Re: regexp: how to split a cellstr array into substring arrays, each matching regular expressions

[Philip Nienhuis]

Answering myself (was easier then I thought):

Philip Nienhuis wrote:
> Having a cellstr array like this:
>
> octave:178> ar = {'abcdefguvwxAny' ; 'acegxyzTrailing'; 'vxzJunk'}
> ar =
> {
> [1,1] = abcdefguvwxAny
> [2,1] = acegxyzTrailing
> [3,1] = vxzJunk
> }
>
> how can I efficiently split it into two columns using regular
> expressions like
> '[abcdefg]' and '[uvwxyz]'
>
> to obtain
>
> { 'abcdefg', 'uvwxAny'; 'acegTrailing', 'xyz'; '', 'vxzJunk'} ?
>
> IOW, I'd like to split the cellstr array at the location where
> '[uvwxyz]' matches (even if not present, see far below).
>
>
> The closest I get is:
>
> ## Invert pattern and use 'split' keyword
> octave:179> ss = regexp (ar, '[^abcdefg]', 'split')
> ss =
> {
> [1,1] =
> {
> [1,1] = abcdefg
> [1,2] =
> [1,3] =
> [1,4] =
> [1,5] =
> [1,6] =
> [1,7] =
> [1,8] =
> }
> [2,1] =
> {
> [1,1] = aceg
> [1,2] =
> [1,3] =
> [1,4] =
> [1,5] =
> [1,6] = a
> [1,7] =
> [1,8] =
> [1,9] =
> [1,10] = g
> }
> [3,1] =
> {
> [1,1] =
> [1,2] =
> [1,3] =
> [1,4] =
> [1,5] =
> [1,6] =
> [1,7] =
> [1,8] =
> }
> }
> octave:180> col1 = cellfun (@(x) x{1}, {ss{:}}, 'uni', false)
> col1 =
> {
> [1,1] = abcdefg
> [1,2] = aceg
> [1,3] =
> }
> octave:181> col2 = regexp (ar, '[uvwxyz].*', 'match', 'once')
> tt =
> {
> [1,1] = uvwxAny
> [2,1] = xyzTrailing
> [3,1] = vxzJunk
> }
>
> ## ...or the latter statement, perhaps more robust, as:
> octave:182> tt = regexp (ar, '[uvwxyz].*', 'match')
> tt =
> {
> [1,1] =
> {
> [1,1] = uvwxAny
> }
> [2,1] =
> {
> [1,1] = xyzTrailing
> }
> [3,1] =
> {
> [1,1] = vxzJunk
> }
> }
> octave:183> col2 = cellfun (@(x) [x{:}], {tt{:}}, 'Uni', false)
> tt =
> {
> [1,1] = uvwxAny
> [1,2] = xyzTrailing
> [1,3] = vxzJunk
> }
> octave:184>
>
>
> ( cellfun() was invoked to be able to use repeated indexing; I couldn't
> find another way to extract the first/last entries of ss and tt. )
> I think my method isn't very robust.
> So I hope there's a less convoluted and more reliable way.
>
>
> BTW,
> octave:184> ar = {'abcdefguvwxAny' ; 'acegxyzTrailing'; 'aJunk'}
> ar =
> {
> [1,1] = abcdefguvwxAny
> [2,1] = acegxyzTrailing
> [3,1] = Junk
> }
> octave:186> tt = regexp (ar, '[uvwxyz].*', 'match', 'once')
> tt =
> {
> [1,1] = uvwxAny
> [2,1] = xyzTrailing
> [3,1] = unk
> }
>
> => is this a bug? (swallowing the "J" from the last entry)
>
> Thanks,
>
> Philip

octave-3.6.1.exe:16> ar = {'abcdefguvwxAny' ; 'acegxyzTrailing'; 'Junk'}
ar =
{
  [1,1] = abcdefguvwxAny
  [2,1] = acegxyzTrailing
  [3,1] = Junk
}
octave-3.6.1.exe:17> col_1 = regexp (ar, '^[abcdefg]*', 'once', 'match')
ss =
{
  [1,1] = abcdefg
  [2,1] = aceg
  [3,1] =
}
octave-3.6.1.exe:18> col_2 = regexprep (ar, '^[abcdefg]*', '')
tt =
{
  [1,1] = uvwxAny
  [2,1] = xyzTrailing
  [3,1] = Junk
}

I still have to check with ML on the last question (swallowing characters).

P.