goProject/trunk/goutil/dfaUtil/dfa.go

package dfaUtil

import "strings"

/*
DFA util, is used to verify whether a sentence has invalid words.
The underlying data structure is trie.
https://en.wikipedia.org/wiki/Trie
*/

// dfa util
type DFAUtil struct {
	// The root node
	root *trieNode
}

// 搜索语句
// 由于go不支持tuple，所以为了避免定义多余的struct，特别使用两个list来分别返回匹配的索引的上界和下界
// 在处理此方法的返回值时，需要两者配合使用
// 参数：
//
//	sentence：语句字符串
//
// 返回：
//
//	搜索到的开始位置列表
//	搜索到的结束位置列表
func (this *DFAUtil) SearchSentence(sentence string) (startIndexList, endIndexList []int) {
	sentenceRuneList := []rune(sentence)
	for i := 0; i < len(sentenceRuneList); {
		//按序匹配每个字
		end := this.searchSentenceByStart(i, sentenceRuneList)
		if end < 0 {
			//匹配失败，继续匹配下一个字
			i++
		} else {
			//匹配成功，记录索引位置
			startIndexList = append(startIndexList, i)
			endIndexList = append(endIndexList, end)

			//从匹配到的字后面开始找
			i = end + 1
		}
	}

	return
}

// 从指定的开始位置搜索语句
// 参数：
//
//	start：开始匹配的位置
//	sentenceRuneList：语句字列表
//
// 返回：
//
//	匹配到的结束位置，未匹配到返回-1
func (this *DFAUtil) searchSentenceByStart(start int, sentenceRuneList []rune) (endIndex int) {
	//当前节点，从根节点开始找
	currNode := this.root
	//是否匹配到
	var isMatched bool

	//按顺序匹配字
	for i := start; i < len(sentenceRuneList); {
		child, exists := currNode.children[sentenceRuneList[i]]

		//未匹配到则结束，跳出循环（可能匹配到过词结尾）
		if !exists {
			break
		}

		//是否是词末尾，如果是则先记录下来，因为还可能匹配到更长的词
		//比如["金鳞"、"金鳞岂是池中物"] => 匹配"金鳞岂是池中物"，匹配到"金鳞"不应该停下来，应继续匹配更长的词
		if child.isEndOfWord {
			endIndex = i
			isMatched = true
		}

		//是否已经到词末尾
		if len(child.children) == 0 {
			return endIndex
		} else {
			//继续与后面的字匹配
			currNode = child
		}

		//增加索引匹配下一个位置
		i++
	}

	//匹配结束，若曾经匹配到词末尾，则直接返回匹配到的位置
	if isMatched {
		return endIndex
	} else {
		//没有匹配到词末尾，则返回匹配失败
		return -1
	}
}

// Insert new word into object
func (this *DFAUtil) InsertWord(word []rune) {
	currNode := this.root
	for _, c := range word {
		if cildNode, exist := currNode.children[c]; !exist {
			cildNode = newtrieNode()
			currNode.children[c] = cildNode
			currNode = cildNode
		} else {
			currNode = cildNode
		}
	}

	currNode.isEndOfWord = true
}

// Check if there is any word in the trie that starts with the given prefix.
func (this *DFAUtil) StartsWith(prefix []rune) bool {
	currNode := this.root
	for _, c := range prefix {
		if childNode, exist := currNode.children[c]; !exist {
			return false
		} else {
			currNode = childNode
		}
	}

	return true
}

// Judge if input sentence contains some special caracter
// Return:
// Matc or not
func (this *DFAUtil) IsMatch(sentence string) bool {
	startIndexList, _ := this.SearchSentence(sentence)
	return len(startIndexList) > 0
}

// Handle sentence. Use specified caracter to replace those sensitive caracters.
// input: Input sentence
// replaceCh: candidate
// Return:
// Sentence after manipulation
func (this *DFAUtil) HandleWord(sentence string, replaceCh rune) string {
	startIndexList, endIndexList := this.SearchSentence(sentence)
	if len(startIndexList) == 0 {
		return sentence
	}

	// Manipulate
	sentenceList := []rune(sentence)
	for i := 0; i < len(startIndexList); i++ {
		for index := startIndexList[i]; index <= endIndexList[i]; index++ {
			sentenceList[index] = replaceCh
		}
	}

	return string(sentenceList)
}

// Handle sentence. Use specified caracter to replace those sensitive caracters.
// input: Input sentence
// replaceCh: candidate
// Return:
// Sentence after manipulation
func (this *DFAUtil) HandleWordUseStr(input string, replaceCh string) string {
	input2 := strings.ToUpper(input)

	startIndexList, endIndexList := this.SearchSentence(input2)
	if len(startIndexList) == 0 {
		return input
	}

	// Manipulate
	inputRune := []rune(input)
	replaceChList := []rune(replaceCh)

	//上一次替换掉的数量
	lastReplaceCount := 0

	for i := 0; i < len(startIndexList); i++ {

		//替换字的索引
		index := len(replaceChList)

		//开始位置--加上替换的词的索引
		starIndex := startIndexList[i] + (i * index) - lastReplaceCount

		//结束位置
		endIndex := endIndexList[i] + (i * index) - lastReplaceCount

		//结束字符串
		sentenceAttr := string(inputRune[endIndex+1:])

		//替换范围字符串
		inputRune = append(inputRune[:starIndex], replaceChList...)
		inputRune = append(inputRune, []rune(sentenceAttr)...)
		lastReplaceCount = endIndex + 1 - starIndex
	}

	return string(inputRune)
}

// Create new DfaUtil object
// wordList:word list
func NewDFAUtil(wordList []string) *DFAUtil {
	this := &DFAUtil{
		root: newtrieNode(),
	}

	for _, word := range wordList {
		wordRuneList := []rune(word)
		if len(wordRuneList) > 0 {
			this.InsertWord(wordRuneList)
		}
	}

	return this
}