How to use a two character string as a delimiter in "textscan" function?

I want to use " 'h " as a single delimiter to extract the letters after " 'h " with "textscan" function. I can use " h " as a delimiter, but I cannot use " 'h " as a single delimiter.

Answer 1

[MathWorks Support Team]

For large files with specific delimiter:

You can achieve this using the "MultipleDelimsAsOne" property in 'textscan'.

For example, 

C = textscan(fileID,'%s %s','Delimiter',{'''h'},'MultipleDelimsAsOne',1,'TreatAsEmpty',{'NA','na'},'CommentStyle','//');

For more details, you can access the release-specific documentation by executing the following command in the MATLAB R2020b command window:

>> web(fullfile(docroot, 'matlab/ref/textscan.html?searchHighlight=textscan'))

For smaller text files:

You can use "regexp" instead, as there might be an issue with the order of delimiters.

file = fileread(filename);
pieces = regexp(file, "'h", 'split');
 
if isempty(pieces{end})
   pieces(end) = []; 
end  %special condition for EOF

[Walter Roberson]

Note that if you do this, then it will not pay attention to order, and it will treat singles as delimiters. For example

 'hello henry, it is 3 o'clock
 DD     D               D

each location marked as D would be treated as a delimiter

You cannot reliably create two-character-pair delimiters with textscan. You can reduce the problem by using something like 'delimiter', '''', with format "h%[^']" but that would get you to

 'hello henry, it is 3 o'clock
 Dh.....................DE

That is, Delimiter, then the literal h (that would not appear in the output stream, then all of the character positions marked with period would be included in the output. Then the ' delimiter would be found, ending the %[^'] . And at that point you would have a problem as the "c" does not match the expected literal h

Because of this, unless you have really big files, use a different approach:

 S = fileread('NameOfFile');
 pieces = regexp(S, "'h", 'split');
 if isempty(pieces{end}); pieces(end) = []; end  %special condition for EOF