我希望能够计算给定的 java.util.regex.Pattern 实例可以与字符串中的第一个字符匹配的所有字符的集合。 更正式地说,给定 DFA 相当于某个正则表达式,我想要从起始状态开始的所有传出转换的集合。
示例:
Pattern p = Pattern.compile("[abc]def|daniel|chris|\\s+");
Set<Character> first = getFirstSet(p);
集合 first
应包含以下元素:
{ 'a', 'b', 'c', 'd', ' ', '\n', '\r', '\t' }
有什么想法吗? 我很清楚我可以自己构建 DFA 并以这种方式确定相关状态,但我想避免这种麻烦(阅读:这对我来说不值得那么多)。 请注意,我的主机语言实际上是 Scala,因此我可以访问所有核心 Scala 库(无论其价值如何)。
I would like to be able to compute the set of all characters which may be matched as the first character in a string by a given instance of java.util.regex.Pattern
. More formally, given the DFA equivalent to a certain regular expression, I want the set of all outgoing transitions from the start state.
An example:
Pattern p = Pattern.compile("[abc]def|daniel|chris|\\s+");
Set<Character> first = getFirstSet(p);
The set first
should contain the following elements:
{ 'a', 'b', 'c', 'd', ' ', '\n', '\r', '\t' }
Any ideas? I'm well aware that I could construct the DFA myself and determine the relevant states that way, but I'd like to avoid that kind of hassle (read: it's not worth that much to me). Note that my host language is actually Scala, so I have access to all of the core Scala libs (for what it's worth).
发布评论
评论(2)
我认为您可以解析正则表达式并定义一些递归函数,该函数以从左到右的方式对解析后的正则表达式进行操作,从而构建这样一组第一。
有些事情很简单:
...
将其扩展到您的正则表达式方言知道的所有原语和特殊标志,您就可以开始了。
I think you could parse the regular expression and define some recursive function which operates on the parsed regular expression in a left-to-right-manner, building up such a set of firsts.
Some things are simple:
...
Extend this to all primitives and special flags your regular expression dialect knows and you are good to go.
你可以递归地解决它......
这个想法可能有很多错误,但这就是我会尝试的。 你必须去掉断言、组名和数千个其他东西。 如果你发现像 [^0-9] 这样的倒置字符类,你就必须输出很多字符。
所以我认为这确实是一个复杂的问题。
You could solve it recursivly ...
There are probably a lot of errors in this idea, but this is what I would try. You have to strip out assertion, group names and thousand other things. And if you find an inverted character class like [^0-9] you have to output a lot of characters.
So I assume it is really a complex problem.