分包优化

src/jy/pretreated/division.go

@@ -3,11 +3,13 @@ package pretreated
 import (
 	"fmt"
 	"jy/util"
+
 	qutil "qfw/util"
 	"regexp"
 	"sort"
 	"strconv"
 	"strings"
+	"unicode/utf8"
 )
 
 //分块、分段功能
@@ -53,6 +55,7 @@ var (
 	regStartWrap       = regexp.MustCompile("^[\r\n]")
 	regEndWrap         = regexp.MustCompile("[\r\n]$")
 	regMoreWrap        = regexp.MustCompile("[\r\n]{2,}")
+	regStrWrap        = regexp.MustCompile("分包名称[：:]")
 	replSerial         = regexp.MustCompile("(\r\n|^)([\\d一二三四五六七八九十][、．.:：，])+\\d")
 	moreColonReg       = regexp.MustCompile("[:：]+")
 	regFilter          = regexp.MustCompile("等$")
@@ -578,6 +581,7 @@ func FindPackageFromText(title string, content string,isSite bool,codeSite strin
 //分块之后分包
 func divisionPackageChild(blockPackage *map[string]*util.BlockPackage, content, title string, isFindWinnerOrder, accuracy bool,isSite bool,codeSite string) (bool, string) {
 	//查找知否有分包
+	content = regStrWrap.ReplaceAllString(content, "\n")
 	content = regMoreWrap.ReplaceAllString(content, "\n")
 	content = regEndWrap.ReplaceAllString(content, "")
 	con, pkg, flag := CheckMultiPackage(content, title)
@@ -781,12 +785,28 @@ func interceptText(indexs []int, indexPkgMap map[int]string, pkgIndexMap map[str
 		} else {
 			text = con[iv:]
 		}
+		tmptext := text
 		if strings.Contains(text, "、") {
 			text = strings.Split(text, "、")[0]
 		} else if strings.Contains(text, "\n") {
-			text = strings.Split(text, "\n")[0]
+			texts := strings.Split(text, "\n")
+			text2 :=""
+			if ik+1 < len(indexs)-1 {
+				if startEndMap[indexs[ik+1+1]] != 0 {
+					text2 = con[startEndMap[indexs[ik+1]]:startEndMap[indexs[ik+1+1]]]
+				} else {
+					text2 = con[indexs[ik+1]:indexs[ik+1+1]]
+				}
+				if texts[len(texts)-1] == text2{
+					text = texts[0]
+				}
+			}
+		}
+		if utf8.RuneCountInString(text)<5{
+			indexTextMap[iv] = tmptext
+		}else {
+			indexTextMap[iv] = text
 		}
-		indexTextMap[iv] = text
 		warpCount := len(regSpliteSegment.FindAllStringIndex(text, -1))
 		if warpCount > maxWarpCount {
 			maxWarpCount = warpCount