A problem while splitting a text input with regexp

I have a text file with the input as

sammy yo yo
yoyo with you
Samyukta

and I tried the following code to put each word into an element of an array.

fid = fopen('test4.txt');
table = fscanf(fid,'%c');
table2 = regexp(table,'\n','split');

this means that when I refer to table2{1}, it returns 'sammy yo yo' then I split every line individually with strsplit and ' ' (whitespace) as the delimiter. Therefore, when I refer to table2{1}{2} , it returns 'ýo'. But, the last word of every line has more number of letters than appears i.e. size(table2{1}{2},2) = 3 rather than 2. But when I strcmp it with '\n' and ' ' or any other thing, it returns 0. So now I don't know what to do.

[Walter Roberson]

What shows up for

table2{1}{2}(end) + 0

I suspect you will find it is 13 (carriage return)

[Samyukta Ramnath]

it gives an error : cell contents reference from a non cell array object

Answer 1

[Cedric]

 >> fprintf('%d,', table) ;  fprintf('\n') ;
 115,97,109,109,121,32,121,111,32,121,111,13,10,121,111,121,111,32,119,105,
 116,104,32,121,111,117,13,10,83,97,109,121,117,107,116,97,13,10,

As you can see, at the end of each line, there are 13 (carriage return: '\r') and 10 (new line: '\n').

If you just want to split words, why don't you split using REGEXP only with a pattern which matches whitespaces? For example:

 >> buffer = fileread('test4.txt') ;
 >> words  = regexp(buffer, '\s+', 'split')
 words = 
    'sammy'    'yo'    'yo'    'yoyo'    'with'    'you'    'Samyukta'    ''

with this, you would just have to delete the last cell when empty (which happens when your file ends with '\r\n'), and you would be done.

[Walter Roberson]

Only if the file was created with an older MS Windows editor. More modern MS Windows editors only put in \n (newline) without \r (carriage return). Linux and OS-X have never used \r . (MacOS before OS-X might have used \r )

[Cedric]

Or MATLAB editor actually (I used it to generate this file on 2012b, Win7/64).

The pattern '\s+' works in all cases though.