1767 lines
49 KiB
PHP
1767 lines
49 KiB
PHP
<?php
|
|
/**
|
|
* @brief Conversion module
|
|
* @author <a href='http://www.invisionpower.com'>Invision Power Services, Inc.</a>
|
|
* @copyright (c) 2001 - SVN_YYYY Invision Power Services, Inc.
|
|
* @license http://www.invisionpower.com/legal/standards/
|
|
* @package IPS Tools
|
|
* @since 4 Sept 2013
|
|
* @version SVN_VERSION_NUMBER
|
|
*/
|
|
|
|
namespace IPSUtf8;
|
|
|
|
/**
|
|
* Conversion class
|
|
*/
|
|
class Convert
|
|
{
|
|
const VERSION_ID = '1.1.17';
|
|
|
|
/**
|
|
* @brief Data Store
|
|
*/
|
|
protected $_data = array();
|
|
|
|
/**
|
|
* @brief Instance
|
|
*/
|
|
protected static $instance = NULL;
|
|
|
|
/**
|
|
* @brief Convertable table fields
|
|
*/
|
|
protected static $convertCols = array( 'text', 'mediumtext', 'longtext', 'varchar', 'char', 'tinytext' );
|
|
|
|
/**
|
|
* @brief Numeric table fields
|
|
*/
|
|
protected static $numericCols = array( 'integer' => 11, 'int' => 11, 'smallint' => 6, 'tinyint' => 4, 'mediumint' => 8, 'bigint' => 20, 'decimal' => null, 'numeric' => null, 'float' => null, 'double' => null );
|
|
|
|
/**
|
|
* @brief Tables we need to convert
|
|
*/
|
|
protected static $nonUtf8Tables = NULL;
|
|
|
|
/**
|
|
* @brief Tables we need to convert
|
|
*/
|
|
protected static $nonUtf8Collations = NULL;
|
|
|
|
/**
|
|
* @brief Number of rows we need to convert
|
|
*/
|
|
protected static $rowsToConvert = NULL;
|
|
|
|
/**
|
|
* @brief Native Database object
|
|
*/
|
|
protected static $db = NULL;
|
|
|
|
/**
|
|
* @brief UTF-8 Database object
|
|
*/
|
|
protected static $utf = NULL;
|
|
|
|
/**
|
|
* @brief UTF-8 table engine
|
|
*/
|
|
protected static $utfTableEngine = NULL;
|
|
|
|
/**
|
|
* @brief Create don't populate
|
|
*/
|
|
protected static $createOnly = array( 'content_cache_posts', 'content_cache_sigs', 'sessions', 'topic_views', 'search_keywords' );
|
|
|
|
/**
|
|
* @brief Problem Tables that can cause an error due to serialization while using "fast" mode
|
|
*/
|
|
protected static $problemTables = array( 'cache_store' );
|
|
|
|
/**
|
|
* @brief MySQL supported character sets
|
|
*/
|
|
protected static $mysqlCharSets = NULL;
|
|
|
|
/**
|
|
* Get instance
|
|
*
|
|
* @return Output
|
|
*/
|
|
public static function i()
|
|
{
|
|
if ( self::$instance === NULL )
|
|
{
|
|
static::$db = \IPSUtf8\Db::i();
|
|
if ( defined( 'SOURCE_DB_CHARSET' ) AND SOURCE_DB_CHARSET !== NULL )
|
|
{
|
|
static::$db->set_charset( SOURCE_DB_CHARSET );
|
|
}
|
|
static::$utf = \IPSUtf8\Db::i('utf8');
|
|
self::$instance = new self();
|
|
self::$instance->init();
|
|
}
|
|
|
|
return self::$instance;
|
|
}
|
|
|
|
/**
|
|
* Get value from data store
|
|
*
|
|
* @param mixed $key Key
|
|
* @return mixed Value from the datastore
|
|
*/
|
|
public function __get( $key )
|
|
{
|
|
if( isset( $this->_data[ $key ] ) )
|
|
{
|
|
return $this->_data[ $key ];
|
|
}
|
|
|
|
return NULL;
|
|
}
|
|
|
|
/**
|
|
* Set value in data store
|
|
*
|
|
* @param mixed $key Key
|
|
* @param mixed $value Value
|
|
* @return void
|
|
*/
|
|
public function __set( $key, $value )
|
|
{
|
|
if( array_key_exists( $key, $this->_data ) )
|
|
{
|
|
$this->_data[ $key ] = $value;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Process a batch
|
|
*
|
|
* @param int|null Number of rows to limit per batch
|
|
* @return boolean True (I converted some) False (I didn't)
|
|
*/
|
|
public function process( $limit=null, $fromError=false )
|
|
{
|
|
/* MB4? */
|
|
$sessionData = \IPSUtf8\Session::i()->json;
|
|
|
|
if ( isset( $sessionData['use_utf8mb4'] ) AND ! empty( $sessionData['use_utf8mb4'] ) )
|
|
{
|
|
static::$utf->set_charset( 'utf8mb4' );
|
|
}
|
|
|
|
if ( \IPSUtf8\Session::i()->current_table === NULL AND \IPSUtf8\Session::i()->is_ipb )
|
|
{
|
|
/* Lock tasks until the end of time so they do not attempt to add / remove things during conversion */
|
|
if( static::$db->checkForTable('task_manager') )
|
|
{
|
|
static::$db->update( "task_manager", array( 'task_locked' => 2147483647 ) );
|
|
}
|
|
else
|
|
{
|
|
static::$db->update( "core_tasks", array( 'next_run' => 2147483647 ) );
|
|
}
|
|
}
|
|
|
|
$convertCount = 0;
|
|
$table = $this->getTable();
|
|
$convertCols = static::getConvertableColumns( $table['name'] );
|
|
$numericCols = static::getNumericColumns( $table['name'] );
|
|
$tableCharSet = \IPSUtf8\Convert::i()->database_charset;
|
|
$currentCharSet = \IPSUtf8\Session::i()->current_charset;
|
|
|
|
if ( ! empty( $table['charset'] ) )
|
|
{
|
|
$tableCharSet = $table['charset'];
|
|
|
|
/* Latin1 is actually cp1252, not ISO-8859-1 @link http://dev.mysql.com/doc/refman/5.0/en/charset-we-sets.html */
|
|
/* But only if our document character set is ISO-8859-1 or UTF-8. Otherwise, we're using the slow method which needs to real charset despite what the table is set too */
|
|
if ( $tableCharSet === 'latin1' AND in_array( mb_strtolower( $currentCharSet ), array( 'iso-8859-1', 'utf-8' ) ) )
|
|
{
|
|
$currentCharSet = 'windows-1252';
|
|
}
|
|
|
|
if ( defined( 'FORCE_CONVERT_CHARSET' ) AND FORCE_CONVERT_CHARSET !== NULL )
|
|
{
|
|
$currentCharSet = FORCE_CONVERT_CHARSET;
|
|
}
|
|
}
|
|
|
|
if ( !defined( 'SOURCE_DB_CHARSET' ) OR SOURCE_DB_CHARSET === NULL )
|
|
{
|
|
static::$db->set_charset( $tableCharSet );
|
|
}
|
|
|
|
/* Create table only */
|
|
if ( IPB_LOCK and in_array( $table['name'], static::$createOnly ) )
|
|
{
|
|
static::log( 'No content conversion of ' . $table['name'] . ' required' );
|
|
|
|
$this->getNextTable( $table['name'] );
|
|
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->json = $sessionData;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
return true;
|
|
}
|
|
|
|
while( $convertCount < $limit )
|
|
{
|
|
$sessionData = \IPSUtf8\Session::i()->json;
|
|
|
|
if ( $table === null )
|
|
{
|
|
if ( count( array_keys( \IPSUtf8\Session::i()->tables ) ) == count( array_keys( \IPSUtf8\Session::i()->completed_json ) ) )
|
|
{
|
|
\IPSUtf8\Session::i()->status = 'completed';
|
|
\IPSUtf8\Session::i()->save();
|
|
return false;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/* Got any fields to convert? */
|
|
if ( count( $convertCols ) === 0 )
|
|
{
|
|
/* No data to convert, lets repopulate the easy way! */
|
|
$this->preInsert( $table );
|
|
|
|
static::$utf->delete( $table['name'] );
|
|
static::$utf->insert( $table['name'], static::$db->select( '*', $table['name'] ) );
|
|
|
|
$this->postInsert( $table );
|
|
|
|
/* Update counts */
|
|
$sessionData['convertedCount'] += $table['count'];
|
|
$convertCount += $table['count'];
|
|
|
|
static::log( 'No columns to convert in ' . $table['name'] . ' INSERT INTO FROM SELECT used' );
|
|
|
|
$table = $this->getNextTable( $table['name'] );
|
|
$convertCols = static::getConvertableColumns( $table['name'] );
|
|
$numericCols = static::getNumericColumns( $table['name'] );
|
|
|
|
if ( $table === null or $convertCount >= $limit )
|
|
{
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->json = $sessionData;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
return true;
|
|
}
|
|
}
|
|
|
|
/* If the previous table had no columns to convert, then the table charset may be lost - reset it if it does not match */
|
|
if ( $table !== NULL AND $tableCharSet !== $table['charset'] )
|
|
{
|
|
$tableCharSet = $table['charset'];
|
|
}
|
|
|
|
/* Are we reeeeeeeeally sure we know the charset? Let's check one more time */
|
|
if ( ! $tableCharSet )
|
|
{
|
|
$createTable = static::$db->query( "SHOW CREATE TABLE `" . static::$db->prefix . $table['name'] . "`" )->fetch_assoc();
|
|
|
|
if ( preg_match( '#\scharset=([a-z0-9]+?)(\s|$)#i', $createTable['Create Table'], $matches ) )
|
|
{
|
|
$tableCharSet = mb_strtolower( $matches[1] );
|
|
}
|
|
}
|
|
|
|
/* Latin1 or ISO-8559-1? */
|
|
if ( ! in_array( $table['name'], static::$problemTables ) AND ( in_array( $tableCharSet, $this->getMysqlCharSets() ) ) and ( \IPSUtf8\Session::i()->current_charset == 'utf-8' or \IPSUtf8\Session::i()->current_charset == 'iso-8859-1' ) AND ( !isset( $sessionData['force_conversion'] ) OR $sessionData['force_conversion'] !== 1 ) )
|
|
{
|
|
/* Convert the fast way! */
|
|
$count = static::$utf->select( 'COUNT(*)', $table['name'] )->first();
|
|
if ( $count > 0 )
|
|
{
|
|
/* TRUNCATE can hang so only do it if the table isn't empty (which it will be 99% of the time) */
|
|
/* @see http://dba.stackexchange.com/questions/28055/truncate-table-statement-sometimes-hangs */
|
|
static::log("TRUNCATE TABLE `" . static::$utf->prefix . $table['name'] . "`");
|
|
|
|
static::$utf->query("TRUNCATE TABLE `" . static::$utf->prefix . $table['name'] . "`" );
|
|
}
|
|
|
|
/* Update counts */
|
|
$sessionData['convertedCount'] += $table['count'];
|
|
$convertCount += $table['count'];
|
|
|
|
$select = array();
|
|
$convertCols = static::getConvertableColumns( $table['name'] );
|
|
$schematic = static::getTableSchematic( $table['name'] );
|
|
|
|
foreach( array_keys( $table['definition']['columns'] ) as $col )
|
|
{
|
|
if ( in_array( $col, $convertCols ) )
|
|
{
|
|
/* Even if the current charset is set to UTF-8, if the Table Charset is latin1, we need to use this method */
|
|
/* ... but only if the actual column isn't using utf8 */
|
|
$columnCollation = $schematic['definition']['columns'][$col]['collation'];
|
|
if ( \IPSUtf8\Session::i()->current_charset != 'utf-8' OR ( $tableCharSet != 'utf8' AND ! in_array( $columnCollation, array( 'utf8_unicode_ci', 'utf8mb4_unicode_ci' ) ) ) OR ( $tableCharSet == 'utf8' AND mb_substr( $columnCollation, 0, 6 ) == 'latin1' ) )
|
|
{
|
|
if ( \IPSUtf8\Session::i()->current_charset == 'iso-8859-1' AND $tableCharSet == 'latin1' )
|
|
{
|
|
/* Casting as binary truncates non-latin characters, so if we are using real Latin1, just convert */
|
|
$select[] = 'CONVERT( `' . $col . '` USING utf8 )';
|
|
}
|
|
else
|
|
{
|
|
/* Otherwise cast to binary */
|
|
$select[] = 'CONVERT( CAST(`' . $col . '` AS BINARY) USING utf8 )';
|
|
}
|
|
}
|
|
else
|
|
{
|
|
/**
|
|
* @note This looks a little weird but bare with me. Semi-old databases (post-2.x but pre-4.x) can have UTF8 Tables and Columns, but contain Latin1 data.
|
|
* We need to try and sniff this out and apply a different type of conversion, because in this instance you cannot just extract
|
|
* as BINARY and then insert into the source. Instead, you need to convert to latin1, then to binary, then to utf8. But you can't
|
|
* do that on actual UTF8 data because it will then truncate multibyte characters so we need to compare the conversion against what
|
|
* is stored.
|
|
*/
|
|
$collation = 'utf8_unicode_ci';
|
|
|
|
if ( $tableCharSet == 'utf8mb4' OR $columnCollation == 'utf8mb4_unicode_ci' )
|
|
{
|
|
$collation = 'utf8mb4_unicode_ci';
|
|
}
|
|
|
|
$source_charset = 'latin1';
|
|
if ( defined( 'SOURCE_DB_CHARSET' ) AND SOURCE_DB_CHARSET !== NULL )
|
|
{
|
|
$source_charset = SOURCE_DB_CHARSET;
|
|
}
|
|
|
|
$select[] = 'CASE WHEN STRCMP(CONVERT(CONVERT(CONVERT(`' . $col . '` USING ' . $source_charset . ') USING BINARY) USING utf8), `' . $col . '` COLLATE ' . $collation . ') = 0 THEN CONVERT(`' . $col . '` USING BINARY) ELSE CONVERT(CONVERT(CONVERT(`' . $col . '` USING ' . $source_charset . ') USING BINARY) USING utf8) END';
|
|
}
|
|
}
|
|
else
|
|
{
|
|
$select[] = '`' . $col . '`';
|
|
}
|
|
}
|
|
|
|
$this->preInsert( $table );
|
|
|
|
$sql = "INSERT IGNORE INTO `" . static::$utf->prefix . $table['name'] . "` SELECT " . implode( ',', $select ) . " FROM `" . static::$db->prefix . $table['name'] . "`";
|
|
|
|
static::log( $sql );
|
|
|
|
static::$utf->query( $sql );
|
|
|
|
$this->postInsert( $table );
|
|
|
|
$table = $this->getNextTable( $table['name'] );
|
|
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->json = $sessionData;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
/* Always reset so we can test for no convertable columns above */
|
|
return true;
|
|
}
|
|
|
|
/* If we have a work table, do we have a PKEY and a current row? */
|
|
if ( \IPSUtf8\Session::i()->current_pkey )
|
|
{
|
|
$start = static::$utf->select('MAX(`' . \IPSUtf8\Session::i()->current_pkey . '`) as max', \IPSUtf8\Session::i()->current_table )->setKeyField('max')->first();
|
|
|
|
$rows = static::$db->select(
|
|
'*',
|
|
$table['name'],
|
|
array( \IPSUtf8\Session::i()->current_pkey . ' > ?', intval( $start ) ),
|
|
\IPSUtf8\Session::i()->current_pkey . ' ASC',
|
|
( $limit ? array( 0, ( $limit - $convertCount ) ) : null )
|
|
);
|
|
}
|
|
else
|
|
{
|
|
$start = static::$utf->select('COUNT(*) as count', \IPSUtf8\Session::i()->current_table )->setKeyField('count')->first();
|
|
|
|
/* Fetch via offset */
|
|
$rows = static::$db->select(
|
|
'*',
|
|
$table['name'],
|
|
null,
|
|
null,
|
|
( $limit ? array( $start, ( $limit - $convertCount ) ) : array( $start, 18446744073709551615 ) ) # No really, this is what MySQL recommends for offset, no limit (http://dev.mysql.com/doc/refman/5.1/en/select.html#id4651990)
|
|
);
|
|
}
|
|
|
|
if ( !defined( 'FORCE_CONVERT_METHOD' ) OR FORCE_CONVERT_METHOD === NULL )
|
|
{
|
|
\IPSUtf8\Text\Charset::$method = 'internal';
|
|
|
|
/* Optimise for latin character sets */
|
|
if ( function_exists( 'mb_convert_encoding' ) )
|
|
{
|
|
if ( in_array( strtolower( $currentCharSet ), array_map( 'strtolower', mb_list_encodings() ) ) )
|
|
{
|
|
\IPSUtf8\Text\Charset::$method = 'mb';
|
|
}
|
|
}
|
|
}
|
|
|
|
$rowCount = 0;
|
|
$gotRows = count( $rows );
|
|
$batch = array();
|
|
$batchBytes = 0;
|
|
$max = ( $limit ) ?: 250;
|
|
|
|
/* If we have a text column then the data set gets large */
|
|
if ( $limit > 50 AND static::hasTextColumn( $table['name'] ) )
|
|
{
|
|
$max = 50;
|
|
}
|
|
|
|
/* Throttle inserts for other reasons? */
|
|
$throttle = static::throttleInserts( $table['name'] );
|
|
|
|
if ( $throttle !== false )
|
|
{
|
|
$max = ( $throttle < $limit ) ? $throttle : $limit;
|
|
}
|
|
|
|
/* Anything fetched? */
|
|
if ( $gotRows === 0 )
|
|
{
|
|
try
|
|
{
|
|
static::$utf->query("ALTER TABLE `" . static::$utf->prefix . "{$table['name']}` ENABLE KEYS;");
|
|
static::log( "{$table['name']} keys enabled." );
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} enable keys exception caught." );
|
|
}
|
|
|
|
$table = $this->getNextTable( $table['name'] );
|
|
$convertCols = static::getConvertableColumns( $table['name'] );
|
|
$numericCols = static::getNumericColumns( $table['name'] );
|
|
|
|
continue;
|
|
}
|
|
|
|
foreach( $rows as $row )
|
|
{
|
|
foreach( $convertCols as $col )
|
|
{
|
|
if ( isset( $row[ $col ] ) AND ! empty( $row[ $col ] ) )
|
|
{
|
|
if ( static::isSerialized( $row[ $col ] ) AND \IPSUtf8\Text\Charset::i()->needsConverting( $row[ $col ], $currentCharSet, 'UTF-8' ) )
|
|
{
|
|
/* Store a copy in case it doesn't work */
|
|
$original = $row[ $col ];
|
|
$work = unserialize( $row[ $col ] );
|
|
|
|
array_walk_recursive( $work, function( &$input, $key ) use ($currentCharSet)
|
|
{
|
|
$input = \IPSUtf8\Text\Charset::i()->convert( $input, $currentCharSet, 'UTF-8' );
|
|
} );
|
|
|
|
$row[ $col ] = serialize( $work );
|
|
|
|
/* Did it work? */
|
|
if ( ! @unserialize( $row[ $col ] ) )
|
|
{
|
|
/* Something went wrong... maybe we can figure it out */
|
|
$work = unserialize( $original );
|
|
array_walk_recursive( $work, function ( &$input, $key ) use ( $currentCharSet, $tableCharSet ) {
|
|
if ( function_exists( 'mb_detect_encoding' ) )
|
|
{
|
|
$encoding = mb_detect_encoding( $input );
|
|
$input = \IPSUtf8\Text\Charset::i()->convert( $input, $encoding, 'UTF-8' );
|
|
}
|
|
else
|
|
{
|
|
/* @todo expand */
|
|
if ( $tableCharSet == 'latin1' )
|
|
{
|
|
$encoding = 'ISO-8859-1';
|
|
}
|
|
$input = \IPSUtf8\Text\Charset::i()->convert( $input, $encoding, 'UTF-8' );
|
|
}
|
|
} );
|
|
|
|
$row[ $col ] = serialize( $work );
|
|
}
|
|
}
|
|
else
|
|
{
|
|
$row[ $col ] = \IPSUtf8\Text\Charset::i()->convert( $row[ $col ], $currentCharSet, 'UTF-8' );
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Numeric columns */
|
|
foreach( $numericCols as $col )
|
|
{
|
|
if ( isset( $row[ $col ] ) AND ! empty( $row[ $col ] ) )
|
|
{
|
|
if ( $row[ $col ] === null OR ! is_numeric( $row[ $col ] ) )
|
|
{
|
|
$row[ $col ] = 0;
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Roughly add the size of this insert in bytes */
|
|
$tmp = $row; // Use a copy so the reference in array_walk doesn't overwrite the row to be inserted
|
|
|
|
/* Make sure it's an associative array as topic_views only has 1 column so DB driver returns an indexed array only */
|
|
array_walk_recursive( $tmp, function( &$input, $key )
|
|
{
|
|
$input = htmlentities( (string) $input, ENT_QUOTES | ENT_IGNORE, 'utf-8', false );
|
|
} );
|
|
|
|
$batchBytes += mb_strlen( @json_encode( $tmp ), '8bit');
|
|
unset( $tmp );
|
|
$batch[] = $row;
|
|
$nextBatch = null;
|
|
|
|
/* Give us an error margin to account for MySQL syntax */
|
|
if ( $batchBytes > ( $this->_data['max_allowed_packet'] - ( ( $this->_data['max_allowed_packet'] / 100 ) * ( count( $batch ) * 0.3 ) ) ) )
|
|
{
|
|
/* Remove last row */
|
|
$nextBatch = array_pop( $batch );
|
|
|
|
static::log( "Max packet hit with " . $batchBytes . 'b with ' . count( $batch ) . ' rows' );
|
|
}
|
|
|
|
/* If we have moved the only row from $batch to $nextbatch, then process that now */
|
|
if ( count( $nextBatch ) and ! count( $batch ) )
|
|
{
|
|
$batch = $nextBatch;
|
|
$nextBatch = array();
|
|
}
|
|
|
|
/* Got a batch to write? */
|
|
if ( $nextBatch !== null OR ( count( $batch ) === $max ) OR ( $gotRows === count( $batch ) ) )
|
|
{
|
|
try
|
|
{
|
|
/* Optimise */
|
|
$this->preInsert( $table );
|
|
|
|
$insertId = static::$utf->insert( $table['name'], $batch, false, true, $table );
|
|
|
|
if ( ! empty( static::$utf->error ) )
|
|
{
|
|
throw new \RuntimeException( static::$utf->error );
|
|
}
|
|
|
|
$this->postInsert( $table );
|
|
|
|
/* Update counts */
|
|
$sessionData['convertedCount'] += count( $batch );
|
|
$convertCount += count( $batch );
|
|
|
|
if ( \IPSUtf8\Session::i()->current_pkey )
|
|
{
|
|
\IPSUtf8\Session::i()->current_row = $insertId;
|
|
}
|
|
else
|
|
{
|
|
\IPSUtf8\Session::i()->current_row += count( $batch );
|
|
}
|
|
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->json = $sessionData;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
static::log( count( $batch ) . " rows batch inserted using conversion method: " . \IPSUtf8\Text\Charset::$method . ', last insert ID ' . $insertId );
|
|
|
|
$batch = ( $nextBatch !== null ) ? array( $nextBatch ) : array();
|
|
$batchBytes = 0;
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
$msg = $e->getMessage();
|
|
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
/* Trying to insert duplicate data, just let it go to the next in case there is a small overlap */
|
|
if ( mb_stristr( $msg, 'duplicate entry' ) )
|
|
{
|
|
static::log( "Duplicate key failure on insert\n{$msg}\n" . var_export( $batch, true ) );
|
|
return $this->heal( $limit, $convertCount );
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Anything fetched? */
|
|
if ( $gotRows === 0 )
|
|
{
|
|
try
|
|
{
|
|
static::$utf->query("ALTER TABLE `" . static::$utf->prefix . "{$table['name']}` ENABLE KEYS;");
|
|
static::log( "{$table['name']} keys enabled." );
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} enable keys exception caught." );
|
|
}
|
|
|
|
$table = $this->getNextTable( $table['name'] );
|
|
$convertCols = static::getConvertableColumns( $table['name'] );
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
|
|
/**
|
|
* Pre insert
|
|
*
|
|
* @param array $table Table definition data
|
|
* @return void
|
|
*/
|
|
public function preInsert( $table )
|
|
{
|
|
$row = static::$utf->query( "SHOW TABLE STATUS LIKE '" . static::$utf->prefix . $table['name'] . "'" )->fetch_assoc();
|
|
static::$utfTableEngine = $row['Engine'];
|
|
|
|
if ( empty( static::$utfTableEngine ) )
|
|
{
|
|
static::$utfTableEngine = static::$utf->defaultEngine();
|
|
}
|
|
|
|
if ( \strtolower( static::$utfTableEngine ) === 'myisam' )
|
|
{
|
|
static::log( "Pre inserts for MyISAM table " . $table['name'] );
|
|
|
|
try
|
|
{
|
|
//static::$utf->query("ALTER TABLE " . static::$utf->prefix . "{$table['name']} DISABLE KEYS;");
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} disable keys exception caught." );
|
|
}
|
|
}
|
|
else if ( \strtolower( static::$utfTableEngine ) === 'innodb' )
|
|
{
|
|
static::log( "Pre inserts for InnoDB table " . $table['name'] );
|
|
|
|
try
|
|
{
|
|
static::$utf->query("SET autocommit=0");
|
|
static::$utf->query("SET unique_checks=0");
|
|
static::$utf->query("SET foreign_key_checks=0");
|
|
}
|
|
catch ( Exception $e ) { }
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Post insert
|
|
*
|
|
* @param array $table Table definition data
|
|
* @return void
|
|
*/
|
|
public function postInsert( $table )
|
|
{
|
|
if ( \strtolower( static::$utfTableEngine ) === 'myisam' )
|
|
{
|
|
static::log( "Post inserts for MyISAM table " . $table['name'] );
|
|
|
|
try
|
|
{
|
|
//static::$utf->query("ALTER TABLE " . static::$utf->prefix . "{$table['name']} ENABLE KEYS;");
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} enable keys exception caught." );
|
|
}
|
|
}
|
|
else if ( \strtolower( static::$utfTableEngine ) === 'innodb' )
|
|
{
|
|
static::log( "Post inserts for InnoDB table " . $table['name'] );
|
|
|
|
try
|
|
{
|
|
static::$utf->query("SET autocommit=1");
|
|
static::$utf->query("SET unique_checks=1");
|
|
static::$utf->query("SET foreign_key_checks=1");
|
|
}
|
|
catch ( Exception $e ) { }
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Finish the conversion
|
|
*
|
|
* @return null
|
|
*/
|
|
public function finish()
|
|
{
|
|
$charset = 'utf8';
|
|
$collation = 'utf8_unicode_ci';
|
|
|
|
/* MB4? */
|
|
$sessionData = \IPSUtf8\Session::i()->json;
|
|
|
|
if ( isset( $sessionData['use_utf8mb4'] ) AND ! empty( $sessionData['use_utf8mb4'] ) )
|
|
{
|
|
static::$utf->set_charset( 'utf8mb4' );
|
|
|
|
$charset = 'utf8mb4';
|
|
$collation = 'utf8mb4_unicode_ci';
|
|
}
|
|
|
|
/* IPB 3? */
|
|
if ( isset( \IPSUtf8\Session::i()->tables['core_sys_conf_settings'] ) AND isset( \IPSUtf8\Session::i()->tables['cache_store'] ) )
|
|
{
|
|
if ( static::$utf->checkForTable('core_sys_conf_settings' ) )
|
|
{
|
|
static::$utf->update( 'core_sys_conf_settings', array( 'conf_value' => '' ), array( 'conf_key=?', 'gb_char_set' ) );
|
|
|
|
$rows = static::$utf->select( '*', 'core_sys_conf_settings', array( 'conf_add_cache=?', 1 ) );
|
|
|
|
$settings = array();
|
|
foreach( $rows as $row ) #row your boat
|
|
{
|
|
$value = $row['conf_value'] != "" ? $row['conf_value'] : $row['conf_default'];
|
|
|
|
if ( $value == '{blank}' )
|
|
{
|
|
$value = '';
|
|
}
|
|
|
|
$settings[ $row['conf_key'] ] = $value;
|
|
}
|
|
|
|
static::$utf->update( 'cache_store', array( 'cs_value' => serialize( $settings ) ), array( 'cs_key=?', 'settings' ) );
|
|
}
|
|
|
|
/* Update Language Locales */
|
|
if ( static::$utf->checkForTable('core_sys_lang') )
|
|
{
|
|
/* Store current */
|
|
$currentLocale = setlocale( LC_ALL, '0' );
|
|
|
|
/* Loop through languages */
|
|
$languages = static::$utf->select( '*', 'core_sys_lang' );
|
|
foreach( $languages AS $language )
|
|
{
|
|
$locale = explode( '.', $language['lang_short'] ); # We want to update even if a charset is already set.
|
|
foreach( array( "{$locale[0]}.UTF8", "{$locale[0]}.UTF-8", "{$locale[0]}.utf8" ) AS $test )
|
|
{
|
|
$verify = setlocale( LC_ALL, $test );
|
|
|
|
if ( $verify !== FALSE )
|
|
{
|
|
static::$utf->update( 'core_sys_lang', array( 'lang_short' => $test ), array( 'lang_id=?', $language['lang_id'] ) );
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
foreach( explode( ";", $currentLocale ) as $locale )
|
|
{
|
|
$parts = explode( "=", $locale );
|
|
if( in_array( $parts[0], array( 'LC_ALL', 'LC_COLLATE', 'LC_CTYPE', 'LC_MONETARY', 'LC_NUMERIC', 'LC_TIME' ) ) )
|
|
{
|
|
setlocale( constant( $parts[0] ), $parts[1] );
|
|
}
|
|
}
|
|
|
|
/* Clear Cache so it regenerates */
|
|
static::$utf->update( 'cache_store', array( 'cs_value' => '' ), array( 'cs_key=?', 'lang_data' ) );
|
|
}
|
|
|
|
/* Unlock Tasks */
|
|
if( static::$utf->checkForTable('task_manager') )
|
|
{
|
|
static::$utf->update( "task_manager", array( 'task_locked' => 0 ) );
|
|
}
|
|
else if( static::$utf->checkForTable('core_tasks') )
|
|
{
|
|
static::$utf->update( "core_tasks", array( 'next_run' => time() ) );
|
|
}
|
|
}
|
|
|
|
/* Change database collation */
|
|
require( ROOT_PATH . '/conf_global.php' );
|
|
|
|
static::$utf->query( "ALTER DATABASE `" . $INFO['sql_database'] . "` DEFAULT CHARACTER SET " . $charset . " COLLATE " . $collation );
|
|
}
|
|
|
|
/**
|
|
* Restore the original tables.
|
|
*
|
|
* @return null
|
|
*/
|
|
public function restoreOriginalTables()
|
|
{
|
|
/* Grab all tables to convert */
|
|
$stmt = static::$db->prepare( "SHOW TABLES" );
|
|
$stmt->execute();
|
|
$stmt->bind_result( $name );
|
|
|
|
$tables = array();
|
|
|
|
while ( $stmt->fetch() === true )
|
|
{
|
|
$tables[] = $name;
|
|
}
|
|
|
|
foreach( $tables as $name )
|
|
{
|
|
if ( mb_substr( $name, 0, 5 ) === 'orig_' )
|
|
{
|
|
$plainName = mb_substr( $name, 5 );
|
|
|
|
static::$db->query( "DROP TABLE IF EXISTS `" . $plainName . "`" );
|
|
|
|
$rename = "RENAME TABLE `" . $name . '` TO `' . $plainName . '`';
|
|
|
|
static::$db->query( $rename );
|
|
|
|
static::log( $rename );
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Delete the original tables.
|
|
*
|
|
* @return null
|
|
*/
|
|
public function deleteOriginalTables()
|
|
{
|
|
/* Grab all tables to convert */
|
|
$stmt = static::$utf->prepare( "SHOW TABLES" );
|
|
$stmt->execute();
|
|
$stmt->bind_result( $name );
|
|
|
|
$tables = array();
|
|
|
|
while ( $stmt->fetch() === true )
|
|
{
|
|
$tables[] = $name;
|
|
}
|
|
|
|
foreach( $tables as $name )
|
|
{
|
|
if ( mb_substr( $name, 0, 5 ) === 'orig_' )
|
|
{
|
|
static::$db->query( "DROP TABLE IF EXISTS `" . $name . "`" );
|
|
|
|
static::log( "DROP TABLE IF EXISTS `" . $name . "`" );
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Rename the tables
|
|
*
|
|
* @return null
|
|
*/
|
|
public function renameTables()
|
|
{
|
|
/* Grab all tables to convert */
|
|
$stmt = static::$db->prepare( "SHOW TABLES" );
|
|
$stmt->execute();
|
|
$stmt->bind_result( $name );
|
|
|
|
$tables = array();
|
|
$origPrefix = static::$db->prefix;
|
|
|
|
if ( mb_substr( $origPrefix, 0, 6 ) === 'x_utf_' )
|
|
{
|
|
$origPrefix = mb_substr( $origPrefix, 6 );
|
|
}
|
|
|
|
while ( $stmt->fetch() === true )
|
|
{
|
|
if( mb_substr( $name, 0, 6 ) === 'x_utf_' )
|
|
{
|
|
$tables[] = $name;
|
|
}
|
|
}
|
|
|
|
foreach( $tables as $name )
|
|
{
|
|
$tableNameNoPrefix = $name;
|
|
$tableName = mb_substr( $name, mb_strlen( static::$utf->prefix ) );
|
|
|
|
if ( $tableName !== 'convert_session' AND $tableName !== 'convert_session_tables' )
|
|
{
|
|
if( static::$utf->checkForTable( $tableName ) )
|
|
{
|
|
static::$db->query( "DROP TABLE IF EXISTS `" . $origPrefix . $tableName . '`' );
|
|
static::log( "DROP TABLE IF EXISTS `" . $origPrefix . $tableName . '`' );
|
|
}
|
|
|
|
static::$utf->query( "RENAME TABLE `" . $tableNameNoPrefix . '` TO `' . $origPrefix . $tableName . '`' );
|
|
static::log( "RENAME TABLE `" . $tableNameNoPrefix . '` TO `' . $origPrefix . $tableName . '`' );
|
|
}
|
|
}
|
|
|
|
/* Update table data */
|
|
\IPSUtf8\Session::i()->updateTableData();
|
|
}
|
|
|
|
/**
|
|
* Go through the DB and fix the collation of UTF8 tables
|
|
*
|
|
* @return null
|
|
*/
|
|
public function fixCollation()
|
|
{
|
|
$charset = 'utf8';
|
|
$collation = 'utf8_unicode_ci';
|
|
|
|
/* MB4? */
|
|
$sessionData = \IPSUtf8\Session::i()->json;
|
|
|
|
if ( isset( $sessionData['use_utf8mb4'] ) AND ! empty( $sessionData['use_utf8mb4'] ) )
|
|
{
|
|
static::$utf->set_charset( 'utf8mb4' );
|
|
|
|
$charset = 'utf8mb4';
|
|
$collation = 'utf8mb4_unicode_ci';
|
|
}
|
|
|
|
/* Change database collation */
|
|
require( ROOT_PATH . '/conf_global.php' );
|
|
|
|
static::$utf->query( "ALTER DATABASE `" . $INFO['sql_database'] . "` DEFAULT CHARACTER SET " . $charset . " COLLATE " . $collation );
|
|
|
|
foreach( \IPSUtf8\Session::i()->tables as $name => $data )
|
|
{
|
|
if ( in_array( $name, $this->getNonUtf8CollationTables() ) )
|
|
{
|
|
/* Skip if it's an _old table */
|
|
if( mb_substr( $name, -4 ) == '_old' )
|
|
{
|
|
continue;
|
|
}
|
|
|
|
$cols = static::getConvertableColumns( $data['name'] );
|
|
$tableData = \IPSUtf8\Session::i()->tables[ $data['name'] ];
|
|
|
|
if ( count( $cols ) )
|
|
{
|
|
$dropIndexes = array();
|
|
$addIndexes = array();
|
|
$modify = array();
|
|
$hasUnique = FALSE;
|
|
|
|
/* Changing collation can cause issues for duplicate enries in unique keys */
|
|
if ( isset( $tableData['definition']['indexes'] ) )
|
|
{
|
|
foreach( $tableData['definition']['indexes'] as $key => $index )
|
|
{
|
|
if ( $index['type'] === 'unique' )
|
|
{
|
|
/* Make sure none of the columns in this index are auto_increment */
|
|
foreach( $index['columns'] AS $k => $idxcol )
|
|
{
|
|
if ( $tableData['definition']['columns'][ $idxcol ]['auto_increment'] === TRUE )
|
|
{
|
|
continue 2;
|
|
}
|
|
}
|
|
|
|
if ( static::$db->checkForIndex( $name, $key ) )
|
|
{
|
|
$hasUnique = TRUE;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* We don't need to drop any indexes - just create a copy of the table and insert ignore */
|
|
if ( $hasUnique === TRUE )
|
|
{
|
|
static::$utf->query( "RENAME TABLE `" . static::$db->prefix . $name . "` TO `" . static::$db->prefix . $name . "_temp`" );
|
|
static::$utf->query( "CREATE TABLE `" . static::$db->prefix . $name . "` LIKE `" . static::$db->prefix . $name . "_temp`" );
|
|
static::$utf->query( "INSERT IGNORE INTO `" . static::$db->prefix . $name . "` SELECT * FROM `" . static::$db->prefix . $name . "_temp`" );
|
|
static::$utf->query( "DROP TABLE `" . static::$db->prefix . $name . "_temp`" );
|
|
}
|
|
|
|
static::$utf->query( "ALTER TABLE `" . static::$db->prefix . $name . "` CONVERT TO CHARACTER SET " . $charset . " COLLATE " . $collation );
|
|
|
|
if ( count( $cols ) )
|
|
{
|
|
foreach( $cols as $col )
|
|
{
|
|
if ( isset( $tableData['definition']['columns'][ $col ] ) )
|
|
{
|
|
$colData = $tableData['definition']['columns'][ $col ];
|
|
|
|
$modify[] = " MODIFY `" . $col . "` " . $colData['type'] . ( ( is_numeric( $colData['length'] ) AND $colData['length'] > 0 ) ? "(" . $colData['length'] . ")" : '' ) . " CHARACTER SET " . $charset . " COLLATE " . $collation;
|
|
}
|
|
}
|
|
}
|
|
|
|
if ( count( $cols ) )
|
|
{
|
|
if ( count( $modify ) )
|
|
{
|
|
$query = "ALTER TABLE `" . static::$db->prefix . $name . "` " . implode( ',', $modify );
|
|
|
|
static::log( $query );
|
|
static::$utf->query( $query );
|
|
}
|
|
|
|
if ( /*count( $modify ) and*/ count( $addIndexes ) )
|
|
{
|
|
$query = "ALTER TABLE `" . static::$db->prefix . $name . "` " . implode( ',', $addIndexes );
|
|
|
|
static::log( $query );
|
|
static::$utf->query( $query );
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
/* Update table data */
|
|
\IPSUtf8\Session::i()->updateTableData();
|
|
}
|
|
|
|
/**
|
|
* Something went wrong, so try and heal to restart progress
|
|
*
|
|
* @param int $limit Process() method limit
|
|
* @param int $cycleCount Items processed if this was in the middle of process()
|
|
* @return void
|
|
*/
|
|
public function heal( $limit=250, $cycleCount=0 )
|
|
{
|
|
if ( \IPSUtf8\Session::i()->current_table )
|
|
{
|
|
$table = \IPSUtf8\Session::i()->tables[ \IPSUtf8\Session::i()->current_table ];
|
|
|
|
/* Table doesn't exist, so rewind back to the start of this table */
|
|
if ( ! static::$utf->checkForTable( $table['name'] ) )
|
|
{
|
|
if ( static::$utf->createTable( $table['definition'] ) === false )
|
|
{
|
|
throw new \RuntimeException( static::$utf->error . "\n" . var_export( $table['definition'], true ) );
|
|
}
|
|
|
|
$completed = \IPSUtf8\Session::i()->completed_json;
|
|
|
|
if ( isset( $completed[ \IPSUtf8\Session::i()->current_table ] ) )
|
|
{
|
|
unset( $completed[ \IPSUtf8\Session::i()->current_table ] );
|
|
}
|
|
|
|
/* Got a primary key so we can use a WHERE N > X query rather than a limit for efficiency? */
|
|
if ( isset( $table['definition']['indexes']['PRIMARY'] ) )
|
|
{
|
|
$pkey = $table['definition']['indexes']['PRIMARY']['columns'][0];
|
|
|
|
/* Is it numeric? */
|
|
if ( mb_stristr( $table['definition']['columns'][ $pkey ]['type'], 'int' ) )
|
|
{
|
|
\IPSUtf8\Session::i()->current_pkey = $pkey;
|
|
}
|
|
}
|
|
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->completed_json = $completed;
|
|
\IPSUtf8\Session::i()->current_row = 0;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
/* Run again */
|
|
return $this->process( ( $limit - $cycleCount ) );
|
|
}
|
|
|
|
if ( \IPSUtf8\Session::i()->current_pkey )
|
|
{
|
|
/* Get the latest row */
|
|
$max = static::$utf->select('MAX(`' . \IPSUtf8\Session::i()->current_pkey . '`) as max', \IPSUtf8\Session::i()->current_table )->setKeyField('max')->first();
|
|
}
|
|
else
|
|
{
|
|
/* Get the count */
|
|
$max = static::$utf->select('COUNT(*) as count', \IPSUtf8\Session::i()->current_table )->setKeyField('count')->first();
|
|
}
|
|
|
|
\IPSUtf8\Session::i()->status = 'processing';
|
|
\IPSUtf8\Session::i()->current_row = $max;
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
/* Run again */
|
|
return $this->process( ( $limit - $cycleCount ), TRUE );
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Set the current work table
|
|
*
|
|
* @return void
|
|
*/
|
|
public function getTable()
|
|
{
|
|
$table = null;
|
|
|
|
if ( \IPSUtf8\Session::i()->current_table AND isset( \IPSUtf8\Session::i()->tables[ \IPSUtf8\Session::i()->current_table ] ) )
|
|
{
|
|
/* Table selected and convertion in progress */
|
|
$table = \IPSUtf8\Session::i()->tables[ \IPSUtf8\Session::i()->current_table ];
|
|
|
|
static::log( "Continuing with " . $table['name'] . ' (PKEY: ' . \IPSUtf8\Session::i()->current_pkey . ')' );
|
|
|
|
if ( ! static::$utf->checkForTable( $table['name'] ) )
|
|
{
|
|
if ( static::$utf->createTable( $this->checkTable( $table['definition'] ) ) === false )
|
|
{
|
|
throw new \RuntimeException( static::$utf->error . "\n" . var_export( $table['definition'], true ) );
|
|
}
|
|
try
|
|
{
|
|
static::$utf->query("ALTER TABLE `" . static::$utf->prefix . "{$table['name']}` DISABLE KEYS;");
|
|
static::log( "{$table['name']} keys disabled." );
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} disable keys exception caught." );
|
|
}
|
|
}
|
|
|
|
return $table;
|
|
}
|
|
else
|
|
{
|
|
/* need to resolve an issue where if you only need to convert a few tables, these few will be named x_utf_ while others won't,
|
|
so for now, just convert all */
|
|
$nonUtf8Tables = array_keys( \IPSUtf8\Session::i()->tables );//$this->getNonUtf8Tables();
|
|
|
|
/* No table selected */
|
|
if ( is_array( $nonUtf8Tables ) and count( $nonUtf8Tables ) )
|
|
{
|
|
if ( is_array( \IPSUtf8\Session::i()->completed_json ) )
|
|
{
|
|
$diff = array_diff( $nonUtf8Tables, array_keys( \IPSUtf8\Session::i()->completed_json ) );
|
|
|
|
if ( count( $diff ) )
|
|
{
|
|
$table = \IPSUtf8\Session::i()->tables[ array_shift( $diff ) ];
|
|
}
|
|
else
|
|
{
|
|
return null;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
$table = array_shift( $nonUtf8Tables );
|
|
}
|
|
|
|
if ( isset( $table['name'] ) )
|
|
{
|
|
\IPSUtf8\Session::i()->current_table = $table['name'];
|
|
|
|
/* Got a primary key so we can use a WHERE N > X query rather than a limit for efficiency? */
|
|
if ( isset( $table['definition']['indexes']['PRIMARY'] ) )
|
|
{
|
|
$pkey = $table['definition']['indexes']['PRIMARY']['columns'][0];
|
|
|
|
/* Is it numeric? */
|
|
if ( mb_stristr( $table['definition']['columns'][ $pkey ]['type'], 'int' ) )
|
|
{
|
|
\IPSUtf8\Session::i()->current_pkey = $pkey;
|
|
}
|
|
}
|
|
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
/* Create the table for UTF8 goodness */
|
|
static::$utf->dropTable( $table['name'], true );
|
|
|
|
if ( static::$utf->createTable( $this->checkTable( $table['definition'] ) ) === false )
|
|
{
|
|
throw new \RuntimeException( static::$utf->error . "\n" . var_export( $table['definition'], true ) );
|
|
}
|
|
|
|
static::log( "Created UTF8 table " . $table['name'] . ' (PKEY: ' . \IPSUtf8\Session::i()->current_pkey . ')' );
|
|
|
|
try
|
|
{
|
|
static::$utf->query("ALTER TABLE `" . static::$utf->prefix . "{$table['name']}` DISABLE KEYS;");
|
|
static::log( "{$table['name']} keys disabled." );
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$table['name']} disable keys exception caught." );
|
|
}
|
|
|
|
return $table;
|
|
}
|
|
}
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
/**
|
|
* Attempt to fix issues with keys longer than 1000bytes
|
|
*
|
|
* @param array $definition Table definition
|
|
* @return array
|
|
*/
|
|
public function checkTable( $definition )
|
|
{
|
|
/* MB4? */
|
|
$sessionData = \IPSUtf8\Session::i()->json;
|
|
$length = 0;
|
|
$multiplier = ( ! empty( $sessionData['use_utf8mb4'] ) ) ? 4 : 3;
|
|
$needsFixing = array();
|
|
$maxLen = 1000;
|
|
|
|
if ( \mb_strtolower( $definition['engine'] ) === 'innodb' )
|
|
{
|
|
$maxLen = 767;
|
|
}
|
|
|
|
if ( isset( $definition['indexes'] ) )
|
|
{
|
|
foreach( $definition['indexes'] as $key => $index )
|
|
{
|
|
foreach( $index['columns'] as $i => $column )
|
|
{
|
|
if ( isset( $definition['columns'][ $column ] ) and ( ( ! empty( $definition['columns'][ $column ]['length'] ) or in_array( mb_strtolower( $definition['columns'][ $column ]['type'] ), array( 'longtext', 'mediumtext', 'text' ) ) ) ) )
|
|
{
|
|
$length += (int) ( $definition['columns'][ $column ]['length'] ) ? $definition['columns'][ $column ]['length'] : 250;
|
|
}
|
|
}
|
|
|
|
if ( $length * $multiplier > $maxLen )
|
|
{
|
|
foreach( $index['columns'] as $i => $column )
|
|
{
|
|
if ( isset( $definition['columns'][ $column ] ) and ( ( ! empty( $definition['columns'][ $column ]['length'] ) or in_array( mb_strtolower( $definition['columns'][ $column ]['type'] ), array( 'longtext', 'mediumtext', 'text' ) ) ) ) )
|
|
{
|
|
/* Column name, column length, column type */
|
|
$needsFixing[ $key ][ $i ] = array( $column, ( (int) $definition['columns'][ $column ]['length'] ) ? $definition['columns'][ $column ]['length'] : 250, $definition['columns'][ $column ]['type'] );
|
|
}
|
|
}
|
|
}
|
|
|
|
$length = 0;
|
|
}
|
|
}
|
|
|
|
if ( count( $needsFixing ) )
|
|
{
|
|
foreach( $needsFixing as $key => $i )
|
|
{
|
|
$totalLength = 0;
|
|
$maxChars = $maxLen / $multiplier;
|
|
|
|
foreach( $i as $vals )
|
|
{
|
|
$totalLength += $vals[1];
|
|
}
|
|
|
|
if ( $totalLength > $maxChars )
|
|
{
|
|
/* Check each column can be reduced by the amount we need reducing */
|
|
$debt = 0;
|
|
|
|
$reduceEachBy = ( ( 100 / $totalLength ) * $maxChars ) / 100;
|
|
|
|
/* Apply debt if we have any. We do not reduce integers */
|
|
foreach( $i as $x => $vals )
|
|
{
|
|
if ( array_key_exists( mb_strtolower( $vals[2] ), static::$numericCols ) AND static::$numericCols[ mb_strtolower( $vals[2] ) ] !== NULL )
|
|
{
|
|
/* Don't go by "display" space, go by storage space, which is static regardless of supplied value ... an INT(4) still uses 11 chars */
|
|
$debt += static::$numericCols[ mb_strtolower( $vals[2] ) ];
|
|
}
|
|
}
|
|
|
|
/* Recalculate value to multiply index sub lengths with (subtracting debt) */
|
|
$reduceEachBy = ( ( 100 / ( $totalLength - $debt ) ) * ( $maxChars - $debt ) ) / 100;
|
|
|
|
foreach( $i as $x => $vals )
|
|
{
|
|
/* No length? */
|
|
if ( empty( $vals[1] ) )
|
|
{
|
|
$vals[1] = 250;
|
|
}
|
|
|
|
if ( array_key_exists( mb_strtolower( $vals[2] ), static::$numericCols ) AND static::$numericCols[ mb_strtolower( $vals[2] ) ] !== NULL )
|
|
{
|
|
/* Preserve col len */
|
|
continue;
|
|
}
|
|
|
|
$vals[1] = floor( $vals[1] * $reduceEachBy );
|
|
$i[ $x ] = $vals;
|
|
}
|
|
}
|
|
|
|
foreach( $i as $x => $vals )
|
|
{
|
|
if ( $definition['columns'][ $definition['indexes'][ $key ]['columns'][ $x ] ]['length'] != $vals[1] )
|
|
{
|
|
$definition['indexes'][ $key ]['length'][ $x ] = $vals[1];
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
return $definition;
|
|
}
|
|
|
|
/**
|
|
* Fetches the next table to process or false if nothing left to process
|
|
*
|
|
* @param string $currentTable Name of table just processed
|
|
* @return array Array of table data
|
|
*/
|
|
public function getNextTable( $currentTable )
|
|
{
|
|
try
|
|
{
|
|
static::$utf->query("ALTER TABLE `" . static::$utf->prefix . "{$currentTable}` ENABLE KEYS;");
|
|
static::log( "{$table['name']} keys enabled." );
|
|
}
|
|
catch( \IPSUtf8\Db\Exception $e )
|
|
{
|
|
static::log( static::$utf->prefix . "{$currentTable} enable keys exception caught." );
|
|
}
|
|
|
|
$completed = \IPSUtf8\Session::i()->completed_json;
|
|
$completed[ $currentTable ] = $currentTable;
|
|
|
|
\IPSUtf8\Session::i()->completed_json = $completed;
|
|
\IPSUtf8\Session::i()->current_pkey = null;
|
|
\IPSUtf8\Session::i()->current_table = null;
|
|
\IPSUtf8\Session::i()->current_row = 0;
|
|
|
|
\IPSUtf8\Session::i()->save();
|
|
|
|
return $this->getTable();
|
|
}
|
|
|
|
/**
|
|
* Init this class
|
|
*/
|
|
public function init()
|
|
{
|
|
/* Grab all tables to convert */
|
|
$stmt = static::$db->query( "show variables" );
|
|
|
|
$tables = array();
|
|
|
|
while ( $row = $stmt->fetch_array( MYSQLI_ASSOC ) )
|
|
{
|
|
$key = $row['Variable_name'];
|
|
$value = $row['Value'];
|
|
|
|
if ( $key == 'character_set_database' )
|
|
{
|
|
$this->_data['database_charset'] = strtolower( $value );
|
|
}
|
|
|
|
if ( $key == 'bulk_insert_buffer_size' )
|
|
{
|
|
$this->_data['bulk_insert_buffer_size'] = $value;
|
|
}
|
|
|
|
if ( $key == 'max_allowed_packet' )
|
|
{
|
|
$this->_data['max_allowed_packet'] = $value;
|
|
}
|
|
}
|
|
|
|
if ( ! empty( $this->_data['max_allowed_packet'] ) AND ! empty( $this->_data['bulk_insert_buffer_size'] ) AND ( $this->_data['bulk_insert_buffer_size'] < $this->_data['max_allowed_packet'] ) )
|
|
{
|
|
$this->_data['max_allowed_packet'] = $this->_data['bulk_insert_buffer_size'];
|
|
}
|
|
|
|
if ( empty( $this->_data['max_allowed_packet'] ) AND empty( $this->_data['bulk_insert_buffer_size'] ) )
|
|
{
|
|
/* No value so use MySQL default of 1MB */
|
|
$this->_data['max_allowed_packet'] = 1048576;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Returns true if the table has a text or medium text column
|
|
*
|
|
* @param string $name Name of table
|
|
* @return boolean
|
|
*/
|
|
public static function hasTextColumn( $name )
|
|
{
|
|
$table = \IPSUtf8\Session::i()->tables[ $name ];
|
|
|
|
foreach( $table['definition']['columns'] as $col => $val )
|
|
{
|
|
if ( in_array( mb_strtolower( $val['type'] ), array( 'mediumtext', 'text' ) ) )
|
|
{
|
|
return true;
|
|
}
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
|
|
/**
|
|
* Tables with lots of dynamic columns can be slow
|
|
*
|
|
* @param string $name Name of table
|
|
* @return boolean
|
|
*/
|
|
public static function throttleInserts( $name )
|
|
{
|
|
$table = \IPSUtf8\Session::i()->tables[ $name ];
|
|
$total = 0;
|
|
$var = 0;
|
|
|
|
foreach( $table['definition']['columns'] as $col => $val )
|
|
{
|
|
$total++;
|
|
|
|
if ( in_array( mb_strtolower( $val['type'] ), array( 'varchar' ) ) )
|
|
{
|
|
$var++;
|
|
}
|
|
}
|
|
|
|
if ( $total > 70 )
|
|
{
|
|
return 50;
|
|
}
|
|
else if ( $total > 30 AND $var > 0 )
|
|
{
|
|
if ( ( $var / $total ) * 100 > 25 )
|
|
{
|
|
return 100;
|
|
}
|
|
}
|
|
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the DB is UTF8 already
|
|
*
|
|
* @return bool
|
|
*/
|
|
public function databaseIsUtf8()
|
|
{
|
|
$yeahProbably = false;
|
|
|
|
if ( \IPSUtf8\Convert::i()->database_charset == 'utf8mb4' OR \IPSUtf8\Convert::i()->database_charset == 'utf8' OR \IPSUtf8\Session::i()->current_charset == 'utf-8' )
|
|
{
|
|
$yeahProbably = true;
|
|
}
|
|
|
|
/* Another check */
|
|
if ( $yeahProbably === true )
|
|
{
|
|
/* Best check the tables, then */
|
|
$json = \IPSUtf8\Session::i()->json;
|
|
|
|
if ( count( $this->getNonUtf8Tables() ) )
|
|
{
|
|
$yeahProbably = false;
|
|
}
|
|
}
|
|
|
|
return $yeahProbably;
|
|
}
|
|
|
|
/**
|
|
* Grab the character sets from MySQL
|
|
*
|
|
* @return array
|
|
*/
|
|
public function getMysqlCharSets()
|
|
{
|
|
if ( static::$mysqlCharSets === NULL )
|
|
{
|
|
static::$mysqlCharSets = array( 'latin1', 'utf8' );
|
|
|
|
$stmt = static::$db->query( "show character set" );
|
|
|
|
while ( $row = $stmt->fetch_array( MYSQLI_ASSOC ) )
|
|
{
|
|
static::$mysqlCharSets[] = $row['Charset'];
|
|
}
|
|
}
|
|
|
|
return static::$mysqlCharSets;
|
|
}
|
|
|
|
/**
|
|
* Return tables that need converting
|
|
*
|
|
* @param boolean $force Force a recount
|
|
* @return array
|
|
*/
|
|
public function getNonUtf8Tables( $force=false )
|
|
{
|
|
if ( static::$nonUtf8Tables === NULL or $force === TRUE )
|
|
{
|
|
static::$nonUtf8Tables = array();
|
|
|
|
/* Best check the tables, then */
|
|
$json = \IPSUtf8\Session::i()->json;
|
|
|
|
foreach( $json['charSets'] as $charSet => $tablesArray )
|
|
{
|
|
if ( ! empty( $json['force_conversion'] ) or ( $charSet != 'utf8' and $charSet != 'utf8mb4' ) )
|
|
{
|
|
static::$nonUtf8Tables = array_merge( static::$nonUtf8Tables, $tablesArray );
|
|
}
|
|
}
|
|
}
|
|
|
|
return static::$nonUtf8Tables;
|
|
}
|
|
|
|
/**
|
|
* Returns total rows to convert
|
|
*
|
|
* @return array
|
|
*/
|
|
public function getTotalRowsToConvert()
|
|
{
|
|
if ( static::$rowsToConvert === NULL )
|
|
{
|
|
static::$rowsToConvert = 0;
|
|
|
|
foreach( array_keys( \IPSUtf8\Session::i()->tables ) as $table )
|
|
{
|
|
static::$rowsToConvert += intval( \IPSUtf8\Session::i()->tables[ $table ]['count'] );
|
|
}
|
|
}
|
|
|
|
return static::$rowsToConvert;
|
|
}
|
|
|
|
/**
|
|
* Returns whether the DB has the correct collation
|
|
*
|
|
* @return bool
|
|
*/
|
|
public function getNonUtf8CollationTables()
|
|
{
|
|
if ( static::$nonUtf8Collations === null )
|
|
{
|
|
static::$nonUtf8Collations = array();
|
|
|
|
/* Best check the tables, then */
|
|
foreach( \IPSUtf8\Session::i()->tables as $table => $definition )
|
|
{
|
|
/* If the table itself is not utf8_unicode_ci, don't bother check columns */
|
|
if ( $definition['collation'] AND !in_array( $definition['collation'], array( 'utf8_unicode_ci', 'utf8mb4_unicode_ci' ) ) )
|
|
{
|
|
static::$nonUtf8Collations[] = $table;
|
|
continue;
|
|
}
|
|
|
|
foreach( $definition['definition']['columns'] as $name => $column )
|
|
{
|
|
if ( $column['collation'] and ! in_array( $column['collation'], array( 'utf8_unicode_ci', 'utf8mb4_unicode_ci' ) ) )
|
|
{
|
|
static::$nonUtf8Collations[] = $table;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
return static::$nonUtf8Collations;
|
|
}
|
|
|
|
/**
|
|
* Fetch debug information
|
|
*
|
|
* @return array
|
|
*/
|
|
public function getDebugString()
|
|
{
|
|
$data = array();
|
|
$tableCharSet = \IPSUtf8\Convert::i()->database_charset;
|
|
$currentCharSet = \IPSUtf8\Session::i()->current_charset;
|
|
|
|
$json = \IPSUtf8\Session::i()->json;
|
|
$data[] = "IP.Board Character Set: " . $currentCharSet;
|
|
$data[] = "Database Character Set: " . $tableCharSet;
|
|
$data[] = "Original table prefix: " . static::$db->prefix;
|
|
$data[] = "Converted table prefix: " . static::$utf->prefix;
|
|
|
|
foreach( $json['charSets'] as $charSet => $tablesArray )
|
|
{
|
|
$data[] = count( $tablesArray ) . ' tables are ' . $charSet;
|
|
}
|
|
|
|
$data[] = count( $this->getNonUtf8CollationTables() ) . " tables have incorrect collations";
|
|
$data[] = "Can use 'dump' method: " . var_export( $this->canDump(), true );
|
|
|
|
return $data;
|
|
}
|
|
|
|
/**
|
|
* Detect whether this is a serialized string or not
|
|
*
|
|
* @param string $string The actual string
|
|
* @return boolean
|
|
*/
|
|
public static function isSerialized( $string )
|
|
{
|
|
/* If it looks nothing like a serialized string, then return it */
|
|
if ( ! preg_match( '#^a:\d+:\{#', $string ) )
|
|
{
|
|
return false;
|
|
}
|
|
|
|
/* Now make sure it's actually a serialized string */
|
|
return ( boolean ) @unserialize( $string );
|
|
}
|
|
|
|
/**
|
|
* Log data
|
|
*/
|
|
public static function log( $message )
|
|
{
|
|
$file = THIS_PATH . '/tmp/log_' . date('Y-m-d') . '.cgi';
|
|
$isNew = false;
|
|
|
|
if ( ! is_file( $file ) )
|
|
{
|
|
$isNew = true;
|
|
}
|
|
|
|
@file_put_contents( $file, "\n" . str_repeat( '-', 48 ) . "\n" . date('r') . "\n" . $message, FILE_APPEND );
|
|
|
|
if ( $isNew )
|
|
{
|
|
@chmod( $file, 0777 );
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Return a list of numeric columns that need checking
|
|
*
|
|
* @param string $name Table
|
|
* @return array Array of columns that need checking as they can contain INT
|
|
*/
|
|
public static function getNumericColumns( $name )
|
|
{
|
|
if ( ! is_string( $name ) )
|
|
{
|
|
return false;
|
|
}
|
|
|
|
$table = \IPSUtf8\Session::i()->tables[ $name ];
|
|
$return = array();
|
|
|
|
foreach( $table['definition']['columns'] as $col => $val )
|
|
{
|
|
if ( in_array( mb_strtolower( $val['type'] ), static::$numericCols ) )
|
|
{
|
|
$return[] = $val['name'];
|
|
}
|
|
}
|
|
|
|
return $return;
|
|
}
|
|
|
|
/**
|
|
* Return a list of columns that need converting
|
|
*
|
|
* @param string $name Table
|
|
* @return array Array of columns that need converting as they can contain text
|
|
*/
|
|
public static function getConvertableColumns( $name )
|
|
{
|
|
if ( ! is_string( $name ) )
|
|
{
|
|
return false;
|
|
}
|
|
|
|
$table = \IPSUtf8\Session::i()->tables[ $name ];
|
|
$return = array();
|
|
|
|
foreach( $table['definition']['columns'] as $col => $val )
|
|
{
|
|
if ( in_array( mb_strtolower( $val['type'] ), static::$convertCols ) )
|
|
{
|
|
$return[] = $val['name'];
|
|
}
|
|
}
|
|
|
|
return $return;
|
|
}
|
|
|
|
/**
|
|
* Can we use the fast dump method?
|
|
*
|
|
* @return boolean
|
|
*/
|
|
public static function canDump()
|
|
{
|
|
/* Best check the tables, then */
|
|
$json = \IPSUtf8\Session::i()->json;
|
|
$tablesOk = true;
|
|
|
|
foreach( $json['charSets'] as $charSet => $tablesArray )
|
|
{
|
|
if ( $charSet != 'utf8' and $charSet != 'utf8mb4' and mb_substr( $charSet, 0, 5) != 'latin' and $charSet != 'windows-1252' )
|
|
{
|
|
$tablesOk = false;
|
|
break;
|
|
}
|
|
}
|
|
|
|
if (
|
|
( $tablesOk ) AND
|
|
( \IPSUtf8\Session::i()->current_charset == 'utf-8' or \IPSUtf8\Session::i()->current_charset == 'iso-8859-1' ) AND
|
|
( ( is_callable( 'exec' ) AND false === stripos( ini_get( 'disable_functions' ), 'exec' ) ) )
|
|
)
|
|
{
|
|
try
|
|
{
|
|
@exec("iconv -l", $output );
|
|
}
|
|
catch( \ErrorException $e )
|
|
{
|
|
return false;
|
|
}
|
|
|
|
if ( ! array( $output ) or ! count( $output ) or ( ! stristr( implode( "\n", $output ), 'latin' ) ) )
|
|
{
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
else
|
|
{
|
|
return false;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Get Table Schematic
|
|
*
|
|
* @return array
|
|
*/
|
|
protected static function getTableSchematic( $table )
|
|
{
|
|
return json_decode( \IPSUtf8\Db::i('utf8')->select( 'table_schema', 'convert_session_tables', array( "table_name=?", $table ) )->first(), TRUE );
|
|
}
|
|
}
|