Extract numbers between two underscores.

Hi;

I am new to MATLAB and I need to extract numbers between two underscores. The only way I think this can be done is using the 'regexp' function. However, I am not sure about the expression that I need to the match it with. Can anyone help me?

Thanks;

Examples of my strings are:

    Color_84_2014-01-31-16-49-31-702.jpg
    Color_85_2014-01-31-16-49-31-732.jpg
    Color_86_2014-01-31-16-49-31-762.jpg
    Color_87_2014-01-31-16-49-31-792.jpg

So I just need the number 84,85,86,87,...

Thanks Again!

[Cedric]

Do you have all these strings in e.g. a file and you want to extract all the numbers in one shot, or are you processing these string one at a time?

Answer 1

[dpb]

>> s='Color_84_2014-01-31-16-49-31-702.jpg';
>> sscanf(s,'Color_%d_%d-%d-%d-%d-%d-%d-%d.jpg')
ans =
        84
      2014
         1
        31
        16
        49
        31
       702
>>

Answer 2

[Geoff Hayes]

You could just use the strfind function to find the underscores, and then extract the data in between those two values:

 str  = 'Color_84_2014-01-31-16-49-31-702.jpg';
 idcs = strfind(str,'_');
 num  = str2num(str(idcs(1)+1:idcs(2)-1));

A regexp alternative probably exists. Try the above and see what happens!

[Cedric]

 num = str2double( regexp( str, '(?<=_)\d+(?=_)', 'match' )) ;

or

 tokens = regexp( str, '_(\d+)_', 'tokens' ) ;
 num    = str2double( cat( 1, tokens{:} )) ;

[Geoff Hayes]

Neat - I wondered about that!

[Cedric]

The pattern in the first approach matches:

• One or more numeric characters: '\d+', where \d is the wildcard/operator for numeric characters and + is the quantifier "one or more".
• Preceded by an underscore: '(?<=_)' where (?<=..) is a positive "lookbehind" for ...
• Followed by an underscore: '(?=_)' where (?=..) is a positive "lookahead" for ...

Note that the "lookaround" operators are not included in the output.

The second approach is based on tokens extraction. The pattern says: an underscore followed by one or more numeric characters followed by an underscore, which is the following match '_\d+_', but we group the \d+ in a token (defined by the framing parentheses) and we ask REGEXP to extract just the token (with the 3rd 'tokens' argument in the call).

Answer 3

[Jos (10584)]

str = 'Color_84_2014-01-31-16-49-31-702.jpg'
num = sscanf(str,'Color_%d')

If your strings are stored in a cell array of strings C, use cell fun

num = cellfun(@(str) sscanf(str,'Color_%d') , C)

[dpb]

>> sscanf(s,'Color_%d')
ans =
  84

Gotta have enough formats to convert. Also note that despite the title there are dashes as well as underscores in the string.