Problem sending images to a folder with accented names

Question

I am sending an image to a folder and saving only the location (path) of it in the database. If for example the name I assign the image has no accents works perfectly, only when it has accents it doesn’t work. When storing in the database and showing on the website shows everything correct.

When storing in the folder I cannot fetch the respective image because the path (path) is for example "Products/joe" and so the name appears inside the folder z&A

Sending code to the database and folder:

    if(isset($_POST['upload']) && isset($_FILES['file-image'])) {
    $filetmp = mysqli_real_escape_string($_FILES["file-image"]["tmp_name"]);
    $filename = mysqli_real_escape_string($dbc, $_FILES["file-image"]["name"]);
    $filetype = mysqli_real_escape_string($dbc, $_FILES["file-image"]["type"]);
    $ProdName = mysqli_real_escape_string($dbc, $_POST['nameProduct']);
    $ProdPrice = mysqli_real_escape_string($dbc, $_POST['productPrice']);
    $ProdDescr = mysqli_real_escape_string($dbc, $_POST['descrProduct']);

    $filepath = "Products/" . $ProdName ;
    $info = getimagesize($filetmp);
    if ($info == FALSE or (empty($ProdName) or empty($ProdPrice) or empty($ProdDescr))) {
        alertError();
    }elseif ($info == FALSE and (empty($ProdName) or empty($ProdPrice) or empty($ProdDescr))) {
        alertError();
    }elseif ($info == TRUE and (empty($ProdName) or empty($ProdPrice) or empty($ProdDescr))) {
        alertError();
    }elseif ($info == TRUE and (!(empty($ProdName) or empty($ProdPrice) or empty($ProdDescr)))) {
        move_uploaded_file($filetmp, $filepath);
        $result = mysqli_query($dbc, "INSERT INTO images (img_name,img_path,img_type) Values('$ProdName','$filepath','$filetype')") or die(errorAdmin());
        $last_id = $dbc->insert_id;
        $insertProduct = mysqli_query($dbc, "INSERT INTO products (name_Product,prod_description,prod_price,img_id)
                                        Values('$ProdName','$ProdDescr','$ProdPrice','$last_id')") or die(errorAdmin());
        mysqli_query($dbc, "SET NAMES 'utf8'") or die(mysqli_error($dbc));
        if ($insertProduct == TRUE and $result == TRUE) {
            ?>
            <script type="text/javascript">
                swal({
                    title: 'Good Job',
                    text: 'Produto Criado Com Sucesso',
                    type: 'success',
                    confirmButtonText: 'Feito'
                });
            </script>
            <?php
        }
    }
    mysqli_close($dbc);
}

Answer 1

Using accent on filenames is a bit tricky and will hardly be a portable solution.

Operating systems handle character encoding in different ways.

Linux

---

Linux uses binary for filenames. This means that it does not require a specific encoding to save the filename. For example if we create two files with different encodings in their names, but which are equivalent (we will talk more about this):

$ echo "foo" > $'z\xc3\xa9.txt'
$ echo "bar" > $'ze\xcc\x81.txt'

As we execute the command ls, we will have the following output:

$ ls 
zé.txt zé.txt

This means that the program that uses the filesystem needs to worry thoroughly with the encoding and normalization of characters, because if there is more than one file with the same name, but with different normalizations, which file should the program consider correct? Apart from security issues.

OSX

---

OSX already uses UTF-16 to encode file names. This is done transparently for the user. That is, if your program saves the file name using UTF-8, OSX will translate it to UTF-16.

Another point to consider when using OSX is how it normalizes characters. The Unicode specification defines something called equivalence and standardisation. This concept defines how the combination of characters is done in the different types of normalizations (NFD, NFC, NFKD, NFKC) for the same character. To illustrate this concept, just save a file with the correct sequence in UTF-8 of the character is (\xC3\xA9):

<?php
file_put_contents("z\xC3\xA9.txt","xyz");
?>

And we’ll check what was actually recorded:

<?php
list($file) = glob('z*');
echo urlencode($file) . "\n";
?>

We will have the following output:

ze%CC%81.txt

The OSX recorded e0xCC81 instead of 0xCCA9. This shows us that the OSX uses a normalization that uses three bytes, the e and two more bytes combined result in acute accent. These three bytes together form the letter é.

Windows

---

Windows also uses UTF-16 (using NTFS), with the OSX-like behavior of transparent character translation. Regarding normalization, it behaves different from OSX, if we run the same code we run on OSX we will see the correct characters in the output.

Completion

---

We already have here two equivalent ways of writing the same letter in Unicode. We have two more, which are beyond the scope here. The fact is that if you really want to work with accents on behalf of files, you will need some care and probably use an intermediary to save those files.

The extent lntl PHP offers a class called Normalize which allows you to control the normalization of characters. It is an interesting option if you want to intermediate these encodings.

My suggestion is that you translate accented characters to pure ASCII whenever you can.