Find if any string in a cell array is contained in a string

I have a cell array of strings, for example, C = {'bob','jack','john'}.

I have a test string T = 'The quick brown jack jumped'.

In this case, I'd like my output to be 2, since 'jack' is contained in T. There is a guarantee that only one string in the cell array will match any part of the test string, i.e. the output will be a single index.

What's the best way to go about implementing this code quickly? I don't want to loop through the cell array and check each time since this function is called as the user types an input in a GUI.

Answer 1

[Star Strider]

You have to split ‘T’ into its component words first, but that’s easy enough with a regexp call:

T = 'The quick brown jack jumped';
Tc = regexp(T, ' ', 'split');
C = {'bob','jack','john'};

CLoc = find(ismember(C,Tc))

CLoc =
     2

‘CLoc’ is the location in ‘C’ of the matching word.

[TastyPastry]

What happens if C is multiple words? For example, if C = {'bob','jack jumped','john'}. I'm trying to make a program which detects the user typing a phrase and then autofills the rest by detecting the phrase from some bank, in this case, C.

[Star Strider]

The same logic holds, but because ‘T’ has to be split into its component words, there would be no match with the phrase.

However, if ‘C’ were:

C = {'bob','jack', 'jumped','john'};

CLoc = 
     2     3

(I just verified that.)

So if you wanted to match phrases, you would have to figure out a way to parse the phrase from ‘T’ rather than the words. That would require you to create logic to recognise phrases. The ismember match would then work.

Answer 2

[Kirby Fears]

Have you tested the speed of a for-loop or cellfun()? It's probably your best bet here unless C contains a really large number of strings. You can make the computation faster by ordering C from most-likely to least-likely to be typed.

for-loop:

idx=[];
for c=1:numel(C),
      temp=regexp(T,C{c},'once');
      if ~isempty(temp),
          idx=c;
          break;
      end,
end,

or cellfun:

idx2=find(cellfun(@(c)~isempty(regexp(T,c,'once')),C));

If you involve Java classes, you could possibly speed up string comparisons. However, you can't avoid the need to search for every string in C within T. If C is large enough to try Java code, clever use of an iterable class might be the solution you're looking for.

Hope this helps.

[TastyPastry]

I'm not sure if the looping is causing the program to slow down enough so that the callback isn't functioning properly.

I have the program set up so that the string checker is the KeyPressFcn of an edit box in a GUI. However, it appears that unless I use a breakpoint before the for loop and manually step through, the callback fails to execute properly and does not replace matched text. I've got no idea what's going wrong here.

[Kirby Fears]

Perhaps the "replacement" part is what's not working. Try setting your KeyPressFcn to replace text with a fixed string like str='mystring' any time a key is pressed to confirm that replacement is working.

Hope this helps.