RowIterator::rewind() - Code Metrics - Inspection of "Add Performance tests" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Failed Conditions

Push — perf-tests ( 50942d...2fc93e )

by Adrien

created 2016-06-16 17:11 UTC

RowIterator::rewind() A

↳ Parent: RowIterator

Complexity

Conditions	2
Paths	2

Size

Total Lines	16
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	0

Metric	Value
c	1
b	0
f	0
dl	0
loc	16
rs	9.4285
cc	2
eloc	8
nc	2
nop	0

<?php

namespace Box\Spout\Reader\ODS;

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

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

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

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

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

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

    /** @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;

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

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

        $this->hasAlreadyBeenRewound = true;
        $this->numReadRows = 0;
        $this->rowDataBuffer = null;
        $this->hasReachedEndOfFile = false;

        $this->next();
    }

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

    /**
     * Move forward to next element. Empty rows will be skipped.
     * @link http://php.net/manual/en/iterator.next.php
     *
     * @return void
     * @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
     * @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
     */
    public function next()
    {
        $rowData = [];
        $cellValue = null;
        $numColumnsRepeated = 1;
        $numCellsRead = 0;
        $hasAlreadyReadOneCell = false;

        try {
            while ($this->xmlReader->read()) {
                if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
                    // Start of a cell description
                    $currentNumColumnsRepeated = $this->getNumColumnsRepeatedForCurrentNode();

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

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

                    $cellValue = $currentCellValue;
                    $numColumnsRepeated = $currentNumColumnsRepeated;

                    $numCellsRead++;
                    $hasAlreadyReadOneCell = true;

                } else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
                    // End of the row description
                    $isEmptyRow = ($numCellsRead <= 1 && $this->isEmptyCellValue($cellValue));
                    if ($isEmptyRow) {
                        // skip empty rows
                        $this->next();
                        return;
                    }

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

                } else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_TABLE)) {
                    // The closing "</table:table>" marks the end of the file
                    $this->hasReachedEndOfFile = true;
                    break;
                }
            }

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

        $this->rowDataBuffer = $rowData;
    }

    /**
     * @return int The value of "table:number-columns-repeated" attribute of the current node, or 1 if attribute missing
     */
    protected function getNumColumnsRepeatedForCurrentNode()
    {
        $numColumnsRepeated = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_COLUMNS_REPEATED);
        return ($numColumnsRepeated !== null) ? intval($numColumnsRepeated) : 1;
    }

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

    /**
     * empty() replacement that honours 0 as a valid value
     *
     * @param string|int|float|bool|\DateTime|\DateInterval|null $value The cell value
     * @return bool
     */
    protected function isEmptyCellValue($value)
    {
        return (!isset($value) || trim($value) === '');
    }

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

    /**
     * Return the key of the current element
     * @link http://php.net/manual/en/iterator.key.php
     *
     * @return int
     */
    public function key()
    {
        return $this->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\ODS;
4
5			use Box\Spout\Common\Exception\IOException;
6			use Box\Spout\Reader\Exception\IteratorNotRewindableException;
7			use Box\Spout\Reader\Exception\XMLProcessingException;
8			use Box\Spout\Reader\IteratorInterface;
9			use Box\Spout\Reader\ODS\Helper\CellValueFormatter;
10			use Box\Spout\Reader\Wrapper\XMLReader;
11
12			/**
13			* Class RowIterator
14			*
15			* @package Box\Spout\Reader\ODS
16			*/
17			class RowIterator implements IteratorInterface
18			{
19			/** Definition of XML nodes names used to parse data */
20			const XML_NODE_TABLE = 'table:table';
21			const XML_NODE_ROW = 'table:table-row';
22			const XML_NODE_CELL = 'table:table-cell';
23			const MAX_COLUMNS_EXCEL = 16384;
24
25			/** Definition of XML attribute used to parse data */
26			const XML_ATTRIBUTE_NUM_COLUMNS_REPEATED = 'table:number-columns-repeated';
27
28			/** @var \Box\Spout\Reader\Wrapper\XMLReader The XMLReader object that will help read sheet's XML data */
29			protected $xmlReader;
30
31			/** @var Helper\CellValueFormatter Helper to format cell values */
32			protected $cellValueFormatter;
33
34			/** @var bool Whether the iterator has already been rewound once */
35			protected $hasAlreadyBeenRewound = false;
36
37			/** @var int Number of read rows */
38			protected $numReadRows = 0;
39
40			/** @var array\|null Buffer used to store the row data, while checking if there are more rows to read */
41			protected $rowDataBuffer = null;
42
43			/** @var bool Indicates whether all rows have been read */
44			protected $hasReachedEndOfFile = false;
45
46			/**
47			* @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
48			* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
49			*/
50			public function __construct($xmlReader, $shouldFormatDates)
51			{
52			$this->xmlReader = $xmlReader;
53			$this->cellValueFormatter = new CellValueFormatter($shouldFormatDates);
54			}
55
56			/**
57			* Rewind the Iterator to the first element.
58			* NOTE: It can only be done once, as it is not possible to read an XML file backwards.
59			* @link http://php.net/manual/en/iterator.rewind.php
60			*
61			* @return void
62			* @throws \Box\Spout\Reader\Exception\IteratorNotRewindableException If the iterator is rewound more than once
63			*/
64			public function rewind()
65			{
66			// Because sheet and row data is located in the file, we can't rewind both the
67			// sheet iterator and the row iterator, as XML file cannot be read backwards.
68			// Therefore, rewinding the row iterator has been disabled.
69			if ($this->hasAlreadyBeenRewound) {
70			throw new IteratorNotRewindableException();
71			}
72
73			$this->hasAlreadyBeenRewound = true;
74			$this->numReadRows = 0;
75			$this->rowDataBuffer = null;
76			$this->hasReachedEndOfFile = false;
77
78			$this->next();
79			}
80
81			/**
82			* Checks if current position is valid
83			* @link http://php.net/manual/en/iterator.valid.php
84			*
85			* @return boolean
86			*/
87			public function valid()
88			{
89			return (!$this->hasReachedEndOfFile);
90			}
91
92			/**
93			* Move forward to next element. Empty rows will be skipped.
94			* @link http://php.net/manual/en/iterator.next.php
95			*
96			* @return void
97			* @throws \Box\Spout\Reader\Exception\SharedStringNotFoundException If a shared string was not found
98			* @throws \Box\Spout\Common\Exception\IOException If unable to read the sheet data XML
99			*/
100			public function next()
101			{
102			$rowData = [];
103			$cellValue = null;
104			$numColumnsRepeated = 1;
105			$numCellsRead = 0;
106			$hasAlreadyReadOneCell = false;
107
108			try {
109			while ($this->xmlReader->read()) {
110			if ($this->xmlReader->isPositionedOnStartingNode(self::XML_NODE_CELL)) {
111			// Start of a cell description
112			$currentNumColumnsRepeated = $this->getNumColumnsRepeatedForCurrentNode();
113
114			$node = $this->xmlReader->expand();
115			$currentCellValue = $this->getCellValue($node);
116
117			// process cell N only after having read cell N+1 (see below why)
118			if ($hasAlreadyReadOneCell) {
119			for ($i = 0; $i < $numColumnsRepeated; $i++) {
120			$rowData[] = $cellValue;
121			}
122			}
123
124			$cellValue = $currentCellValue;
125			$numColumnsRepeated = $currentNumColumnsRepeated;
126
127			$numCellsRead++;
128			$hasAlreadyReadOneCell = true;
129
130			} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_ROW)) {
131			// End of the row description
132			$isEmptyRow = ($numCellsRead <= 1 && $this->isEmptyCellValue($cellValue));
133			if ($isEmptyRow) {
134			// skip empty rows
135			$this->next();
136			return;
137			}
138
139			// Only add the value if the last read cell is not a trailing empty cell repeater in Excel.
140			// The current count of read columns is determined by counting the values in $rowData.
141			// This is to avoid creating a lot of empty cells, as Excel adds a last empty "<table:table-cell>"
142			// with a number-columns-repeated value equals to the number of (supported columns - used columns).
143			// In Excel, the number of supported columns is 16384, but we don't want to returns rows with
144			// always 16384 cells.
145			if ((count($rowData) + $numColumnsRepeated) !== self::MAX_COLUMNS_EXCEL) {
146			for ($i = 0; $i < $numColumnsRepeated; $i++) {
147			$rowData[] = $cellValue;
148			}
149			$this->numReadRows++;
150			}
151			break;
152
153			} else if ($this->xmlReader->isPositionedOnEndingNode(self::XML_NODE_TABLE)) {
154			// The closing "</table:table>" marks the end of the file
155			$this->hasReachedEndOfFile = true;
156			break;
157			}
158			}
159
160			} catch (XMLProcessingException $exception) {
161			throw new IOException("The sheet's data cannot be read. [{$exception->getMessage()}]");
162			}
163
164			$this->rowDataBuffer = $rowData;
165			}
166
167			/**
168			* @return int The value of "table:number-columns-repeated" attribute of the current node, or 1 if attribute missing
169			*/
170			protected function getNumColumnsRepeatedForCurrentNode()
171			{
172			$numColumnsRepeated = $this->xmlReader->getAttribute(self::XML_ATTRIBUTE_NUM_COLUMNS_REPEATED);
173			return ($numColumnsRepeated !== null) ? intval($numColumnsRepeated) : 1;
174			}
175
176			/**
177			* Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
178			*
179			* @param \DOMNode $node
180			* @return string\|int\|float\|bool\|\DateTime\|\DateInterval\|null The value associated with the cell, empty string if cell's type is void/undefined, null on error
181			*/
182			protected function getCellValue($node)
183			{
184			return $this->cellValueFormatter->extractAndFormatNodeValue($node);
185			}
186
187			/**
188			* empty() replacement that honours 0 as a valid value
189			*
190			* @param string\|int\|float\|bool\|\DateTime\|\DateInterval\|null $value The cell value
191			* @return bool
192			*/
193			protected function isEmptyCellValue($value)
194			{
195			return (!isset($value) \|\| trim($value) === '');
196			}
197
198			/**
199			* Return the current element, from the buffer.
200			* @link http://php.net/manual/en/iterator.current.php
201			*
202			* @return array\|null
203			*/
204			public function current()
205			{
206			return $this->rowDataBuffer;
207			}
208
209			/**
210			* Return the key of the current element
211			* @link http://php.net/manual/en/iterator.key.php
212			*
213			* @return int
214			*/
215			public function key()
216			{
217			return $this->numReadRows;
218			}
219
220
221			/**
222			* Cleans up what was created to iterate over the object.
223			*
224			* @return void
225			*/
226			public function end()
227			{
228			$this->xmlReader->close();
229			}
230			}
231

box / spout

Push — perf-tests ( 50942d...2fc93e )

RowIterator::rewind() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like