Regular expression Help : Skip lines containing some word

[kuldeep sidhu]

Hi
Could you please guide me in writing a regular expression that should match the following criteria..
Sample Text

NULL0 current status: UP
Line protocol current status: UP (spoofing)
Sample101 current status: UP
Line protocol current status: UP

From above sample i need to select llines having current status in those, but if protocol word is there on left, it should skip this line.

I tried a few ways
1.

\s*[Cc]urrent status:\s(.*)

- This selects all the lines.
2.

.*protocol\s[Cc]urrent status:\s(.*)

- This selects 2 and 4 line.
I want opposite of that. I want 1 and 3 lines to be selected.

I could not find how to negate the 2nd regular expression so that it only select lines 1 and 3.

[Stephan van Hulst]

You can do this with a zero width negative lookahead. Note that you can also make regular expressions case insensitive with the help of flags.

[fred rosenberger]

Does it have to be a single regex? I always find that if I go to a huge effort to write a very complicated regex to meet all the criteria, then a week/month later when the criteria changes I am screwed.

Can you instead write a regex that finds all lines that have "current status", and then have a second regex to filter those that have "protocol"? If your criteria changes, it's pretty easy to add another that also filters (for example) "media", or that also allows "previous status".

[kuldeep sidhu]

fred rosenberger wrote:Does it have to be a single regex? I always find that if I go to a huge effort to write a very complicated regex to meet all the criteria, then a week/month later when the criteria changes I am screwed.

Can you instead write a regex that finds all lines that have "current status", and then have a second regex to filter those that have "protocol"? If your criteria changes, it's pretty easy to add another that also filters (for example) "media", or that also allows "previous status".

Thanks Fred.
No. I can not use multiple regex. I need to put this in already existing system which only accepts a single regex.

[kuldeep sidhu]

Stephan van Hulst wrote:You can do this with a zero width negative lookahead. Note that you can also make regular expressions case insensitive with the help of flags.

Thanks Stephan. I got a idea from your reply.
Actually i used negative lookbehind to achieve the desired output.
So here is the final regex i used -

\s*(?<!protocol\s)[Cc]urrent status:\s(\w*)$

This only selects the status value (UP in this case) from line 1 and 3.

Result

1. 	current status: UP	 	 
 	1.1. Group: UP
2. 	current status: UP	 	 
 	2.1. Group: UP

[Stephan van Hulst]

Great job!

If you're going to use the ^ and $ boundary matchers, I recommend using the Pattern.MULTILINE flag, because otherwise they match the begin and end of input, and not the begin and end of individual lines.

Pattern pattern = Pattern.compile(
  "(?<name> \\w+) \\s+ (?<!protocol\\ ) current\\ status \\s*:\\s* (?<status> .*?)$",
  CASE_INSENSITIVE | COMMENTS | MULTILINE
);
 
while (scanner.findWithinHorizon(pattern, 0) != null) {
  MatchResult match = scanner.match();
  String name   = match.group(1);
  String status = match.group(2);
  System.out.println(name + ": " + status);
}

[kuldeep sidhu]

Thanks Stephan. As per the logic we are putting these lines in arraylist(line per entry). So it iterates arraylist and run regex on that.

Could you please suggest something for the following scenario -
Sample

Link speed type is OK, link duplex type is DOWN
Unknown-speed mode, Link speed type is OK, Flow-control is disabled

From this i want to extract the line starting with Link speed(It can be on new line or middle of line) upto the Flow-Control...(this part can or cannot be present)

So the desired output would be

Link speed type is OK, link duplex type is DOWN
Link speed type is OK

means first line full and from second line only link speed part exculding comma and Flow-control part..
I tried few ways..

regex

^\s*(?:.*,)?\s*(Link speed.*)

Problem is this selects 2nd line till the end.

regex

^\s*(?:.*,)?\s*(Link speed.*(?=\W*Flow-control))

this also capture the comma before flow-control..

I believe the trick is that Flow-control.... line can be present at the end or not.
I know it would be easy, but i could not get it properly. I could not understand the working of this properly.

[Stephan van Hulst]

You could just add a optional comma to your lookahead, and make the quantifier after Link speed reluctant.

[kuldeep sidhu]

Stephan van Hulst wrote:You could just add a optional comma to your lookahead, and make the quantifier after Link speed reluctant.

I tried but could not get it working..could you please post the regex you think will work.

[Stephan van Hulst]

You should show us what you tried first.

[kuldeep sidhu]

@Stephan
I tried this as you suggested

^\s*(?:.*,)?\s*(Link speed.*?(?=[,]\W*Flow-control))

This selects the 2nd line properly but nothing is captured from first line.

[Stephan van Hulst]

This completely depends on how the regular expression is used. In the examples I wrote so far, I used a scanner and the findInLine() or findWithinHorizon() methods. In order to help you, you have to provide us with an SSCCE illustrating your current problem.

[kuldeep sidhu]

@Stephan
I am able to get the desired output using the following regex -

^\s*(?:.*,)?\s*(Link speed.*?)(?=,\s*Flow-control|$)

This will either check for flow-control or end of line.
The captured text would be between Link speed...to flow-control or end of the line.

[Stephan van Hulst]

Glad you worked it out!