如何记录正则表达式本身？

Question

I have a regular expression eg at regex101

(\/+[^\/\[\]]+(?:\[[^\]']*(?:'[^']*')\])?)+

I have verified that it matches my test string

//SapButton[@automationId='tbar[0]/btn[15]']

Since the Regex cannot be understood rightaway, I tried the documentation feature using (?#) , so I changed the Regex to also at regex101

((?# Capturing group for the type name)
\/+(?# Start with / or // )
[^\/\[\]]+(?# Type name exclusing start of attribute and next type)
(?:(?# Non-capturing group for the attribute)
\[(?# Start of an attribute)
[^\]']*(?# Anything but end of attribute or start of string)
(?:(?# non-capturing group for string)
'(?# string start)
[^']*(?# anything inside the string, except end of string)
'(?# string end)
)(?# end of string group)
\](?# end of attribute)
)?(?# Attribute can occur 0 or one time)
)+(?# Type can occur once or many times)

But now the regex does not match my test string any more. The reason are the newlines. Changing the Regex to

((?# Capturing group for the type name)\/+(?# Start with / or // )[^\/\[\]]+(?# Type name exclusing start of attribute and next type)(?:(?# Non-capturing group for the attribute)\[(?# Start of an attribute)[^\]']*(?# Anything but end of attribute or start of string)(?:(?# non-capturing group for string)'(?# string start)[^']*(?# anything inside the string, except end of string)'(?# string end))(?# end of string group)\](?# end of attribute))?(?# Attribute can occur 0 or one time))+(?# Type can occur once or many times)

works. But it is unreadable again.

How do I document a regular expression in itself properly?

Note that I want to avoid doing it in the comment of the C# method since this has too much potential for not being updated when the regex is changed.

IMHO, it would be best done in a verbatim string with multiple lines (but it still has to work, of course).

Answer 1

There is the Ignore White Space option

The problem is that then you'll have to escape spaces and # with a \\ . The good news is that # will begin a comment, like // in C#

You can activate it with RegexOptions.IgnorePatternWhitespace or with (?x) at the beginning of the regex.

(?x) is supported by https://regex101.com/

Answer 2

It isn't exactly a desirable solution, but you can store your documented regexp in a String and, before match, you can replace all \\r\\n from your String.

Then, you will have a readable representation in your code, and a correct regexp at runtime.

I don't know other way, but what you with is laudable.

Regards