Find index of cells containing my string

Hi, I have a cell aray (40,000X1)in which every cell contains a string. I would like to find the indexes of the cells containing a specific string.

I used the following:

Index = strfind(Mycellarray, 'Bla');

I get the following error: ??? Error using ==> cell.strfind at 35 If any of the input arguments are cell arrays, the first must be a cell array of strings and the second must be a character array.

What is wrong, in the help file strfind accepts cell arrays and a pattern...?

Thank you

[Walter Roberson]

What version are you using?

>> Mycellarray = {'hello';'what';'is';'Bla';'about'};

>> strfind(Mycellarray,'Bla')

ans =

[]

[]

[]

[1]

[]

I will wildly speculate that you have a cell array in which each element is a cell array that contains a string.

[Jos (10584)]

Are you really sure you have a cell array of strings? Check it with ISCELLSTR

If this is not true, then you need to convert it first to be able to use STRFIND. For instance, if Walter is right and you have a cell array of a cell array of strings,

C = { {'a'}, {'b'}, {'c'}, {'a'}, {'a'} }

iscellstr(C) % false !

% ... but in this particular case it is easy to convert:

C2 = [C{:}]

and perhaps STRCMP better suits your needs as it will return a logical array directly

strcmp(C2,'a')

[New]

Hi

Thank you for your answers. I used C2 = [C{:}] as advised but then still trying to get the list of Indices I used:

Index = strfind(C2,'bla')

I get a cell array in which every cell is either empty [] or 1 but no list of indices.

K.

[Felipe]

I found out that if the cell has empty indexes it wont work. Example:

a =

    'hey'
    'oh'
       []

>> strfind(a,'hey')

??? Error using ==> cell.strfind at 35 If any of the input arguments are cell arrays, the first must be a cell array of strings and the second must be a character array.

>> a{3} = 'letsGo'

a =

    'hey'
    'oh'
    'letsGo'

>> strfind(a,'hey')

ans =

    [1]
    []
    []

It is not considered as a cell string if all the cells in the array have a string in them.

[Jan]

@Felipe: Exactly. A cell is a cell string, if it contains strings only. And STRFIND works on strings and cell strings only.

[Yusuf Arslan]

Hello, I have got a similar Problem.

- find(strcmp(rawdata,'ggiBoundaries1(1,1)'))

I want to find in rawdata the first entry of ggiBoundaries1. But it doesn't work. The error message is =

"0×1 empty double column vector".

When I enter "ggiBoundaries1(1,1)" the output is a char. Maybe that's the problem or not?

[Jan]

This is not an error message. It simply tells you, that the string is not found.

Are you really looking for the string 'ggiBoundaries1(1,1)' ? Or do you mean:

find(strcmp(rawdata, ggiBoundaries1(1,1)))

to search for occurrences of the first character of the variable ggiBoundaries1?

Please do not append a new question to an existing thread. Better open a new one. Thanks.

Answer 1

[Jan]

Do you want to search for 'bla' within the text in each element of the cell array, or for elements that are 'bla' exactly? If you explain this detail, then your question would be easier to answer.

If you are searching for text that has 'bla' as part of the text, then starting in R2016b you can use the “contains” function, as Alexander Cranney pointed out.

Index = find(contains(C,'bla'));

In previous versions of MATLAB, you can use the “strfind” function. However, “strfind” returns a cell array of indices. For any input cell whose text does not contain 'bla', “strfind” returns an empty cell. Use “isempty” and “cellfun” with the “find” function to find the empty cells.

IndexC = strfind(C,'bla');
Index = find(not(cellfun('isempty',IndexC)))

If you are searching for text that is exactly 'bla', then see Jos’ answer.

[Walter Roberson]

Possibly bla might occur multiple times and the indices of all of the positions is what is desired?

[Pooya89]

A little correction to Jan's answer:

   IndexC = strfind(C, 'bla'); % not C2
   ...

No need to convert C into string from cell : C2 = [C{:}];

[Alexander Cranney]

As of MATLAB 2016b, there is a new function "contains" that does exactly this!

    % code
Index = find(contains(C,'bla'));

[xiangxue wang]

Just to add my experience of running this. If the cell C contains some entry will NaN, it will gave the error like "First argument must be a string array, character vector, or cell array of character vectors" make sure your cell doesn't contain NaN

[Duc Minh Nguyen]

This answer will cause problem if there are other strings containing 'bla' in the array. For exact searching, use strcmp instead:

idx = find(strcmp(C, 'bla'))

[Jan]

@Duc Minh Nguyen: As said in the answer, it is the purpose of the code to find any occurrences of 'bla', not just the string 'bla'.

[ANA ROSHAN]

it was very useful. thanks!

[Nnamdi Njoku]

Just what i needed, thanks

[Aditi Bishnoi]

what to use if I want to find the exact string match 'bla'.

My cell array has both 'bla' and 'blah' elements, but i want to pick out only 'bla'.

[Walter Roberson]

exact_match_mask = strcmp(YourCell, 'bla')
exact_match_locations = find(exact_match_mask)

provided that your cell entries are all character vectors.

Answer 2

[Jay]

This worked for me:

idx = find(ismember(C, 'bla'))

[Narayanan Rengaswamy]

Great! Works clean. Thanks.

[Cameron Kimbrough]

Nice, just what I was looking for.

[Omer Moussaffi]

And this is faster, too. On my lenovo T460s, a search of a 3000-long cellstring takes about 0.5ms, compared with 2ms with

idx = find(contains(C,'bla'))

[juliette Grimaldi]

Great, thanks! Old stuff still makes it !

[Einat]

Yes! Worked just like I needed it to.

[ANKUR WADHWA]

Doesn't work on this one

C = { {'a'}, {'b'}, {'c'}, {'a'}, {'a'};{'b'}, {'a'}, {'c'}, {'a'}, {'c'} }
idx = find(strcmp([C{:}], 'a'))

Returns

idx = 1     4     7     8     9

Any suggestion on how to solve it.

[per isakson]

It works as documented.

First see Row- and column-major order

Matlab uses column-major order. (I fail to find a page to link to in the documentation.)

Try

%%
C = { {'a'}, {'b'}, {'c'}, {'a'}, {'a'}
      {'b'}, {'a'}, {'c'}, {'a'}, {'c'} };
%%  
cac = [C{:}]
idx = find(strcmp( cac, 'a' ))
%%
C1 = permute( C, [2,1] ); % switch rows and columns
idx = find(strcmp( [C1{:}], 'a' ))

it outputs

cac =
  1×10 cell array
  Columns 1 through 8
    {'a'}    {'b'}    {'b'}    {'a'}    {'c'}    {'c'}    {'a'}    {'a'}
  Columns 9 through 10
    {'a'}    {'c'}
idx =
     1     4     7     8     9
idx =
     1     4     5     7     9

Answer 3

[Matt B]

I realize this question is old now, but a simple way of doing this is to define an inline function:

cellfind = @(string)(@(cell_contents)(strcmp(string,cell_contents)));

You can then use this with cellfun to return a boolean value for each element of the cell. For example:

cell_array={1,eye(2),true,'foo',10};
string='foo'
logical_cells = cellfun(cellfind('foo'),cell_array)
logical_cells = 
                 [0,0,0,1,0]

[Sepp]

Great answer. Works very well.

[josh gore]

The inline function was a life saver!

[Jan]

@Matt B: strcmp accepts a cell array directly, so you can avoid the complicated cellfun approach with the expensive anonymous function:

cell_array = {1,eye(2),true,'foo',10}
strcmp(cell_array, 'foo')
>> [0,0,0,1,0]

Answer 4

[Jos (10584)]

So, your original array was a cell array of cells with a single string. Use STRCMP and FIND to get the indices of the cells with a cell containing the specified string

C = { {'a'}, {'b'}, {'c'}, {'a'}, {'a'} } % data
idx = find(strcmp([C{:}], 'a')) % single line engine

[AATIQ OUSSAMA]

Thanks Jos , It works fine !

Answer 5

[Omer Moussaffi]

Faster options: count startsWith endsWith

E,g, Index = count(Mycellarray, 'Bla');

[Walter Roberson]

Yes, this method should work well starting from R2016b.

Answer 6

[Jan]

You can check if your cell is a cell string:

iscellstr(Mycellarray);

This displays the indices and contents of the non-strings:

Index = find(~cellfun('isclass', Mycellarray, 'char'))
disp(Mycellarray(Index));

Another idea is, that some strings are multi-row CHAR matrices:

Index = find(cellfun('size', Mycellarray, 1) > 1)

Answer 7

[Peter Farkas]

You can also explicitelly define the index matrix:

[rw, ~] = size(cellArray);

ind = [1:1:rw];

idx = strcmp(cellArray, stringToSearchFor);

yourResult = ind(idx);

It is kind of verbose, if you review the code in 2 years time, you will still know what is going on.

Answer 8

[Mukesh Jadhav]

Haven't tested but this should work fine.

word_to_find=strfind(strarray,'stringtofind');
starray.index(word_to_find);