CellValueFormatter::extractAndFormatNodeValue() - Code Metrics - Inspection of "Adding support for <text:s> inside <text:span> (fi..." - box/spout - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#667)

unknown

created 2019-07-10 12:18 UTC

CellValueFormatter::extractAndFormatNodeValue() B

↳ Parent: CellValueFormatter

Complexity

Conditions	9
Paths	9

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	18
CRAP Score	9

Importance

Changes

Metric	Value
dl	0
loc	24
ccs	18
cts	18
cp	1
rs	8.0555
c	0
b	0
f	0
cc	9
nc	9
nop	1
crap	9

<?php

namespace Box\Spout\Reader\ODS\Helper;

use Box\Spout\Reader\Exception\InvalidValueException;

/**
 * Class CellValueFormatter
 * This class provides helper functions to format cell values
 */
class CellValueFormatter
{
    /** Definition of all possible cell types */
    const CELL_TYPE_STRING = 'string';
    const CELL_TYPE_FLOAT = 'float';
    const CELL_TYPE_BOOLEAN = 'boolean';
    const CELL_TYPE_DATE = 'date';
    const CELL_TYPE_TIME = 'time';
    const CELL_TYPE_CURRENCY = 'currency';
    const CELL_TYPE_PERCENTAGE = 'percentage';
    const CELL_TYPE_VOID = 'void';

    /** Definition of XML nodes names used to parse data */
    const XML_NODE_P = 'p';
    const XML_NODE_S = 'text:s';
    const XML_NODE_A = 'text:a';
    const XML_NODE_SPAN = 'text:span';

    /** Definition of XML attributes used to parse data */
    const XML_ATTRIBUTE_TYPE = 'office:value-type';
    const XML_ATTRIBUTE_VALUE = 'office:value';
    const XML_ATTRIBUTE_BOOLEAN_VALUE = 'office:boolean-value';
    const XML_ATTRIBUTE_DATE_VALUE = 'office:date-value';
    const XML_ATTRIBUTE_TIME_VALUE = 'office:time-value';
    const XML_ATTRIBUTE_CURRENCY = 'office:currency';
    const XML_ATTRIBUTE_C = 'text:c';

    /** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
    protected $shouldFormatDates;

    /** @var \Box\Spout\Common\Helper\Escaper\ODS Used to unescape XML data */
    protected $escaper;

    /**
     * @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
     * @param \Box\Spout\Common\Helper\Escaper\ODS $escaper Used to unescape XML data
     */
    public function __construct($shouldFormatDates, $escaper)
    {
        $this->shouldFormatDates = $shouldFormatDates;
        $this->escaper = $escaper;
    }

    /**
     * Returns the (unescaped) correctly marshalled, cell value associated to the given XML node.
     * @see http://docs.oasis-open.org/office/v1.2/os/OpenDocument-v1.2-os-part1.html#refTable13
     *
     * @param \DOMNode $node
     * @throws InvalidValueException If the node value is not valid
     * @return string|int|float|bool|\DateTime|\DateInterval The value associated with the cell, empty string if cell's type is void/undefined
     */
    public function extractAndFormatNodeValue($node)
    {
        $cellType = $node->getAttribute(self::XML_ATTRIBUTE_TYPE);

        switch ($cellType) {
            case self::CELL_TYPE_STRING:
                return $this->formatStringCellValue($node);
            case self::CELL_TYPE_FLOAT:
                return $this->formatFloatCellValue($node);
            case self::CELL_TYPE_BOOLEAN:
                return $this->formatBooleanCellValue($node);
            case self::CELL_TYPE_DATE:
                return $this->formatDateCellValue($node);
            case self::CELL_TYPE_TIME:
                return $this->formatTimeCellValue($node);
            case self::CELL_TYPE_CURRENCY:
                return $this->formatCurrencyCellValue($node);
            case self::CELL_TYPE_PERCENTAGE:
                return $this->formatPercentageCellValue($node);
            case self::CELL_TYPE_VOID:
            default:
                return '';
        }
    }

    /**
     * Returns the cell String value.
     *
     * @param \DOMNode $node
     * @return string The value associated with the cell
     */
    protected function formatStringCellValue($node)
    {
        $pNodeValues = [];
        $pNodes = $node->getElementsByTagName(self::XML_NODE_P);

        foreach ($pNodes as $pNode) {
            $pNodeValues[] = $this->extractTextFromNode($pNode);
        }

        $escapedCellValue = implode("\n", $pNodeValues);
        $cellValue = $this->escaper->unescape($escapedCellValue);

        return $cellValue;
    }

    /**
     * @param \DOMNode $pNode
     * @return string
     */
    protected function extractTextFromNode($pNode)
    {
        $currentPValue = '';

        foreach ($pNode->childNodes as $childNode) {
            if ($childNode instanceof \DOMText) {
                $currentPValue .= $childNode->nodeValue;
            } elseif ($childNode->nodeName === self::XML_NODE_S) {
                $spaceAttribute = $childNode->getAttribute(self::XML_ATTRIBUTE_C);
                $numSpaces = (!empty($spaceAttribute)) ? (int) $spaceAttribute : 1;
                $currentPValue .= str_repeat(' ', $numSpaces);
            } elseif ($childNode->nodeName === self::XML_NODE_A || $childNode->nodeName === self::XML_NODE_SPAN) {
                $currentPValue .= $this->extractTextFromNode($childNode);
            }
        }

        return $currentPValue;
    }

    /**
     * Returns the cell Numeric value from the given node.
     *
     * @param \DOMNode $node
     * @return int|float The value associated with the cell
     */
    protected function formatFloatCellValue($node)
    {
        $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);

        $nodeIntValue = (int) $nodeValue;
        $nodeFloatValue = (float) $nodeValue;
        $cellValue = ((float) $nodeIntValue === $nodeFloatValue) ? $nodeIntValue : $nodeFloatValue;

        return $cellValue;
    }

    /**
     * Returns the cell Boolean value from the given node.
     *
     * @param \DOMNode $node
     * @return bool The value associated with the cell
     */
    protected function formatBooleanCellValue($node)
    {
        $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_BOOLEAN_VALUE);

        return (bool) $nodeValue;
    }

    /**
     * Returns the cell Date value from the given node.
     *
     * @param \DOMNode $node
     * @throws InvalidValueException If the value is not a valid date
     * @return \DateTime|string The value associated with the cell
     */
    protected function formatDateCellValue($node)
    {
        // The XML node looks like this:
        // <table:table-cell calcext:value-type="date" office:date-value="2016-05-19T16:39:00" office:value-type="date">
        //   <text:p>05/19/16 04:39 PM</text:p>
        // </table:table-cell>

        if ($this->shouldFormatDates) {
            // The date is already formatted in the "p" tag
            $nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
            $cellValue = $nodeWithValueAlreadyFormatted->nodeValue;
        } else {
            // otherwise, get it from the "date-value" attribute
            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_DATE_VALUE);
            try {
                $cellValue = new \DateTime($nodeValue);
            } catch (\Exception $e) {
                throw new InvalidValueException($nodeValue);
            }
        }

        return $cellValue;
    }

    /**
     * Returns the cell Time value from the given node.
     *
     * @param \DOMNode $node
     * @throws InvalidValueException If the value is not a valid time
     * @return \DateInterval|string The value associated with the cell
     */
    protected function formatTimeCellValue($node)
    {
        // The XML node looks like this:
        // <table:table-cell calcext:value-type="time" office:time-value="PT13H24M00S" office:value-type="time">
        //   <text:p>01:24:00 PM</text:p>
        // </table:table-cell>

        if ($this->shouldFormatDates) {
            // The date is already formatted in the "p" tag
            $nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
            $cellValue = $nodeWithValueAlreadyFormatted->nodeValue;
        } else {
            // otherwise, get it from the "time-value" attribute
            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_TIME_VALUE);
            try {
                $cellValue = new \DateInterval($nodeValue);
            } catch (\Exception $e) {
                throw new InvalidValueException($nodeValue);
            }
        }

        return $cellValue;
    }

    /**
     * Returns the cell Currency value from the given node.
     *
     * @param \DOMNode $node
     * @return string The value associated with the cell (e.g. "100 USD" or "9.99 EUR")
     */
    protected function formatCurrencyCellValue($node)
    {
        $value = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
        $currency = $node->getAttribute(self::XML_ATTRIBUTE_CURRENCY);

        return "$value $currency";
    }

    /**
     * Returns the cell Percentage value from the given node.
     *
     * @param \DOMNode $node
     * @return int|float The value associated with the cell
     */
    protected function formatPercentageCellValue($node)
    {
        // percentages are formatted like floats
        return $this->formatFloatCellValue($node);
    }
}


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

box / spout

Pull Request — master (#667)

CellValueFormatter::extractAndFormatNodeValue() B

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like