RowIterator::processCellStartingNode() - Code Metrics - Inspection of "Introduce XMLProcessor to reduce ODS,XLSX readers'..." - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — master ( 73d5d0...a19231 )

by Adrien

created 2016-10-18 23:28 UTC

RowIterator::processCellStartingNode() A

↳ Parent: RowIterator

Complexity

Conditions	1
Paths	1

Size

Total Lines	10
Code Lines	6

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	10
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();

        $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();
        }
    }

    /**
     * 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)) {
            $lastCellIndex = $matches[1];
            $this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 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);

        $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 $sharedStringsHelper Helper to work with shared strings
86		*/
87	96	public function __construct($filePath, $sheetDataXMLFilePath, $options, $sharedStringsHelper)
88		{
89	96	$this->filePath = $filePath;
90	96	$this->sheetDataXMLFilePath = $this->normalizeSheetDataXMLFilePath($sheetDataXMLFilePath);
91
92	96	$this->xmlReader = new XMLReader();
93
94	96	$this->styleHelper = new StyleHelper($filePath);
95	96	$this->cellValueFormatter = new CellValueFormatter($sharedStringsHelper, $this->styleHelper, $options->shouldFormatDates());
96
97	96	$this->shouldPreserveEmptyRows = $options->shouldPreserveEmptyRows();
98
99		// Register all callbacks to process different nodes when reading the XML file
100	96	$this->xmlProcessor = new XMLProcessor($this->xmlReader);
101	96	$this->xmlProcessor->registerCallback(self::XML_NODE_DIMENSION, XMLProcessor::NODE_TYPE_START, [$this, 'processDimensionStartingNode']);
102	96	$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_START, [$this, 'processRowStartingNode']);
103	96	$this->xmlProcessor->registerCallback(self::XML_NODE_CELL, XMLProcessor::NODE_TYPE_START, [$this, 'processCellStartingNode']);
104	96	$this->xmlProcessor->registerCallback(self::XML_NODE_ROW, XMLProcessor::NODE_TYPE_END, [$this, 'processRowEndingNode']);
105	96	$this->xmlProcessor->registerCallback(self::XML_NODE_WORKSHEET, XMLProcessor::NODE_TYPE_END, [$this, 'processWorksheetEndingNode']);
106	96	}
107
108		/**
109		* @param string $sheetDataXMLFilePath Path of the sheet data XML file as in [Content_Types].xml
110		* @return string Path of the XML file containing the sheet data,
111		* without the leading slash.
112		*/
113	96	protected function normalizeSheetDataXMLFilePath($sheetDataXMLFilePath)
114		{
115	96	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	93	public function rewind()
128		{
129	93	$this->xmlReader->close();
130
131	93	$sheetDataFilePath = 'zip://' . $this->filePath . '#' . $this->sheetDataXMLFilePath;
132	93	if ($this->xmlReader->open($sheetDataFilePath) === false) {
133	3	throw new IOException("Could not open \"{$this->sheetDataXMLFilePath}\".");
134		}
135
136	90	$this->numReadRows = 0;
137	90	$this->lastRowIndexProcessed = 0;
138	90	$this->nextRowIndexToBeProcessed = 0;
139	90	$this->rowDataBuffer = null;
140	90	$this->hasReachedEndOfFile = false;
141	90	$this->numColumns = 0;
142
143	90	$this->next();
144	90	}
145
146		/**
147		* Checks if current position is valid
148		* @link http://php.net/manual/en/iterator.valid.php
149		*
150		* @return bool
151		*/
152	90	public function valid()
153		{
154	90	return (!$this->hasReachedEndOfFile);
155		}
156
157		/**
158		* Move forward to next element. Reads data describing the next unprocessed row.
159		* @link http://php.net/manual/en/iterator.next.php
160		*
161		* @return void
162		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
163		* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
164		*/
165	90	public function next()
166		{
167	90	$this->nextRowIndexToBeProcessed++;
168
169	90	if ($this->doesNeedDataForNextRowToBeProcessed()) {
170	90	$this->readDataForNextRow();
171	90	}
172	90	}
173
174		/**
175		* Returns whether we need data for the next row to be processed.
176		* We don't need to read data if:
177		* we have already read at least one row
178		* AND
179		* we need to preserve empty rows
180		* AND
181		* the last row that was read is not the row that need to be processed
182		* (i.e. if we need to return empty rows)
183		*
184		* @return bool Whether we need data for the next row to be processed.
185		*/
186	90	protected function doesNeedDataForNextRowToBeProcessed()
187		{
188	90	$hasReadAtLeastOneRow = ($this->lastRowIndexProcessed !== 0);
189
190		return (
191	90	!$hasReadAtLeastOneRow \|\|
192	84	!$this->shouldPreserveEmptyRows \|\|
193	3	$this->lastRowIndexProcessed < $this->nextRowIndexToBeProcessed
194	90	);
195		}
196
197		/**
198		* @return void
199		* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
200		* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
201		*/
202	90	protected function readDataForNextRow()
203		{
204	90	$this->currentlyProcessedRowData = [];
205
206		try {
207	90	$this->xmlProcessor->readUntilStopped();
208	90	} catch (XMLProcessingException $exception) {
209		throw new IOException("The {$this->sheetDataXMLFilePath} file cannot be read. [{$exception->getMessage()}]");
210		}
211
212	90	$this->rowDataBuffer = $this->currentlyProcessedRowData;
213	90	}
214
215		/**
216		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<dimension>" starting node
217		* @return int A return code that indicates what action should the processor take next
218		*/
219	45	protected function processDimensionStartingNode($xmlReader)
220		{
221		// Read dimensions of the sheet
222	45	$dimensionRef = $xmlReader->getAttribute(self::XML_ATTRIBUTE_REF); // returns 'A1:M13' for instance (or 'A1' for empty sheet)
223	45	if (preg_match('/[A-Z\d]+:([A-Z\d]+)/', $dimensionRef, $matches)) {
224	36	$lastCellIndex = $matches[1];
225	36	$this->numColumns = CellHelper::getColumnIndexFromCellIndex($lastCellIndex) + 1;
226	36	}
227
228	45	return XMLProcessor::PROCESSING_CONTINUE;
229		}
230
231		/**
232		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<row>" starting node
233		* @return int A return code that indicates what action should the processor take next
234		*/
235	87	protected function processRowStartingNode($xmlReader)
236		{
237		// Reset index of the last processed column
238	87	$this->lastColumnIndexProcessed = -1;
239
240		// Mark the last processed row as the one currently being read
241	87	$this->lastRowIndexProcessed = $this->getRowIndex($xmlReader);
242
243		// Read spans info if present
244	87	$numberOfColumnsForRow = $this->numColumns;
245	87	$spans = $xmlReader->getAttribute(self::XML_ATTRIBUTE_SPANS); // returns '1:5' for instance
246	87	if ($spans) {
247	30	list(, $numberOfColumnsForRow) = explode(':', $spans);
248	30	$numberOfColumnsForRow = intval($numberOfColumnsForRow);
249	30	}
250
251	87	$this->currentlyProcessedRowData = ($numberOfColumnsForRow !== 0) ? array_fill(0, $numberOfColumnsForRow, '') : [];
252
253	87	return XMLProcessor::PROCESSING_CONTINUE;
254		}
255
256		/**
257		* @param \Box\Spout\Reader\Wrapper\XMLReader $xmlReader XMLReader object, positioned on a "<cell>" starting node
258		* @return int A return code that indicates what action should the processor take next
259		*/
260	87	protected function processCellStartingNode($xmlReader)
261		{
262	87	$currentColumnIndex = $this->getColumnIndex($xmlReader);
263
264	87	$node = $xmlReader->expand();
265	87	$this->currentlyProcessedRowData[$currentColumnIndex] = $this->getCellValue($node);
266	87	$this->lastColumnIndexProcessed = $currentColumnIndex;
267
268	87	return XMLProcessor::PROCESSING_CONTINUE;
269		}
270
271		/**
272		* @return int A return code that indicates what action should the processor take next
273		*/
274	87	protected function processRowEndingNode()
275		{
276		// if the fetched row is empty and we don't want to preserve it..,
277	87	if (!$this->shouldPreserveEmptyRows && $this->isEmptyRow($this->currentlyProcessedRowData)) {
278		// ... skip it
279		return XMLProcessor::PROCESSING_CONTINUE;
280		}
281
282	87	$this->numReadRows++;
283
284		// If needed, we fill the empty cells
285	87	if ($this->numColumns === 0) {
286	51	$this->currentlyProcessedRowData = CellHelper::fillMissingArrayIndexes($this->currentlyProcessedRowData);
287	51	}
288
289		// at this point, we have all the data we need for the row
290		// so that we can populate the buffer
291	87	return XMLProcessor::PROCESSING_STOP;
292		}
293
294		/**
295		* @return int A return code that indicates what action should the processor take next
296		*/
297	87	protected function processWorksheetEndingNode()
298		{
299		// The closing "</worksheet>" marks the end of the file
300	87	$this->hasReachedEndOfFile = true;
301
302	87	return XMLProcessor::PROCESSING_STOP;
303		}
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

Push — master ( 73d5d0...a19231 )

RowIterator::processCellStartingNode() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like