderstand["patterns"].Strings()
		}
		lines := strings.Split(b.Detail, "\n")
		for lineNum, lineVal := range lines {
			if r.matchAnyPattern(b.Id, lineVal, vv["patterns"].Strings()) {
				return modeName, nil
			}
			for _, vvv := range lineUnderstandPatterns {
				countLine := len(strings.Split(vvv, "+"))
				if countLine > 1 {
					countLine += vv["space"].Int()
				}
				newText := []string{}
				for i := lineNum; i < lineNum+countLine && i < len(lines); i++ {
					newText = append(newText, lines[i])
				}
				if r.matchAnyPattern(b.Id, strings.Join(newText, "\n"), []string{vvv}) {
					return modeName, nil
				}
			}
		}
		//
		if modelPatterns := vv["modelPatterns"].String(); modelPatterns != "" && r.matchAnyPattern(b.Id, b.Detail, []string{modelPatterns}) {
			bigModel[modeName] = true
		}
	}
	return "", bigModel
}

// 识别中标联合体
func (r *Rule) bidCommonwealth(b *BidInfo) int {
	// Step 1: 排除黑名单
	if r.matchAnyPattern(b.Id, b.Detail, bidCommonwealth_blacklistPatterns) {
		return -1
	}
	for _, v := range b.KvText {
		if r.matchAnyPattern(b.Id, v, bidCommonwealth_whitelistPatterns) {
			return 1
		}
	}
	if b.FirstWinner != "" && regexp.MustCompile(bidCommonwealth_firstWinnerOrder).MatchString(b.FirstWinner) {
		log.Println(b.Id, "第一中标候选人", b.FirstWinner, bidCommonwealth_firstWinnerOrder)
		return 1
	} else if r.matchOnlyOnce(b.Id, b.Detail) {
		return 1
	} else if r.matchAnyPattern(b.Id, b.Detail, bidCommonwealth_whitelistPatterns) {
		return 1
	} else if b.Multipackage < 2 && len(moreWinnerReg.Split(b.Winner, -1)) > 1 {
		log.Println(b.Id, "单包多个中标单位")
		return 1
	}
	index := strings.Index(b.Detail, "中标")
	if index != -1 {
		start := max(0, index-50)
		end := min(len(b.Detail), index+50)
		contextAroundWin := b.Detail[start:end]
		if strings.Contains(contextAroundWin, "联合体") {
			log.Println(b.Id, "”中标“附件有”联合体“")
			return 1
		}
	}
	if r.matchAnyPattern(b.Id, b.Detail, bidCommonwealth_modelPatterns) {
		return -2
	}
	return -1
}

// 匹配任意模式
func (r *Rule) matchAnyPattern(_id string, text string, patterns []string) bool {
	for _, pattern := range patterns {
		if strings.HasPrefix(pattern, "全部包含：") {
			vs := strings.Split(strings.TrimPrefix(pattern, "全部包含："), "+")
			index := 0
			for _, v := range vs {
				if strings.Contains(text, v) {
					index++
				}
			}
			if index == len(vs) {
				log.Println(_id, "matchAnyPattern", pattern)
				return true
			}
			continue
		}
		if matched, _ := regexp.MatchString(pattern, text); matched {
			log.Println(_id, "matchAnyPattern", pattern)
			return true
		}
	}
	return false
}

// 全部匹配
func (r *Rule) matchAllPattern(text string, patterns []string) bool {
	count := 0
	for _, pattern := range patterns {
		if matched, _ := regexp.MatchString(pattern, text); matched {
			count++
		}
	}
	if count > 0 && count == len(patterns) {
		return true
	}
	return false
}

func (r *Rule) matchOnlyOnce(_id, text string) bool {
	for _, v := range showOnlyOnce {
		count := 0
		for _, vv := range v {
			if len(regexp.MustCompile(vv).FindAllString(text, -1)) != 1 {
				break
			}
			count++
		}
		if count > 0 && count == len(v) {
			log.Println(_id, v)
			return true
		}
	}
	return false
}