RowIterator::processCellStartingNode() - Code Metrics - Inspection of "Option to keep empty rows" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#331)

by Adrien

created 2016-10-13 21:29 UTC

RowIterator::processCellStartingNode() A

↳ Parent: RowIterator

Complexity

Conditions	1
Paths	1

Size

Total Lines	11
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	11
ccs	6
cts	6
cp	1
rs	9.4285
c	0
b	0
f	0
cc	1
eloc	6
nc	1
nop	2
crap	1

<?php

namespace Box\Spout\Reader\XLSX;

use Box\Spout\Common\Exception\IOException;
use Box\Spout\Reader\Exception\XMLProcessingException;
use Box\Spout\Reader\IteratorInterface;
use Box\Spout\Reader\Wrapper\XMLReader;
use Box\Spout\Reader\XLSX\Helper\CellHelper;
use Box\Spout\Reader\XLSX\Helper\CellValueFormatter;
use Box\Spout\Reader\XLSX\Helper\StyleHelper;

/**
 * Class RowIterator
 *
 * @package Box\Spout\Reader\XLSX
 */
class RowIterator implements IteratorInterface
{
    /** Definition of XML nodes names used to parse data */
    const XML_NODE_DIMENSION = 'dimension';
    const XML_NODE_WORKSHEET = 'worksheet';
    const XML_NODE_ROW = 'row';
    const XML_NODE_CELL = 'c';

    /** Definition of XML attributes used to parse data */
    const XML_ATTRIBUTE_REF = 'ref';
    const XML_ATTRIBUTE_SPANS = 'spans';
    const XML_ATTRIBUTE_ROW_INDEX = 'r';
    const XML_ATTRIBUTE_CELL_INDEX = 'r';

    /** @var string Path of the XLSX file being read */
    protected $filePath;

    /** @var string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml */
    protected $sheetDataXMLFilePath;

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

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

    /** @var Helper\StyleHelper $styleHelper Helper to work with styles */
    protected $styleHelper;

    /**
     * TODO: This variable can be deleted when row indices get preserved
     * @var int Number of read rows
     */
    protected $numReadRows = 0;

    /** @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 The number of columns the sheet has (0 meaning undefined) */
    protected $numColumns = 0;

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

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

    /** @var int Row index to be processed next (one-based) */
    protected $nextRowIndexToBeProcessed = 0;

    /** @var int Last column index processed (zero-based) */
    protected $lastColumnIndexProcessed = -1;

    /**
     * @param string $filePath Path of the XLSX file being read
     * @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
     * @param Helper\SharedStringsHelper $sharedStringsHelper Helper to work with shared strings
     * @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
     * @param bool $shouldPreserveEmptyRows Whether empty rows should be returned or skipped
     */
    public function __construct($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $shouldFormatDates, $shouldPreserveEmptyRows)
    {
        $this->filePath = $filePath;
        $this->sheetDataXMLFilePath = $this->normalizeSheetDataXMLFilePath($sheetDataXMLFilePath);

        $this->xmlReader = new XMLReader();

        $this->styleHelper = new StyleHelper($filePath);
        $this->cellValueFormatter = new CellValueFormatter($sharedStringsHelper, $this->styleHelper, $shouldFormatDates);

        $this->shouldPreserveEmptyRows = $shouldPreserveEmptyRows;
    }

    /**
     * @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
     * @return string Path of the XML file containing the sheet data,
     *                without the leading slash.
     */
    protected function normalizeSheetDataXMLFilePath($sheetDataXMLFilePath)
    {
        return ltrim($sheetDataXMLFilePath, '/');
    }

    /**
     * Rewind the Iterator to the first element.
     * Initializes the XMLReader object that reads the associated sheet data.
     * The XMLReader is configured to be safe from billion laughs attack.
     * @link http://php.net/manual/en/iterator.rewind.php
     *
     * @return void
     * @throws \Box\Spout\Common\Exception\IOException If the sheet data XML cannot be read
     */
    public function rewind()
    {
        $this->xmlReader->close();

        $sheetDataFilePath = 'zip://' . $this->filePath . '#' . $this->sheetDataXMLFilePath;
        if ($this->xmlReader->open($sheetDataFilePath) === false) {
            throw new IOException("Could not open \"{$this->sheetDataXMLFilePath}\".");
        }

        $this->numReadRows = 0;
        $this->lastRowIndexProcessed = 0;
        $this->nextRowIndexToBeProcessed = 0;
        $this->rowDataBuffer = null;
        $this->hasReachedEndOfFile = false;
        $this->numColumns = 0;

        $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. Reads data describing the next unprocessed row.
     * @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()
    {
        $this->nextRowIndexToBeProcessed++;

        if ($this->doesNeedDataForNextRowToBeProcessed()) {
            $this->readDataForNextRow($this->xmlReader);
        }
    }

    /**
     * Returns whether we need data for the next row to be processed.
     * We don't need to read data if:
     *   we have already read at least one row
     *     AND
     *   we need to preserve empty rows
     *     AND
     *   the last row that was read is not the row that need to be processed
     *   (i.e. if we need to return empty rows)
     *
     * @return bool Whether we need data for the next row to be processed.
     */
    protected function doesNeedDataForNextRowToBeProcessed()
    {
        $hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);

        return (
            !$hasReadAtLeastOneRow ||
            !$this->shouldPreserveEmptyRows ||
            $this->lastRowIndexProcessed < $this->nextRowIndexToBeProcessed
        );
    }

    /**
     * @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_DIMENSION)) {
                    $this->processDimensionStartingNode($xmlReader);

                } else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_ROW)) {
                    $rowData = $this->processRowStartingNode($xmlReader);

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

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

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

                    // 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_WORKSHEET)) {
                    $this->processWorksheetEndingNode();
                    break;
                }
            }

        } catch (XMLProcessingException $exception) {
            throw new IOException("The {$this->sheetDataXMLFilePath} file cannot be read. [{$exception->getMessage()}]");
        }

        $this->rowDataBuffer = $rowData;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<dimension>" starting node
     * @return void
     */
    protected function processDimensionStartingNode($xmlReader)
    {
        // Read dimensions of the sheet
        $dimensionRef = $xmlReader->getAttribute(self::XML_ATTRIBUTE_REF); // returns 'A1:M13' for instance (or 'A1' for empty sheet)
        if (preg_match('/[A-Z\d]+:([A-Z\d]+)/', $dimensionRef, $matches)) {
            $lastCellIndex = $matches[1];
            $this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
        }
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" starting node
     * @return array
     */
    protected function processRowStartingNode($xmlReader)
    {
        // Reset index of the last processed column
        $this->lastColumnIndexProcessed = -1;

        // Mark the last processed row as the one currently being read
        $this->lastRowIndexProcessed = $this->getRowIndex($xmlReader);

        // Read spans info if present
        $numberOfColumnsForRow = $this->numColumns;
        $spans = $xmlReader->getAttribute(self::XML_ATTRIBUTE_SPANS); // returns '1:5' for instance
        if ($spans) {
            list(, $numberOfColumnsForRow) = explode(':', $spans);
            $numberOfColumnsForRow = intval($numberOfColumnsForRow);
        }

        return ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
    }

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

        $node = $xmlReader->expand();
        $rowData[$currentColumnIndex] = $this->getCellValue($node);

        $this->lastColumnIndexProcessed = $currentColumnIndex;

        return $rowData;
    }

    /**
     * @param array $rowData Data of all cells read so far (key = cell index, value = cell value)
     * @return array
     */
    protected function processRowEndingNode($rowData)
    {
        $this->numReadRows++;

        // If needed, we fill the empty cells
        return ($this->numColumns !== 0) ? $rowData : CellHelper::fillMissingArrayIndexes($rowData);
    }

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

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" node
     * @return int Row index
     * @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
     */
    protected function getRowIndex($xmlReader)
    {
        // Get "r" attribute if present (from something like <row r="3"...>
        $currentRowIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_ROW_INDEX);

        return ($currentRowIndex !== null) ?
                intval($currentRowIndex) :
                $this->lastRowIndexProcessed + 1;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<c>" node
     * @return int Column index
     * @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
     */
    protected function getColumnIndex($xmlReader)
    {
        // Get "r" attribute if present (from something like <c r="A1"...>
        $currentCellIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_CELL_INDEX);

        return ($currentCellIndex !== null) ?
                CellHelper::getColumnIndexFromCellIndex($currentCellIndex) :
                $this->lastColumnIndexProcessed + 1;
    }

    /**
     * Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
     *
     * @param \DOMNode $node
     * @return string|int|float|bool|\DateTime|null The value associated with the cell (null when the cell has an error)
     */
    protected function getCellValue($node)
    {
        return $this->cellValueFormatter->extractAndFormatNodeValue($node);
    }

    /**
     * @param array $rowData
     * @return bool Whether the given row is empty
     */
    protected function isEmptyRow($rowData)
    {
        return (count($rowData) === 1 && $rowData[0] === '');
    }

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

        if ($this->shouldPreserveEmptyRows) {
            // when we need to preserve empty rows, we will either return
            // an empty row or the last row read. This depends whether the
            // index of last row that was read matches the index of the last
            // row whose value should be returned.
            if ($this->lastRowIndexProcessed !== $this->nextRowIndexToBeProcessed) {
                // return empty row if mismatch between last processed row
                // and the row that needs to be returned
                $rowDataForRowToBeProcessed = [''];
            }
        }

        return $rowDataForRowToBeProcessed;
    }

    /**
     * Return the key of the current element. Here, the row index.
     * @link http://php.net/manual/en/iterator.key.php
     *
     * @return int
     */
    public function key()
    {
        // TODO: This should return $this->nextRowIndexToBeProcessed
        //       but to avoid a breaking change, the return value for
        //       this function has been kept as the number of rows read.
        return $this->shouldPreserveEmptyRows ?
                $this->nextRowIndexToBeProcessed :
                $this->numReadRows;
    }


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


1		<?php
2
3		namespace Box\Spout\Reader\XLSX;
4
5		use Box\Spout\Common\Exception\IOException;
6		use Box\Spout\Reader\Exception\XMLProcessingException;
7		use Box\Spout\Reader\IteratorInterface;
8		use Box\Spout\Reader\Wrapper\XMLReader;
9		use Box\Spout\Reader\XLSX\Helper\CellHelper;
10		use Box\Spout\Reader\XLSX\Helper\CellValueFormatter;
11		use Box\Spout\Reader\XLSX\Helper\StyleHelper;
12
13		/**
14		* Class RowIterator
15		*
16		* @package Box\Spout\Reader\XLSX
17		*/
18		class RowIterator implements IteratorInterface
19		{
20		/** Definition of XML nodes names used to parse data */
21		const XML_NODE_DIMENSION = 'dimension';
22		const XML_NODE_WORKSHEET = 'worksheet';
23		const XML_NODE_ROW = 'row';
24		const XML_NODE_CELL = 'c';
25
26		/** Definition of XML attributes used to parse data */
27		const XML_ATTRIBUTE_REF = 'ref';
28		const XML_ATTRIBUTE_SPANS = 'spans';
29		const XML_ATTRIBUTE_ROW_INDEX = 'r';
30		const XML_ATTRIBUTE_CELL_INDEX = 'r';
31
32		/** @var string Path of the XLSX file being read */
33		protected $filePath;
34
35		/** @var string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml */
36		protected $sheetDataXMLFilePath;
37
38		/** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
39		protected $xmlReader;
40
41		/** @var Helper\CellValueFormatter Helper to format cell values */
42		protected $cellValueFormatter;
43
44		/** @var Helper\StyleHelper $styleHelper Helper to work with styles */
45		protected $styleHelper;
46
47		/**
48		* TODO: This variable can be deleted when row indices get preserved
49		* @var int Number of read rows
50		*/
51		protected $numReadRows = 0;
52
53		/** @var array\|null Buffer used to store the row data, while checking if there are more rows to read */
54		protected $rowDataBuffer = null;
55
56		/** @var bool Indicates whether all rows have been read */
57		protected $hasReachedEndOfFile = false;
58
59		/** @var int The number of columns the sheet has (0 meaning undefined) */
60		protected $numColumns = 0;
61
62		/** @var bool Whether empty rows should be returned or skipped */
63		protected $shouldPreserveEmptyRows;
64
65		/** @var int Last row index processed (one-based) */
66		protected $lastRowIndexProcessed = 0;
67
68		/** @var int Row index to be processed next (one-based) */
69		protected $nextRowIndexToBeProcessed = 0;
70
71		/** @var int Last column index processed (zero-based) */
72		protected $lastColumnIndexProcessed = -1;
73
74		/**
75		* @param string $filePath Path of the XLSX file being read
76		* @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
77		* @param Helper\SharedStringsHelper $sharedStringsHelper Helper to work with shared strings
78		* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
79		* @param bool $shouldPreserveEmptyRows Whether empty rows should be returned or skipped
80		*/
81	96	public function __construct($filePath, $sheetDataXMLFilePath, $sharedStringsHelper, $shouldFormatDates, $shouldPreserveEmptyRows)
82		{
83	96	$this->filePath = $filePath;
84	96	$this->sheetDataXMLFilePath = $this->normalizeSheetDataXMLFilePath($sheetDataXMLFilePath);
85
86	96	$this->xmlReader = new XMLReader();
87
88	96	$this->styleHelper = new StyleHelper($filePath);
89	96	$this->cellValueFormatter = new CellValueFormatter($sharedStringsHelper, $this->styleHelper, $shouldFormatDates);
90
91	96	$this->shouldPreserveEmptyRows = $shouldPreserveEmptyRows;
92	96	}
93
94		/**
95		* @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
96		* @return string Path of the XML file containing the sheet data,
97		* without the leading slash.
98		*/
99	96	protected function normalizeSheetDataXMLFilePath($sheetDataXMLFilePath)
100		{
101	96	return ltrim($sheetDataXMLFilePath, '/');
102		}
103
104		/**
105		* Rewind the Iterator to the first element.
106		* Initializes the XMLReader object that reads the associated sheet data.
107		* The XMLReader is configured to be safe from billion laughs attack.
108		* @link http://php.net/manual/en/iterator.rewind.php
109		*
110		* @return void
111		* @throws \Box\Spout\Common\Exception\IOException If the sheet data XML cannot be read
112		*/
113	93	public function rewind()
114		{
115	93	$this->xmlReader->close();
116
117	93	$sheetDataFilePath = 'zip://' . $this->filePath . '#' . $this->sheetDataXMLFilePath;
118	93	if ($this->xmlReader->open($sheetDataFilePath) === false) {
119	3	throw new IOException("Could not open \"{$this->sheetDataXMLFilePath}\".");
120		}
121
122	90	$this->numReadRows = 0;
123	90	$this->lastRowIndexProcessed = 0;
124	90	$this->nextRowIndexToBeProcessed = 0;
125	90	$this->rowDataBuffer = null;
126	90	$this->hasReachedEndOfFile = false;
127	90	$this->numColumns = 0;
128
129	90	$this->next();
130	90	}
131
132		/**
133		* Checks if current position is valid
134		* @link http://php.net/manual/en/iterator.valid.php
135		*
136		* @return bool
137		*/
138	90	public function valid()
139		{
140	90	return (!$this->hasReachedEndOfFile);
141		}
142
143		/**
144		* Move forward to next element. Reads data describing the next unprocessed row.
145		* @link http://php.net/manual/en/iterator.next.php
146		*
147		* @return void
148		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
149		* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
150		*/
151	90	public function next()
152		{
153	90	$this->nextRowIndexToBeProcessed++;
154
155	90	if ($this->doesNeedDataForNextRowToBeProcessed()) {
156	90	$this->readDataForNextRow($this->xmlReader);
157	90	}
158	90	}
159
160		/**
161		* Returns whether we need data for the next row to be processed.
162		* We don't need to read data if:
163		* we have already read at least one row
164		* AND
165		* we need to preserve empty rows
166		* AND
167		* the last row that was read is not the row that need to be processed
168		* (i.e. if we need to return empty rows)
169		*
170		* @return bool Whether we need data for the next row to be processed.
171		*/
172	90	protected function doesNeedDataForNextRowToBeProcessed()
173		{
174	90	$hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);
175
176		return (
177	90	!$hasReadAtLeastOneRow \|\|
178	84	!$this->shouldPreserveEmptyRows \|\|
179	3	$this->lastRowIndexProcessed < $this->nextRowIndexToBeProcessed
180	90	);
181		}
182
183		/**
184		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object
185		* @return void
186		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
187		* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
188		*/
189	90	protected function readDataForNextRow($xmlReader)
190		{
191	90	$rowData = [];
192
193		try {
194	90	while ($xmlReader->read()) {
195	90	if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_DIMENSION)) {
196	45	$this->processDimensionStartingNode($xmlReader);
197
198	90	} else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_ROW)) {
199	87	$rowData = $this->processRowStartingNode($xmlReader);
200
201	90	} else if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
202	87	$rowData = $this->processCellStartingNode($xmlReader, $rowData);
203
204	90	} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
205		// if the fetched row is empty and we don't want to preserve it..,
206	87	if (!$this->shouldPreserveEmptyRows && $this->isEmptyRow($rowData)) {
207		// ... skip it
208		continue;
209		}
210
211	87	$rowData = $this->processRowEndingNode($rowData);
212
213		// at this point, we have all the data we need for the row
214		// so that we can populate the buffer
215	87	break;
216
217	90	} else if ($xmlReader->isPositionedOnEndingNode(self::XML_NODE_WORKSHEET)) {
218	87	$this->processWorksheetEndingNode();
219	87	break;
220		}
221	90	}
222
223	90	} catch (XMLProcessingException $exception) {
224		throw new IOException("The {$this->sheetDataXMLFilePath} file cannot be read. [{$exception->getMessage()}]");
225		}
226
227	90	$this->rowDataBuffer = $rowData;
228	90	}
229
230		/**
231		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<dimension>" starting node
232		* @return void
233		*/
234	45	protected function processDimensionStartingNode($xmlReader)
235		{
236		// Read dimensions of the sheet
237	45	$dimensionRef = $xmlReader->getAttribute(self::XML_ATTRIBUTE_REF); // returns 'A1:M13' for instance (or 'A1' for empty sheet)
238	45	if (preg_match('/[A-Z\d]+:([A-Z\d]+)/', $dimensionRef, $matches)) {
239	36	$lastCellIndex = $matches[1];
240	36	$this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
241	36	}
242	45	}
243
244		/**
245		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" starting node
246		* @return array
247		*/
248	87	protected function processRowStartingNode($xmlReader)
249		{
250		// Reset index of the last processed column
251	87	$this->lastColumnIndexProcessed = -1;
252
253		// Mark the last processed row as the one currently being read
254	87	$this->lastRowIndexProcessed = $this->getRowIndex($xmlReader);
255
256		// Read spans info if present
257	87	$numberOfColumnsForRow = $this->numColumns;
258	87	$spans = $xmlReader->getAttribute(self::XML_ATTRIBUTE_SPANS); // returns '1:5' for instance
259	87	if ($spans) {
260	30	list(, $numberOfColumnsForRow) = explode(':', $spans);
261	30	$numberOfColumnsForRow = intval($numberOfColumnsForRow);
262	30	}
263
264	87	return ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
265		}
266
267		/**
268		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<cell>" starting node
269		* @param array $rowData Data of all cells read so far (key = cell index, value = cell value)
270		* @return array Original row data + data for the cell that was just read (key = cell index, value = cell value)
271		*/
272	87	protected function processCellStartingNode($xmlReader, $rowData)
273		{
274	87	$currentColumnIndex = $this->getColumnIndex($xmlReader);
275
276	87	$node = $xmlReader->expand();
277	87	$rowData[$currentColumnIndex] = $this->getCellValue($node);
278
279	87	$this->lastColumnIndexProcessed = $currentColumnIndex;
280
281	87	return $rowData;
282		}
283
284		/**
285		* @param array $rowData Data of all cells read so far (key = cell index, value = cell value)
286		* @return array
287		*/
288	87	protected function processRowEndingNode($rowData)
289		{
290	87	$this->numReadRows++;
291
292		// If needed, we fill the empty cells
293	87	return ($this->numColumns !== 0) ? $rowData : CellHelper::fillMissingArrayIndexes($rowData);
294		}
295
296		/**
297		* @return void
298		*/
299	87	protected function processWorksheetEndingNode()
300		{
301		// The closing "</worksheet>" marks the end of the file
302	87	$this->hasReachedEndOfFile = true;
303	87	}
304
305		/**
306		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" node
307		* @return int Row index
308		* @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
309		*/
310	87	protected function getRowIndex($xmlReader)
311		{
312		// Get "r" attribute if present (from something like <row r="3"...>
313	87	$currentRowIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_ROW_INDEX);
314
315	87	return ($currentRowIndex !== null) ?
316	87	intval($currentRowIndex) :
317	87	$this->lastRowIndexProcessed + 1;
318		}
319
320		/**
321		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<c>" node
322		* @return int Column index
323		* @throws \Box\Spout\Common\Exception\InvalidArgumentException When the given cell index is invalid
324		*/
325	87	protected function getColumnIndex($xmlReader)
326		{
327		// Get "r" attribute if present (from something like <c r="A1"...>
328	87	$currentCellIndex = $xmlReader->getAttribute(self::XML_ATTRIBUTE_CELL_INDEX);
329
330	87	return ($currentCellIndex !== null) ?
331	87	CellHelper::getColumnIndexFromCellIndex($currentCellIndex) :
332	87	$this->lastColumnIndexProcessed + 1;
333		}
334
335		/**
336		* Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
337		*
338		* @param \DOMNode $node
339		* @return string\|int\|float\|bool\|\DateTime\|null The value associated with the cell (null when the cell has an error)
340		*/
341	87	protected function getCellValue($node)
342		{
343	87	return $this->cellValueFormatter->extractAndFormatNodeValue($node);
344		}
345
346		/**
347		* @param array $rowData
348		* @return bool Whether the given row is empty
349		*/
350	84	protected function isEmptyRow($rowData)
351		{
352	84	return (count($rowData) === 1 && $rowData[0] === '');
353		}
354
355		/**
356		* Return the current element, either an empty row or from the buffer.
357		* @link http://php.net/manual/en/iterator.current.php
358		*
359		* @return array\|null
360		*/
361	87	public function current()
362		{
363	87	$rowDataForRowToBeProcessed = $this->rowDataBuffer;
364
365	87	if ($this->shouldPreserveEmptyRows) {
366		// when we need to preserve empty rows, we will either return
367		// an empty row or the last row read. This depends whether the
368		// index of last row that was read matches the index of the last
369		// row whose value should be returned.
370	3	if ($this->lastRowIndexProcessed !== $this->nextRowIndexToBeProcessed) {
371		// return empty row if mismatch between last processed row
372		// and the row that needs to be returned
373	3	$rowDataForRowToBeProcessed = [''];
374	3	}
375	3	}
376
377	87	return $rowDataForRowToBeProcessed;
378		}
379
380		/**
381		* Return the key of the current element. Here, the row index.
382		* @link http://php.net/manual/en/iterator.key.php
383		*
384		* @return int
385		*/
386	84	public function key()
387		{
388		// TODO: This should return $this->nextRowIndexToBeProcessed
389		// but to avoid a breaking change, the return value for
390		// this function has been kept as the number of rows read.
391	84	return $this->shouldPreserveEmptyRows ?
392	84	$this->nextRowIndexToBeProcessed :
393	84	$this->numReadRows;
394		}
395
396
397		/**
398		* Cleans up what was created to iterate over the object.
399		*
400		* @return void
401		*/
402	93	public function end()
403		{
404	93	$this->xmlReader->close();
405	93	}
406		}
407

box / spout

Pull Request — master (#331)

RowIterator::processCellStartingNode() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like