Box\Spout\Reader\ODS\RowIterator

Complexity

Total Complexity

37

Size/Duplication

Total Lines	339
Duplicated Lines	0 %

Coupling/Cohesion

Components	1
Dependencies	2

Test Coverage

Coverage

95.96%

Importance

Changes

0

17 Methods

Rating	Name	Duplication	Size	Complexity
A	__construct()	0	6	1
A	rewind()	0	17	2
A	valid()	0	4	1
A	next()	0	8	2
A	doesNeedDataForNextRowToBeProcessed()	0	9	2
D	readDataForNextRow()	0	39	9
A	processRowStartingNode()	0	8	1
A	processCellStartingNode()	0	20	3
B	processRowEndingNode()	0	23	4
A	processTableEndingNode()	0	5	1
A	getNumRowsRepeatedForCurrentNode()	0	5	2
A	getNumColumnsRepeatedForCurrentNode()	0	5	2
A	getCellValue()	0	4	1
A	isEmptyRow()	0	7	3
A	current()	0	4	1
A	key()	0	4	1
A	end()	0	4	1

<?php

namespace Box\Spout\Reader\ODS;

use Box\Spout\Common\Exception\IOException;
use Box\Spout\Reader\Exception\IteratorNotRewindableException;
use Box\Spout\Reader\Exception\XMLProcessingException;
use Box\Spout\Reader\IteratorInterface;
use Box\Spout\Reader\ODS\Helper\CellValueFormatter;
use Box\Spout\Reader\Wrapper\XMLReader;

/**
 * Class RowIterator
 *
 * @package Box\Spout\Reader\ODS
 */
class RowIterator implements IteratorInterface
{
    /** Definition of XML nodes names used to parse data */
    const XML_NODE_TABLE = 'table:table';
    const XML_NODE_ROW = 'table:table-row';
    const XML_NODE_CELL = 'table:table-cell';
    const MAX_COLUMNS_EXCEL = 16384;

    /** Definition of XML attribute used to parse data */
    const XML_ATTRIBUTE_NUM_ROWS_REPEATED = 'table:number-rows-repeated';
    const XML_ATTRIBUTE_NUM_COLUMNS_REPEATED = 'table:number-columns-repeated';

    /** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
    protected $xmlReader;

    /** @var bool Whether empty rows should be returned or skipped */
    protected $shouldPreserveEmptyRows;

    /** @var Helper\CellValueFormatter Helper to format cell values */
    protected $cellValueFormatter;

    /** @var bool Whether the iterator has already been rewound once */
    protected $hasAlreadyBeenRewound = false;

    /** @var array|null Buffer used to store the row data, while checking if there are more rows to read */
    protected $rowDataBuffer = null;

    /** @var bool Indicates whether all rows have been read */
    protected $hasReachedEndOfFile = false;

    /** @var int Last row index processed (one-based) */
    protected $lastRowIndexProcessed = 0;

    /** @var int Row index to be processed next (one-based) */
    protected $nextRowIndexToBeProcessed = 1;

    /** @var mixed|null Value of the last processed cell (because when reading cell at column N+1, cell N is processed) */
    protected $lastProcessedCellValue = null;

    /** @var int Number of times the last processed row should be repeated */
    protected $numRowsRepeated = 1;

    /** @var int Number of times the last cell value should be copied to the cells on its right */
    protected $numColumnsRepeated = 1;

    /** @var bool Whether at least one cell has been read for the row currently being processed */
    protected $hasAlreadyReadOneCellInCurrentRow = false;


    /**
     * @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
     * @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
     * @param bool $shouldPreserveEmptyRows Whether empty rows should be returned or skipped
     */
    public function __construct($xmlReader, $shouldFormatDates, $shouldPreserveEmptyRows)
    {
        $this->xmlReader = $xmlReader;
        $this->shouldPreserveEmptyRows = $shouldPreserveEmptyRows;
        $this->cellValueFormatter = new CellValueFormatter($shouldFormatDates);
    }

    /**
     * Rewind the Iterator to the first element.
     * NOTE: It can only be done once, as it is not possible to read an XML file backwards.
     * @link http://php.net/manual/en/iterator.rewind.php
     *
     * @return void
     * @throws \Box\Spout\Reader\Exception\IteratorNotRewindableException If the iterator is rewound more than once
     */
    public function rewind()
    {
        // Because sheet and row data is located in the file, we can't rewind both the
        // sheet iterator and the row iterator, as XML file cannot be read backwards.
        // Therefore, rewinding the row iterator has been disabled.
        if ($this->hasAlreadyBeenRewound) {
            throw new IteratorNotRewindableException();
        }

        $this->hasAlreadyBeenRewound = true;
        $this->lastRowIndexProcessed = 0;
        $this->nextRowIndexToBeProcessed = 1;
        $this->rowDataBuffer = null;
        $this->hasReachedEndOfFile = false;

        $this->next();
    }

    /**
     * Checks if current position is valid
     * @link http://php.net/manual/en/iterator.valid.php
     *
     * @return bool
     */
    public function valid()
    {
        return (!$this->hasReachedEndOfFile);
    }

    /**
     * Move forward to next element. Empty rows will be skipped.
     * @link http://php.net/manual/en/iterator.next.php
     *
     * @return void
     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
     * @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
     */
    public function next()
    {
        if ($this->doesNeedDataForNextRowToBeProcessed()) {
            $this->readDataForNextRow($this->xmlReader);
        }

        $this->lastRowIndexProcessed++;
    }

    /**
     * Returns whether we need data for the next row to be processed.
     * We DO need to read data if:
     *   - we have not read any rows yet
     *      OR
     *   - the next row to be processed immediately follows the last read row
     *
     * @return bool Whether we need data for the next row to be processed.
     */
    protected function doesNeedDataForNextRowToBeProcessed()
    {
        $hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);

        return (
            !$hasReadAtLeastOneRow ||
            $this->lastRowIndexProcessed === $this->nextRowIndexToBeProcessed - 1
        );
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object
     * @return void
     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
     * @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
     */
    protected function readDataForNextRow($xmlReader)
    {
        $rowData = [];

        try {
            while ($xmlReader->read()) {
                if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_ROW)) {
                    $this->processRowStartingNode($xmlReader);

                } else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
                    $rowData = $this->processCellStartingNode($xmlReader, $rowData);

                } else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
                    $isEmptyRow = $this->isEmptyRow($rowData, $this->lastProcessedCellValue);

                    // if the fetched row is empty and we don't want to preserve it...
                    if (!$this->shouldPreserveEmptyRows && $isEmptyRow) {
                        // ... skip it
                        continue;
                    }

                    $rowData = $this->processRowEndingNode($rowData, $isEmptyRow);

                    // at this point, we have all the data we need for the row
                    // so that we can populate the buffer
                    break;

                } else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_TABLE)) {
                    $this->processTableEndingNode();
                    break;
                }
            }

        } catch (XMLProcessingException $exception) {
            throw new IOException("The sheet's data cannot be read. [{$exception->getMessage()}]");
        }

        $this->rowDataBuffer = $rowData;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-row>" starting node
     * @return void
     */
    protected function processRowStartingNode($xmlReader)
    {
        // Reset data from current row
        $this->hasAlreadyReadOneCellInCurrentRow = false;
        $this->lastProcessedCellValue = null;
        $this->numColumnsRepeated = 1;
        $this->numRowsRepeated = $this->getNumRowsRepeatedForCurrentNode($xmlReader);
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-cell>" starting node
     * @param array $rowData Data of all cells read so far
     * @return array Original row data + data for the cell that was just read
     */
    protected function processCellStartingNode($xmlReader, $rowData)
    {
        $currentNumColumnsRepeated = $this->getNumColumnsRepeatedForCurrentNode($xmlReader);

        $node = $xmlReader->expand();
        $currentCellValue = $this->getCellValue($node);

        // process cell N only after having read cell N+1 (see below why)
        if ($this->hasAlreadyReadOneCellInCurrentRow) {
            for ($i = 0; $i < $this->numColumnsRepeated; $i++) {
                $rowData[] = $this->lastProcessedCellValue;
            }
        }

        $this->hasAlreadyReadOneCellInCurrentRow = true;
        $this->lastProcessedCellValue = $currentCellValue;
        $this->numColumnsRepeated = $currentNumColumnsRepeated;

        return $rowData;
    }

    /**
     * @param array $rowData Data of all cells read so far
     * @param bool $isEmptyRow Whether the given row is empty
     * @return array
     */
    protected function processRowEndingNode($rowData, $isEmptyRow)
    {
        // if the row is empty, we don't want to return more than one cell
        $actualNumColumnsRepeated = (!$isEmptyRow) ? $this->numColumnsRepeated : 1;

        // Only add the value if the last read cell is not a trailing empty cell repeater in Excel.
        // The current count of read columns is determined by counting the values in $rowData.
        // This is to avoid creating a lot of empty cells, as Excel adds a last empty "<table:table-cell>"
        // with a number-columns-repeated value equals to the number of (supported columns - used columns).
        // In Excel, the number of supported columns is 16384, but we don't want to returns rows with
        // always 16384 cells.
        if ((count($rowData) + $actualNumColumnsRepeated) !== self::MAX_COLUMNS_EXCEL) {
            for ($i = 0; $i < $actualNumColumnsRepeated; $i++) {
                $rowData[] = $this->lastProcessedCellValue;
            }
        }

        // If we are processing row N and the row is repeated M times,
        // then the next row to be processed will be row (N+M).
        $this->nextRowIndexToBeProcessed += $this->numRowsRepeated;

        return $rowData;
    }

    /**
     * @return void
     */
    protected function processTableEndingNode()
    {
        // The closing "</table:table>" marks the end of the file
        $this->hasReachedEndOfFile = true;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-row>" starting node
     * @return int The value of "table:number-rows-repeated" attribute of the current node, or 1 if attribute missing
     */
    protected function getNumRowsRepeatedForCurrentNode($xmlReader)
    {
        $numRowsRepeated = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_ROWS_REPEATED);
        return ($numRowsRepeated !== null) ? intval($numRowsRepeated) : 1;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<table:table-cell>" starting node
     * @return int The value of "table:number-columns-repeated" attribute of the current node, or 1 if attribute missing
     */
    protected function getNumColumnsRepeatedForCurrentNode($xmlReader)
    {
        $numColumnsRepeated = $xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_COLUMNS_REPEATED);
        return ($numColumnsRepeated !== null) ? intval($numColumnsRepeated) : 1;
    }

    /**
     * Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
     *
     * @param \DOMNode $node
     * @return string|int|float|bool|\DateTime|\DateInterval|null The value associated with the cell, empty string if cell's type is void/undefined, null on error
     */
    protected function getCellValue($node)
    {
        return $this->cellValueFormatter->extractAndFormatNodeValue($node);
    }

    /**
     * After finishing processing each cell, a row is considered empty if it contains
     * no cells or if the value of the last read cell is an empty string.
     * After finishing processing each cell, the last read cell is not part of the
     * row data yet (as we still need to apply the "num-columns-repeated" attribute).
     *
     * @param array $rowData
     * @param string|int|float|bool|\DateTime|\DateInterval|null The value of the last read cell
     * @return bool Whether the row is empty
     */
    protected function isEmptyRow($rowData, $lastReadCellValue)
    {
        return (
            count($rowData) === 0 &&
            (!isset($lastReadCellValue) || trim($lastReadCellValue) === '')
        );
    }

    /**
     * Return the current element, from the buffer.
     * @link http://php.net/manual/en/iterator.current.php
     *
     * @return array|null
     */
    public function current()
    {
        return $this->rowDataBuffer;
    }

    /**
     * Return the key of the current element
     * @link http://php.net/manual/en/iterator.key.php
     *
     * @return int
     */
    public function key()
    {
        return $this->lastRowIndexProcessed;
    }


    /**
     * Cleans up what was created to iterate over the object.
     *
     * @return void
     */
    public function end()
    {
        $this->xmlReader->close();
    }
}