fluent-cart / vendor / openspout / openspout / src / Reader / CSV / RowIterator.php

RowIterator.php in FluentCart A New Era of eCommerce – Faster, Lighter, and Simpler 1.3.19, at vendor/openspout/openspout/src/Reader/CSV/RowIterator.php

208 lines 8.3 KB

Raw Download Zip

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