Characters corrupted when inserted in database - php

Question

I have a PHP script that gets a XML, selects some data, and insert them in my database (MySQL). All my fields are utf8_bin. This XML is ISO-8859-1, and can't change this because is another site that sends it to me.
Example: the string "NG Contábil" is set "NG Contábil" in my db. This is my script:
<?php
header('Content-Type: text/html; charset=utf-8');
mysql_query("SET NAMES 'utf8'");
mysql_query('SET character_set_connection=utf8');
mysql_query('SET character_set_client=utf8');
mysql_query('SET character_set_results=utf8');
include '/PagSeguroLibrary/PagSeguroLibrary.php';
include '/PagSeguroLibrary/domain/PagSeguroAccountCredentials.class.php';
include 'conexao.php';
include 'alias_array.php';
include 'retorna_cpf.php';
$conexao = ConectaBD::get_instance();
$conexao->conectar_pronto();
$conexao->BD_pronto();
//(...)
$xml = simplexml_load_file('arquivo.xml');
if($xml === null)
$xml = simplexml_load_string($resposta_transacao);
$status = $xml->status;
$nome = $xml->sender->name;
$email = $xml->sender->email;
$codigo = $xml->code;
$vetor = $xml->items->item;
$cpf = retorna_cpf($email);
foreach($vetor as $v)
{
$nome_produto = $v->description;
if($nome_produto != 'frete')
{
//Retorna o id do produto a partir da descrição
$result = mysql_query('SELECT id_product
FROM ps_product_lang
WHERE name = "'.$nome_produto.'"');
$array = Array();
while($row = mysql_fetch_alias_array($result))
{
foreach($row as $linha)
array_push($array, $linha);
}
mysql_query('INSERT INTO pagamento(Status, Nome, Email, CPF, idproduto, Codigo, Inscrito, id, Enviado, NomeProduto)
VALUES ("'.$status.'", "'.$nome.'", "'.$email.'", "'.$cpf.'", "'.$array[0].'", "'.$codigo.'", 0, null, 0, "'.$nome_produto.'")');
}
}
fclose($arquivo);
unlink('arquivo.xml');
?>
Thanks for any answer!

Answer 1

You have overlooked just one little nuance:
mysql_query("SET NAMES 'utf8'"); is not a magical spell that have to be cast in order to make proper encoding, but actually an SQL query, which needs to be run in the same instance you are actually using to run SQL queries.
So, if you are connecting to your mysql database using ConectaBD::get_instance(); you have to run SET NAMES utf8 query after that call, not before.
I don't know why, but adding utf8_decode() solves the problem
I know.
simplexml's output is always utf-8.
While, as I pointed out above, you don't set your client connection into utf8.
So, it remains default latin1
With (quite useless) utf8_decode() call you're casting your utf-8 data back into latin1 and thus it correctly stored into database.

Answer 2

This should solve your issue:
...
$xml = simplexml_load_file('arquivo.xml');
$xml = iconv('ISO-8859-1', 'UTF-8', $xml);
...