is it possible to compare strings completely written in Greek characters ? for example:
str1='ΑΒΓΔ'
str2='ΕΖΗΘ'
strcmp(str1,str2)
is it possible to apply the above when i read strings in Greek characters from a file? for example:
line='ΑΒΓΔ;ΕΖΗΘ'
[str1 str2] = strread(line,'%s %s','delimiter',';')
strcmp(str1,str2)
MATLAB and unicode still don't yet mix well, AFAIK. Your first example sadly returns a false positive. your second example works, but not without its caveats.
Take for example test.txt (using UTF-8 encoding):
ΑΒΓΔ;ΕΖΗΘ
Then:
%// First try:
fid = fopen('test.txt', 'r', 'n', 'UTF-8');
A = textscan(fid, '%s', 'delimiter', ';');
fclose(fid);
A{1}{1}+0
%// apparently, textscan does not read the correct number of bits per
%// character...
%// Let's try manually:
fid = fopen('test.txt', 'r', 'n', 'UTF-8');
txt = fgetl(fid);
fclose(fid);
%// That makes things even worse! All characters get converted to the
%// placeholder character 26, causing false positives in string
%// comparisons:
D = textscan(txt, '%s', 'delimiter', ';');
D{1}{1}+0
strcmp(D{1}{2}, D{1}{2})
%// Regexp works better; it preserves the right character codes and yields
%// a (correct) negative on comparison:
C = regexp(txt, ';', 'split');
C{1}+0
strcmp(C{1}, C{2})
%// So, the "best" way:
fid = fopen('test.txt', 'r', 'n', 'UTF-8');
D = {};
while ~feof(fid)
line = fgetl(fid);
D{end+1} = regexp(line, ';', 'split'); %#ok<SAGROW>
end
fclose(fid);
The disp still does not display them correctly, unless you have specifically selected a font for the command window/editor window which supports unicode.
If you're on Linux, unicode display works OK if you call MATLAB from bash or similar. But that has more to do with your shell than with MATLAB...
Also, looking at help unicode2native:
%// Reading
fid = fopen('japanese.txt', 'r', 'n', 'Shift_JIS');
str = fread(fid, '*char')';
fclose(fid);
disp(str);
%// Writing
fid = fopen('japanese_out.txt', 'w', 'n', 'Shift_JIS');
fwrite(fid, str, 'char');
fclose(fid);
The disp fails over here (R2010a), but the write is OK...
String encoding is not one of MATLAB's strong points. For working with Unicode, there are unicode2native and native2unicode. However, string literals that you enter in your code are ASCII-only, as far as I can tell. This means that you need to read your non-ASCII strings from a file that uses a suitable encoding using fread and then use native2unicode to convert the raw bytes to Unicode.
