filehash.php

См. документацию.

<?php
namespace Bitrix\Clouds;
 
use Bitrix\Main\Localization\Loc;
use Bitrix\Main\ORM\Data\DataManager;
use Bitrix\Main\ORM\Fields\DatetimeField;
use Bitrix\Main\ORM\Fields\IntegerField;
use Bitrix\Main\ORM\Fields\StringField;
use Bitrix\Main\ORM\Fields\Validators\LengthValidator;
use Bitrix\Main\ORM\Fields\ExpressionField;
 
Loc::loadMessages(__FILE__);

 
class FileHashTable extends DataManager
{
    public static function getTableName()
    {
        return 'b_clouds_file_hash';
    }

    public static function getMap()
    {
        return [
            new IntegerField(
                'ID',
                [
                    'primary' => true,
                    'autocomplete' => true,
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_ID_FIELD'),
                ]
            ),
            new IntegerField(
                'BUCKET_ID',
                [
                    'required' => true,
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_BUCKET_ID_FIELD'),
                ]
            ),
            new StringField(
                'FILE_PATH',
                [
                    'required' => true,
                    'validation' => [__CLASS__, 'validateFilePath'],
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_FILE_PATH_FIELD'),
                ]
            ),
            new IntegerField(
                'FILE_SIZE',
                [
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_FILE_SIZE_FIELD'),
                ]
            ),
            new DatetimeField(
                'FILE_MTIME',
                [
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_FILE_MTIME_FIELD'),
                ]
            ),
            new StringField(
                'FILE_HASH',
                [
                    'validation' => [__CLASS__, 'validateFileHash'],
                    'title' => Loc::getMessage('FILE_HASH_ENTITY_FILE_HASH_FIELD'),
                ]
            ),
        ];
    }

    public static function validateFilePath()
    {
        return [
            new LengthValidator(null, 760),
        ];
    }

    public static function validateFileHash()
    {
        return [
            new LengthValidator(null, 50),
        ];
    }

    public static function addList($bucketId, array $files)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $helper = $connection->getSqlHelper();
        $values = [];
        foreach ($files['file'] as $i => $file)
        {
            $fileSize = $files['file_size'][$i];
            $fileMtime = \CCloudUtil::gmtTimeToDateTime($files['file_mtime'][$i]);
            $fileHash = $files['file_hash'][$i];
            $values [] = '('
                . $bucketId
                . ",'" . $helper->forSql($file) . "'"
                . ',' . intval($fileSize)
                . ",'" . $fileMtime->format('Y-m-d h:i:s') . "'"
                . ",'" . $helper->forSql($fileHash) . "'"
                . ')'
            ;
        }
        $sql = '
            INSERT INTO ' . static::getTableName() . '
            (BUCKET_ID, FILE_PATH, FILE_SIZE, FILE_MTIME, FILE_HASH)
            VALUES
            ' . implode(",\n", $values) . '
        ';
        return $connection->query($sql);
    }

    public static function syncList($bucketId, $path, array $files, $prevLastKey)
    {
        $result = null;
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $helper = $connection->getSqlHelper();
 
        $index = [];
        foreach ($files['file'] as $i => $filePath)
        {
            $index[$path . $filePath] = $i;
        }
 
        $filter = [
            '=BUCKET_ID' => $bucketId,
        ];
        if ($prevLastKey)
        {
            $filter['>FILE_PATH'] = $path . $prevLastKey;
        }
        if ($files['last_key'])
        {
            $filter['<=FILE_PATH'] = $path . $files['last_key'];
        }
        $fileList = static::getList([
            'select' => ['ID', 'FILE_PATH', 'FILE_SIZE', 'FILE_HASH'],
            'filter' => $filter,
        ]);
        while ($fileInfo = $fileList->fetch())
        {
            if (
                array_key_exists($fileInfo['FILE_PATH'], $index)
                && ($files['file_size'][$index[$fileInfo['FILE_PATH']]] == $fileInfo['FILE_SIZE'])
                && ($files['file_hash'][$index[$fileInfo['FILE_PATH']]] == $fileInfo['FILE_HASH'])
            )
            {
                unset($files['file'][$index[$fileInfo['FILE_PATH']]]);
            }
            else
            {
                $deleteResult = static::delete($fileInfo['ID']);
            }
        }
 
        $values = [];
        foreach ($files['file'] as $i => $file)
        {
            $fileSize = $files['file_size'][$i];
            $fileMtime = \CCloudUtil::gmtTimeToDateTime($files['file_mtime'][$i]);
            $fileHash = $files['file_hash'][$i];
            $values [] = '('
                . $bucketId
                . ",'" . $helper->forSql($path . $file) . "'"
                . ',' . intval($fileSize)
                . ",'" . $fileMtime->format('Y-m-d h:i:s') . "'"
                . ",'" . $helper->forSql($fileHash) . "'"
                . ')'
            ;
        }
 
        $insertSize = 1000;
        while ($values)
        {
            $sql = '
                INSERT INTO ' . static::getTableName() . '
                (BUCKET_ID, FILE_PATH, FILE_SIZE, FILE_MTIME, FILE_HASH)
                VALUES
                ' . implode(",\n", array_splice($values, 0, $insertSize)) . '
            ';
            $result = $connection->query($sql);
        }
 
        return $result;
    }

    public static function syncEnd($bucketId, $path, $prevLastKey)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $sqlHelper = $connection->getSqlHelper();
        $delete = '
            DELETE from ' . static::getTableName() . '
            WHERE BUCKET_ID = ' . $bucketId . '
            AND FILE_PATH like \'' . $sqlHelper->forSql($path) . '%\'
            AND FILE_PATH > \'' . $sqlHelper->forSql($path . $prevLastKey) . '\'
        ';
        $result = $connection->query($delete);
        return $result;
    }

    public static function addFile($bucketId, $path, array $fileInfo)
    {
        return static::add([
            'BUCKET_ID' => $bucketId,
            'FILE_PATH' => $path,
            'FILE_SIZE' => $fileInfo['size'],
            'FILE_MTIME' => \CCloudUtil::gmtTimeToDateTime($fileInfo['mtime']),
            'FILE_HASH' => $fileInfo['hash'],
        ]);
    }

    public static function getLastKey($bucketId)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $sql = 'SELECT max(FILE_PATH) LAST_KEY from ' . static::getTableName() . ' WHERE BUCKET_ID=' . $bucketId;
        $last = $connection->query($sql)->fetch();
        return $last && $last['LAST_KEY'] ? $last['LAST_KEY'] : '';
    }

    public static function deleteAll($bucketId)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $delete = 'DELETE from ' . static::getTableName() . ' WHERE BUCKET_ID=' . $bucketId;
        $result = $connection->query($delete);
        return $result;
    }

    public static function deleteByFilePath($bucketId, $filePath)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $sqlHelper = $connection->getSqlHelper();
        $delete = '
            DELETE from ' . static::getTableName() . '
            WHERE BUCKET_ID = ' . $bucketId . "
            AND FILE_PATH = '" . $sqlHelper->forSql($filePath) . "'
        ";
        $result = $connection->query($delete);
        return $result;
    }

    public static function dirList($bucketId, $path, $order, $filter)
    {
        $connection = \Bitrix\Main\Application::getConnection();
        $sqlHelper = $connection->getSqlHelper();
 
        $query = \Bitrix\Clouds\FileHashTable::query();
        $query->setSelect([
            new ExpressionField(
                'FILE_TYPE',
                'case when position(\'/\' in substring(%s, length(\'' . $sqlHelper->forSql($path) . '\')+1)) > 0 then \'D\' else \'F\' end',
                ['FILE_PATH']
            ),
            new ExpressionField(
                'NAME',
                'substring_index(substring(%s, length(\'' . $sqlHelper->forSql($path) . '\')+1), \'/\', 1)',
                ['FILE_PATH']
            ),
            new ExpressionField(
                'SUM_FILE_SIZE',
                'SUM(%s)',
                ['FILE_SIZE']
            ),
            new ExpressionField(
                'MAX_FILE_MTIME',
                'MAX(%s)',
                ['FILE_MTIME']
            ),
            new ExpressionField(
                'FILE_COUNT',
                'COUNT(1)'
            ),
        ]);
 
        $filter['=BUCKET_ID'] = $bucketId;
        $filter['%=FILE_PATH'] = $path . '%';
        $query->setFilter($filter);
        $query->setGroup(['FILE_TYPE', 'NAME']);
        $query->setOrder($order);
 
        $sql = $query->getQuery();
 
        return $connection->query($sql);
    }

    public static function duplicateList($bucketId, $filter, $order, $limit = 0)
    {
        $connection = \Bitrix\Main\Application::getConnection();
        $helper = $connection->getSqlHelper();
        $query = \Bitrix\Main\File\Internal\FileHashTable::query();
        $query->registerRuntimeField('FILE_PATH', [
            'expression' => [
                $helper->getConcatFunction('%s', "'/'", '%s'), 'FILE.SUBDIR', 'FILE.FILE_NAME'
            ]
        ]);
        $query->setSelect([
            'FILE_HASH',
            'FILE_SIZE',
            new ExpressionField(
                'FILE_COUNT',
                'COUNT(distinct %s)',
                ['FILE_PATH']
            ),
            new ExpressionField(
                'FILE_ID_MIN',
                'MIN(%s)',
                ['FILE_ID']
            ),
            new ExpressionField(
                'FILE_ID_MAX',
                'MAX(%s)',
                ['FILE_ID']
            ),
        ]);
 
        $filter['=FILE.HANDLER_ID'] = $bucketId;
        $filter['>FILE_COUNT'] = 1;
        $query->setFilter($filter);
        $query->setGroup(['FILE_HASH', 'FILE_SIZE']);
        $query->setOrder($order);
        if ($limit > 0)
        {
            $query->setLimit($limit);
        }
 
        $sql = $query->getQuery();
 
        return $connection->query($sql);
    }

    public static function getFileDuplicates($bucketId, $fileHash, $fileSize)
    {
        $query = \Bitrix\Main\File\Internal\FileHashTable::getList([
            'select' => [
                'FILE_ID',
            ],
            'filter' => [
                '=FILE.HANDLER_ID' => $bucketId,
                '=FILE_HASH' => $fileHash,
                '=FILE_SIZE' => $fileSize,
            ],
            'order' => [
                'FILE_ID' => 'ASC',
            ],
        ]);
 
        $result = [];
        while ($fileDuplicate = $query->fetch())
        {
            $result[] = $fileDuplicate['FILE_ID'];
        }
        return $result;
    }

    public static function getDuplicatesStat($bucketId)
    {
        $bucketId = intval($bucketId);
        $connection = \Bitrix\Main\Application::getConnection();
        $helper = $connection->getSqlHelper();
        $sql = '
            select sum(DUP_COUNT) DUP_COUNT, sum(DUP_SIZE) DUP_SIZE
            from (
                select
                    b_file_hash.FILE_SIZE
                    ,b_file_hash.FILE_HASH
                    ,count(distinct ' . $helper->getConcatFunction('b_file.SUBDIR', "'/'", 'b_file.FILE_NAME') . ')-1 DUP_COUNT
                    ,(count(distinct ' . $helper->getConcatFunction('b_file.SUBDIR', "'/'", 'b_file.FILE_NAME') . ")-1) * b_file_hash.FILE_SIZE DUP_SIZE
                from
                    b_file_hash
                    inner join b_file on
                        b_file.ID = b_file_hash.FILE_ID
                where
                    b_file.HANDLER_ID = '" . $bucketId . "'
                group by
                    b_file_hash.FILE_SIZE, b_file_hash.FILE_HASH
                having
                    count(distinct " . $helper->getConcatFunction('b_file.SUBDIR', "'/'", 'b_file.FILE_NAME') . ') > 1
            ) t
        ';
 
        return $connection->query($sql)->fetch();
    }

    public static function copyToFileHash($lastKey, $pageSize)
    {
        global $DB;
        $connection = \Bitrix\Main\Application::getConnection();
        $helper = $connection->getSqlHelper();
 
        $lastKey = (int)$lastKey;
        $pageSize = (int)$pageSize;
        $sql = '
            SELECT
                b_file.ID as FILE_ID
                ,b_clouds_file_hash.FILE_SIZE as FILE_SIZE
                ,b_clouds_file_hash.FILE_HASH as FILE_HASH
            FROM
                b_file
                INNER JOIN b_clouds_file_hash ON
                    b_clouds_file_hash.BUCKET_ID = ' . $DB->ToNumber('b_file.HANDLER_ID') . '
                    AND b_clouds_file_hash.FILE_PATH = ' . $helper->getConcatFunction("'/'", 'b_file.SUBDIR', "'/'", 'b_file.FILE_NAME') . '
                LEFT JOIN b_file_duplicate ON
                    b_file_duplicate.DUPLICATE_ID = b_file.ID
            WHERE
                b_file.ID > ' . $lastKey . '
                AND b_file_duplicate.DUPLICATE_ID is null
            ORDER BY b_file.ID
            LIMIT ' . $pageSize . '
        ';
 
        $fileIds = $connection->query('
            SELECT
                min(FILE_ID) as FILE_ID_MIN
                ,max(FILE_ID) as FILE_ID_MAX
                ,count(FILE_ID) FILE_ID_CNT
            FROM (' . $sql . ') t
        ')->fetch();
 
        if ($fileIds['FILE_ID_CNT'] > 0)
        {
            $sql = $helper->getInsertIgnore(
                'b_file_hash',
                '(FILE_ID, FILE_SIZE, FILE_HASH)',
                $sql
            );
 
            $connection->query($sql);
        }
 
        return $fileIds;
    }

    public static function prepareDuplicates($bucketId, &$fileIds)
    {
        $originalId = false;
        if ($fileIds)
        {
            //Exclude any id that is already a duplicate
            $duplicates = \Bitrix\Main\File\Internal\FileDuplicateTable::getList([
                'select' => ['DUPLICATE_ID'],
                'filter' => [
                    '=DUPLICATE_ID' => $fileIds,
                ],
            ]);
            while ($duplicate = $duplicates->fetch())
            {
                //Others will be excluded from the process
                $p = array_search($duplicate['DUPLICATE_ID'], $fileIds);
                if ($p !== false)
                {
                    unset($fileIds[$p]);
                }
            }
        }
 
        if ($fileIds)
        {
            //Search throught file id for any existing originals
            $originals = \Bitrix\Main\File\Internal\FileDuplicateTable::getList([
                'select' => ['ORIGINAL_ID'],
                'filter' => [
                    '=ORIGINAL_ID' => $fileIds,
                ],
                'order' => ['ORIGINAL_ID' => 'ASC'],
            ]);
            while ($original = $originals->fetch())
            {
                //First will be used for future deduplication
                if ($originalId === false)
                {
                    $originalId = $original['ORIGINAL_ID'];
                }
 
                //Others will be excluded from the process
                $p = array_search($original['ORIGINAL_ID'], $fileIds);
                if ($p !== false)
                {
                    unset($fileIds[$p]);
                }
            }
            //None originals exists yet
            if ($originalId === false)
            {
                $originalId = array_shift($fileIds);
            }
        }
 
        return $originalId;
    }
}