处理从Java到PHP到MySQL的字符编码

Question

In Java I pass a String to PHP.

In PHP I take that String and do a search for it with a MySQL query.

Here is php code:

    $query = $database->escape_value(trim($_POST['query'])); 
    $result = mysqli_query($dbconnection, Data::getSearchQuery($query));
    while ($row = mysqli_fetch_assoc($result)) {
        $output[] = $row;
    }
    print(json_encode($output));

    mysqli_close($dbconnection);


  public static function getSearchQuery($item_query) {

        $query = "
            SELECT i.item, i.item_id, c.category, c.cat_id
            FROM items as i
            LEFT JOIN master_cat AS c
                    ON (c.cat_id = i.cat_id)
            WHERE i.item LIKE '%{$item_query}%' 

            ORDER BY i.item ASC;";

        return $query;
    }

This always works if I use regular characters on my US keyboard. But the moment I start using irregular characters, the search turns empty.

I can verify that MySQL stores the data AS THE USER ENTERS IT. So if they typed Beyoncè , that is how database stores it.

But when I search for Beyoncè (or whatever) in the above code, it returns empty.

How should I handle the char. encoding here?

Answer 1

Three points to think of:

1) The $item_query variable could come in wrong encoding.

2) >> I can verify that MySQL stores the data AS THE USER ENTERS IT

This can get tricky. If one writes an iso8859-1 encoded string to an utf-8 database, the string is obviously stored incorrectly. If that string is read with a client (ie phpmyadmin or mysql command line tool) configured to iso8859-1, the string is correctly returned - although its representation in the database is clearly wrong.

3) The MySql settings: Have your set utf-8 for the connection itself? What about charsets and collations for the database/the table?

https://dev.mysql.com/doc/refman/5.5/en/charset-syntax.html

UPDATE: I assume you want everything to be UTF-8. Kind of quick hack to test:

• Beyoncé has 7 characters (see MySQL CHAR_LENGTH function)
• in UTF-8, it occupies 8 bytes (see MySQL LENGTH function). The eight bytes are, represented in a one-byte-per-character encoding like windows-1252, something like Beyoncé .

This leads to the following diagnostic tests ...

1. The PHP-issued SQL command

    "SELECT CHAR_LENGTH($item_query), LENGTH($item_query);" 

   should then return a result of (7, 8) to show us that the $item_query variable is probably correctly encoded and the database likes UTF-8 . (7, 7) would mean $item_query wasn't UTF-8, and (8, 8) would mean the database doesn't want to deal with UTF-8 yet. If the latter is the case, then perhaps issue a SET NAMES 'UTF8'; before the query.

2. Similarly, the PHP-issued SQL command

    SELECT CHAR_LENGTH('Beyoncé'), LENGTH('Beyoncé'); 

   should return the result (7, 8) to show us that your PHP editor is configured to edit UTF-8 php files .

3. Repeat the previous step with phpmyadmin (or any SQL client) to be sure that this client uses UTF-8 , too.

4. No table was involved yet! The SQL command

    SELECT CHAR_LENGTH(somecolumn), LENGTH(somecolumn) FROM sometable; 

   (with sometable having UTF-8 character encoding and somecolumn containing some diacritical characters) should tell you if UTF-8 was used when storing values to the table .

5. If all previous tests passed, test again with LIKE. Even 'Beyoncé' LIKE 'Beyonce' should work then. For more information, google MySQL collation.