problems with a regex

Hi.

I'm trying to create a regular expression to match and extract some information. Two examples of the source string

example one: 10/0/leaf.nr.0 is a Projection error - touches edge - 3D points.csv

example two: 10/2/leaf.nr.2 is a Projection error - 3D points.csv

I want to extract the string between "is a " and " - touches edge" OR " - 3D" In both example strings this would be "Projection error" but this can be something else.

Currently I have the pattern:

'.*is\sa\s(?<type>.*)(?:\s\-\stouches\sedge)?(?:\s\-\s3D).*.csv'

for example one this returns (not expected):

'Projection error - touches edge'

but for example two it returns(expected):

'Projection error'

IF I change the pattern to:

'.*is\sa\s(?<type>.*)(?:\s\-\stouches\sedge)(?:\s\-\s3D).*.csv'

so I require the (?:\s\-\stouches\sedge) to be matched it returns (correctly):

'Projection error'

for example one but now example two (that dont have the the "touches edge" part ) will not match(of cause).

I dont get why example one also contains the " - touches edge" in the result using the first pattern when I ask it to match this pattern 0 or 1 times.

Any help will be highly appreciated.

Best regards, Thomas

[Thomas]

My current solution is to use this pattern instead:

'.*is\sa\s(?<type>[\w\s]*)(?:\s\-\s)?.*'

It results in the needed information except an extra space character are added. So the result for both example one and two are now:

"Projection error "

Answer 1

[Muthu Annamalai]

A simple solution to parse the string with rule

   "is a " and ( " - touches edge" OR " - 3D" )

is to use sequential regexp().

That way you know "is a" bit of your source is split out, and then you can search for which of 2 alternatives are present in your case.

Also see the 'NOT' exclusion class operators in regexp, and 'split' mode of regexp.

http://www.mathworks.com/help/matlab/ref/regexp.html

[Thomas]

Thanks for your response.

My task is not to match either of the two cases - its simply to extract the string between "is a " and the first " - " (This is a new, shorter, formulation of my problem that I just realized)

Splitting would be a way to go but I would like to know if its possible to create a regex for it.

Answer 2

[per isakson]

to extract the string between "is a " and the first " - " This formulation is close to a pseudo-code for the expression we search.

    ex1 = '10/0/leaf.nr.0 is a Projection error - touches edge - 3D points.csv';
    ex2 = '10/2/leaf.nr.2 is a Projection error - 3D points.csv';
    regexp( ex1, '(?<=is a )[^\-]+(?= \- )', 'match' )
    regexp( ex2, '(?<=is a )[^\-]+(?= \- )', 'match' )

returns

    ans = 
        'Projection error'
    ans = 
        'Projection error'

Search the doc for "Lookaround Assertions" or just "Lookaround". Lookahead Assertions in Regular Expressions

PS. '\-' or just '-' ; a backslash (escape) too many seldom hurts and I've problems to remember when it's needed.

.

OR according to the requirement of the OP

    regexp( ex1, '(?<=is a ).+?(?= ((\- touches edge)|(\- 3D)))', 'match' )
    regexp( ex2, '(?<=is a ).+?(?= ((\- touches edge)|(\- 3D)))', 'match' )

The extra parentheses, (), makes the expression more readable - imo.

The "?" in ".+?" is the

    Lazy expression: match as few characters as necessary.