How to capture an optional expression using regular expressions?

Dear all,

I would like to use regular expressions to capture and transform expressions of the form

name

or

name(string,digits)

where name belongs to a list of NAMES and digits is an integer: 1, 2, 3,...

That is, "name" is optionally followed by

1. an opening parenthesis: (
2. a string
3. some numbers : 1, 2, 3,...
4. a closing parenthesis: )

For that purpose I wrote the following regular expression that does not work

 expr='(\w+)((\w+),(\d+))?'
 replace='${convertMe($1,$3,$4)}';
 result=regexprep(cellarray,expr,replace)

I have written a convertMe function taking 3 inputs but only the first input gets in. The other inputs the function receives are $3 and $4 instead of the second string and the digits.

Any suggestions?

Answer 1

[Walter Roberson]

For the longer case, expr='(\w+)(?:\(\w+),\(d+)\))' replace='${convertMe($1,$2,$3)}'

For the case with no argument supplied, it is not clear what you would like passed to convertMe or if you want convertMe to be called at all.

[Patrick Mboma]

Hi Walter, Thanks for the answer. It seems to me that what you suggest has some problems. I tried to correct your suggestion as follows

 expr='(\w+)(?:\((\w+),(\d+)\))'

But it still doesn't work. I do capture the first \w+ but not the second and not the \d+ . The second and third arguments received by convertMe are $2 and $3.

In the short case, I would like to capture only the first \w+ only. I would be fine if in that case convertMe receives $2 and $3 as the second and third input arguments respectively.

[Walter Roberson]

Odd. I had it working the other day, but now it doesn't.

Answer 2

[Cedric]

Another option is to parse all entries first, and then to rebuild relevant expressions:

 entries = {'name1(John, 48)', 'name2', 'name3(Doo)'} ;  
 tokens  = regexp( entries, '([\w\d_-]+)\(?(\w+)?,?\s*(\d+)?', 'tokens', 'once' ) ;
 parsed  = vertcat( tokens{:} ) ;

With that you get

 >> parsed
 parsed = 
    'name1'    'John'    '48'
    'name2'    ''        ''  
    'name3'    'Doo'     ''

which is easy to post process for building whatever you need.