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

Completed

Pull Request — master (#557)

by Adrien

created 2018-06-03 19:08 UTC

RowIterator::rewind() A

↳ Parent: RowIterator

Complexity

Conditions	1
Paths	1

Size

Total Lines	9
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	1

Importance

Changes

Metric	Value
dl	0
loc	9
ccs	6
cts	6
cp	1
rs	9.6666
c	0
b	0
f	0
cc	1
eloc	5
nc	1
nop	0
crap	1

<?php

namespace Box\Spout\Reader\CSV;

use Box\Spout\Common\Entity\Row;
use Box\Spout\Common\Helper\EncodingHelper;
use Box\Spout\Common\Helper\GlobalFunctionsHelper;
use Box\Spout\Common\Manager\OptionsManagerInterface;
use Box\Spout\Reader\Common\Entity\Options;
use Box\Spout\Reader\CSV\Creator\InternalEntityFactory;
use Box\Spout\Reader\IteratorInterface;

/**
 * Class RowIterator
 * Iterate over CSV rows.
 */
class RowIterator implements IteratorInterface
{
    /**
     * Value passed to fgetcsv. 0 means "unlimited" (slightly slower but accomodates for very long lines).
     */
    const MAX_READ_BYTES_PER_LINE = 0;

    /** @var resource Pointer to the CSV file to read */
    protected $filePointer;

    /** @var int Number of read rows */
    protected $numReadRows = 0;

    /** @var Row|null Buffer used to store the current row, while checking if there are more rows to read */
    protected $rowBuffer;

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

    /** @var string Defines the character used to delimit fields (one character only) */
    protected $fieldDelimiter;

    /** @var string Defines the character used to enclose fields (one character only) */
    protected $fieldEnclosure;

    /** @var string Encoding of the CSV file to be read */
    protected $encoding;

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

    /** @var \Box\Spout\Common\Helper\EncodingHelper Helper to work with different encodings */
    protected $encodingHelper;

    /** @var \Box\Spout\Reader\CSV\Creator\InternalEntityFactory Factory to create entities */
    protected $entityFactory;

    /** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
    protected $globalFunctionsHelper;

    /**
     * @param resource $filePointer Pointer to the CSV file to read
     * @param OptionsManagerInterface $optionsManager
     * @param EncodingHelper $encodingHelper
     * @param InternalEntityFactory $entityFactory
     * @param GlobalFunctionsHelper $globalFunctionsHelper
     */
    public function __construct(
        $filePointer,
        OptionsManagerInterface $optionsManager,
        EncodingHelper $encodingHelper,
        InternalEntityFactory $entityFactory,
        GlobalFunctionsHelper $globalFunctionsHelper
    ) {
        $this->filePointer = $filePointer;
        $this->fieldDelimiter = $optionsManager->getOption(Options::FIELD_DELIMITER);
        $this->fieldEnclosure = $optionsManager->getOption(Options::FIELD_ENCLOSURE);
        $this->encoding = $optionsManager->getOption(Options::ENCODING);
        $this->shouldPreserveEmptyRows = $optionsManager->getOption(Options::SHOULD_PRESERVE_EMPTY_ROWS);
        $this->encodingHelper = $encodingHelper;
        $this->entityFactory = $entityFactory;
        $this->globalFunctionsHelper = $globalFunctionsHelper;
    }

    /**
     * Rewind the Iterator to the first element
     * @see http://php.net/manual/en/iterator.rewind.php
     *
     * @return void
     */
    public function rewind()
    {
        $this->rewindAndSkipBom();

        $this->numReadRows = 0;
        $this->rowBuffer = null;

        $this->next();
    }

    /**
     * This rewinds and skips the BOM if inserted at the beginning of the file
     * by moving the file pointer after it, so that it is not read.
     *
     * @return void
     */
    protected function rewindAndSkipBom()
    {
        $byteOffsetToSkipBom = $this->encodingHelper->getBytesOffsetToSkipBOM($this->filePointer, $this->encoding);

        // sets the cursor after the BOM (0 means no BOM, so rewind it)
        $this->globalFunctionsHelper->fseek($this->filePointer, $byteOffsetToSkipBom);
    }

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

    /**
     * Move forward to next element. Reads data for the next unprocessed row.
     * @see http://php.net/manual/en/iterator.next.php
     *
     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
     * @return void
     */
    public function next()
    {
        $this->hasReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);

        if (!$this->hasReachedEndOfFile) {
            $this->readDataForNextRow();
        }
    }

    /**
     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
     * @return void
     */
    protected function readDataForNextRow()
    {
        do {
            $rowData = $this->getNextUTF8EncodedRow();
        } while ($this->shouldReadNextRow($rowData));

        if ($rowData !== false) {
            // str_replace will replace NULL values by empty strings
            $rowDataBufferAsArray = str_replace(null, null, $rowData);
            $this->rowBuffer = $this->entityFactory->createRowFromArray($rowDataBufferAsArray);
            $this->numReadRows++;
        } else {
            // If we reach this point, it means end of file was reached.
            // This happens when the last lines are empty lines.
            $this->hasReachedEndOfFile = true;
        }
    }

    /**
     * @param array|bool $currentRowData
     * @return bool Whether the data for the current row can be returned or if we need to keep reading
     */
    protected function shouldReadNextRow($currentRowData)
    {
        $hasSuccessfullyFetchedRowData = ($currentRowData !== false);
        $hasNowReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
        $isEmptyLine = $this->isEmptyLine($currentRowData);

        return (
            (!$hasSuccessfullyFetchedRowData && !$hasNowReachedEndOfFile) ||
            (!$this->shouldPreserveEmptyRows && $isEmptyLine)
        );
    }

    /**
     * Returns the next row, converted if necessary to UTF-8.
     * As fgetcsv() does not manage correctly encoding for non UTF-8 data,
     * we remove manually whitespace with ltrim or rtrim (depending on the order of the bytes)
     *
     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
     * @return array|false The row for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
     */
    protected function getNextUTF8EncodedRow()
    {
        $encodedRowData = $this->globalFunctionsHelper->fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
        if ($encodedRowData === false) {
            return false;
        }

        foreach ($encodedRowData as $cellIndex => $cellValue) {
            switch ($this->encoding) {
                case EncodingHelper::ENCODING_UTF16_LE:
                case EncodingHelper::ENCODING_UTF32_LE:
                    // remove whitespace from the beginning of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
                    $cellValue = ltrim($cellValue);
                    break;

                case EncodingHelper::ENCODING_UTF16_BE:
                case EncodingHelper::ENCODING_UTF32_BE:
                    // remove whitespace from the end of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
                    $cellValue = rtrim($cellValue);
                    break;
            }

            $encodedRowData[$cellIndex] = $this->encodingHelper->attemptConversionToUTF8($cellValue, $this->encoding);
        }

        return $encodedRowData;
    }

    /**
     * @param array|bool $lineData Array containing the cells value for the line
     * @return bool Whether the given line is empty
     */
    protected function isEmptyLine($lineData)
    {
        return (is_array($lineData) && count($lineData) === 1 && $lineData[0] === null);
    }

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

    /**
     * Return the key of the current element
     * @see 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()
    {
        // do nothing
    }
}


1		<?php
2
3		namespace Box\Spout\Reader\CSV;
4
5		use Box\Spout\Common\Entity\Row;
6		use Box\Spout\Common\Helper\EncodingHelper;
7		use Box\Spout\Common\Helper\GlobalFunctionsHelper;
8		use Box\Spout\Common\Manager\OptionsManagerInterface;
9		use Box\Spout\Reader\Common\Entity\Options;
10		use Box\Spout\Reader\CSV\Creator\InternalEntityFactory;
11		use Box\Spout\Reader\IteratorInterface;
12
13		/**
14		* Class RowIterator
15		* Iterate over CSV rows.
16		*/
17		class RowIterator implements IteratorInterface
18		{
19		/**
20		* Value passed to fgetcsv. 0 means "unlimited" (slightly slower but accomodates for very long lines).
21		*/
22		const MAX_READ_BYTES_PER_LINE = 0;
23
24		/** @var resource Pointer to the CSV file to read */
25		protected $filePointer;
26
27		/** @var int Number of read rows */
28		protected $numReadRows = 0;
29
30		/** @var Row\|null Buffer used to store the current row, while checking if there are more rows to read */
31		protected $rowBuffer;
32
33		/** @var bool Indicates whether all rows have been read */
34		protected $hasReachedEndOfFile = false;
35
36		/** @var string Defines the character used to delimit fields (one character only) */
37		protected $fieldDelimiter;
38
39		/** @var string Defines the character used to enclose fields (one character only) */
40		protected $fieldEnclosure;
41
42		/** @var string Encoding of the CSV file to be read */
43		protected $encoding;
44
45		/** @var bool Whether empty rows should be returned or skipped */
46		protected $shouldPreserveEmptyRows;
47
48		/** @var \Box\Spout\Common\Helper\EncodingHelper Helper to work with different encodings */
49		protected $encodingHelper;
50
51		/** @var \Box\Spout\Reader\CSV\Creator\InternalEntityFactory Factory to create entities */
52		protected $entityFactory;
53
54		/** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
55		protected $globalFunctionsHelper;
56
57		/**
58		* @param resource $filePointer Pointer to the CSV file to read
59		* @param OptionsManagerInterface $optionsManager
60		* @param EncodingHelper $encodingHelper
61		* @param InternalEntityFactory $entityFactory
62		* @param GlobalFunctionsHelper $globalFunctionsHelper
63		*/
64	27	public function __construct(
65		$filePointer,
66		OptionsManagerInterface $optionsManager,
67		EncodingHelper $encodingHelper,
68		InternalEntityFactory $entityFactory,
69		GlobalFunctionsHelper $globalFunctionsHelper
70		) {
71	27	$this->filePointer = $filePointer;
72	27	$this->fieldDelimiter = $optionsManager->getOption(Options::FIELD_DELIMITER);
73	27	$this->fieldEnclosure = $optionsManager->getOption(Options::FIELD_ENCLOSURE);
74	27	$this->encoding = $optionsManager->getOption(Options::ENCODING);
75	27	$this->shouldPreserveEmptyRows = $optionsManager->getOption(Options::SHOULD_PRESERVE_EMPTY_ROWS);
76	27	$this->encodingHelper = $encodingHelper;
77	27	$this->entityFactory = $entityFactory;
78	27	$this->globalFunctionsHelper = $globalFunctionsHelper;
79	27	}
80
81		/**
82		* Rewind the Iterator to the first element
83		* @see http://php.net/manual/en/iterator.rewind.php
84		*
85		* @return void
86		*/
87	26	public function rewind()
88		{
89	26	$this->rewindAndSkipBom();
90
91	26	$this->numReadRows = 0;
92	26	$this->rowBuffer = null;
93
94	26	$this->next();
95	26	}
96
97		/**
98		* This rewinds and skips the BOM if inserted at the beginning of the file
99		* by moving the file pointer after it, so that it is not read.
100		*
101		* @return void
102		*/
103	26	protected function rewindAndSkipBom()
104		{
105	26	$byteOffsetToSkipBom = $this->encodingHelper->getBytesOffsetToSkipBOM($this->filePointer, $this->encoding);
106
107		// sets the cursor after the BOM (0 means no BOM, so rewind it)
108	26	$this->globalFunctionsHelper->fseek($this->filePointer, $byteOffsetToSkipBom);
109	26	}
110
111		/**
112		* Checks if current position is valid
113		* @see http://php.net/manual/en/iterator.valid.php
114		*
115		* @return bool
116		*/
117	26	public function valid()
118		{
119	26	return ($this->filePointer && !$this->hasReachedEndOfFile);
120		}
121
122		/**
123		* Move forward to next element. Reads data for the next unprocessed row.
124		* @see http://php.net/manual/en/iterator.next.php
125		*
126		* @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
127		* @return void
128		*/
129	26	public function next()
130		{
131	26	$this->hasReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
132
133	26	if (!$this->hasReachedEndOfFile) {
134	26	$this->readDataForNextRow();
135		}
136	26	}
137
138		/**
139		* @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
140		* @return void
141		*/
142	26	protected function readDataForNextRow()
143		{
144		do {
145	26	$rowData = $this->getNextUTF8EncodedRow();
146	26	} while ($this->shouldReadNextRow($rowData));
147
148	26	if ($rowData !== false) {
149		// str_replace will replace NULL values by empty strings
150	24	$rowDataBufferAsArray = str_replace(null, null, $rowData);
151	24	$this->rowBuffer = $this->entityFactory->createRowFromArray($rowDataBufferAsArray);
152	24	$this->numReadRows++;
153		} else {
154		// If we reach this point, it means end of file was reached.
155		// This happens when the last lines are empty lines.
156	9	$this->hasReachedEndOfFile = true;
157		}
158	26	}
159
160		/**
161		* @param array\|bool $currentRowData
162		* @return bool Whether the data for the current row can be returned or if we need to keep reading
163		*/
164	26	protected function shouldReadNextRow($currentRowData)
165		{
166	26	$hasSuccessfullyFetchedRowData = ($currentRowData !== false);
167	26	$hasNowReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);
168	26	$isEmptyLine = $this->isEmptyLine($currentRowData);
169
170		return (
171	26	(!$hasSuccessfullyFetchedRowData && !$hasNowReachedEndOfFile) \|\|
172	26	(!$this->shouldPreserveEmptyRows && $isEmptyLine)
173		);
174		}
175
176		/**
177		* Returns the next row, converted if necessary to UTF-8.
178		* As fgetcsv() does not manage correctly encoding for non UTF-8 data,
179		* we remove manually whitespace with ltrim or rtrim (depending on the order of the bytes)
180		*
181		* @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
182		* @return array\|false The row for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
183		*/
184	26	protected function getNextUTF8EncodedRow()
185		{
186	26	$encodedRowData = $this->globalFunctionsHelper->fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
187	26	if ($encodedRowData === false) {
188	9	return false;
189		}
190
191	25	foreach ($encodedRowData as $cellIndex => $cellValue) {
192	25	switch ($this->encoding) {
193	25	case EncodingHelper::ENCODING_UTF16_LE:
194	22	case EncodingHelper::ENCODING_UTF32_LE:
195		// remove whitespace from the beginning of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
196	4	$cellValue = ltrim($cellValue);
197	4	break;
198
199	21	case EncodingHelper::ENCODING_UTF16_BE:
200	20	case EncodingHelper::ENCODING_UTF32_BE:
201		// remove whitespace from the end of a string as fgetcsv() add extra whitespace when it try to explode non UTF-8 data
202	2	$cellValue = rtrim($cellValue);
203	2	break;
204		}
205
206	25	$encodedRowData[$cellIndex] = $this->encodingHelper->attemptConversionToUTF8($cellValue, $this->encoding);
207		}
208
209	25	return $encodedRowData;
210		}
211
212		/**
213		* @param array\|bool $lineData Array containing the cells value for the line
214		* @return bool Whether the given line is empty
215		*/
216	26	protected function isEmptyLine($lineData)
217		{
218	26	return (is_array($lineData) && count($lineData) === 1 && $lineData[0] === null);
219		}
220
221		/**
222		* Return the current element from the buffer
223		* @see http://php.net/manual/en/iterator.current.php
224		*
225		* @return Row\|null
226		*/
227	24	public function current()
228		{
229	24	return $this->rowBuffer;
230		}
231
232		/**
233		* Return the key of the current element
234		* @see http://php.net/manual/en/iterator.key.php
235		*
236		* @return int
237		*/
238	18	public function key()
239		{
240	18	return $this->numReadRows;
241		}
242
243		/**
244		* Cleans up what was created to iterate over the object.
245		*
246		* @return void
247		*/
248		public function end()
249		{
250		// do nothing
251		}
252		}
253

box / spout

Pull Request — master (#557)

RowIterator::rewind() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like