fluent-cart / vendor / openspout / openspout / src / Reader / ODS / Helper / CellValueFormatter.php

CellValueFormatter.php in FluentCart A New Era of eCommerce – Faster, Lighter, and Simpler 1.5.4, at vendor/openspout/openspout/src/Reader/ODS/Helper/CellValueFormatter.php

267 lines 10.2 KB

Raw Download Zip

1	<?php
2
3	namespace FluentCart\OpenSpout\Reader\ODS\Helper;
4
5	use FluentCart\OpenSpout\Reader\Exception\InvalidValueException;
6	/**
7	* This class provides helper functions to format cell values.
8	*/
9	class CellValueFormatter
10	{
11	/** Definition of all possible cell types */
12	public const CELL_TYPE_STRING = 'string';
13	public const CELL_TYPE_FLOAT = 'float';
14	public const CELL_TYPE_BOOLEAN = 'boolean';
15	public const CELL_TYPE_DATE = 'date';
16	public const CELL_TYPE_TIME = 'time';
17	public const CELL_TYPE_CURRENCY = 'currency';
18	public const CELL_TYPE_PERCENTAGE = 'percentage';
19	public const CELL_TYPE_VOID = 'void';
20	/** Definition of XML nodes names used to parse data */
21	public const XML_NODE_P = 'p';
22	public const XML_NODE_TEXT_A = 'text:a';
23	public const XML_NODE_TEXT_SPAN = 'text:span';
24	public const XML_NODE_TEXT_S = 'text:s';
25	public const XML_NODE_TEXT_TAB = 'text:tab';
26	public const XML_NODE_TEXT_LINE_BREAK = 'text:line-break';
27	/** Definition of XML attributes used to parse data */
28	public const XML_ATTRIBUTE_TYPE = 'office:value-type';
29	public const XML_ATTRIBUTE_VALUE = 'office:value';
30	public const XML_ATTRIBUTE_BOOLEAN_VALUE = 'office:boolean-value';
31	public const XML_ATTRIBUTE_DATE_VALUE = 'office:date-value';
32	public const XML_ATTRIBUTE_TIME_VALUE = 'office:time-value';
33	public const XML_ATTRIBUTE_CURRENCY = 'office:currency';
34	public const XML_ATTRIBUTE_C = 'text:c';
35	/** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
36	protected $shouldFormatDates;
37	/** @var \OpenSpout\Common\Helper\Escaper\ODS Used to unescape XML data */
38	protected $escaper;
39	/** @var array List of XML nodes representing whitespaces and their corresponding value */
40	private static $WHITESPACE_XML_NODES = [self::XML_NODE_TEXT_S => ' ', self::XML_NODE_TEXT_TAB => "\t", self::XML_NODE_TEXT_LINE_BREAK => "\n"];
41	/**
42	* @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
43	* @param \OpenSpout\Common\Helper\Escaper\ODS $escaper Used to unescape XML data
44	*/
45	public function __construct($shouldFormatDates, $escaper)
46	{
47	$this->shouldFormatDates = $shouldFormatDates;
48	$this->escaper = $escaper;
49	}
50	/**
51	* Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
52	*
53	* @see http://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part1.html#refTable13
54	*
55	* @param \DOMElement $node
56	*
57	* @throws InvalidValueException If the node value is not valid
58	*
59	* @return bool\|\DateInterval\|\DateTime\|float\|int\|string The value associated with the cell, empty string if cell's type is void/undefined
60	*/
61	public function extractAndFormatNodeValue($node)
62	{
63	$cellType = $node->getAttribute(self::XML_ATTRIBUTE_TYPE);
64	switch ($cellType) {
65	case self::CELL_TYPE_STRING:
66	return $this->formatStringCellValue($node);
67	case self::CELL_TYPE_FLOAT:
68	return $this->formatFloatCellValue($node);
69	case self::CELL_TYPE_BOOLEAN:
70	return $this->formatBooleanCellValue($node);
71	case self::CELL_TYPE_DATE:
72	return $this->formatDateCellValue($node);
73	case self::CELL_TYPE_TIME:
74	return $this->formatTimeCellValue($node);
75	case self::CELL_TYPE_CURRENCY:
76	return $this->formatCurrencyCellValue($node);
77	case self::CELL_TYPE_PERCENTAGE:
78	return $this->formatPercentageCellValue($node);
79	case self::CELL_TYPE_VOID:
80	default:
81	return '';
82	}
83	}
84	/**
85	* Returns the cell String value.
86	*
87	* @param \DOMElement $node
88	*
89	* @return string The value associated with the cell
90	*/
91	protected function formatStringCellValue($node)
92	{
93	$pNodeValues = [];
94	$pNodes = $node->getElementsByTagName(self::XML_NODE_P);
95	foreach ($pNodes as $pNode) {
96	$pNodeValues[] = $this->extractTextValueFromNode($pNode);
97	}
98	$escapedCellValue = \implode("\n", $pNodeValues);
99	return $this->escaper->unescape($escapedCellValue);
100	}
101	/**
102	* Returns the cell Numeric value from the given node.
103	*
104	* @param \DOMElement $node
105	*
106	* @return float\|int The value associated with the cell
107	*/
108	protected function formatFloatCellValue($node)
109	{
110	$nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
111	$nodeIntValue = (int) $nodeValue;
112	$nodeFloatValue = (float) $nodeValue;
113	return (float) $nodeIntValue === $nodeFloatValue ? $nodeIntValue : $nodeFloatValue;
114	}
115	/**
116	* Returns the cell Boolean value from the given node.
117	*
118	* @param \DOMElement $node
119	*
120	* @return bool The value associated with the cell
121	*/
122	protected function formatBooleanCellValue($node)
123	{
124	$nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_BOOLEAN_VALUE);
125	return (bool) $nodeValue;
126	}
127	/**
128	* Returns the cell Date value from the given node.
129	*
130	* @param \DOMElement $node
131	*
132	* @throws InvalidValueException If the value is not a valid date
133	*
134	* @return \DateTime\|string The value associated with the cell
135	*/
136	protected function formatDateCellValue($node)
137	{
138	// The XML node looks like this:
139	// <table:table-cell calcext:value-type="date" office:date-value="2016-05-19T16:39:00" office:value-type="date">
140	// <text:p>05/19/16 04:39 PM</text:p>
141	// </table:table-cell>
142	if ($this->shouldFormatDates) {
143	// The date is already formatted in the "p" tag
144	$nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
145	$cellValue = $nodeWithValueAlreadyFormatted->nodeValue;
146	} else {
147	// otherwise, get it from the "date-value" attribute
148	$nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_DATE_VALUE);
149	try {
150	$cellValue = new \DateTime($nodeValue);
151	} catch (\Exception $e) {
152	throw new InvalidValueException($nodeValue);
153	}
154	}
155	return $cellValue;
156	}
157	/**
158	* Returns the cell Time value from the given node.
159	*
160	* @param \DOMElement $node
161	*
162	* @throws InvalidValueException If the value is not a valid time
163	*
164	* @return \DateInterval\|string The value associated with the cell
165	*/
166	protected function formatTimeCellValue($node)
167	{
168	// The XML node looks like this:
169	// <table:table-cell calcext:value-type="time" office:time-value="PT13H24M00S" office:value-type="time">
170	// <text:p>01:24:00 PM</text:p>
171	// </table:table-cell>
172	if ($this->shouldFormatDates) {
173	// The date is already formatted in the "p" tag
174	$nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
175	$cellValue = $nodeWithValueAlreadyFormatted->nodeValue;
176	} else {
177	// otherwise, get it from the "time-value" attribute
178	$nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_TIME_VALUE);
179	try {
180	$cellValue = new \DateInterval($nodeValue);
181	} catch (\Exception $e) {
182	throw new InvalidValueException($nodeValue);
183	}
184	}
185	return $cellValue;
186	}
187	/**
188	* Returns the cell Currency value from the given node.
189	*
190	* @param \DOMElement $node
191	*
192	* @return string The value associated with the cell (e.g. "100 USD" or "9.99 EUR")
193	*/
194	protected function formatCurrencyCellValue($node)
195	{
196	$value = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
197	$currency = $node->getAttribute(self::XML_ATTRIBUTE_CURRENCY);
198	return "{$value} {$currency}";
199	}
200	/**
201	* Returns the cell Percentage value from the given node.
202	*
203	* @param \DOMElement $node
204	*
205	* @return float\|int The value associated with the cell
206	*/
207	protected function formatPercentageCellValue($node)
208	{
209	// percentages are formatted like floats
210	return $this->formatFloatCellValue($node);
211	}
212	/**
213	* @param \DOMNode $pNode
214	*
215	* @return string
216	*/
217	private function extractTextValueFromNode($pNode)
218	{
219	$textValue = '';
220	foreach ($pNode->childNodes as $childNode) {
221	if ($childNode instanceof \DOMText) {
222	$textValue .= $childNode->nodeValue;
223	} elseif ($this->isWhitespaceNode($childNode->nodeName)) {
224	$textValue .= $this->transformWhitespaceNode($childNode);
225	} elseif (self::XML_NODE_TEXT_A === $childNode->nodeName \|\| self::XML_NODE_TEXT_SPAN === $childNode->nodeName) {
226	$textValue .= $this->extractTextValueFromNode($childNode);
227	}
228	}
229	return $textValue;
230	}
231	/**
232	* Returns whether the given node is a whitespace node. It must be one of these:
233	* - <text:s />
234	* - <text:tab />
235	* - <text:line-break />.
236	*
237	* @param string $nodeName
238	*
239	* @return bool
240	*/
241	private function isWhitespaceNode($nodeName)
242	{
243	return isset(self::$WHITESPACE_XML_NODES[$nodeName]);
244	}
245	/**
246	* The "<text:p>" node can contain the string value directly
247	* or contain child elements. In this case, whitespaces contain in
248	* the child elements should be replaced by their XML equivalent:
249	* - space => <text:s />
250	* - tab => <text:tab />
251	* - line break => <text:line-break />.
252	*
253	* @see https://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part1.html#__RefHeading__1415200_253892949
254	*
255	* @param \DOMElement $node The XML node representing a whitespace
256	*
257	* @return string The corresponding whitespace value
258	*/
259	private function transformWhitespaceNode($node)
260	{
261	$countAttribute = $node->getAttribute(self::XML_ATTRIBUTE_C);
262	// only defined for "<text:s>"
263	$numWhitespaces = !empty($countAttribute) ? (int) $countAttribute : 1;
264	return \str_repeat(self::$WHITESPACE_XML_NODES[$node->nodeName], $numWhitespaces);
265	}
266	}
267