RowIterator::getEncodedEOLDelimiter() - Code Metrics - Inspection of "Refactor readers for better di" - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — develop_3.0 (#457)

by Adrien

created 2017-08-26 21:44 UTC

RowIterator::getEncodedEOLDelimiter() A

↳ Parent: RowIterator

Complexity

Conditions	2
Paths	2

Size

Total Lines	8
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	6

Importance

Changes

Metric	Value
dl	0
loc	8
ccs	0
cts	4
cp	0
rs	9.4285
c	0
b	0
f	0
cc	2
eloc	4
nc	2
nop	0
crap	6

<?php

namespace Box\Spout\Reader\CSV;

use Box\Spout\Common\Creator\HelperFactory;
use Box\Spout\Reader\Common\Entity\Options;
use Box\Spout\Reader\CSV\Creator\EntityFactory;
use Box\Spout\Reader\IteratorInterface;
use Box\Spout\Common\Helper\EncodingHelper;

/**
 * Class RowIterator
 * Iterate over CSV rows.
 *
 * @package Box\Spout\Reader\CSV
 */
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 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 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\GlobalFunctionsHelper Helper to work with global functions */
    protected $globalFunctionsHelper;

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

    /**
     * @param resource $filePointer Pointer to the CSV file to read
     * @param \Box\Spout\Common\Manager\OptionsManagerInterface $optionsManager
     * @param \Box\Spout\Common\Helper\EncodingHelper $encodingHelper
     * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
     */
    public function __construct($filePointer, $optionsManager, $encodingHelper, $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->globalFunctionsHelper = $globalFunctionsHelper;
    }

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

        $this->numReadRows = 0;
        $this->rowDataBuffer = 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
     * @link 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.
     * @link http://php.net/manual/en/iterator.next.php
     *
     * @return void
     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
     */
    public function next()
    {
        $this->hasReachedEndOfFile = $this->globalFunctionsHelper->feof($this->filePointer);

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

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

        if ($rowData !== false) {
            // str_replace will replace NULL values by empty strings
            $this->rowDataBuffer = str_replace(null, null, $rowData);
            $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)
     *
     * @return array|false The row for the current file pointer, encoded in UTF-8 or FALSE if nothing to read
     * @throws \Box\Spout\Common\Exception\EncodingConversionException If unable to convert data to UTF-8
     */
    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
     * @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()
    {
        // do nothing
    }
}


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

box / spout

Pull Request — develop_3.0 (#457)

RowIterator::getEncodedEOLDelimiter() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like