Parser.php at 12.5.2 · Jetpack – WP Security, Backup, Speed, & Growth

jetpack / vendor / nojimage / twitter-text-php / lib / Twitter / Text / Parser.php

Parser.php

191 lines 5.8 KB

Raw Download Zip

1	<?php
2
3	/**
4	* @author Takashi Nojima
5	* @copyright Copyright 2018, Takashi Nojima
6	* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0
7	* @package Twitter.Text
8	*/
9
10	namespace Twitter\Text;
11
12	/**
13	* Twitter Text Parser
14	*
15	* @author Takashi Nojima
16	* @copyright Copyright 2018, Takashi Nojima
17	* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0
18	* @package Twitter.Text
19	*/
20	class Parser
21	{
22	/**
23	* @var Configuration
24	*/
25	private $config;
26
27	/**
28	* Create a Parser
29	*
30	* @param Configuration $config
31	* @return Parser
32	*/
33	public static function create(Configuration $config = null)
34	{
35	return new self($config);
36	}
37
38	/**
39	* construct
40	*
41	* @param Configuration $config
42	*/
43	public function __construct(Configuration $config = null)
44	{
45	if ($config === null) {
46	$config = new Configuration();
47	}
48
49	$this->config = $config;
50	}
51
52	/**
53	* Parses a given tweet text with the weighted character count configuration
54	*
55	* @param string $tweet which is to be parsed
56	* @return ParseResults
57	*/
58	public function parseTweet($tweet)
59	{
60	if ($tweet === null \|\| '' === $tweet) {
61	return new ParseResults();
62	}
63
64	$normalizedTweet = StringUtils::normalizeFromNFC($tweet);
65	$normalizedTweetLength = StringUtils::strlen($normalizedTweet);
66
67	$emojiParsingEnabled = $this->config->getEmojiParsingEnabled();
68	$maxWeightedTweetLength = $this->config->getScaledMaxWeightedTweetLength();
69	$transformedUrlWeight = $this->config->getScaledTransformedURLLength();
70
71	$extractor = new Extractor();
72	$urlEntitiesMap = $this->transformEntitiesToHash($extractor->extractURLsWithIndices($normalizedTweet));
73	$emojiEntitiesMap = $emojiParsingEnabled
74	? $this->transformEntitiesToHash($extractor->extractEmojiWithIndices($normalizedTweet))
75	: array();
76
77	$hasInvalidCharacters = false;
78	$weightedCount = 0;
79	$offset = 0;
80	$displayOffset = 0;
81	$validOffset = 0;
82
83	while ($offset < $normalizedTweetLength) {
84	if (isset($urlEntitiesMap[$offset])) {
85	list($urlStart, $urlEnd) = $urlEntitiesMap[$offset]['indices'];
86	$urlLength = $urlEnd - $urlStart;
87
88	$weightedCount += $transformedUrlWeight;
89	$offset += $urlLength;
90	$displayOffset += $urlLength;
91	if ($weightedCount <= $maxWeightedTweetLength) {
92	$validOffset += $urlLength;
93	}
94	} elseif ($emojiParsingEnabled && isset($emojiEntitiesMap[$offset])) {
95	$emoji = $emojiEntitiesMap[$offset]['emoji'];
96	$emojiLength = StringUtils::strlen($emoji);
97	$charCount = StringUtils::charCount($emoji);
98
99	$weightedCount += $this->config->defaultWeight;
100	$offset += $emojiLength;
101	$displayOffset += $charCount;
102	if ($weightedCount <= $maxWeightedTweetLength) {
103	$validOffset += $charCount;
104	}
105	} else {
106	$char = StringUtils::substr($normalizedTweet, $offset, 1);
107
108	$hasInvalidCharacters = $hasInvalidCharacters \|\| $this->hasInvalidCharacters($char);
109	$charCount = StringUtils::strlen($char);
110	$charWidth = StringUtils::isSurrogatePair($char) ? 2 : 1;
111
112	$weightedCount += $this->getCharacterWeight($char, $this->config);
113	$offset += $charCount;
114	$displayOffset += $charWidth;
115
116	if (!$hasInvalidCharacters && $weightedCount <= $maxWeightedTweetLength) {
117	$validOffset += $charWidth;
118	}
119	}
120	}
121
122	$scaledWeightedLength = $weightedCount / $this->config->scale;
123	$permillage = $scaledWeightedLength * 1000 / $this->config->maxWeightedTweetLength;
124	$isValid = !$hasInvalidCharacters && $weightedCount <= $maxWeightedTweetLength;
125
126	$normalizedTweetOffset = StringUtils::strlen($tweet) - $normalizedTweetLength;
127	$displayTextRange = array(0, $displayOffset + $normalizedTweetOffset - 1);
128	$validTextRange = array(0, $validOffset + $normalizedTweetOffset - 1);
129
130	return new ParseResults($scaledWeightedLength, $permillage, $isValid, $displayTextRange, $validTextRange);
131	}
132
133	/**
134	* Convert to Hash by indices start
135	*
136	* @param array $entities
137	* @return array
138	*/
139	private function transformEntitiesToHash(array $entities)
140	{
141	return array_reduce($entities, function ($map, $entity) {
142	$map[$entity['indices'][0]] = $entity;
143
144	return $map;
145	}, array());
146	}
147
148	/**
149	* Get the character weight from ranges
150	*
151	* @param string $char the Character
152	* @param Configuration $config the parse configuration
153	* @return int
154	*/
155	private function getCharacterWeight($char, Configuration $config)
156	{
157	$codePoint = StringUtils::ord($char);
158
159	foreach ($config->ranges as $range) {
160	if ($this->inRange($codePoint, $range)) {
161	return $range['weight'];
162	}
163	}
164
165	return $config->defaultWeight;
166	}
167
168	/**
169	* check codepoint in range
170	*
171	* @param int $codePoint
172	* @param array $range
173	* @return boolean
174	*/
175	private function inRange($codePoint, array $range)
176	{
177	return ($codePoint >= $range['start'] && $codePoint <= $range['end']);
178	}
179
180	/**
181	* check has invalid characters
182	*
183	* @param string $char
184	* @return bool
185	*/
186	private function hasInvalidCharacters($char)
187	{
188	return preg_match(Regex::getInvalidCharactersMatcher(), $char);
189	}
190	}
191