KMP表构建算法

时间:2014-09-10 15:47:52

标签: algorithm pattern-matching knuth-morris-pratt

我检查了KMP table-building algorithm from Wikipedia,但我不理解第二种情况循环背后的逻辑

(second case: it doesn't, but we can fall back)
        else if cnd > 0 then
            let cnd ← T[cnd]

我尝试使用此算法构建一个表,它运行得很好。我知道cnd ← T[cnd]有助于找到合适的后缀长度。我不明白的是"怎么"它做到了吗?

用例子解释会很好。

谢谢!

编辑:我刚发现我的问题与此处的问题重复:"Partial match" table (aka "failure function") in KMP (on wikipedia)

我想我现在得到了答案。不过,还有一个解释会有所帮助。谢谢!

1 个答案:

答案 0 :(得分:3)

假设您有一个字符串Hello World!!!,并且您想要搜索Head Up

Hello World!!!
Head Up
  ^

当您处于第一个和第二个字符时,第一个条件应用(first case: the substring continues),对于标记位置,字符不匹配但您已经在子字符串匹配内(2个字符)匹配到那里),这种情况对应于第二个条件(second case: it doesn't, but we can fall back)。第三种情况是你错过匹配模式的第一个字符。

第二个条件是必要的,因为您可以使用匹配字符的信息直到未命中匹配,以避免您已经知道结果的不必要的比较(跳过您已经知道的string的字符开始部分模式不匹配)。

示例:使用字符串HeHello World!!!并搜索Hello

HeHello World!!!
Hello
  ^ when you miss match this character using the table of KMP you known that 
    could skip 2 characters because

HeHello World!!!
 Hello
 ^ this would miss match

在为模式HeHello构建模式表的情况下。假设^cnd*pos。起点为pos = 2cnd = 0(但在使用pos - 1 = 1检查模式时)。

HeHeHello     T [-1,0,0,0,0,0,0,0,0]
^* comparing 0 with 1 go to condition 3        cnd = 0, pos = 2
                        _
HeHeHello     T [-1,0,0,1,0,0,0,0,0]
^ * comparing 0 with 2 go to condition 1       cnd = 0, pos = 3
                          _
HeHeHello     T [-1,0,0,1,2,0,0,0,0]
 ^ * comparing 1 with 3 go to condition 1      cnd = 1, pos = 4
                            _
HeHeHello     T [-1,0,0,1,2,3,0,0,0]
  ^ * comparing 2 with 4 go to condition 1     cnd = 2, pos = 5
                              _
HeHeHello     T [-1,0,0,1,2,3,4,0,0]
   ^ * comparing 3 with 5 go to condition 1    cnd = 3, pos = 6

HeHeHello     T [-1,0,0,1,2,3,4,0,0]
    ^ * comparing 4 with 6 go to condition 2 (cnd = T[cnd], cnd = T[4] = 2)

HeHeHello     T [-1,0,0,1,2,3,4,0,0]
  ^   * comparing 2 with 6 go to condition 2 (cnd = T[cnd], cnd = T[2] = 0)

...