How to compare letters if they are same

Hello

I have a problem with a code. I would like to compare two files with letters/names. I would like to find which name of file 2 corresponds to the name of file 1 , depending on the name similarity.

I mean I would like to match names from one file name to another file name.

Which command should I use?

Thank you in advance

[dpb]

Give us some examples...exact matches use strfind and friends or the new(ish) string functions contains et al.

Pattern matching would use regular expressions which is a whole new world...

[Rik]

Should those be detected as the same? Maybe the unique function will even do the trick (if you put the file names in a cellstr)

[Ivan Mich]

Excuse me but I do not understand. what do you mean wtih the unique function will even do the trick (if you put the file names in a cellstr) ?

[Ivan Mich]

Which command should I use?

[Walter Roberson]

MATLAB would say that upper('Αθήνα') is ΑΘΉΝΑ not ΑΘΗΝΑ

[Ivan Mich]

So is there a way to recognise the same as lower as upper letters ? (Despite of lower and upper letters, I want to match same words between these two files) ?

[Stephen Cobeldick]

Disregarding case, do the names have to match exactly, or just "depending on the name similarity" as you wrote?

If you are need matching based on similarity (as you wrote), then you could use some edit distance metric (e.g. the Levenshtein distance).

How should diacritics be treated: as different characters, or ignored?

[Ivan Mich]

I would like diacritics to be ignored if is it possible

[Walter Roberson]

Which letters need to be supported? Do you need to support removing diacritics from letters other than U+0386 to U+0390 ? https://www.compart.com/en/unicode/block/U+0370 ? There are a lot of them... https://en.wikipedia.org/wiki/Greek_script_in_Unicode

[Stephen Cobeldick]

Orignal question and comments:

http://web.archive.org/web/20210121123045/https://webcache.googleusercontent.com/search?q=cache%3ANPddy0hW4ewJ%3Ahttps%3A%2F%2Fwww.mathworks.com%2Fmatlabcentral%2Fanswers%2F721344-how-to-compare-greek-letters-if-they-are-same

http://web.archive.org/web/20210121122700/https://www.mathworks.com/matlabcentral/answers/721344-how-to-compare-letters-if-they-are-same

[Rik]

@Ivan Mich You know this forum. After 89 questions you should know we don't like people deleting parts of the question. Please don't do that. You are only giving people work to recover what you deleted, making your deletion pointless as well.

Answer 1

[Walter Roberson]

Typically the easiest way to handle situations like this, that are not plain upper / lower case (use upper() or lower() or strcmpi() for those), is to create a mapping table,

%I only partly filled this out; see
%https://en.wikipedia.org/wiki/Greek_script_in_Unicode

map = char(0x001:0x03ff);

%https://www.unicode.org/charts/PDF/U0370.pdf
map(0x0391:0x03a9) = 0x03b1:0x03c9;   %alpha to omega upper to lower ΑΩ αω
map(0x0386) = 0x03b1; % Ά
map(0x0388) = 0x03b5; % Έ
map(0x0389) = 0x03b7; % Ή
map(0x038A) = 0x03b9; % Ί
map(0x038C) = 0x03b9; % Ό
map(0x038E) = 0x03c5; % Ύ
map(0x038F) = 0x03c9; % Ώ
map(0x03AC) = 0x03b1; % ἀ
map(0x03AD) = 0x03b5; % ἐ
map(0x03AE) = 0x03b7; % ἠ
map(0x03AF) = 0x03b9; % ἰ

%https://www.unicode.org/charts/PDF/U0080.pdf
map(0x00b5) = 0x03bc;  %mu

%https://en.wikipedia.org/wiki/Greek_Extended
map(0x1f00:0x1f0f) = 0x03b1;   %alpha extended
map(0x1f10:0x1f1f) = 0x03b5;   %epsilon extended
map(0x1f20:0x1f2f) = 0x03b7;   %eta extended
%and more

After which you take

map('Αθήνα')

ans = 'αθηνα'

map('ΑΘΗΝΑ')

ans = 'αθηνα'

[Ivan Mich]

Thank you , but command window shows me :

Error: File: greek_letters.m Line: 26 Column: 13

Invalid expression. Check for missing multiplication operator, missing or unbalanced

delimiters, or other syntax error. To construct matrices, use brackets instead of

parentheses.

[Walter Roberson]

Which MATLAB release are you using?

In your file, is line 26 the one that is

map(0x1f20:0x1f2f) = 0x03b7;   %eta extended

?

The code was tested in R2020b.

[Ivan Mich]

I am using MATLAB 2019a.

line 26 is map = char(0x001:0x03ff);

[Walter Roberson]

Your version did not have hex input yet.

In each place that I coded 0x followed by digits, convert that to a call to hex2dec() with the digits in quotes. You might need to remove the 0x part. For example

map = char(hex2dec('0001'):hex2dec('03ff'));

You can skip the leading 0, such as hex2dec('3ff') but using the leading 0 helps to emphasize that you are using Unicode code points, which by convention are given in 4 digit hex until 0x10000

[Walter Roberson]

%I only partly filled this out; see
%https://en.wikipedia.org/wiki/Greek_script_in_Unicode

H = @hex2dec;

map = char(H('0001'):H('03ff'));

%https://www.unicode.org/charts/PDF/U0370.pdf
map(H('0391'):H('03a9')) = H('03b1:H('03c9;   %alpha to omega upper to lower ΑΩ αω
map(H('0386')) = H('03b1'); % Ά
map(H('0388')) = H('03b5'); % Έ
map(H('0389')) = H('03b7'); % Ή
map(H('038A')) = H('03b9'); % Ί
map(H('038C')) = H('03b9'); % Ό
map(H('038E')) = H('03c5'); % Ύ
map(H('038F')) = H('03c9'); % Ώ
map(H('03AC')) = H('03b1'); % ἀ
map(H('03AD')) = H('03b5'); % ἐ
map(H('03AE')) = H('03b7'); % ἠ
map(H('03AF')) = H('03b9'); % ἰ

%https://www.unicode.org/charts/PDF/U0080.pdf
map(H('00b5')) = H('03bc');  %mu

%https://en.wikipedia.org/wiki/Greek_Extended
map(H('1f00'):H('1f0f')) = H('03b1');   %alpha extended
map(H('1f10'):H('1f1f')) = H('03b5');   %epsilon extended
map(H('1f20'):H('1f2f')) = H('03b7');   %eta extended
%and more

Answer 2

[Stephen Cobeldick]

Rather than building maps by hand, I would get Python to do the heavy lifting, e.g.:

baz = @(v)char(v(1)); % only need the first decomposed character.
fun = @(c)baz(py.unicodedata.normalize('NFKD',c)); % to remove diacritics.
in1 = 'Αθήνα';
in2 = 'ΑΘΗΝΑ';
st1 = arrayfun(fun,in1) % remove diacritics

st1 = 'Αθηνα'

st2 = arrayfun(fun,in2) % remove diacritics

st2 = 'ΑΘΗΝΑ'

strcmpi(st1,st2) % case-insensitive comparison

ans = logical
   1

[Stephen Cobeldick]

"My version is 2019a . Do I need to to something in order to run?"

You will need Python installed.

You might need to use pyenv tell MATLAB about your Python installation. For an overview:

https://blogs.mathworks.com/racing-lounge/2020/09/14/using-matlab-and-python-together/

https://www.mathworks.com/help/matlab/call-python-libraries.html

Note that Python module names and command names can change, you will need to check the actual names/syntax used for the version that you have installed (my answer uses Python 2.7, which is what this website supports).

EDIT: I note that pyenv was introduced in R2019b. I do not know when this style of Python access was introduced.