RowIterator::processCellStartingNode() - Code Metrics - Inspection of "Use openFileInZip() only (#421)" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 606103...99816b )

by Adrien

created 2017-05-22 12:39 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	1
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;
use Box\Spout\Reader\Common\XMLProcessor;

/**
 * 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 \Box\Spout\Reader\Common\XMLProcessor Helper Object to process XML nodes */
    protected $xmlProcessor;

    /** @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 Contains the data for the currently processed row (key = cell index, value = cell value) */
    protected $currentlyProcessedRowData = [];

    /** @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 \Box\Spout\Reader\XLSX\ReaderOptions $options Reader's current options
     * @param Helper\SharedStringsHelper $sharedStringsHelper Helper to work with shared strings
     */
    public function __construct($filePath, $sheetDataXMLFilePath, $options, $sharedStringsHelper)
    {
        $this->filePath = $filePath;
        $this->sheetDataXMLFilePath = $this->normalizeSheetDataXMLFilePath($sheetDataXMLFilePath);

        $this->xmlReader = new XMLReader();

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

        $this->shouldPreserveEmptyRows = $options->shouldPreserveEmptyRows();

        // Register all callbacks to process different nodes when reading the XML file
        $this->xmlProcessor = new XMLProcessor($this->xmlReader);
        $this->xmlProcessor->registerCallback(self::XML_NODE_DIMENSION, XMLProcessor::NODE_TYPE_START, [$this, 'processDimensionStartingNode']);
        $this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_START, [$this, 'processRowStartingNode']);
        $this->xmlProcessor->registerCallback(self::XML_NODE_CELL, XMLProcessor::NODE_TYPE_START, [$this, 'processCellStartingNode']);
        $this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_END, [$this, 'processRowEndingNode']);
        $this->xmlProcessor->registerCallback(self::XML_NODE_WORKSHEET, XMLProcessor::NODE_TYPE_END, [$this, 'processWorksheetEndingNode']);
    }

    /**
     * @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();

        if ($this->xmlReader->openFileInZip($this->filePath, $this->sheetDataXMLFilePath) === false) {
            throw new IOException("Could not open \"{$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();
        }
    }

    /**
     * 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
        );
    }

    /**
     * @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()
    {
        $this->currentlyProcessedRowData = [];

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

        $this->rowDataBuffer = $this->currentlyProcessedRowData;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<dimension>" starting node
     * @return int A return code that indicates what action should the processor take next
     */
    protected function processDimensionStartingNode($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)) {
            $this->numColumns = CellHelper::getColumnIndexFromCellIndex($matches[1]) + 1;
        }

        return XMLProcessor::PROCESSING_CONTINUE;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" starting node
     * @return int A return code that indicates what action should the processor take next
     */
    protected function processRowStartingNode($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);
        }

        $this->currentlyProcessedRowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];

        return XMLProcessor::PROCESSING_CONTINUE;
    }

    /**
     * @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<cell>" starting node
     * @return int A return code that indicates what action should the processor take next
     */
    protected function processCellStartingNode($xmlReader)
    {
        $currentColumnIndex = $this->getColumnIndex($xmlReader);

        // NOTE: expand() will automatically decode all XML entities of the child nodes
        $node = $xmlReader->expand();
        $this->currentlyProcessedRowData[$currentColumnIndex] = $this->getCellValue($node);
        $this->lastColumnIndexProcessed = $currentColumnIndex;

        return XMLProcessor::PROCESSING_CONTINUE;
    }

    /**
     * @return int A return code that indicates what action should the processor take next
     */
    protected function processRowEndingNode()
    {
        // if the fetched row is empty and we don't want to preserve it..,
        if (!$this->shouldPreserveEmptyRows && $this->isEmptyRow($this->currentlyProcessedRowData)) {
            // ... skip it
            return XMLProcessor::PROCESSING_CONTINUE;
        }

        $this->numReadRows++;

        // If needed, we fill the empty cells
        if ($this->numColumns === 0) {
            $this->currentlyProcessedRowData = CellHelper::fillMissingArrayIndexes($this->currentlyProcessedRowData);
        }

        // at this point, we have all the data we need for the row
        // so that we can populate the buffer
        return XMLProcessor::PROCESSING_STOP;
    }

    /**
     * @return int A return code that indicates what action should the processor take next
     */
    protected function processWorksheetEndingNode()
    {
        // The closing "</worksheet>" marks the end of the file
        $this->hasReachedEndOfFile = true;

        return XMLProcessor::PROCESSING_STOP;
    }

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

box / spout

Push — master ( 606103...99816b )

RowIterator::processCellStartingNode() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like