C# Regex - only match if substring exists?

Question

Ok, so I think I've got a handle on negation - now what about only selecting a match that has a specified substring within it?

Given:

This is a random bit of information from 0 to 1.
  This is a non-random bit of information I do NOT want to match
This is the end of this bit

This is a random bit of information from 0 to 1.
  This is a random bit of information I do want to match
This is the end of this bit

And attempting the following regex:

/(?s)This is a random bit(?:(?=This is a random).)*?This is the end/g

Why isn't this working? What am I missing?

I'm using regexstorm.com for testing...

Answer 1

You ruined a tempered greedy token by turning the negative lookahead into a positive one. It won't work that way because the positive lookahead requires the text to equal This is a random at each position after This is a random bit .

You need:

• Match the leading delimiter ( This is a random bit )
• Match all 0+ text that is not the leading/closing delimiters and not the required random text inside this block
• Match the specific string inside ( This is a random )
• Match all 0+ text that is not the leading/closing delimiters
• Match the closing delimiter ( This is the end )

So, use

(?s)This is a random bit(?:(?!This is a random bit|This is the end|This is a random).)*This is a random(?:(?!This is a random bit|This is the end).)*This is the end

See the regex demo

• (?s) - DOTALL mode on ( . matches a newline)
• This is a random bit - Leading delimiter
• (?: # Start of the tempered greedy token (?!This is a random bit # Leading delimiter | This is the end # Trailing delimiter | This is a random) # Sepcific string inside . # Any character )* # End of tempered greedy token
• This is a random - specified substring
• (?:(?!This is a random bit|This is the end).)* - Another tempered greedy token matching any text not leading/closing delimiters up to the first...
• This is the end - trailing delimiter

Answer 2

I hope you understand this (?:(?=This is a random).) can only match once, never twice if it were quantified. For example Th can satisfy the lookahead. When the T is consumed, the next character is h which will never satisfy the lookahhead Th . The next expression is evaluated, never to return to the lookahead again. Use a negative lookahead instead.