csvfile.php

См. документацию.

<?php
 
namespace Bitrix\Translate\IO;
 
use Bitrix\Main;
use Bitrix\Translate;
 
class CsvFile
    extends Translate\IO\File
{
    // fields type with delimiter,
    public const FIELDS_TYPE_FIXED_WIDTH = 'F';
    // fields type fixed width
    public const FIELDS_TYPE_WITH_DELIMITER = 'R';
 
    public const ERROR_32K_FIELD_LENGTH = '32k_field_length';

    protected $fieldsType = self::FIELDS_TYPE_WITH_DELIMITER;
 
    // field delimiter
    public const DELIMITER_TAB = "\t";
    public const DELIMITER_ZPT = ',';
    public const DELIMITER_SPS = ' ';
    public const DELIMITER_TZP = ';';

    protected $fieldDelimiter = self::DELIMITER_TZP;
 
    // UTF-8 Byte-Order Mark
    public const BOM_TYPE_UTF8 = "\xEF\xBB\xBF";

    protected $bomMark = self::BOM_TYPE_UTF8;

    protected $hasBom = false;
 
    public const LINE_DELIMITER_WIN = "\r\n";
    public const LINE_DELIMITER_UNIX = "\r";

    protected $rowDelimiter = self::LINE_DELIMITER_WIN;

    protected $widthMap = [];

    protected $firstHeader = false;

    private $fileSize;
    private $currentPosition = 0;
    private $buffer = '';
    private $bufferPosition = 0;
    private $bufferSize = 0;
 

    public function openLoad(): bool
    {
        if ($this->isExists())
        {
            $this->open(Main\IO\FileStreamOpenMode::READ);
 
            $this->fileSize = $this->getSize();
            $this->checkUtf8Bom();
        }
 
        return $this->isExists() && $this->isReadable();
    }

    public function openWrite(string $mode = Main\IO\FileStreamOpenMode::WRITE): bool 
    {
        $this->open($mode);
 
        if (\is_resource($this->filePointer))
        {
            if ($mode === Main\IO\FileStreamOpenMode::WRITE)
            {
                $this->fileSize = 0;
                if ($this->hasBom)
                {
                    $this->fileSize = $this->write($this->bomMark);
                }
            }
            else
            {
                $this->fileSize = $this->getSize();
            }
 
            return true;
        }
 
        return false;
    }

    public function setUtf8Bom(string $mark = self::BOM_TYPE_UTF8): self
    {
        $this->bomMark = $mark;
 
        return $this;
    }

    public function hasUtf8Bom(): bool
    {
        return $this->hasBom;
    }

    public function prefaceWithUtf8Bom(bool $exists = true): self
    {
        $this->hasBom = $exists;
 
        return $this;
    }

    public function checkUtf8Bom(): bool
    {
        $this->seek(0);
        $bom = $this->read(strlen($this->bomMark));
        if($bom === $this->bomMark)
        {
            $this->hasBom = true;
        }
 
        if ($this->hasBom)
        {
            $this->seek(strlen($this->bomMark));
        }
        else
        {
            $this->seek(0);
        }
 
        return $this->hasBom;
    }

    public function setFieldsType(string $fieldsType = self::FIELDS_TYPE_WITH_DELIMITER): self
    {
        $this->fieldsType =
            ($fieldsType === self::FIELDS_TYPE_FIXED_WIDTH ? self::FIELDS_TYPE_FIXED_WIDTH : self::FIELDS_TYPE_WITH_DELIMITER);
 
        return $this;
    }

    public function setFieldDelimiter(string $fieldDelimiter = self::DELIMITER_TZP): self
    {
        $this->fieldDelimiter = (\mb_strlen($fieldDelimiter) > 1 ? \mb_substr($fieldDelimiter, 0, 1) : $fieldDelimiter);
 
        return $this;
    }

    public function setRowDelimiter(string $rowDelimiter = self::LINE_DELIMITER_WIN): self
    {
        $this->rowDelimiter = $rowDelimiter;
 
        return $this;
    }

    public function setFirstHeader(bool $firstHeader = false): self
    {
        $this->firstHeader = $firstHeader;
 
        return $this;
    }

    public function getFirstHeader(): bool 
    {
        return $this->firstHeader;
    }

    public function setWidthMap(array $mapFields): self
    {
        $this->widthMap = [];
        for ($i = 0, $n = \count($mapFields); $i < $n; $i++)
        {
            $this->widthMap[$i] = (int)$mapFields[$i];
        }
 
        return $this;
    }

    protected function fetchDelimiter(): ?array
    {
        $isInside = false;
        $str = '';
        $result = [];
        while ($this->currentPosition <= $this->fileSize)
        {
            $ch = $this->buffer[$this->bufferPosition];
            if ($ch === "\r" || $ch === "\n")
            {
                if (!$isInside)
                {
                    while ($this->currentPosition <= $this->fileSize)
                    {
                        $this->incrementCurrentPosition();
                        $ch = $this->buffer[$this->bufferPosition];
                        if ($ch !== "\r" && $ch !== "\n")
                        {
                            break;
                        }
                    }
                    if ($this->firstHeader)
                    {
                        $this->firstHeader = false;
                        $result = [];
                        $str = '';
                        continue;
                    }
 
                    $result[] = $str;
 
                    return $result;
                }
            }
            elseif ($ch === "\"")
            {
                if (!$isInside)
                {
                    $isInside = true;
                    $this->incrementCurrentPosition();
                    continue;
                }
 
                $this->incrementCurrentPosition();
                if ($this->buffer[$this->bufferPosition] !== "\"")
                {
                    $isInside = false;
                    continue;
                }
            }
            elseif ($ch === $this->fieldDelimiter)
            {
                if (!$isInside)
                {
                    $result[] = $str;
                    $str = '';
                    $this->incrementCurrentPosition();
                    continue;
                }
            }
 
            //inline "call"
            $this->currentPosition ++;
            $this->bufferPosition ++;
            if ($this->bufferPosition >= $this->bufferSize)
            {
                $this->buffer = $this->read(1024 * 1024);
                $this->bufferSize = strlen($this->buffer);
                $this->bufferPosition = 0;
            }
 
            $str .= $ch;
        }
 
        if ($str !== '')
        {
            $result[] = $str;
        }
 
        if ($result === [])
        {
            $result = null;
        }
 
        return $result;
    }

    protected function fetchWidth(): ?array
    {
        $str = '';
        $ind = 1;
        $jnd = 0;
        $result = [];
 
        while ($this->currentPosition <= $this->fileSize)
        {
            $ch = $this->buffer[$this->bufferPosition];
            if ($ch === "\r" || $ch === "\n")
            {
                while ($this->currentPosition <= $this->fileSize)
                {
                    $this->incrementCurrentPosition();
                    $ch = $this->buffer[$this->bufferPosition];
                    if ($ch !== "\r" && $ch !== "\n")
                    {
                        break;
                    }
                }
                if ($this->firstHeader)
                {
                    $this->firstHeader = false;
                    $result = [];
                    $ind = 1;
                    $str = '';
                    continue;
                }
 
                $result[] = $str;
 
                return $result;
            }
            if ($ind === $this->widthMap[$jnd])
            {
                $result[] = $str. $ch;
                $str = '';
                $this->incrementCurrentPosition();
                $ind ++;
                $jnd ++;
                continue;
            }
 
            //inline "call"
            $this->currentPosition ++;
            $this->bufferPosition ++;
            if($this->bufferPosition >= $this->bufferSize)
            {
                $this->buffer = $this->read( 1024 * 1024);
                $this->bufferSize = strlen($this->buffer);
                $this->bufferPosition = 0;
            }
 
            $ind ++;
            $str .= $ch;
        }
 
        if ($str !== '')
        {
            $result[] = $str;
        }
 
        if ($result === [])
        {
            $result = null;
        }
 
        return $result;
    }

    public function fetch(): ?array
    {
        if ($this->fieldsType === self::FIELDS_TYPE_WITH_DELIMITER)
        {
            if ($this->fieldDelimiter === '')
            {
                return null;
            }
 
            return $this->fetchDelimiter();
        }
 
        if (empty($this->widthMap))
        {
            return null;
        }
 
        return $this->fetchWidth();
    }

    protected function incrementCurrentPosition(): void
    {
        $this->currentPosition ++;
        $this->bufferPosition ++;
        if ($this->bufferPosition >= $this->bufferSize)
        {
            $this->buffer = $this->read( 1024 * 1024);
            $this->bufferSize = strlen($this->buffer);
            $this->bufferPosition = 0;
        }
    }

    public function moveFirst(): void
    {
        $this->setPos(0);
    }

    public function getPos(): int
    {
        return $this->currentPosition;
    }

    public function setPos(int $position = 0): void
    {
        if ($position <= $this->fileSize)
        {
            $this->currentPosition = $position;
        }
        else
        {
            $this->currentPosition = $this->fileSize;
        }
 
        $pos = $this->currentPosition;
        if($this->hasBom)
        {
            $pos += 3;
        }
        $this->seek($pos);
 
        $this->buffer = $this->read(1024 * 1024);
 
        $this->bufferSize = strlen($this->buffer);
        $this->bufferPosition = 0;
    }

    public function put(array $fields): bool
    {
        $length = false;
        $throw32KWarning = false;
        if ($this->fieldsType == self::FIELDS_TYPE_WITH_DELIMITER)
        {
            $content = '';
            for ($i = 0, $n = \count($fields); $i < $n; $i++)
            {
                if ($i>0)
                {
                    $content .= $this->fieldDelimiter;
                }
                //$pos1 = strpos($fields[$i], $this->fieldDelimiter);
                //$pos2 = $pos1 || strpos($fields[$i], "\"");
                //$pos3 = $pos2 || strpos($fields[$i], "\n");
                //$pos4 = $pos3 || strpos($fields[$i], "\r");
                //if ($pos1 !== false || $pos2 !== false || $pos3 !== false || $pos4 !== false)
                if ($fields[$i] === null)
                {
                    $fields[$i] = '';
                }
                elseif (\preg_match("#[\"\n\r]+#u", $fields[$i]))
                {
                    $fields[$i] = \str_replace("\"", "\"\"", $fields[$i]);
                    //$fields[$i] = str_replace("\\", "\\\\", $fields[$i]);
                }
                $content .= "\"";
                $content .= $fields[$i];
                $content .= "\"";
 
                // ms excel las limitation with total number of characters that a cell can contain 32767 characters
                if ($throw32KWarning !== true && strlen($fields[$i]) > 32767)
                {
                    $throw32KWarning = true;
                }
            }
            if ($content !== '')
            {
                $content .= $this->rowDelimiter;
 
                $length = $this->write($content);
                if ($length !== false)
                {
                    $this->fileSize += $length;
                }
            }
        }
        // todo: $this->fieldsType == self::FIELDS_TYPE_FIXED_WIDTH
 
        if ($throw32KWarning)
        {
            if (!$this->hasError(self::ERROR_32K_FIELD_LENGTH))
            {
                $this->addError(new Main\Error(
                    'Excel has limit when the total number of characters that a cell can contain is 32767 characters.',
                    self::ERROR_32K_FIELD_LENGTH
                ));
            }
        }
 
        return ($length !== false);
    }
}