I am looking for certain entries with special words in a string. The string looks like this.

entry 1: hello
entry 2: world
entry 3: this
is a multiline
that makes it hard
entry 4: here we have a special entry
entry 5: here
we
have 
another special entry
in a multiline
entry 6: end

Because it is an multiline problem I use Java's DOTALL so that the . matches also newline characters.

I am looking for entries that have the word special in it.

First I tried to find a regex, that captures a full entry: entry \d : .*?(?=\s*(entry \d: )|\Z). That is like a simplified version of this

Then I thought, ok I just have to exchange the .*? for the regex I need to find. But entry \d : .*?special.*?(?=\s*(entry \d: )|\Z) does not work, probably because the special breaks the greed of the expression.

Does anyone know a better solution?

CodePudding user response：

You can use a tempered greedy token:

(?s)entry \d : (?:(?!entry \d : ).)*special.*?(?=\s*entry \d : |$)

See the regex demo. Details:

• entry \d : - entry space one or more digits, :, space
• (?:(?!entry \d : ).)* - any char, repeated zero or more times, that does not start the entry space one or more digits, :, space sequence
• special - a fixed string
• .*? - any zero or more chars as few as possible
• (?=\s*entry \d : |$) - a positive lookahead that matches a location in string that is immediately followed with zero or more whitespaces, entry, space, one or more digits, : and space, or end of the string.

NOTE: Do not use Pattern.MULTILINE with this regex. Or, keep on using \Z (end of the string, or position right before the trailing newline, LF char).

CodePudding user response：

If you use words and space classes instead of dots then it seems to work

/entry \d : [\w\s]*special[\w\s]*?(?=\s*(?:entry \d :)|$)/gm

It seems that if you allow the colon : in your text, it breaks the expression.

And also you have \Z in your expression but it seems to me that end of line $ is more suited here

CodePudding user response：

[Edit:] I unfortunately missed the multiline nature of entries, so this answer is valid for single line entries but will return only the first line for multiline entries. I think one could overcome this by setting a certain regex for delimiter, though.

I'd suggest you use a Scanner to deal with the multi line aspect. This will give you a stream of tokens (the lines). You can use a String.contains(...) or a String.matches(...) to filter tokens then.

var result = new Scanner(myMultiLineString).tokens()
                                           .useDelimiter("\\n")
                                           // alternatively use String.contains(...)
                                           // if you're looking for a constant
                                           // rather than a complex rule.
                                           .filter(s -> s.matches(regex))
                                           .collect(Collectors.toList());