【自動機】簡單的正則表達式匹配

題目來源
https://leetcode.com/problems/regular-expression-matching/

內容:實現一個能支持”.” “*”的模式匹配程序,並判斷是否整個都匹配上。例如:

isMatch("aa", "a") -> false
isMatch("aab", "c*a*b") -> true

做法比較傳統,用自動機來做。

自動機形如下方(其表達式爲c*aa*bb*):
《【自動機】簡單的正則表達式匹配》
當狀態爲s1的時候,如果此處讀到的數據是a的話,可以進入狀態s2。

那麼對於 表達式爲”a*ab”怎麼辦?
它既要適配”ab”, “aab”, “aaab”…
其實可以用一個棧stackSavePoint保存這種(a*)既可以讀當前的數據, 也可以直接跳過到下一個這種類型的節點,保存當前數據的index,然後直接將當前數據交由下一狀態處理。如果遇到當前狀態並不能處理,則從stackSavePoint棧中彈出一節點,當前狀態轉到這一節點,並將當前數據的指針指向(index + 1),也就是回退N-1步,效率有點低,但還是前進了一步。

具體代碼如下:

class Solution {
public:

// c*cab

    struct Node {
        bool isEverything;
        bool isSelfContain;
        char accept;
        struct Node* son;
    };

    struct Node* addNode(struct Node* parent, char ch) { // return current node after added.

        struct Node* son = parent->son;

        if (ch == '*') {
            parent->isSelfContain = true;
            return parent;
        }

        struct Node* newSon = (Node*)malloc(sizeof(struct Node));
        newSon->accept = ch;
        newSon->isEverything = '.' == ch;
        newSon->isSelfContain = false;
        newSon->son = NULL;
        parent->son = newSon;
        return newSon;
    }

    // to merge something like a*a*a*a*b => a*b or .*a*b*c*d => .*d
    struct Node* mergeSameNeighbour(struct Node* parent) {
        struct Node* header = parent;
        while (parent && parent->son) {
            if (parent->isSelfContain && parent->son->isSelfContain) {
                bool isEverything = parent->isEverything || parent->son->isEverything;
                if (isEverything || parent->accept == parent->son->accept) {
                    parent->isEverything = isEverything;
                    struct Node* tmp = parent->son;
                    parent->son = parent->son->son;
                    free(tmp);
                }
            }
            parent = parent->son;
        }
        return header;
    }

    bool isMatch(string s, string p) {

        if (p.empty()) {
            return s.empty();
        }

        Node* emptyHeader = (Node*) malloc(sizeof(struct Node));
        Node* current = emptyHeader;
        int index = 0;
        int len = p.length();
        while (index < len) {
            current = addNode(current, p.at(index++));
        }
        mergeSameNeighbour(emptyHeader->son);
        index = 0;
        len = s.length();
        stack<Node*> savePoint;
        stack<int> saveIndex;
        current = emptyHeader->son;
        while (index < len) {
            char ch = s[index];
            if (!current) {
                goto helpme;
            }
            // printf("char[%d]: %c, current: %c\n", index, ch, current->accept);
            if (current->isSelfContain) {
                if (current->isEverything) {
                    // call me out when you cannot handle it
                    savePoint.push(current);
                    saveIndex.push(index);
                } else if (current->accept == ch) {
                    savePoint.push(current);
                    saveIndex.push(index);
                } 
                current = current->son;
                continue;
            }

            if (current->isEverything || current->accept == ch) {
                index++;
                current = current->son;
                continue;
            }
            helpme:
            if (savePoint.size()) {
                current = savePoint.top();
                savePoint.pop();
                int lastIndex = saveIndex.top();
                saveIndex.pop();
                index = lastIndex + 1;
            } else {
                return false;
            }
        }
        free(emptyHeader);
        /* in here, the test word run out, so we just find whethere the rest is NULL, or all are selfContain, which can be regarded as 0 */
        while(current) {
            if (current->isSelfContain) {
                current = current->son;
            } else {
                return false;
            }
        }
        return !current;
    }
};

這裏有個問題, 如果需要支持[a-zA-Z.*-+]等的話,估計在Node.accept 中修改成keymap的形式,估計問題應該不是很大。

点赞