Extracting the comments and the line of comments from a matlab script

Is there anyway to extract the comments and the line of comments from a matlab script and save it in a text file?

[Bob Thompson]

You can open the entire script as a text file, and then parse it with something like regexp.

[Walter Roberson]

@Jan has functions that parse scripts.

Finding comments is not easy, because of continuation markers and quoted strings and block comments.

[Jan]

@Walter Roberson: After I've inserted block comments in my parser, the modern double quotes for strings are a tedious new job.

a = "'%no comment'"      % comment
b = '"%no comment"'      % comment
c = '"'...  comment "    % comment
d = "asd"'  % comment '
e = '''%no comment'
f = e' %comment

The serious problem is not finding the start of the comment, but to recognize strings and char vectors.

[Steven Lord]

Can you share a little more information about why you're trying to find and extract all the comments from a MATLAB program file? As others have stated this isn't going to be easy, and they were only considering files with the .m extension. The question of what constitutes a comment in a Live Script with the .mlx extension is more complicated -- does anything that's in a text section count as a comment (since it's not going to be executed) or only comments inside a code section?

[Zahra Sheikhbahaee]

@Steven LordI have a vey long well documented code from other people and I need to make some changes to make it useful for my application. Being able to extract comments with the line of comments give me a summary of the code and where I should do my modifications because I would like to get a sense of algorithm like inner loop and outer loops to update some local and global parameters. That is my main reason.

Answer 1

[DGM]

Normally, I'd just use sed

sed -n -e 's/^[[:blank:]]*//' -e '/^%/p' myscriptfile.m > pileofcomments.txt

I guess you could also use it in a system() call. I don't know if there's anything similar in Windows.

If you're looking for a strictly Matlab way, have fun. I have no familiarity with doing this in Matlab, so the solution I came up with is hardly elegant. Even so, I doubt the most elegant way of doing this can come close to the sed example.

fid = fopen('myscriptfile.m','r');

% read the file into a cell array, one cell per line
i = 1;
tline = fgetl(fid);
A{i} = tline;
while ischar(tline)
    i = i+1;
    tline = fgetl(fid);
    A{i} = tline;
end
fclose(fid);
A = A.';

% get rid of empty lines
mk = cellfun(@ischar,A) & ~cellfun(@isempty,A);
A = A(mk);
% find matching lines
B = regexp(A,'^\s*%.*','match');
B = vertcat(B{:})

% optionally, get rid of leading whitespace
[~,C] = regexp(B,'^\s+','match','split')
for m = 1:numel(C)
	C{m} = C{m}{end}; 
end

C % these are all the final lines

% write the comments to a new file
fid = fopen('pileofcomments.txt','w+');
for m = 1:numel(C)
	% you may have to set the appropriate newline/cr characters
	fprintf(fid,'%s\n',C{m});
end
fclose(fid);

EDIT:

It's worth noting that both of these solutions target lines which begin with comments (ignoring leading whitespace). They will not find comments at the end of lines like so:

% it will find this comment
            % it will also find this comment
x = 2; % it will not find this comment

[Zahra Sheikhbahaee]

Thanks for the answer! Well I was hoping I can get those comments too. Is there anyway to extract those cooments as well?

[Walter Roberson]

Ask @Jan for parsing code https://www.mathworks.com/matlabcentral/answers/823990-get-all-used-variable-names-from-a-script#comment_1507490

[DGM]

I second Walter's recommendation (I was looking on the FEX for it).

Adding that one extra requirement would greatly complicate things.

[Jan]

See my comment above : The % masked in strings and CHAR vectors are a problem. Comments are started by "..." also. Block comments need to be considered also: %{ ... %}

@DGM: An easier method to read a text as a cell string:

S = fileread(FileName);
C = strsplit(S, '\n');

Removing leading and trailing blanks:

C = strtrim(C);