利用?防止过度匹配

Preventing Over Matching

? matches are limited in scope (zero or one match only), and so are interval matches when using exact amounts or ranges. But the other forms of repetition described in this lesson can match an unlimited number of matches—sometimes too many.

All the examples thus far were carefully chosen so as not to run into over matching, but consider this next example. The text that follows is part of a Web page and contains text with embedded HTML tags. The regular expression needs to match any text within tags (perhaps so as to be able to replace the formatting). Here's the example:

Text

This offer is not available to customers

living in AK and HI.

RegEx

<[Bb]>.*

Result

This offer is not available to customers

living in AK and HI
.

Analysis

<[Bb]> matches the opening tag (in either uppercase or lowercase), and matches the closing tag (also in either uppercase or lowercase). But instead of two matches, only one was found; the .* matched everything after the first until the last so that the text AK and HI was matched. This includes the text we wanted matched, but also other instances of the tags as well.

The reason for this is that metacharacters such as * and + are greedy; that is, they look for the greatest possible match as opposed to the smallest. It is almost as if the matching starts from the end of the text, working backward until the next match is found, in contrast to starting from the beginning. This is deliberate and by design, quantifiers are greedy.

But what if you don't want greedy matching? The solution is to use lazy versions of these quantifiers (they are referred to as being lazy because they match the fewest characters instead of the most). Lazy quantifiers are defined by appending an ? to the quantifier being used, and each of the greedy quantifiers has a lazy equivalent as listed in Table 5.1.

Table 5.1. Greedy and Lazy Quantifiers

Greedy

Lazy

*

*?

+

+?

{n,}

{n,}?

*? is the lazy version of *, so let's revisit our example, this time using *?:

Text

This offer is not available to customers

living in AK and HI.

RegEx

<[Bb]>.*?

Result

This offer is not available to customers

living in AK

 and HI
.

Analysis

That worked, by using the lazy *? only AK, was matched in the first match allowing HI to be matched independently.