find index of cell array

Hello all, suppose i have cell array 6x1 which contains abstracts of articles, so each row of the cell contains 800 character. if i wan to find the position of str='b e h i n d' from that cell array,only b will be generated, so what is wrong my code. the cell arrays data is in the attachment or myabs.mat

load ('myabs.mat');

mydata=X;

[row,column]=size(mydata);

distance=zeros(row,column);

str='behind using a cover';

j=1;

for n=1:row

          gec=char(mydata{n});

          %for j=1:length(str)

          for m=1:800

          while j<length(str)

          %for j=1:length(str)

          if (gec(m)==str(j))

          indexx(j)=m;

          distance(n,m)=indexx(j);

          end

          break;

j=j+1;
         end

    end

end

[Stephen23]

Please edit your question and format the code correctly using the {} Code button that you will find above the textbox.

Also it would be easier if you simply uploaded the .MAT file that you are referring to, for us to try out ourselves.

[Mohamuud hassan]

hi stephen, i uploaded the .MAT file and i formatted the cod as {}code. simply what i want is. to find str's characters indexce sequentially one by one from the cell array which holds 6 rows of articles abstract and each row contains 800 characters. for example the first character of str is 'b', so we get it in the index 135 according to the first abstracts i mean first row of my cell array, the next character of str is 'e', and it will be 149....so those was estimated indexes, but may code was not doing correctly. please help me.

[Mohamuud hassan]

simply what i want is. to find str's characters indexces sequentially one by one from the cell array which holds 6 rows of articles abstract and each row contains 800 characters. for example the first character of str is 'b', so we get it in the index 135 according to the first abstracts i mean first row of my cell array, after that we will find the next character of str which is 'e', and it was estimated to be found in index 149 of first row but the result is zero....so what is wrong from my code. that is my question? please help me.

[Stephen23]

This question is continued here (with bug fix!):

http://www.mathworks.com/matlabcentral/answers/218539-when-the-secret-text-contain-pair-doubke-character-like-dd-mm-ll-the-y-vector-reswtores-as-n

Answer 1

[Stephen23]

There might be simpler ways to achieve this task, but this seems to work:

str = 'behind';
load myabs
for m = 1:numel(X)
    Y = bsxfun(@eq,X{m}(:),str);
    for n = 2:size(Y,2)
        Y(:,n) = Y(:,n) & cumsum(Y(:,n-1))>0;
    end
    Z = find(any(diff([false(size(str));cumsum(Y)>0],1),2))
end

and when run it prints this in the command window:

Z =
   135
   149
   235
   236
   239
   266
Z =
    72
    73
   127
   136
   160
   174
Z =
   163
   170
   172
   173
   177
   221
Z =
    40
    53
    65
    66
    76
   112
Z =
   216
   217
   267
   279
   295
   319
Z =
   170
   175
   178
   196
   197
   234

[Mohamuud hassan]

thanks Stephen Cobeldick that was my point, but this statement

Z = find(any(diff([false(size(str));cumsum(Y)>0],1),2)) will hold on or generate one row, so,how i can generate all rows in one loop with one output.
thank you for your support

[Stephen23]

To store the value of Z just preallocate the output array before the loop and us indexing to allocate the values on each loop iteration:

str = 'behind';
load myabs
Z = nan(size(X,1),numel(str));
for m = 1:numel(X)
    Y = bsxfun(@eq,X{m}(:),str);
    for n = 2:size(Y,2)
        Y(:,n) = Y(:,n) & cumsum(Y(:,n-1))>0;
    end
    tmp = find(any(diff([false(size(str));cumsum(Y)>0],1),2));
    Z(m,1:numel(tmp)) = tmp;
end

and then each row of Z corresponds to one cell of X:

>> Z
Z =
 135   149   235   236   239   266
  72    73   127   136   160   174
 163   170   172   173   177   221
  40    53    65    66    76   112
 216   217   267   279   295   319
 170   175   178   196   197   234

Or you might like to consider using my third solution using regexp, which detects all instances of the pattern string, not just the first one.

[Mohamuud hassan]

Thank you Stephen Cobeldick, this step, is best way that i can achieve my goal. again and again thank you for your continued support

Answer 2

[Stephen23]

This task could be performed much faster and more robustly by using inbuilt functions. Why not just use strfind instead of these slow and buggy nested loops:

>> A = {'some line of words.','more text to search','some words with other words too','different information here'};
>> B = strfind(A,'words');
>> B{:}
ans =
  14
ans =
   []
ans =
   6    23
ans =
   []

which returns a cell array of indices giving the locations in each string where the search text 'words' was found. It correctly found the one instance in the first string, and two instances in the third string.

You should avoid calling variables i or j, as these are both names of the inbuilt imaginary unit.

[Mohamuud hassan]

thank you Stephen Cobeldick. But, what i want to find is not string, is character. as i highlighted before, i have one vector which is called str='behind the cover'. so what i wan is to fin fisrt charter of str's vector form the cell array and if we get that character to store character's position into indexx vector and continue searching character one by one until last character which is r.

[Stephen23]

It is not clear what the difference is: you want to match every character sequentially, whereas strfind matches the whole string at once. What is the difference in your eyes? Do these characters have to be sequential?

Can you please give an example of one string to be searched and the pattern to be found, and show us how the pattern should be matched to the string.

[Mohamuud hassan]

simply what i want is. to find str's characters indexces sequentially one by one from the cell array which holds 6 rows of articles abstract and each row contains 800 characters. for example the first character of str is 'b', so we get it in the index 135 according to the first abstracts i mean first row of my cell array, after that we will find the next character of str which is 'e', and it was estimated to be found in index 149 of first row but the result is zero....so what is wrong from my code. that is my question? please help me.