encoding - mysql substitute html special chars with UTF equivalents

Question

I have a database where some of the elements consist of HTML special characters:

| Universidad Tecnológica Nacional - UTN                                                  |
| Instituto Tecnológico de Buenos Aires                                                   |
| Instituto Superior del Profesorado "Dr. Joaquín V. González" |
| Escuela Nacional de Náutica "Manuel Belgrano"                         |
| Conservatorio Nacional de Música "Carlos López Buchardo"     |
| Instituto Argentino de Computacion - IAC                                                         |
| Conservatorio de Superior de Música "Manuel de Falla"                 |

I need to convert it to a proper UTF format. Can I do better than just iterating through the database, and having a mapping from each code to the equivalent symbol?

á -> 'á'
" -> '"'
...

Answer 1

As mentioned in my comment above, it's terribly unclear what you're trying to do in your own case.

Can I do better than just iterating through the database, and having a mapping from each code to the equivalent symbol?

Well, yes. You can replace character code entities (e.g. { and ƫ) with their replacement characters without having to lookup the character code in a "mapping". But named entities (e.g. ") will always need to be looked up.

Here's my attempt to solve the general case:

1. Create a table to store named character entities defined in HTML:

   CREATE TABLE ents (
     ref VARCHAR(8) NOT NULL COLLATE utf8_bin,
     rep CHAR(1)    NOT NULL,
     PRIMARY KEY (ref)
   );
   

2. Populate this table - I suggest using a script, for example from PHP:

   $dbh = new PDO("mysql:dbname=$dbname", $username, $password);
   $dbh->setAttribute(PDO::ATTR_EMULATE_PREPARES, FALSE);
   $ins = $dbh->prepare('INSERT INTO ents (ref, rep) VALUES (?, ?)');
   $t = get_html_translation_table(HTML_ENTITIES);
   foreach ($t as $k => $v) $ins->execute([substr($v, 1, -1), $k]);
   

3. Define an SQL function to perform entity replacements (using this table where applicable, or else by character code):

   DELIMITER ;;
   
   CREATE FUNCTION dhe(s TEXT) RETURNS TEXT
   BEGIN
     DECLARE n, p, i, t INT DEFAULT 0;
     DECLARE r VARCHAR(12);
     entity_search: LOOP
       SET n := LOCATE('&', s, n+1);
       IF (!n) THEN
         LEAVE entity_search;
       END IF;
   
       IF (SUBSTRING(s, n+1, 1) = '#') THEN
         CASE
           WHEN SUBSTRING(s, n+2, 1) RLIKE '[[:digit:]]' THEN
             SET t := 2, p := n+2, r := '[[:digit:]]';
           WHEN SUBSTRING(s, n+2, 1) = 'x' THEN
             SET t := 3, p := n+3, r := '[[:xdigit:]]';
           ELSE ITERATE entity_search;
         END CASE;
       ELSE
         SET t := 1, p := n+1, r := '[[:alnum:]_]';
       END IF;
   
       SET i := 0;
       reference: LOOP
         IF SUBSTRING(s, p+i, 1) NOT RLIKE r THEN
           IF SUBSTRING(s, p+i, 1) RLIKE '[[:alnum:]_]' THEN
             ITERATE entity_search;
           END IF;
           LEAVE reference;
         END IF;
         IF i = 8 THEN ITERATE entity_search; END IF;
         SET i := i + 1;
       END LOOP reference;
   
       SET s := CONCAT(
         LEFT(s, n-1),
         CASE t
           WHEN 1 THEN COALESCE(
             (SELECT rep FROM ents WHERE ref = SUBSTRING(s, p, i))
           , SUBSTRING(s, n, i + IF(SUBSTRING(s, p+i, 1)=';',1,0))
           )
           WHEN 2 THEN CHAR(SUBSTRING(s, p, i))
           WHEN 3 THEN CHAR(CONV(SUBSTRING(s, p, i), 16, 10))
         END,
         SUBSTRING(s, p + i + IF(SUBSTRING(s, p+i, 1)=';',1,0))
       );
     END LOOP entity_search;
     RETURN s;
   END;;
   
   DELIMITER ;
   

4. Apply this function twice to decode your (apparently) doubly-encoded table:

   UPDATE my_table SET my_column = dhe(dhe(my_column));