visualizer / vendor / openspout / openspout / src / Reader / CSV / RowIterator.php

RowIterator.php in Visualizer – Tables & Charts Manager with Built-in AI Generator 3.11.10, at vendor/openspout/openspout/src/Reader/CSV/RowIterator.php

250 lines 8.3 KB

Raw Download Zip

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