Bump and prep 1.34.1
[lhc/web/wiklou.git] / languages / Language.php
1 <?php
2 /**
3 * Internationalisation code.
4 * See https://www.mediawiki.org/wiki/Special:MyLanguage/Localisation for more information.
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
10 *
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public License along
17 * with this program; if not, write to the Free Software Foundation, Inc.,
18 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
19 * http://www.gnu.org/copyleft/gpl.html
20 *
21 * @file
22 * @ingroup Language
23 */
24
25 /**
26 * @defgroup Language Language
27 */
28
29 use CLDRPluralRuleParser\Evaluator;
30 use MediaWiki\Languages\LanguageNameUtils;
31 use MediaWiki\MediaWikiServices;
32
33 /**
34 * Internationalisation code
35 * @ingroup Language
36 */
37 class Language {
38 /**
39 * Return autonyms in fetchLanguageName(s).
40 * @since 1.32
41 * @deprecated since 1.34, LanguageNameUtils::AUTONYMS
42 */
43 const AS_AUTONYMS = LanguageNameUtils::AUTONYMS;
44
45 /**
46 * Return all known languages in fetchLanguageName(s).
47 * @since 1.32
48 * @deprecated since 1.34, use LanguageNameUtils::ALL
49 */
50 const ALL = LanguageNameUtils::ALL;
51
52 /**
53 * Return in fetchLanguageName(s) only the languages for which we have at
54 * least some localisation.
55 * @since 1.32
56 * @deprecated since 1.34, use LanguageNameUtils::SUPPORTED
57 */
58 const SUPPORTED = LanguageNameUtils::SUPPORTED;
59
60 /**
61 * @var LanguageConverter|FakeConverter
62 */
63 public $mConverter;
64
65 public $mVariants, $mCode, $mLoaded = false;
66 public $mMagicExtensions = [];
67 private $mHtmlCode = null;
68 /** @var Language|false */
69 private $mParentLanguage = false;
70
71 public $dateFormatStrings = [];
72 public $mExtendedSpecialPageAliases;
73
74 /** @var array|null */
75 protected $namespaceNames;
76 protected $mNamespaceIds, $namespaceAliases;
77
78 /**
79 * ReplacementArray object caches
80 */
81 public $transformData = [];
82
83 /** @var LocalisationCache */
84 private $localisationCache;
85
86 /** @var LanguageNameUtils */
87 private $langNameUtils;
88
89 public static $mLangObjCache = [];
90
91 /**
92 * Return a fallback chain for messages in getFallbacksFor
93 * @since 1.32
94 */
95 const MESSAGES_FALLBACKS = 0;
96
97 /**
98 * Return a strict fallback chain in getFallbacksFor
99 * @since 1.32
100 */
101 const STRICT_FALLBACKS = 1;
102
103 // TODO Make these const once we drop HHVM support (T192166)
104 public static $mWeekdayMsgs = [
105 'sunday', 'monday', 'tuesday', 'wednesday', 'thursday',
106 'friday', 'saturday'
107 ];
108
109 public static $mWeekdayAbbrevMsgs = [
110 'sun', 'mon', 'tue', 'wed', 'thu', 'fri', 'sat'
111 ];
112
113 public static $mMonthMsgs = [
114 'january', 'february', 'march', 'april', 'may_long', 'june',
115 'july', 'august', 'september', 'october', 'november',
116 'december'
117 ];
118 public static $mMonthGenMsgs = [
119 'january-gen', 'february-gen', 'march-gen', 'april-gen', 'may-gen', 'june-gen',
120 'july-gen', 'august-gen', 'september-gen', 'october-gen', 'november-gen',
121 'december-gen'
122 ];
123 public static $mMonthAbbrevMsgs = [
124 'jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul', 'aug',
125 'sep', 'oct', 'nov', 'dec'
126 ];
127
128 public static $mIranianCalendarMonthMsgs = [
129 'iranian-calendar-m1', 'iranian-calendar-m2', 'iranian-calendar-m3',
130 'iranian-calendar-m4', 'iranian-calendar-m5', 'iranian-calendar-m6',
131 'iranian-calendar-m7', 'iranian-calendar-m8', 'iranian-calendar-m9',
132 'iranian-calendar-m10', 'iranian-calendar-m11', 'iranian-calendar-m12'
133 ];
134
135 public static $mHebrewCalendarMonthMsgs = [
136 'hebrew-calendar-m1', 'hebrew-calendar-m2', 'hebrew-calendar-m3',
137 'hebrew-calendar-m4', 'hebrew-calendar-m5', 'hebrew-calendar-m6',
138 'hebrew-calendar-m7', 'hebrew-calendar-m8', 'hebrew-calendar-m9',
139 'hebrew-calendar-m10', 'hebrew-calendar-m11', 'hebrew-calendar-m12',
140 'hebrew-calendar-m6a', 'hebrew-calendar-m6b'
141 ];
142
143 public static $mHebrewCalendarMonthGenMsgs = [
144 'hebrew-calendar-m1-gen', 'hebrew-calendar-m2-gen', 'hebrew-calendar-m3-gen',
145 'hebrew-calendar-m4-gen', 'hebrew-calendar-m5-gen', 'hebrew-calendar-m6-gen',
146 'hebrew-calendar-m7-gen', 'hebrew-calendar-m8-gen', 'hebrew-calendar-m9-gen',
147 'hebrew-calendar-m10-gen', 'hebrew-calendar-m11-gen', 'hebrew-calendar-m12-gen',
148 'hebrew-calendar-m6a-gen', 'hebrew-calendar-m6b-gen'
149 ];
150
151 public static $mHijriCalendarMonthMsgs = [
152 'hijri-calendar-m1', 'hijri-calendar-m2', 'hijri-calendar-m3',
153 'hijri-calendar-m4', 'hijri-calendar-m5', 'hijri-calendar-m6',
154 'hijri-calendar-m7', 'hijri-calendar-m8', 'hijri-calendar-m9',
155 'hijri-calendar-m10', 'hijri-calendar-m11', 'hijri-calendar-m12'
156 ];
157
158 /**
159 * @since 1.20
160 * @var array
161 */
162 public static $durationIntervals = [
163 'millennia' => 31556952000,
164 'centuries' => 3155695200,
165 'decades' => 315569520,
166 'years' => 31556952, // 86400 * ( 365 + ( 24 * 3 + 25 ) / 400 )
167 'weeks' => 604800,
168 'days' => 86400,
169 'hours' => 3600,
170 'minutes' => 60,
171 'seconds' => 1,
172 ];
173
174 /**
175 * Cache for language fallbacks.
176 * @see Language::getFallbacksIncludingSiteLanguage
177 * @since 1.21
178 * @var array
179 */
180 private static $fallbackLanguageCache = [];
181
182 /**
183 * Cache for grammar rules data
184 * @var MapCacheLRU|null
185 */
186 private static $grammarTransformations;
187
188 /**
189 * Unicode directional formatting characters, for embedBidi()
190 */
191 private static $lre = "\u{202A}"; // U+202A LEFT-TO-RIGHT EMBEDDING
192 private static $rle = "\u{202B}"; // U+202B RIGHT-TO-LEFT EMBEDDING
193 private static $pdf = "\u{202C}"; // U+202C POP DIRECTIONAL FORMATTING
194
195 /**
196 * Directionality test regex for embedBidi(). Matches the first strong directionality codepoint:
197 * - in group 1 if it is LTR
198 * - in group 2 if it is RTL
199 * Does not match if there is no strong directionality codepoint.
200 *
201 * The form is '/(?:([strong ltr codepoint])|([strong rtl codepoint]))/u' .
202 *
203 * Generated by UnicodeJS (see tools/strongDir) from the UCD; see
204 * https://phabricator.wikimedia.org/diffusion/GUJS/ .
205 */
206 // @codeCoverageIgnoreStart
207 // phpcs:ignore Generic.Files.LineLength
208 private static $strongDirRegex = '/(?:([\x{41}-\x{5a}\x{61}-\x{7a}\x{aa}\x{b5}\x{ba}\x{c0}-\x{d6}\x{d8}-\x{f6}\x{f8}-\x{2b8}\x{2bb}-\x{2c1}\x{2d0}\x{2d1}\x{2e0}-\x{2e4}\x{2ee}\x{370}-\x{373}\x{376}\x{377}\x{37a}-\x{37d}\x{37f}\x{386}\x{388}-\x{38a}\x{38c}\x{38e}-\x{3a1}\x{3a3}-\x{3f5}\x{3f7}-\x{482}\x{48a}-\x{52f}\x{531}-\x{556}\x{559}-\x{55f}\x{561}-\x{587}\x{589}\x{903}-\x{939}\x{93b}\x{93d}-\x{940}\x{949}-\x{94c}\x{94e}-\x{950}\x{958}-\x{961}\x{964}-\x{980}\x{982}\x{983}\x{985}-\x{98c}\x{98f}\x{990}\x{993}-\x{9a8}\x{9aa}-\x{9b0}\x{9b2}\x{9b6}-\x{9b9}\x{9bd}-\x{9c0}\x{9c7}\x{9c8}\x{9cb}\x{9cc}\x{9ce}\x{9d7}\x{9dc}\x{9dd}\x{9df}-\x{9e1}\x{9e6}-\x{9f1}\x{9f4}-\x{9fa}\x{a03}\x{a05}-\x{a0a}\x{a0f}\x{a10}\x{a13}-\x{a28}\x{a2a}-\x{a30}\x{a32}\x{a33}\x{a35}\x{a36}\x{a38}\x{a39}\x{a3e}-\x{a40}\x{a59}-\x{a5c}\x{a5e}\x{a66}-\x{a6f}\x{a72}-\x{a74}\x{a83}\x{a85}-\x{a8d}\x{a8f}-\x{a91}\x{a93}-\x{aa8}\x{aaa}-\x{ab0}\x{ab2}\x{ab3}\x{ab5}-\x{ab9}\x{abd}-\x{ac0}\x{ac9}\x{acb}\x{acc}\x{ad0}\x{ae0}\x{ae1}\x{ae6}-\x{af0}\x{af9}\x{b02}\x{b03}\x{b05}-\x{b0c}\x{b0f}\x{b10}\x{b13}-\x{b28}\x{b2a}-\x{b30}\x{b32}\x{b33}\x{b35}-\x{b39}\x{b3d}\x{b3e}\x{b40}\x{b47}\x{b48}\x{b4b}\x{b4c}\x{b57}\x{b5c}\x{b5d}\x{b5f}-\x{b61}\x{b66}-\x{b77}\x{b83}\x{b85}-\x{b8a}\x{b8e}-\x{b90}\x{b92}-\x{b95}\x{b99}\x{b9a}\x{b9c}\x{b9e}\x{b9f}\x{ba3}\x{ba4}\x{ba8}-\x{baa}\x{bae}-\x{bb9}\x{bbe}\x{bbf}\x{bc1}\x{bc2}\x{bc6}-\x{bc8}\x{bca}-\x{bcc}\x{bd0}\x{bd7}\x{be6}-\x{bf2}\x{c01}-\x{c03}\x{c05}-\x{c0c}\x{c0e}-\x{c10}\x{c12}-\x{c28}\x{c2a}-\x{c39}\x{c3d}\x{c41}-\x{c44}\x{c58}-\x{c5a}\x{c60}\x{c61}\x{c66}-\x{c6f}\x{c7f}\x{c82}\x{c83}\x{c85}-\x{c8c}\x{c8e}-\x{c90}\x{c92}-\x{ca8}\x{caa}-\x{cb3}\x{cb5}-\x{cb9}\x{cbd}-\x{cc4}\x{cc6}-\x{cc8}\x{cca}\x{ccb}\x{cd5}\x{cd6}\x{cde}\x{ce0}\x{ce1}\x{ce6}-\x{cef}\x{cf1}\x{cf2}\x{d02}\x{d03}\x{d05}-\x{d0c}\x{d0e}-\x{d10}\x{d12}-\x{d3a}\x{d3d}-\x{d40}\x{d46}-\x{d48}\x{d4a}-\x{d4c}\x{d4e}\x{d57}\x{d5f}-\x{d61}\x{d66}-\x{d75}\x{d79}-\x{d7f}\x{d82}\x{d83}\x{d85}-\x{d96}\x{d9a}-\x{db1}\x{db3}-\x{dbb}\x{dbd}\x{dc0}-\x{dc6}\x{dcf}-\x{dd1}\x{dd8}-\x{ddf}\x{de6}-\x{def}\x{df2}-\x{df4}\x{e01}-\x{e30}\x{e32}\x{e33}\x{e40}-\x{e46}\x{e4f}-\x{e5b}\x{e81}\x{e82}\x{e84}\x{e87}\x{e88}\x{e8a}\x{e8d}\x{e94}-\x{e97}\x{e99}-\x{e9f}\x{ea1}-\x{ea3}\x{ea5}\x{ea7}\x{eaa}\x{eab}\x{ead}-\x{eb0}\x{eb2}\x{eb3}\x{ebd}\x{ec0}-\x{ec4}\x{ec6}\x{ed0}-\x{ed9}\x{edc}-\x{edf}\x{f00}-\x{f17}\x{f1a}-\x{f34}\x{f36}\x{f38}\x{f3e}-\x{f47}\x{f49}-\x{f6c}\x{f7f}\x{f85}\x{f88}-\x{f8c}\x{fbe}-\x{fc5}\x{fc7}-\x{fcc}\x{fce}-\x{fda}\x{1000}-\x{102c}\x{1031}\x{1038}\x{103b}\x{103c}\x{103f}-\x{1057}\x{105a}-\x{105d}\x{1061}-\x{1070}\x{1075}-\x{1081}\x{1083}\x{1084}\x{1087}-\x{108c}\x{108e}-\x{109c}\x{109e}-\x{10c5}\x{10c7}\x{10cd}\x{10d0}-\x{1248}\x{124a}-\x{124d}\x{1250}-\x{1256}\x{1258}\x{125a}-\x{125d}\x{1260}-\x{1288}\x{128a}-\x{128d}\x{1290}-\x{12b0}\x{12b2}-\x{12b5}\x{12b8}-\x{12be}\x{12c0}\x{12c2}-\x{12c5}\x{12c8}-\x{12d6}\x{12d8}-\x{1310}\x{1312}-\x{1315}\x{1318}-\x{135a}\x{1360}-\x{137c}\x{1380}-\x{138f}\x{13a0}-\x{13f5}\x{13f8}-\x{13fd}\x{1401}-\x{167f}\x{1681}-\x{169a}\x{16a0}-\x{16f8}\x{1700}-\x{170c}\x{170e}-\x{1711}\x{1720}-\x{1731}\x{1735}\x{1736}\x{1740}-\x{1751}\x{1760}-\x{176c}\x{176e}-\x{1770}\x{1780}-\x{17b3}\x{17b6}\x{17be}-\x{17c5}\x{17c7}\x{17c8}\x{17d4}-\x{17da}\x{17dc}\x{17e0}-\x{17e9}\x{1810}-\x{1819}\x{1820}-\x{1877}\x{1880}-\x{18a8}\x{18aa}\x{18b0}-\x{18f5}\x{1900}-\x{191e}\x{1923}-\x{1926}\x{1929}-\x{192b}\x{1930}\x{1931}\x{1933}-\x{1938}\x{1946}-\x{196d}\x{1970}-\x{1974}\x{1980}-\x{19ab}\x{19b0}-\x{19c9}\x{19d0}-\x{19da}\x{1a00}-\x{1a16}\x{1a19}\x{1a1a}\x{1a1e}-\x{1a55}\x{1a57}\x{1a61}\x{1a63}\x{1a64}\x{1a6d}-\x{1a72}\x{1a80}-\x{1a89}\x{1a90}-\x{1a99}\x{1aa0}-\x{1aad}\x{1b04}-\x{1b33}\x{1b35}\x{1b3b}\x{1b3d}-\x{1b41}\x{1b43}-\x{1b4b}\x{1b50}-\x{1b6a}\x{1b74}-\x{1b7c}\x{1b82}-\x{1ba1}\x{1ba6}\x{1ba7}\x{1baa}\x{1bae}-\x{1be5}\x{1be7}\x{1bea}-\x{1bec}\x{1bee}\x{1bf2}\x{1bf3}\x{1bfc}-\x{1c2b}\x{1c34}\x{1c35}\x{1c3b}-\x{1c49}\x{1c4d}-\x{1c7f}\x{1cc0}-\x{1cc7}\x{1cd3}\x{1ce1}\x{1ce9}-\x{1cec}\x{1cee}-\x{1cf3}\x{1cf5}\x{1cf6}\x{1d00}-\x{1dbf}\x{1e00}-\x{1f15}\x{1f18}-\x{1f1d}\x{1f20}-\x{1f45}\x{1f48}-\x{1f4d}\x{1f50}-\x{1f57}\x{1f59}\x{1f5b}\x{1f5d}\x{1f5f}-\x{1f7d}\x{1f80}-\x{1fb4}\x{1fb6}-\x{1fbc}\x{1fbe}\x{1fc2}-\x{1fc4}\x{1fc6}-\x{1fcc}\x{1fd0}-\x{1fd3}\x{1fd6}-\x{1fdb}\x{1fe0}-\x{1fec}\x{1ff2}-\x{1ff4}\x{1ff6}-\x{1ffc}\x{200e}\x{2071}\x{207f}\x{2090}-\x{209c}\x{2102}\x{2107}\x{210a}-\x{2113}\x{2115}\x{2119}-\x{211d}\x{2124}\x{2126}\x{2128}\x{212a}-\x{212d}\x{212f}-\x{2139}\x{213c}-\x{213f}\x{2145}-\x{2149}\x{214e}\x{214f}\x{2160}-\x{2188}\x{2336}-\x{237a}\x{2395}\x{249c}-\x{24e9}\x{26ac}\x{2800}-\x{28ff}\x{2c00}-\x{2c2e}\x{2c30}-\x{2c5e}\x{2c60}-\x{2ce4}\x{2ceb}-\x{2cee}\x{2cf2}\x{2cf3}\x{2d00}-\x{2d25}\x{2d27}\x{2d2d}\x{2d30}-\x{2d67}\x{2d6f}\x{2d70}\x{2d80}-\x{2d96}\x{2da0}-\x{2da6}\x{2da8}-\x{2dae}\x{2db0}-\x{2db6}\x{2db8}-\x{2dbe}\x{2dc0}-\x{2dc6}\x{2dc8}-\x{2dce}\x{2dd0}-\x{2dd6}\x{2dd8}-\x{2dde}\x{3005}-\x{3007}\x{3021}-\x{3029}\x{302e}\x{302f}\x{3031}-\x{3035}\x{3038}-\x{303c}\x{3041}-\x{3096}\x{309d}-\x{309f}\x{30a1}-\x{30fa}\x{30fc}-\x{30ff}\x{3105}-\x{312d}\x{3131}-\x{318e}\x{3190}-\x{31ba}\x{31f0}-\x{321c}\x{3220}-\x{324f}\x{3260}-\x{327b}\x{327f}-\x{32b0}\x{32c0}-\x{32cb}\x{32d0}-\x{32fe}\x{3300}-\x{3376}\x{337b}-\x{33dd}\x{33e0}-\x{33fe}\x{3400}-\x{4db5}\x{4e00}-\x{9fd5}\x{a000}-\x{a48c}\x{a4d0}-\x{a60c}\x{a610}-\x{a62b}\x{a640}-\x{a66e}\x{a680}-\x{a69d}\x{a6a0}-\x{a6ef}\x{a6f2}-\x{a6f7}\x{a722}-\x{a787}\x{a789}-\x{a7ad}\x{a7b0}-\x{a7b7}\x{a7f7}-\x{a801}\x{a803}-\x{a805}\x{a807}-\x{a80a}\x{a80c}-\x{a824}\x{a827}\x{a830}-\x{a837}\x{a840}-\x{a873}\x{a880}-\x{a8c3}\x{a8ce}-\x{a8d9}\x{a8f2}-\x{a8fd}\x{a900}-\x{a925}\x{a92e}-\x{a946}\x{a952}\x{a953}\x{a95f}-\x{a97c}\x{a983}-\x{a9b2}\x{a9b4}\x{a9b5}\x{a9ba}\x{a9bb}\x{a9bd}-\x{a9cd}\x{a9cf}-\x{a9d9}\x{a9de}-\x{a9e4}\x{a9e6}-\x{a9fe}\x{aa00}-\x{aa28}\x{aa2f}\x{aa30}\x{aa33}\x{aa34}\x{aa40}-\x{aa42}\x{aa44}-\x{aa4b}\x{aa4d}\x{aa50}-\x{aa59}\x{aa5c}-\x{aa7b}\x{aa7d}-\x{aaaf}\x{aab1}\x{aab5}\x{aab6}\x{aab9}-\x{aabd}\x{aac0}\x{aac2}\x{aadb}-\x{aaeb}\x{aaee}-\x{aaf5}\x{ab01}-\x{ab06}\x{ab09}-\x{ab0e}\x{ab11}-\x{ab16}\x{ab20}-\x{ab26}\x{ab28}-\x{ab2e}\x{ab30}-\x{ab65}\x{ab70}-\x{abe4}\x{abe6}\x{abe7}\x{abe9}-\x{abec}\x{abf0}-\x{abf9}\x{ac00}-\x{d7a3}\x{d7b0}-\x{d7c6}\x{d7cb}-\x{d7fb}\x{e000}-\x{fa6d}\x{fa70}-\x{fad9}\x{fb00}-\x{fb06}\x{fb13}-\x{fb17}\x{ff21}-\x{ff3a}\x{ff41}-\x{ff5a}\x{ff66}-\x{ffbe}\x{ffc2}-\x{ffc7}\x{ffca}-\x{ffcf}\x{ffd2}-\x{ffd7}\x{ffda}-\x{ffdc}\x{10000}-\x{1000b}\x{1000d}-\x{10026}\x{10028}-\x{1003a}\x{1003c}\x{1003d}\x{1003f}-\x{1004d}\x{10050}-\x{1005d}\x{10080}-\x{100fa}\x{10100}\x{10102}\x{10107}-\x{10133}\x{10137}-\x{1013f}\x{101d0}-\x{101fc}\x{10280}-\x{1029c}\x{102a0}-\x{102d0}\x{10300}-\x{10323}\x{10330}-\x{1034a}\x{10350}-\x{10375}\x{10380}-\x{1039d}\x{1039f}-\x{103c3}\x{103c8}-\x{103d5}\x{10400}-\x{1049d}\x{104a0}-\x{104a9}\x{10500}-\x{10527}\x{10530}-\x{10563}\x{1056f}\x{10600}-\x{10736}\x{10740}-\x{10755}\x{10760}-\x{10767}\x{11000}\x{11002}-\x{11037}\x{11047}-\x{1104d}\x{11066}-\x{1106f}\x{11082}-\x{110b2}\x{110b7}\x{110b8}\x{110bb}-\x{110c1}\x{110d0}-\x{110e8}\x{110f0}-\x{110f9}\x{11103}-\x{11126}\x{1112c}\x{11136}-\x{11143}\x{11150}-\x{11172}\x{11174}-\x{11176}\x{11182}-\x{111b5}\x{111bf}-\x{111c9}\x{111cd}\x{111d0}-\x{111df}\x{111e1}-\x{111f4}\x{11200}-\x{11211}\x{11213}-\x{1122e}\x{11232}\x{11233}\x{11235}\x{11238}-\x{1123d}\x{11280}-\x{11286}\x{11288}\x{1128a}-\x{1128d}\x{1128f}-\x{1129d}\x{1129f}-\x{112a9}\x{112b0}-\x{112de}\x{112e0}-\x{112e2}\x{112f0}-\x{112f9}\x{11302}\x{11303}\x{11305}-\x{1130c}\x{1130f}\x{11310}\x{11313}-\x{11328}\x{1132a}-\x{11330}\x{11332}\x{11333}\x{11335}-\x{11339}\x{1133d}-\x{1133f}\x{11341}-\x{11344}\x{11347}\x{11348}\x{1134b}-\x{1134d}\x{11350}\x{11357}\x{1135d}-\x{11363}\x{11480}-\x{114b2}\x{114b9}\x{114bb}-\x{114be}\x{114c1}\x{114c4}-\x{114c7}\x{114d0}-\x{114d9}\x{11580}-\x{115b1}\x{115b8}-\x{115bb}\x{115be}\x{115c1}-\x{115db}\x{11600}-\x{11632}\x{1163b}\x{1163c}\x{1163e}\x{11641}-\x{11644}\x{11650}-\x{11659}\x{11680}-\x{116aa}\x{116ac}\x{116ae}\x{116af}\x{116b6}\x{116c0}-\x{116c9}\x{11700}-\x{11719}\x{11720}\x{11721}\x{11726}\x{11730}-\x{1173f}\x{118a0}-\x{118f2}\x{118ff}\x{11ac0}-\x{11af8}\x{12000}-\x{12399}\x{12400}-\x{1246e}\x{12470}-\x{12474}\x{12480}-\x{12543}\x{13000}-\x{1342e}\x{14400}-\x{14646}\x{16800}-\x{16a38}\x{16a40}-\x{16a5e}\x{16a60}-\x{16a69}\x{16a6e}\x{16a6f}\x{16ad0}-\x{16aed}\x{16af5}\x{16b00}-\x{16b2f}\x{16b37}-\x{16b45}\x{16b50}-\x{16b59}\x{16b5b}-\x{16b61}\x{16b63}-\x{16b77}\x{16b7d}-\x{16b8f}\x{16f00}-\x{16f44}\x{16f50}-\x{16f7e}\x{16f93}-\x{16f9f}\x{1b000}\x{1b001}\x{1bc00}-\x{1bc6a}\x{1bc70}-\x{1bc7c}\x{1bc80}-\x{1bc88}\x{1bc90}-\x{1bc99}\x{1bc9c}\x{1bc9f}\x{1d000}-\x{1d0f5}\x{1d100}-\x{1d126}\x{1d129}-\x{1d166}\x{1d16a}-\x{1d172}\x{1d183}\x{1d184}\x{1d18c}-\x{1d1a9}\x{1d1ae}-\x{1d1e8}\x{1d360}-\x{1d371}\x{1d400}-\x{1d454}\x{1d456}-\x{1d49c}\x{1d49e}\x{1d49f}\x{1d4a2}\x{1d4a5}\x{1d4a6}\x{1d4a9}-\x{1d4ac}\x{1d4ae}-\x{1d4b9}\x{1d4bb}\x{1d4bd}-\x{1d4c3}\x{1d4c5}-\x{1d505}\x{1d507}-\x{1d50a}\x{1d50d}-\x{1d514}\x{1d516}-\x{1d51c}\x{1d51e}-\x{1d539}\x{1d53b}-\x{1d53e}\x{1d540}-\x{1d544}\x{1d546}\x{1d54a}-\x{1d550}\x{1d552}-\x{1d6a5}\x{1d6a8}-\x{1d6da}\x{1d6dc}-\x{1d714}\x{1d716}-\x{1d74e}\x{1d750}-\x{1d788}\x{1d78a}-\x{1d7c2}\x{1d7c4}-\x{1d7cb}\x{1d800}-\x{1d9ff}\x{1da37}-\x{1da3a}\x{1da6d}-\x{1da74}\x{1da76}-\x{1da83}\x{1da85}-\x{1da8b}\x{1f110}-\x{1f12e}\x{1f130}-\x{1f169}\x{1f170}-\x{1f19a}\x{1f1e6}-\x{1f202}\x{1f210}-\x{1f23a}\x{1f240}-\x{1f248}\x{1f250}\x{1f251}\x{20000}-\x{2a6d6}\x{2a700}-\x{2b734}\x{2b740}-\x{2b81d}\x{2b820}-\x{2cea1}\x{2f800}-\x{2fa1d}\x{f0000}-\x{ffffd}\x{100000}-\x{10fffd}])|([\x{590}\x{5be}\x{5c0}\x{5c3}\x{5c6}\x{5c8}-\x{5ff}\x{7c0}-\x{7ea}\x{7f4}\x{7f5}\x{7fa}-\x{815}\x{81a}\x{824}\x{828}\x{82e}-\x{858}\x{85c}-\x{89f}\x{200f}\x{fb1d}\x{fb1f}-\x{fb28}\x{fb2a}-\x{fb4f}\x{10800}-\x{1091e}\x{10920}-\x{10a00}\x{10a04}\x{10a07}-\x{10a0b}\x{10a10}-\x{10a37}\x{10a3b}-\x{10a3e}\x{10a40}-\x{10ae4}\x{10ae7}-\x{10b38}\x{10b40}-\x{10e5f}\x{10e7f}-\x{10fff}\x{1e800}-\x{1e8cf}\x{1e8d7}-\x{1edff}\x{1ef00}-\x{1efff}\x{608}\x{60b}\x{60d}\x{61b}-\x{64a}\x{66d}-\x{66f}\x{671}-\x{6d5}\x{6e5}\x{6e6}\x{6ee}\x{6ef}\x{6fa}-\x{710}\x{712}-\x{72f}\x{74b}-\x{7a5}\x{7b1}-\x{7bf}\x{8a0}-\x{8e2}\x{fb50}-\x{fd3d}\x{fd40}-\x{fdcf}\x{fdf0}-\x{fdfc}\x{fdfe}\x{fdff}\x{fe70}-\x{fefe}\x{1ee00}-\x{1eeef}\x{1eef2}-\x{1eeff}]))/u';
209 // @codeCoverageIgnoreEnd
210
211 /**
212 * Get a cached or new language object for a given language code
213 * @param string $code
214 * @throws MWException
215 * @return Language
216 */
217 static function factory( $code ) {
218 global $wgDummyLanguageCodes, $wgLangObjCacheSize;
219
220 if ( isset( $wgDummyLanguageCodes[$code] ) ) {
221 $code = $wgDummyLanguageCodes[$code];
222 }
223
224 // get the language object to process
225 $langObj = self::$mLangObjCache[$code] ?? self::newFromCode( $code );
226
227 // merge the language object in to get it up front in the cache
228 self::$mLangObjCache = array_merge( [ $code => $langObj ], self::$mLangObjCache );
229 // get rid of the oldest ones in case we have an overflow
230 self::$mLangObjCache = array_slice( self::$mLangObjCache, 0, $wgLangObjCacheSize, true );
231
232 return $langObj;
233 }
234
235 /**
236 * Create a language object for a given language code
237 * @param string $code
238 * @param bool $fallback Whether we're going through language fallback chain
239 * @throws MWException
240 * @return Language
241 */
242 protected static function newFromCode( $code, $fallback = false ) {
243 $langNameUtils = MediaWikiServices::getInstance()->getLanguageNameUtils();
244 if ( !$langNameUtils->isValidCode( $code ) ) {
245 throw new MWException( "Invalid language code \"$code\"" );
246 }
247
248 if ( !$langNameUtils->isValidBuiltInCode( $code ) ) {
249 // It's not possible to customise this code with class files, so
250 // just return a Language object. This is to support uselang= hacks.
251 $lang = new Language;
252 $lang->mCode = $code;
253 return $lang;
254 }
255
256 // Check if there is a language class for the code
257 $class = self::classFromCode( $code, $fallback );
258 // LanguageCode does not inherit Language
259 if ( class_exists( $class ) && is_a( $class, 'Language', true ) ) {
260 $lang = new $class;
261 return $lang;
262 }
263
264 // Keep trying the fallback list until we find an existing class
265 $fallbacks = self::getFallbacksFor( $code );
266 foreach ( $fallbacks as $fallbackCode ) {
267 if ( !$langNameUtils->isValidBuiltInCode( $fallbackCode ) ) {
268 throw new MWException( "Invalid fallback '$fallbackCode' in fallback sequence for '$code'" );
269 }
270
271 $class = self::classFromCode( $fallbackCode );
272 if ( class_exists( $class ) ) {
273 $lang = new $class;
274 $lang->mCode = $code;
275 return $lang;
276 }
277 }
278
279 throw new MWException( "Invalid fallback sequence for language '$code'" );
280 }
281
282 /**
283 * Intended for tests that may change configuration in a way that invalidates caches.
284 *
285 * @since 1.32
286 */
287 public static function clearCaches() {
288 if ( !defined( 'MW_PHPUNIT_TEST' ) && !defined( 'MEDIAWIKI_INSTALL' ) ) {
289 throw new MWException( __METHOD__ . ' must not be used outside tests/installer' );
290 }
291 if ( defined( 'MW_PHPUNIT_TEST' ) ) {
292 MediaWikiServices::getInstance()->resetServiceForTesting( 'LocalisationCache' );
293 MediaWikiServices::getInstance()->resetServiceForTesting( 'LanguageNameUtils' );
294 }
295 self::$mLangObjCache = [];
296 self::$fallbackLanguageCache = [];
297 self::$grammarTransformations = null;
298 }
299
300 /**
301 * Checks whether any localisation is available for that language tag
302 * in MediaWiki (MessagesXx.php exists).
303 *
304 * @deprecated since 1.34, use LanguageNameUtils
305 * @param string $code Language tag (in lower case)
306 * @return bool Whether language is supported
307 * @since 1.21
308 */
309 public static function isSupportedLanguage( $code ) {
310 return MediaWikiServices::getInstance()->getLanguageNameUtils()
311 ->isSupportedLanguage( $code );
312 }
313
314 /**
315 * Returns true if a language code string is a well-formed language tag
316 * according to RFC 5646.
317 * This function only checks well-formedness; it doesn't check that
318 * language, script or variant codes actually exist in the repositories.
319 *
320 * Based on regexes by Mark Davis of the Unicode Consortium:
321 * https://www.unicode.org/repos/cldr/trunk/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
322 *
323 * @param string $code
324 * @param bool $lenient Whether to allow '_' as separator. The default is only '-'.
325 *
326 * @return bool
327 * @since 1.21
328 */
329 public static function isWellFormedLanguageTag( $code, $lenient = false ) {
330 $alpha = '[a-z]';
331 $digit = '[0-9]';
332 $alphanum = '[a-z0-9]';
333 $x = 'x'; # private use singleton
334 $singleton = '[a-wy-z]'; # other singleton
335 $s = $lenient ? '[-_]' : '-';
336
337 $language = "$alpha{2,8}|$alpha{2,3}$s$alpha{3}";
338 $script = "$alpha{4}"; # ISO 15924
339 $region = "(?:$alpha{2}|$digit{3})"; # ISO 3166-1 alpha-2 or UN M.49
340 $variant = "(?:$alphanum{5,8}|$digit$alphanum{3})";
341 $extension = "$singleton(?:$s$alphanum{2,8})+";
342 $privateUse = "$x(?:$s$alphanum{1,8})+";
343
344 # Define certain grandfathered codes, since otherwise the regex is pretty useless.
345 # Since these are limited, this is safe even later changes to the registry --
346 # the only oddity is that it might change the type of the tag, and thus
347 # the results from the capturing groups.
348 # https://www.iana.org/assignments/language-subtag-registry
349
350 $grandfathered = "en{$s}GB{$s}oed"
351 . "|i{$s}(?:ami|bnn|default|enochian|hak|klingon|lux|mingo|navajo|pwn|tao|tay|tsu)"
352 . "|no{$s}(?:bok|nyn)"
353 . "|sgn{$s}(?:BE{$s}(?:fr|nl)|CH{$s}de)"
354 . "|zh{$s}min{$s}nan";
355
356 $variantList = "$variant(?:$s$variant)*";
357 $extensionList = "$extension(?:$s$extension)*";
358
359 $langtag = "(?:($language)"
360 . "(?:$s$script)?"
361 . "(?:$s$region)?"
362 . "(?:$s$variantList)?"
363 . "(?:$s$extensionList)?"
364 . "(?:$s$privateUse)?)";
365
366 # The final breakdown, with capturing groups for each of these components
367 # The variants, extensions, grandfathered, and private-use may have interior '-'
368
369 $root = "^(?:$langtag|$privateUse|$grandfathered)$";
370
371 return (bool)preg_match( "/$root/", strtolower( $code ) );
372 }
373
374 /**
375 * Returns true if a language code string is of a valid form, whether or
376 * not it exists. This includes codes which are used solely for
377 * customisation via the MediaWiki namespace.
378 *
379 * @deprecated since 1.34, use LanguageNameUtils
380 *
381 * @param string $code
382 *
383 * @return bool
384 */
385 public static function isValidCode( $code ) {
386 return MediaWikiServices::getInstance()->getLanguageNameUtils()->isValidCode( $code );
387 }
388
389 /**
390 * Returns true if a language code is of a valid form for the purposes of
391 * internal customisation of MediaWiki, via Messages*.php or *.json.
392 *
393 * @deprecated since 1.34, use LanguageNameUtils
394 *
395 * @param string $code
396 *
397 * @since 1.18
398 * @return bool
399 */
400 public static function isValidBuiltInCode( $code ) {
401 return MediaWikiServices::getInstance()->getLanguageNameUtils()
402 ->isValidBuiltInCode( $code );
403 }
404
405 /**
406 * Returns true if a language code is an IETF tag known to MediaWiki.
407 *
408 * @deprecated since 1.34, use LanguageNameUtils
409 *
410 * @param string $tag
411 *
412 * @since 1.21
413 * @return bool
414 */
415 public static function isKnownLanguageTag( $tag ) {
416 return MediaWikiServices::getInstance()->getLanguageNameUtils()
417 ->isKnownLanguageTag( $tag );
418 }
419
420 /**
421 * Get the LocalisationCache instance
422 *
423 * @deprecated since 1.34, use MediaWikiServices
424 * @return LocalisationCache
425 */
426 public static function getLocalisationCache() {
427 return MediaWikiServices::getInstance()->getLocalisationCache();
428 }
429
430 function __construct() {
431 $this->mConverter = new FakeConverter( $this );
432 // Set the code to the name of the descendant
433 if ( static::class === 'Language' ) {
434 $this->mCode = 'en';
435 } else {
436 $this->mCode = str_replace( '_', '-', strtolower( substr( static::class, 8 ) ) );
437 }
438 $services = MediaWikiServices::getInstance();
439 $this->localisationCache = $services->getLocalisationCache();
440 $this->langNameUtils = $services->getLanguageNameUtils();
441 }
442
443 /**
444 * Reduce memory usage
445 * @suppress PhanTypeSuspiciousNonTraversableForeach
446 */
447 function __destruct() {
448 foreach ( $this as $name => $value ) {
449 unset( $this->$name );
450 }
451 }
452
453 /**
454 * Hook which will be called if this is the content language.
455 * Descendants can use this to register hook functions or modify globals
456 */
457 function initContLang() {
458 }
459
460 /**
461 * @return array
462 * @since 1.19
463 */
464 public function getFallbackLanguages() {
465 return self::getFallbacksFor( $this->mCode );
466 }
467
468 /**
469 * Exports $wgBookstoreListEn
470 * @return array
471 */
472 public function getBookstoreList() {
473 return $this->localisationCache->getItem( $this->mCode, 'bookstoreList' );
474 }
475
476 /**
477 * Returns an array of localised namespaces indexed by their numbers. If the namespace is not
478 * available in localised form, it will be included in English.
479 *
480 * @return string[] List of localized namespace names, indexed by numeric namespace ID.
481 */
482 public function getNamespaces() {
483 if ( is_null( $this->namespaceNames ) ) {
484 global $wgMetaNamespace, $wgMetaNamespaceTalk, $wgExtraNamespaces;
485
486 $validNamespaces = MediaWikiServices::getInstance()->getNamespaceInfo()->
487 getCanonicalNamespaces();
488
489 $this->namespaceNames = $wgExtraNamespaces +
490 $this->localisationCache->getItem( $this->mCode, 'namespaceNames' );
491 $this->namespaceNames += $validNamespaces;
492
493 $this->namespaceNames[NS_PROJECT] = $wgMetaNamespace;
494 if ( $wgMetaNamespaceTalk ) {
495 $this->namespaceNames[NS_PROJECT_TALK] = $wgMetaNamespaceTalk;
496 } else {
497 $talk = $this->namespaceNames[NS_PROJECT_TALK];
498 $this->namespaceNames[NS_PROJECT_TALK] =
499 $this->fixVariableInNamespace( $talk );
500 }
501
502 # Sometimes a language will be localised but not actually exist on this wiki.
503 foreach ( $this->namespaceNames as $key => $text ) {
504 if ( !isset( $validNamespaces[$key] ) ) {
505 unset( $this->namespaceNames[$key] );
506 }
507 }
508
509 # The above mixing may leave namespaces out of canonical order.
510 # Re-order by namespace ID number...
511 ksort( $this->namespaceNames );
512
513 Hooks::run( 'LanguageGetNamespaces', [ &$this->namespaceNames ] );
514 }
515
516 return $this->namespaceNames;
517 }
518
519 /**
520 * Arbitrarily set all of the namespace names at once. Mainly used for testing
521 * @param array $namespaces Array of namespaces (id => name)
522 */
523 public function setNamespaces( array $namespaces ) {
524 $this->namespaceNames = $namespaces;
525 $this->mNamespaceIds = null;
526 }
527
528 /**
529 * Resets all of the namespace caches. Mainly used for testing
530 */
531 public function resetNamespaces() {
532 $this->namespaceNames = null;
533 $this->mNamespaceIds = null;
534 $this->namespaceAliases = null;
535 }
536
537 /**
538 * A convenience function that returns getNamespaces() with spaces instead of underscores
539 * in values. Useful for producing output to be displayed e.g. in `<select>` forms.
540 *
541 * @return array
542 */
543 public function getFormattedNamespaces() {
544 $ns = $this->getNamespaces();
545 foreach ( $ns as $k => $v ) {
546 $ns[$k] = strtr( $v, '_', ' ' );
547 }
548 return $ns;
549 }
550
551 /**
552 * Get a namespace value by key
553 *
554 * <code>
555 * $mw_ns = $lang->getNsText( NS_MEDIAWIKI );
556 * echo $mw_ns; // prints 'MediaWiki'
557 * </code>
558 *
559 * @param int $index The array key of the namespace to return
560 * @return string|bool String if the namespace value exists, otherwise false
561 */
562 public function getNsText( $index ) {
563 $ns = $this->getNamespaces();
564 return $ns[$index] ?? false;
565 }
566
567 /**
568 * A convenience function that returns the same thing as
569 * getNsText() except with '_' changed to ' ', useful for
570 * producing output.
571 *
572 * <code>
573 * $mw_ns = $lang->getFormattedNsText( NS_MEDIAWIKI_TALK );
574 * echo $mw_ns; // prints 'MediaWiki talk'
575 * </code>
576 *
577 * @param int $index The array key of the namespace to return
578 * @return string Namespace name without underscores (empty string if namespace does not exist)
579 */
580 public function getFormattedNsText( $index ) {
581 $ns = $this->getNsText( $index );
582 return strtr( $ns, '_', ' ' );
583 }
584
585 /**
586 * Returns gender-dependent namespace alias if available.
587 * See https://www.mediawiki.org/wiki/Manual:$wgExtraGenderNamespaces
588 * @param int $index Namespace index
589 * @param string $gender Gender key (male, female... )
590 * @return string
591 * @since 1.18
592 */
593 public function getGenderNsText( $index, $gender ) {
594 global $wgExtraGenderNamespaces;
595
596 $ns = $wgExtraGenderNamespaces +
597 (array)$this->localisationCache->getItem( $this->mCode, 'namespaceGenderAliases' );
598
599 return $ns[$index][$gender] ?? $this->getNsText( $index );
600 }
601
602 /**
603 * Whether this language uses gender-dependent namespace aliases.
604 * See https://www.mediawiki.org/wiki/Manual:$wgExtraGenderNamespaces
605 * @return bool
606 * @since 1.18
607 */
608 public function needsGenderDistinction() {
609 global $wgExtraGenderNamespaces, $wgExtraNamespaces;
610 if ( count( $wgExtraGenderNamespaces ) > 0 ) {
611 // $wgExtraGenderNamespaces overrides everything
612 return true;
613 } elseif ( isset( $wgExtraNamespaces[NS_USER] ) && isset( $wgExtraNamespaces[NS_USER_TALK] ) ) {
614 /// @todo There may be other gender namespace than NS_USER & NS_USER_TALK in the future
615 // $wgExtraNamespaces overrides any gender aliases specified in i18n files
616 return false;
617 } else {
618 // Check what is in i18n files
619 $aliases = $this->localisationCache->getItem( $this->mCode, 'namespaceGenderAliases' );
620 return count( $aliases ) > 0;
621 }
622 }
623
624 /**
625 * Get a namespace key by value, case insensitive.
626 * Only matches namespace names for the current language, not the
627 * canonical ones defined in Namespace.php.
628 *
629 * @param string $text
630 * @return int|bool An integer if $text is a valid value otherwise false
631 */
632 function getLocalNsIndex( $text ) {
633 $lctext = $this->lc( $text );
634 $ids = $this->getNamespaceIds();
635 return $ids[$lctext] ?? false;
636 }
637
638 /**
639 * @return array
640 */
641 public function getNamespaceAliases() {
642 if ( is_null( $this->namespaceAliases ) ) {
643 $aliases = $this->localisationCache->getItem( $this->mCode, 'namespaceAliases' );
644 if ( !$aliases ) {
645 $aliases = [];
646 } else {
647 foreach ( $aliases as $name => $index ) {
648 if ( $index === NS_PROJECT_TALK ) {
649 unset( $aliases[$name] );
650 $name = $this->fixVariableInNamespace( $name );
651 $aliases[$name] = $index;
652 }
653 }
654 }
655
656 global $wgExtraGenderNamespaces;
657 $genders = $wgExtraGenderNamespaces + (array)$this->localisationCache
658 ->getItem( $this->mCode, 'namespaceGenderAliases' );
659 foreach ( $genders as $index => $forms ) {
660 foreach ( $forms as $alias ) {
661 $aliases[$alias] = $index;
662 }
663 }
664
665 # Also add converted namespace names as aliases, to avoid confusion.
666 $convertedNames = [];
667 foreach ( $this->getVariants() as $variant ) {
668 if ( $variant === $this->mCode ) {
669 continue;
670 }
671 foreach ( $this->getNamespaces() as $ns => $_ ) {
672 $convertedNames[$this->getConverter()->convertNamespace( $ns, $variant )] = $ns;
673 }
674 }
675
676 $this->namespaceAliases = $aliases + $convertedNames;
677
678 # Filter out aliases to namespaces that don't exist, e.g. from extensions
679 # that aren't loaded here but are included in the l10n cache.
680 # (array_intersect preserves keys from its first argument)
681 $this->namespaceAliases = array_intersect(
682 $this->namespaceAliases,
683 array_keys( $this->getNamespaces() )
684 );
685 }
686
687 return $this->namespaceAliases;
688 }
689
690 /**
691 * @return array
692 */
693 public function getNamespaceIds() {
694 if ( is_null( $this->mNamespaceIds ) ) {
695 global $wgNamespaceAliases;
696 # Put namespace names and aliases into a hashtable.
697 # If this is too slow, then we should arrange it so that it is done
698 # before caching. The catch is that at pre-cache time, the above
699 # class-specific fixup hasn't been done.
700 $this->mNamespaceIds = [];
701 foreach ( $this->getNamespaces() as $index => $name ) {
702 $this->mNamespaceIds[$this->lc( $name )] = $index;
703 }
704 foreach ( $this->getNamespaceAliases() as $name => $index ) {
705 $this->mNamespaceIds[$this->lc( $name )] = $index;
706 }
707 if ( $wgNamespaceAliases ) {
708 foreach ( $wgNamespaceAliases as $name => $index ) {
709 $this->mNamespaceIds[$this->lc( $name )] = $index;
710 }
711 }
712 }
713 return $this->mNamespaceIds;
714 }
715
716 /**
717 * Get a namespace key by value, case insensitive. Canonical namespace
718 * names override custom ones defined for the current language.
719 *
720 * @param string $text
721 * @return int|bool An integer if $text is a valid value otherwise false
722 */
723 public function getNsIndex( $text ) {
724 $lctext = $this->lc( $text );
725 $ns = MediaWikiServices::getInstance()->getNamespaceInfo()->
726 getCanonicalIndex( $lctext );
727 if ( $ns !== null ) {
728 return $ns;
729 }
730 $ids = $this->getNamespaceIds();
731 return $ids[$lctext] ?? false;
732 }
733
734 /**
735 * short names for language variants used for language conversion links.
736 *
737 * @param string $code
738 * @param bool $usemsg Use the "variantname-xyz" message if it exists
739 * @return string
740 */
741 public function getVariantname( $code, $usemsg = true ) {
742 $msg = "variantname-$code";
743 if ( $usemsg && wfMessage( $msg )->exists() ) {
744 return $this->getMessageFromDB( $msg );
745 }
746 $name = $this->langNameUtils->getLanguageName( $code );
747 if ( $name ) {
748 return $name; # if it's defined as a language name, show that
749 } else {
750 # otherwise, output the language code
751 return $code;
752 }
753 }
754
755 /**
756 * @return string[]|bool List of date format preference keys, or false if disabled.
757 */
758 public function getDatePreferences() {
759 return $this->localisationCache->getItem( $this->mCode, 'datePreferences' );
760 }
761
762 /**
763 * @return array
764 */
765 function getDateFormats() {
766 return $this->localisationCache->getItem( $this->mCode, 'dateFormats' );
767 }
768
769 /**
770 * @return array|string
771 */
772 public function getDefaultDateFormat() {
773 $df = $this->localisationCache->getItem( $this->mCode, 'defaultDateFormat' );
774 if ( $df === 'dmy or mdy' ) {
775 global $wgAmericanDates;
776 return $wgAmericanDates ? 'mdy' : 'dmy';
777 } else {
778 return $df;
779 }
780 }
781
782 /**
783 * @return array
784 */
785 public function getDatePreferenceMigrationMap() {
786 return $this->localisationCache->getItem( $this->mCode, 'datePreferenceMigrationMap' );
787 }
788
789 /**
790 * @return array
791 */
792 public function getExtraUserToggles() {
793 wfDeprecated( __METHOD__, '1.34' );
794 return [];
795 }
796
797 /**
798 * @param string $tog
799 * @return string
800 */
801 function getUserToggle( $tog ) {
802 return $this->getMessageFromDB( "tog-$tog" );
803 }
804
805 /**
806 * Get an array of language names, indexed by code.
807 *
808 * @deprecated since 1.34, use LanguageNameUtils::getLanguageNames
809 * @param null|string $inLanguage Code of language in which to return the names
810 * Use self::AS_AUTONYMS for autonyms (native names)
811 * @param string $include One of:
812 * self::ALL all available languages
813 * 'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
814 * self::SUPPORTED only if the language is in 'mw' *and* has a message file
815 * @return array Language code => language name (sorted by key)
816 * @since 1.20
817 */
818 public static function fetchLanguageNames( $inLanguage = self::AS_AUTONYMS, $include = 'mw' ) {
819 return MediaWikiServices::getInstance()->getLanguageNameUtils()
820 ->getLanguageNames( $inLanguage, $include );
821 }
822
823 /**
824 * @deprecated since 1.34, use LanguageNameUtils::getLanguageName
825 * @param string $code The code of the language for which to get the name
826 * @param null|string $inLanguage Code of language in which to return the name
827 * (SELF::AS_AUTONYMS for autonyms)
828 * @param string $include See fetchLanguageNames()
829 * @return string Language name or empty
830 * @since 1.20
831 */
832 public static function fetchLanguageName(
833 $code,
834 $inLanguage = self::AS_AUTONYMS,
835 $include = self::ALL
836 ) {
837 return MediaWikiServices::getInstance()->getLanguageNameUtils()
838 ->getLanguageName( $code, $inLanguage, $include );
839 }
840
841 /**
842 * Get a message from the MediaWiki namespace.
843 *
844 * @param string $msg Message name
845 * @return string
846 */
847 public function getMessageFromDB( $msg ) {
848 return $this->msg( $msg )->text();
849 }
850
851 /**
852 * Get message object in this language. Only for use inside this class.
853 *
854 * @param string $msg Message name
855 * @return Message
856 */
857 protected function msg( $msg ) {
858 return wfMessage( $msg )->inLanguage( $this );
859 }
860
861 /**
862 * @param string $key
863 * @return string
864 */
865 public function getMonthName( $key ) {
866 return $this->getMessageFromDB( self::$mMonthMsgs[$key - 1] );
867 }
868
869 /**
870 * @return array
871 */
872 public function getMonthNamesArray() {
873 $monthNames = [ '' ];
874 for ( $i = 1; $i < 13; $i++ ) {
875 $monthNames[] = $this->getMonthName( $i );
876 }
877 return $monthNames;
878 }
879
880 /**
881 * @param string $key
882 * @return string
883 */
884 public function getMonthNameGen( $key ) {
885 return $this->getMessageFromDB( self::$mMonthGenMsgs[$key - 1] );
886 }
887
888 /**
889 * @param string $key
890 * @return string
891 */
892 public function getMonthAbbreviation( $key ) {
893 return $this->getMessageFromDB( self::$mMonthAbbrevMsgs[$key - 1] );
894 }
895
896 /**
897 * @return array
898 */
899 public function getMonthAbbreviationsArray() {
900 $monthNames = [ '' ];
901 for ( $i = 1; $i < 13; $i++ ) {
902 $monthNames[] = $this->getMonthAbbreviation( $i );
903 }
904 return $monthNames;
905 }
906
907 /**
908 * @param string $key
909 * @return string
910 */
911 public function getWeekdayName( $key ) {
912 return $this->getMessageFromDB( self::$mWeekdayMsgs[$key - 1] );
913 }
914
915 /**
916 * @param string $key
917 * @return string
918 */
919 function getWeekdayAbbreviation( $key ) {
920 return $this->getMessageFromDB( self::$mWeekdayAbbrevMsgs[$key - 1] );
921 }
922
923 /**
924 * @param string $key
925 * @return string
926 */
927 function getIranianCalendarMonthName( $key ) {
928 return $this->getMessageFromDB( self::$mIranianCalendarMonthMsgs[$key - 1] );
929 }
930
931 /**
932 * @param string $key
933 * @return string
934 */
935 function getHebrewCalendarMonthName( $key ) {
936 return $this->getMessageFromDB( self::$mHebrewCalendarMonthMsgs[$key - 1] );
937 }
938
939 /**
940 * @param string $key
941 * @return string
942 */
943 function getHebrewCalendarMonthNameGen( $key ) {
944 return $this->getMessageFromDB( self::$mHebrewCalendarMonthGenMsgs[$key - 1] );
945 }
946
947 /**
948 * @param string $key
949 * @return string
950 */
951 function getHijriCalendarMonthName( $key ) {
952 return $this->getMessageFromDB( self::$mHijriCalendarMonthMsgs[$key - 1] );
953 }
954
955 /**
956 * Pass through result from $dateTimeObj->format()
957 * @param DateTime|bool|null &$dateTimeObj
958 * @param string $ts
959 * @param DateTimeZone|bool|null $zone
960 * @param string $code
961 * @return string
962 */
963 private static function dateTimeObjFormat( &$dateTimeObj, $ts, $zone, $code ) {
964 if ( !$dateTimeObj ) {
965 $dateTimeObj = DateTime::createFromFormat(
966 'YmdHis', $ts, $zone ?: new DateTimeZone( 'UTC' )
967 );
968 }
969 return $dateTimeObj->format( $code );
970 }
971
972 /**
973 * This is a workalike of PHP's date() function, but with better
974 * internationalisation, a reduced set of format characters, and a better
975 * escaping format.
976 *
977 * Supported format characters are dDjlNwzWFmMntLoYyaAgGhHiscrUeIOPTZ. See
978 * the PHP manual for definitions. There are a number of extensions, which
979 * start with "x":
980 *
981 * xn Do not translate digits of the next numeric format character
982 * xN Toggle raw digit (xn) flag, stays set until explicitly unset
983 * xr Use roman numerals for the next numeric format character
984 * xh Use hebrew numerals for the next numeric format character
985 * xx Literal x
986 * xg Genitive month name
987 *
988 * xij j (day number) in Iranian calendar
989 * xiF F (month name) in Iranian calendar
990 * xin n (month number) in Iranian calendar
991 * xiy y (two digit year) in Iranian calendar
992 * xiY Y (full year) in Iranian calendar
993 * xit t (days in month) in Iranian calendar
994 * xiz z (day of the year) in Iranian calendar
995 *
996 * xjj j (day number) in Hebrew calendar
997 * xjF F (month name) in Hebrew calendar
998 * xjt t (days in month) in Hebrew calendar
999 * xjx xg (genitive month name) in Hebrew calendar
1000 * xjn n (month number) in Hebrew calendar
1001 * xjY Y (full year) in Hebrew calendar
1002 *
1003 * xmj j (day number) in Hijri calendar
1004 * xmF F (month name) in Hijri calendar
1005 * xmn n (month number) in Hijri calendar
1006 * xmY Y (full year) in Hijri calendar
1007 *
1008 * xkY Y (full year) in Thai solar calendar. Months and days are
1009 * identical to the Gregorian calendar
1010 * xoY Y (full year) in Minguo calendar or Juche year.
1011 * Months and days are identical to the
1012 * Gregorian calendar
1013 * xtY Y (full year) in Japanese nengo. Months and days are
1014 * identical to the Gregorian calendar
1015 *
1016 * Characters enclosed in double quotes will be considered literal (with
1017 * the quotes themselves removed). Unmatched quotes will be considered
1018 * literal quotes. Example:
1019 *
1020 * "The month is" F => The month is January
1021 * i's" => 20'11"
1022 *
1023 * Backslash escaping is also supported.
1024 *
1025 * Input timestamp is assumed to be pre-normalized to the desired local
1026 * time zone, if any. Note that the format characters crUeIOPTZ will assume
1027 * $ts is UTC if $zone is not given.
1028 *
1029 * @param string $format
1030 * @param string $ts 14-character timestamp
1031 * YYYYMMDDHHMMSS
1032 * 01234567890123
1033 * @param DateTimeZone|null $zone Timezone of $ts
1034 * @param int &$ttl The amount of time (in seconds) the output may be cached for.
1035 * Only makes sense if $ts is the current time.
1036 * @todo handling of "o" format character for Iranian, Hebrew, Hijri & Thai?
1037 *
1038 * @throws MWException
1039 * @return string
1040 */
1041 public function sprintfDate( $format, $ts, DateTimeZone $zone = null, &$ttl = 'unused' ) {
1042 $s = '';
1043 $raw = false;
1044 $roman = false;
1045 $hebrewNum = false;
1046 $dateTimeObj = false;
1047 $rawToggle = false;
1048 $iranian = false;
1049 $hebrew = false;
1050 $hijri = false;
1051 $thai = false;
1052 $minguo = false;
1053 $tenno = false;
1054
1055 $usedSecond = false;
1056 $usedMinute = false;
1057 $usedHour = false;
1058 $usedAMPM = false;
1059 $usedDay = false;
1060 $usedWeek = false;
1061 $usedMonth = false;
1062 $usedYear = false;
1063 $usedISOYear = false;
1064 $usedIsLeapYear = false;
1065
1066 $usedHebrewMonth = false;
1067 $usedIranianMonth = false;
1068 $usedHijriMonth = false;
1069 $usedHebrewYear = false;
1070 $usedIranianYear = false;
1071 $usedHijriYear = false;
1072 $usedTennoYear = false;
1073
1074 if ( strlen( $ts ) !== 14 ) {
1075 throw new MWException( __METHOD__ . ": The timestamp $ts should have 14 characters" );
1076 }
1077
1078 if ( !ctype_digit( $ts ) ) {
1079 throw new MWException( __METHOD__ . ": The timestamp $ts should be a number" );
1080 }
1081
1082 $formatLength = strlen( $format );
1083 for ( $p = 0; $p < $formatLength; $p++ ) {
1084 $num = false;
1085 $code = $format[$p];
1086 if ( $code == 'x' && $p < $formatLength - 1 ) {
1087 $code .= $format[++$p];
1088 }
1089
1090 if ( ( $code === 'xi'
1091 || $code === 'xj'
1092 || $code === 'xk'
1093 || $code === 'xm'
1094 || $code === 'xo'
1095 || $code === 'xt' )
1096 && $p < $formatLength - 1 ) {
1097 $code .= $format[++$p];
1098 }
1099
1100 switch ( $code ) {
1101 case 'xx':
1102 $s .= 'x';
1103 break;
1104 case 'xn':
1105 $raw = true;
1106 break;
1107 case 'xN':
1108 $rawToggle = !$rawToggle;
1109 break;
1110 case 'xr':
1111 $roman = true;
1112 break;
1113 case 'xh':
1114 $hebrewNum = true;
1115 break;
1116 case 'xg':
1117 $usedMonth = true;
1118 $s .= $this->getMonthNameGen( substr( $ts, 4, 2 ) );
1119 break;
1120 case 'xjx':
1121 $usedHebrewMonth = true;
1122 if ( !$hebrew ) {
1123 $hebrew = self::tsToHebrew( $ts );
1124 }
1125 $s .= $this->getHebrewCalendarMonthNameGen( $hebrew[1] );
1126 break;
1127 case 'd':
1128 $usedDay = true;
1129 $num = substr( $ts, 6, 2 );
1130 break;
1131 case 'D':
1132 $usedDay = true;
1133 $s .= $this->getWeekdayAbbreviation(
1134 self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'w' ) + 1
1135 );
1136 break;
1137 case 'j':
1138 $usedDay = true;
1139 $num = intval( substr( $ts, 6, 2 ) );
1140 break;
1141 case 'xij':
1142 $usedDay = true;
1143 if ( !$iranian ) {
1144 $iranian = self::tsToIranian( $ts );
1145 }
1146 $num = $iranian[2];
1147 break;
1148 case 'xmj':
1149 $usedDay = true;
1150 if ( !$hijri ) {
1151 $hijri = self::tsToHijri( $ts );
1152 }
1153 $num = $hijri[2];
1154 break;
1155 case 'xjj':
1156 $usedDay = true;
1157 if ( !$hebrew ) {
1158 $hebrew = self::tsToHebrew( $ts );
1159 }
1160 $num = $hebrew[2];
1161 break;
1162 case 'l':
1163 $usedDay = true;
1164 $s .= $this->getWeekdayName(
1165 self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'w' ) + 1
1166 );
1167 break;
1168 case 'F':
1169 $usedMonth = true;
1170 $s .= $this->getMonthName( substr( $ts, 4, 2 ) );
1171 break;
1172 case 'xiF':
1173 $usedIranianMonth = true;
1174 if ( !$iranian ) {
1175 $iranian = self::tsToIranian( $ts );
1176 }
1177 $s .= $this->getIranianCalendarMonthName( $iranian[1] );
1178 break;
1179 case 'xmF':
1180 $usedHijriMonth = true;
1181 if ( !$hijri ) {
1182 $hijri = self::tsToHijri( $ts );
1183 }
1184 $s .= $this->getHijriCalendarMonthName( $hijri[1] );
1185 break;
1186 case 'xjF':
1187 $usedHebrewMonth = true;
1188 if ( !$hebrew ) {
1189 $hebrew = self::tsToHebrew( $ts );
1190 }
1191 $s .= $this->getHebrewCalendarMonthName( $hebrew[1] );
1192 break;
1193 case 'm':
1194 $usedMonth = true;
1195 $num = substr( $ts, 4, 2 );
1196 break;
1197 case 'M':
1198 $usedMonth = true;
1199 $s .= $this->getMonthAbbreviation( substr( $ts, 4, 2 ) );
1200 break;
1201 case 'n':
1202 $usedMonth = true;
1203 $num = intval( substr( $ts, 4, 2 ) );
1204 break;
1205 case 'xin':
1206 $usedIranianMonth = true;
1207 if ( !$iranian ) {
1208 $iranian = self::tsToIranian( $ts );
1209 }
1210 $num = $iranian[1];
1211 break;
1212 case 'xmn':
1213 $usedHijriMonth = true;
1214 if ( !$hijri ) {
1215 $hijri = self::tsToHijri( $ts );
1216 }
1217 $num = $hijri[1];
1218 break;
1219 case 'xjn':
1220 $usedHebrewMonth = true;
1221 if ( !$hebrew ) {
1222 $hebrew = self::tsToHebrew( $ts );
1223 }
1224 $num = $hebrew[1];
1225 break;
1226 case 'xjt':
1227 $usedHebrewMonth = true;
1228 if ( !$hebrew ) {
1229 $hebrew = self::tsToHebrew( $ts );
1230 }
1231 $num = $hebrew[3];
1232 break;
1233 case 'Y':
1234 $usedYear = true;
1235 $num = substr( $ts, 0, 4 );
1236 break;
1237 case 'xiY':
1238 $usedIranianYear = true;
1239 if ( !$iranian ) {
1240 $iranian = self::tsToIranian( $ts );
1241 }
1242 $num = $iranian[0];
1243 break;
1244 case 'xmY':
1245 $usedHijriYear = true;
1246 if ( !$hijri ) {
1247 $hijri = self::tsToHijri( $ts );
1248 }
1249 $num = $hijri[0];
1250 break;
1251 case 'xjY':
1252 $usedHebrewYear = true;
1253 if ( !$hebrew ) {
1254 $hebrew = self::tsToHebrew( $ts );
1255 }
1256 $num = $hebrew[0];
1257 break;
1258 case 'xkY':
1259 $usedYear = true;
1260 if ( !$thai ) {
1261 $thai = self::tsToYear( $ts, 'thai' );
1262 }
1263 $num = $thai[0];
1264 break;
1265 case 'xoY':
1266 $usedYear = true;
1267 if ( !$minguo ) {
1268 $minguo = self::tsToYear( $ts, 'minguo' );
1269 }
1270 $num = $minguo[0];
1271 break;
1272 case 'xtY':
1273 $usedTennoYear = true;
1274 if ( !$tenno ) {
1275 $tenno = self::tsToYear( $ts, 'tenno' );
1276 }
1277 $num = $tenno[0];
1278 break;
1279 case 'y':
1280 $usedYear = true;
1281 $num = substr( $ts, 2, 2 );
1282 break;
1283 case 'xiy':
1284 $usedIranianYear = true;
1285 if ( !$iranian ) {
1286 $iranian = self::tsToIranian( $ts );
1287 }
1288 $num = substr( $iranian[0], -2 );
1289 break;
1290 case 'xit':
1291 $usedIranianYear = true;
1292 if ( !$iranian ) {
1293 $iranian = self::tsToIranian( $ts );
1294 }
1295 $num = self::$IRANIAN_DAYS[$iranian[1] - 1];
1296 break;
1297 case 'xiz':
1298 $usedIranianYear = true;
1299 if ( !$iranian ) {
1300 $iranian = self::tsToIranian( $ts );
1301 }
1302 $num = $iranian[3];
1303 break;
1304 case 'a':
1305 $usedAMPM = true;
1306 $s .= intval( substr( $ts, 8, 2 ) ) < 12 ? 'am' : 'pm';
1307 break;
1308 case 'A':
1309 $usedAMPM = true;
1310 $s .= intval( substr( $ts, 8, 2 ) ) < 12 ? 'AM' : 'PM';
1311 break;
1312 case 'g':
1313 $usedHour = true;
1314 $h = substr( $ts, 8, 2 );
1315 $num = $h % 12 ?: 12;
1316 break;
1317 case 'G':
1318 $usedHour = true;
1319 $num = intval( substr( $ts, 8, 2 ) );
1320 break;
1321 case 'h':
1322 $usedHour = true;
1323 $h = substr( $ts, 8, 2 );
1324 $num = sprintf( '%02d', $h % 12 ?: 12 );
1325 break;
1326 case 'H':
1327 $usedHour = true;
1328 $num = substr( $ts, 8, 2 );
1329 break;
1330 case 'i':
1331 $usedMinute = true;
1332 $num = substr( $ts, 10, 2 );
1333 break;
1334 case 's':
1335 $usedSecond = true;
1336 $num = substr( $ts, 12, 2 );
1337 break;
1338 case 'c':
1339 case 'r':
1340 $usedSecond = true;
1341 // fall through
1342 case 'e':
1343 case 'O':
1344 case 'P':
1345 case 'T':
1346 $s .= self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1347 break;
1348 case 'w':
1349 case 'N':
1350 case 'z':
1351 $usedDay = true;
1352 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1353 break;
1354 case 'W':
1355 $usedWeek = true;
1356 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1357 break;
1358 case 't':
1359 $usedMonth = true;
1360 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1361 break;
1362 case 'L':
1363 $usedIsLeapYear = true;
1364 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1365 break;
1366 case 'o':
1367 $usedISOYear = true;
1368 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1369 break;
1370 case 'U':
1371 $usedSecond = true;
1372 // fall through
1373 case 'I':
1374 case 'Z':
1375 $num = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, $code );
1376 break;
1377 case '\\':
1378 # Backslash escaping
1379 if ( $p < $formatLength - 1 ) {
1380 $s .= $format[++$p];
1381 } else {
1382 $s .= '\\';
1383 }
1384 break;
1385 case '"':
1386 # Quoted literal
1387 if ( $p < $formatLength - 1 ) {
1388 $endQuote = strpos( $format, '"', $p + 1 );
1389 if ( $endQuote === false ) {
1390 # No terminating quote, assume literal "
1391 $s .= '"';
1392 } else {
1393 $s .= substr( $format, $p + 1, $endQuote - $p - 1 );
1394 $p = $endQuote;
1395 }
1396 } else {
1397 # Quote at end of string, assume literal "
1398 $s .= '"';
1399 }
1400 break;
1401 default:
1402 $s .= $format[$p];
1403 }
1404 if ( $num !== false ) {
1405 if ( $rawToggle || $raw ) {
1406 $s .= $num;
1407 $raw = false;
1408 } elseif ( $roman ) {
1409 $s .= self::romanNumeral( $num );
1410 $roman = false;
1411 } elseif ( $hebrewNum ) {
1412 $s .= self::hebrewNumeral( $num );
1413 $hebrewNum = false;
1414 } else {
1415 $s .= $this->formatNum( $num, true );
1416 }
1417 }
1418 }
1419
1420 if ( $ttl === 'unused' ) {
1421 // No need to calculate the TTL, the caller wont use it anyway.
1422 } elseif ( $usedSecond ) {
1423 $ttl = 1;
1424 } elseif ( $usedMinute ) {
1425 $ttl = 60 - substr( $ts, 12, 2 );
1426 } elseif ( $usedHour ) {
1427 $ttl = 3600 - substr( $ts, 10, 2 ) * 60 - substr( $ts, 12, 2 );
1428 } elseif ( $usedAMPM ) {
1429 $ttl = 43200 - ( substr( $ts, 8, 2 ) % 12 ) * 3600 -
1430 substr( $ts, 10, 2 ) * 60 - substr( $ts, 12, 2 );
1431 } elseif (
1432 $usedDay ||
1433 $usedHebrewMonth ||
1434 $usedIranianMonth ||
1435 $usedHijriMonth ||
1436 $usedHebrewYear ||
1437 $usedIranianYear ||
1438 $usedHijriYear ||
1439 $usedTennoYear
1440 ) {
1441 // @todo Someone who understands the non-Gregorian calendars
1442 // should write proper logic for them so that they don't need purged every day.
1443 $ttl = 86400 - substr( $ts, 8, 2 ) * 3600 -
1444 substr( $ts, 10, 2 ) * 60 - substr( $ts, 12, 2 );
1445 } else {
1446 $possibleTtls = [];
1447 $timeRemainingInDay = 86400 - substr( $ts, 8, 2 ) * 3600 -
1448 substr( $ts, 10, 2 ) * 60 - substr( $ts, 12, 2 );
1449 if ( $usedWeek ) {
1450 $possibleTtls[] =
1451 ( 7 - self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'N' ) ) * 86400 +
1452 $timeRemainingInDay;
1453 } elseif ( $usedISOYear ) {
1454 // December 28th falls on the last ISO week of the year, every year.
1455 // The last ISO week of a year can be 52 or 53.
1456 $lastWeekOfISOYear = DateTime::createFromFormat(
1457 'Ymd',
1458 substr( $ts, 0, 4 ) . '1228',
1459 $zone ?: new DateTimeZone( 'UTC' )
1460 )->format( 'W' );
1461 $currentISOWeek = self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'W' );
1462 $weeksRemaining = $lastWeekOfISOYear - $currentISOWeek;
1463 $timeRemainingInWeek =
1464 ( 7 - self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'N' ) ) * 86400
1465 + $timeRemainingInDay;
1466 $possibleTtls[] = $weeksRemaining * 604800 + $timeRemainingInWeek;
1467 }
1468
1469 if ( $usedMonth ) {
1470 $possibleTtls[] =
1471 ( self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 't' ) -
1472 substr( $ts, 6, 2 ) ) * 86400
1473 + $timeRemainingInDay;
1474 } elseif ( $usedYear ) {
1475 $possibleTtls[] =
1476 ( self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'L' ) + 364 -
1477 self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'z' ) ) * 86400
1478 + $timeRemainingInDay;
1479 } elseif ( $usedIsLeapYear ) {
1480 $year = substr( $ts, 0, 4 );
1481 $timeRemainingInYear =
1482 ( self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'L' ) + 364 -
1483 self::dateTimeObjFormat( $dateTimeObj, $ts, $zone, 'z' ) ) * 86400
1484 + $timeRemainingInDay;
1485 $mod = $year % 4;
1486 if ( $mod || ( !( $year % 100 ) && $year % 400 ) ) {
1487 // this isn't a leap year. see when the next one starts
1488 $nextCandidate = $year - $mod + 4;
1489 if ( $nextCandidate % 100 || !( $nextCandidate % 400 ) ) {
1490 $possibleTtls[] = ( $nextCandidate - $year - 1 ) * 365 * 86400 +
1491 $timeRemainingInYear;
1492 } else {
1493 $possibleTtls[] = ( $nextCandidate - $year + 3 ) * 365 * 86400 +
1494 $timeRemainingInYear;
1495 }
1496 } else {
1497 // this is a leap year, so the next year isn't
1498 $possibleTtls[] = $timeRemainingInYear;
1499 }
1500 }
1501
1502 if ( $possibleTtls ) {
1503 $ttl = min( $possibleTtls );
1504 }
1505 }
1506
1507 return $s;
1508 }
1509
1510 private static $GREG_DAYS = [ 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31 ];
1511 private static $IRANIAN_DAYS = [ 31, 31, 31, 31, 31, 31, 30, 30, 30, 30, 30, 29 ];
1512
1513 /**
1514 * Algorithm by Roozbeh Pournader and Mohammad Toossi to convert
1515 * Gregorian dates to Iranian dates. Originally written in C, it
1516 * is released under the terms of GNU Lesser General Public
1517 * License. Conversion to PHP was performed by Niklas Laxström.
1518 *
1519 * Link: http://www.farsiweb.info/jalali/jalali.c
1520 *
1521 * @param string $ts
1522 *
1523 * @return int[]
1524 */
1525 private static function tsToIranian( $ts ) {
1526 $gy = substr( $ts, 0, 4 ) - 1600;
1527 $gm = substr( $ts, 4, 2 ) - 1;
1528 $gd = substr( $ts, 6, 2 ) - 1;
1529
1530 # Days passed from the beginning (including leap years)
1531 $gDayNo = 365 * $gy
1532 + floor( ( $gy + 3 ) / 4 )
1533 - floor( ( $gy + 99 ) / 100 )
1534 + floor( ( $gy + 399 ) / 400 );
1535
1536 // Add days of the past months of this year
1537 for ( $i = 0; $i < $gm; $i++ ) {
1538 $gDayNo += self::$GREG_DAYS[$i];
1539 }
1540
1541 // Leap years
1542 if ( $gm > 1 && ( ( $gy % 4 === 0 && $gy % 100 !== 0 || ( $gy % 400 == 0 ) ) ) ) {
1543 $gDayNo++;
1544 }
1545
1546 // Days passed in current month
1547 $gDayNo += (int)$gd;
1548
1549 $jDayNo = $gDayNo - 79;
1550
1551 $jNp = floor( $jDayNo / 12053 );
1552 $jDayNo %= 12053;
1553
1554 $jy = 979 + 33 * $jNp + 4 * floor( $jDayNo / 1461 );
1555 $jDayNo %= 1461;
1556
1557 if ( $jDayNo >= 366 ) {
1558 $jy += floor( ( $jDayNo - 1 ) / 365 );
1559 $jDayNo = floor( ( $jDayNo - 1 ) % 365 );
1560 }
1561
1562 $jz = $jDayNo;
1563
1564 for ( $i = 0; $i < 11 && $jDayNo >= self::$IRANIAN_DAYS[$i]; $i++ ) {
1565 $jDayNo -= self::$IRANIAN_DAYS[$i];
1566 }
1567
1568 $jm = $i + 1;
1569 $jd = $jDayNo + 1;
1570
1571 return [ $jy, $jm, $jd, $jz ];
1572 }
1573
1574 /**
1575 * Converting Gregorian dates to Hijri dates.
1576 *
1577 * Based on a PHP-Nuke block by Sharjeel which is released under GNU/GPL license
1578 *
1579 * @see https://phpnuke.org/modules.php?name=News&file=article&sid=8234&mode=thread&order=0&thold=0
1580 *
1581 * @param string $ts
1582 *
1583 * @return int[]
1584 */
1585 private static function tsToHijri( $ts ) {
1586 $year = substr( $ts, 0, 4 );
1587 $month = substr( $ts, 4, 2 );
1588 $day = substr( $ts, 6, 2 );
1589
1590 $zyr = $year;
1591 $zd = $day;
1592 $zm = $month;
1593 $zy = $zyr;
1594
1595 if (
1596 ( $zy > 1582 ) || ( ( $zy == 1582 ) && ( $zm > 10 ) ) ||
1597 ( ( $zy == 1582 ) && ( $zm == 10 ) && ( $zd > 14 ) )
1598 ) {
1599 $zjd = (int)( ( 1461 * ( $zy + 4800 + (int)( ( $zm - 14 ) / 12 ) ) ) / 4 ) +
1600 (int)( ( 367 * ( $zm - 2 - 12 * ( (int)( ( $zm - 14 ) / 12 ) ) ) ) / 12 ) -
1601 (int)( ( 3 * (int)( ( ( $zy + 4900 + (int)( ( $zm - 14 ) / 12 ) ) / 100 ) ) ) / 4 ) +
1602 $zd - 32075;
1603 } else {
1604 $zjd = 367 * $zy - (int)( ( 7 * ( $zy + 5001 + (int)( ( $zm - 9 ) / 7 ) ) ) / 4 ) +
1605 (int)( ( 275 * $zm ) / 9 ) + $zd + 1729777;
1606 }
1607
1608 $zl = $zjd - 1948440 + 10632;
1609 $zn = (int)( ( $zl - 1 ) / 10631 );
1610 $zl = $zl - 10631 * $zn + 354;
1611 $zj = ( (int)( ( 10985 - $zl ) / 5316 ) ) * ( (int)( ( 50 * $zl ) / 17719 ) ) +
1612 ( (int)( $zl / 5670 ) ) * ( (int)( ( 43 * $zl ) / 15238 ) );
1613 $zl = $zl - ( (int)( ( 30 - $zj ) / 15 ) ) * ( (int)( ( 17719 * $zj ) / 50 ) ) -
1614 ( (int)( $zj / 16 ) ) * ( (int)( ( 15238 * $zj ) / 43 ) ) + 29;
1615 $zm = (int)( ( 24 * $zl ) / 709 );
1616 $zd = $zl - (int)( ( 709 * $zm ) / 24 );
1617 $zy = 30 * $zn + $zj - 30;
1618
1619 return [ $zy, $zm, $zd ];
1620 }
1621
1622 /**
1623 * Converting Gregorian dates to Hebrew dates.
1624 *
1625 * Based on a JavaScript code by Abu Mami and Yisrael Hersch
1626 * (abu-mami@kaluach.net, http://www.kaluach.net), who permitted
1627 * to translate the relevant functions into PHP and release them under
1628 * GNU GPL.
1629 *
1630 * The months are counted from Tishrei = 1. In a leap year, Adar I is 13
1631 * and Adar II is 14. In a non-leap year, Adar is 6.
1632 *
1633 * @param string $ts
1634 *
1635 * @return int[]
1636 */
1637 private static function tsToHebrew( $ts ) {
1638 # Parse date
1639 $year = substr( $ts, 0, 4 );
1640 $month = substr( $ts, 4, 2 );
1641 $day = substr( $ts, 6, 2 );
1642
1643 # Calculate Hebrew year
1644 $hebrewYear = $year + 3760;
1645
1646 # Month number when September = 1, August = 12
1647 $month += 4;
1648 if ( $month > 12 ) {
1649 # Next year
1650 $month -= 12;
1651 $year++;
1652 $hebrewYear++;
1653 }
1654
1655 # Calculate day of year from 1 September
1656 $dayOfYear = $day;
1657 for ( $i = 1; $i < $month; $i++ ) {
1658 if ( $i == 6 ) {
1659 # February
1660 $dayOfYear += 28;
1661 # Check if the year is leap
1662 if ( $year % 400 == 0 || ( $year % 4 == 0 && $year % 100 > 0 ) ) {
1663 $dayOfYear++;
1664 }
1665 } elseif ( $i == 8 || $i == 10 || $i == 1 || $i == 3 ) {
1666 $dayOfYear += 30;
1667 } else {
1668 $dayOfYear += 31;
1669 }
1670 }
1671
1672 # Calculate the start of the Hebrew year
1673 $start = self::hebrewYearStart( $hebrewYear );
1674
1675 # Calculate next year's start
1676 if ( $dayOfYear <= $start ) {
1677 # Day is before the start of the year - it is the previous year
1678 # Next year's start
1679 $nextStart = $start;
1680 # Previous year
1681 $year--;
1682 $hebrewYear--;
1683 # Add days since previous year's 1 September
1684 $dayOfYear += 365;
1685 if ( ( $year % 400 == 0 ) || ( $year % 100 != 0 && $year % 4 == 0 ) ) {
1686 # Leap year
1687 $dayOfYear++;
1688 }
1689 # Start of the new (previous) year
1690 $start = self::hebrewYearStart( $hebrewYear );
1691 } else {
1692 # Next year's start
1693 $nextStart = self::hebrewYearStart( $hebrewYear + 1 );
1694 }
1695
1696 # Calculate Hebrew day of year
1697 $hebrewDayOfYear = $dayOfYear - $start;
1698
1699 # Difference between year's days
1700 $diff = $nextStart - $start;
1701 # Add 12 (or 13 for leap years) days to ignore the difference between
1702 # Hebrew and Gregorian year (353 at least vs. 365/6) - now the
1703 # difference is only about the year type
1704 if ( ( $year % 400 == 0 ) || ( $year % 100 != 0 && $year % 4 == 0 ) ) {
1705 $diff += 13;
1706 } else {
1707 $diff += 12;
1708 }
1709
1710 # Check the year pattern, and is leap year
1711 # 0 means an incomplete year, 1 means a regular year, 2 means a complete year
1712 # This is mod 30, to work on both leap years (which add 30 days of Adar I)
1713 # and non-leap years
1714 $yearPattern = $diff % 30;
1715 # Check if leap year
1716 $isLeap = $diff >= 30;
1717
1718 # Calculate day in the month from number of day in the Hebrew year
1719 # Don't check Adar - if the day is not in Adar, we will stop before;
1720 # if it is in Adar, we will use it to check if it is Adar I or Adar II
1721 $hebrewDay = $hebrewDayOfYear;
1722 $hebrewMonth = 1;
1723 $days = 0;
1724 while ( $hebrewMonth <= 12 ) {
1725 # Calculate days in this month
1726 if ( $isLeap && $hebrewMonth == 6 ) {
1727 # Leap year - has Adar I, with 30 days, and Adar II, with 29 days
1728 $days = 30;
1729 if ( $hebrewDay <= $days ) {
1730 # Day in Adar I
1731 $hebrewMonth = 13;
1732 } else {
1733 # Subtract the days of Adar I
1734 $hebrewDay -= $days;
1735 # Try Adar II
1736 $days = 29;
1737 if ( $hebrewDay <= $days ) {
1738 # Day in Adar II
1739 $hebrewMonth = 14;
1740 }
1741 }
1742 } elseif ( $hebrewMonth == 2 && $yearPattern == 2 ) {
1743 # Cheshvan in a complete year (otherwise as the rule below)
1744 $days = 30;
1745 } elseif ( $hebrewMonth == 3 && $yearPattern == 0 ) {
1746 # Kislev in an incomplete year (otherwise as the rule below)
1747 $days = 29;
1748 } else {
1749 # Odd months have 30 days, even have 29
1750 $days = 30 - ( $hebrewMonth - 1 ) % 2;
1751 }
1752 if ( $hebrewDay <= $days ) {
1753 # In the current month
1754 break;
1755 } else {
1756 # Subtract the days of the current month
1757 $hebrewDay -= $days;
1758 # Try in the next month
1759 $hebrewMonth++;
1760 }
1761 }
1762
1763 return [ $hebrewYear, $hebrewMonth, $hebrewDay, $days ];
1764 }
1765
1766 /**
1767 * This calculates the Hebrew year start, as days since 1 September.
1768 * Based on Carl Friedrich Gauss algorithm for finding Easter date.
1769 * Used for Hebrew date.
1770 *
1771 * @param int $year
1772 *
1773 * @return string
1774 */
1775 private static function hebrewYearStart( $year ) {
1776 $a = intval( ( 12 * ( $year - 1 ) + 17 ) % 19 );
1777 $b = intval( ( $year - 1 ) % 4 );
1778 $m = 32.044093161144 + 1.5542417966212 * $a + $b / 4.0 - 0.0031777940220923 * ( $year - 1 );
1779 if ( $m < 0 ) {
1780 $m--;
1781 }
1782 $Mar = intval( $m );
1783 if ( $m < 0 ) {
1784 $m++;
1785 }
1786 $m -= $Mar;
1787
1788 $c = intval( ( $Mar + 3 * ( $year - 1 ) + 5 * $b + 5 ) % 7 );
1789 if ( $c == 0 && $a > 11 && $m >= 0.89772376543210 ) {
1790 $Mar++;
1791 } elseif ( $c == 1 && $a > 6 && $m >= 0.63287037037037 ) {
1792 $Mar += 2;
1793 } elseif ( $c == 2 || $c == 4 || $c == 6 ) {
1794 $Mar++;
1795 }
1796
1797 $Mar += intval( ( $year - 3761 ) / 100 ) - intval( ( $year - 3761 ) / 400 ) - 24;
1798 return $Mar;
1799 }
1800
1801 /**
1802 * Algorithm to convert Gregorian dates to Thai solar dates,
1803 * Minguo dates or Minguo dates.
1804 *
1805 * Link: https://en.wikipedia.org/wiki/Thai_solar_calendar
1806 * https://en.wikipedia.org/wiki/Minguo_calendar
1807 * https://en.wikipedia.org/wiki/Japanese_era_name
1808 *
1809 * @param string $ts 14-character timestamp
1810 * @param string $cName Calender name
1811 * @return array Converted year, month, day
1812 */
1813 private static function tsToYear( $ts, $cName ) {
1814 $gy = substr( $ts, 0, 4 );
1815 $gm = substr( $ts, 4, 2 );
1816 $gd = substr( $ts, 6, 2 );
1817
1818 if ( !strcmp( $cName, 'thai' ) ) {
1819 # Thai solar dates
1820 # Add 543 years to the Gregorian calendar
1821 # Months and days are identical
1822 $gy_offset = $gy + 543;
1823 # fix for dates between 1912 and 1941
1824 # https://en.wikipedia.org/?oldid=836596673#New_year
1825 if ( $gy >= 1912 && $gy <= 1940 ) {
1826 if ( $gm <= 3 ) {
1827 $gy_offset--;
1828 }
1829 $gm = ( $gm - 3 ) % 12;
1830 }
1831 } elseif ( ( !strcmp( $cName, 'minguo' ) ) || !strcmp( $cName, 'juche' ) ) {
1832 # Minguo dates
1833 # Deduct 1911 years from the Gregorian calendar
1834 # Months and days are identical
1835 $gy_offset = $gy - 1911;
1836 } elseif ( !strcmp( $cName, 'tenno' ) ) {
1837 # Nengō dates up to Meiji period
1838 # Deduct years from the Gregorian calendar
1839 # depending on the nengo periods
1840 # Months and days are identical
1841 if ( ( $gy < 1912 )
1842 || ( ( $gy == 1912 ) && ( $gm < 7 ) )
1843 || ( ( $gy == 1912 ) && ( $gm == 7 ) && ( $gd < 31 ) )
1844 ) {
1845 # Meiji period
1846 $gy_gannen = $gy - 1868 + 1;
1847 $gy_offset = $gy_gannen;
1848 if ( $gy_gannen == 1 ) {
1849 $gy_offset = '元';
1850 }
1851 $gy_offset = '明治' . $gy_offset;
1852 } elseif (
1853 ( ( $gy == 1912 ) && ( $gm == 7 ) && ( $gd == 31 ) ) ||
1854 ( ( $gy == 1912 ) && ( $gm >= 8 ) ) ||
1855 ( ( $gy > 1912 ) && ( $gy < 1926 ) ) ||
1856 ( ( $gy == 1926 ) && ( $gm < 12 ) ) ||
1857 ( ( $gy == 1926 ) && ( $gm == 12 ) && ( $gd < 26 ) )
1858 ) {
1859 # Taishō period
1860 $gy_gannen = $gy - 1912 + 1;
1861 $gy_offset = $gy_gannen;
1862 if ( $gy_gannen == 1 ) {
1863 $gy_offset = '元';
1864 }
1865 $gy_offset = '大正' . $gy_offset;
1866 } elseif (
1867 ( ( $gy == 1926 ) && ( $gm == 12 ) && ( $gd >= 26 ) ) ||
1868 ( ( $gy > 1926 ) && ( $gy < 1989 ) ) ||
1869 ( ( $gy == 1989 ) && ( $gm == 1 ) && ( $gd < 8 ) )
1870 ) {
1871 # Shōwa period
1872 $gy_gannen = $gy - 1926 + 1;
1873 $gy_offset = $gy_gannen;
1874 if ( $gy_gannen == 1 ) {
1875 $gy_offset = '元';
1876 }
1877 $gy_offset = '昭和' . $gy_offset;
1878 } elseif (
1879 ( ( $gy == 1989 ) && ( $gm == 1 ) && ( $gd >= 8 ) ) ||
1880 ( ( $gy > 1989 ) && ( $gy < 2019 ) ) ||
1881 ( ( $gy == 2019 ) && ( $gm < 5 ) )
1882 ) {
1883 # Heisei period
1884 $gy_gannen = $gy - 1989 + 1;
1885 $gy_offset = $gy_gannen;
1886 if ( $gy_gannen == 1 ) {
1887 $gy_offset = '元';
1888 }
1889 $gy_offset = '平成' . $gy_offset;
1890 } else {
1891 # Reiwa period
1892 $gy_gannen = $gy - 2019 + 1;
1893 $gy_offset = $gy_gannen;
1894 if ( $gy_gannen == 1 ) {
1895 $gy_offset = '元';
1896 }
1897 $gy_offset = '令和' . $gy_offset;
1898 }
1899 } else {
1900 $gy_offset = $gy;
1901 }
1902
1903 return [ $gy_offset, $gm, $gd ];
1904 }
1905
1906 /**
1907 * Gets directionality of the first strongly directional codepoint, for embedBidi()
1908 *
1909 * This is the rule the BIDI algorithm uses to determine the directionality of
1910 * paragraphs ( https://www.unicode.org/reports/tr9/#The_Paragraph_Level ) and
1911 * FSI isolates ( https://www.unicode.org/reports/tr9/#Explicit_Directional_Isolates ).
1912 *
1913 * TODO: Does not handle BIDI control characters inside the text.
1914 * TODO: Does not handle unallocated characters.
1915 *
1916 * @param string $text Text to test
1917 * @return null|string Directionality ('ltr' or 'rtl') or null
1918 */
1919 private static function strongDirFromContent( $text = '' ) {
1920 if ( !preg_match( self::$strongDirRegex, $text, $matches ) ) {
1921 return null;
1922 }
1923 if ( $matches[1] === '' ) {
1924 return 'rtl';
1925 }
1926 return 'ltr';
1927 }
1928
1929 /**
1930 * Roman number formatting up to 10000
1931 *
1932 * @param int $num
1933 *
1934 * @return string
1935 */
1936 static function romanNumeral( $num ) {
1937 static $table = [
1938 [ '', 'I', 'II', 'III', 'IV', 'V', 'VI', 'VII', 'VIII', 'IX', 'X' ],
1939 [ '', 'X', 'XX', 'XXX', 'XL', 'L', 'LX', 'LXX', 'LXXX', 'XC', 'C' ],
1940 [ '', 'C', 'CC', 'CCC', 'CD', 'D', 'DC', 'DCC', 'DCCC', 'CM', 'M' ],
1941 [ '', 'M', 'MM', 'MMM', 'MMMM', 'MMMMM', 'MMMMMM', 'MMMMMMM',
1942 'MMMMMMMM', 'MMMMMMMMM', 'MMMMMMMMMM' ]
1943 ];
1944
1945 $num = intval( $num );
1946 if ( $num > 10000 || $num <= 0 ) {
1947 return $num;
1948 }
1949
1950 $s = '';
1951 for ( $pow10 = 1000, $i = 3; $i >= 0; $pow10 /= 10, $i-- ) {
1952 if ( $num >= $pow10 ) {
1953 $s .= $table[$i][(int)floor( $num / $pow10 )];
1954 }
1955 $num = $num % $pow10;
1956 }
1957 return $s;
1958 }
1959
1960 /**
1961 * Hebrew Gematria number formatting up to 9999
1962 *
1963 * @param int $num
1964 *
1965 * @return string
1966 */
1967 static function hebrewNumeral( $num ) {
1968 static $table = [
1969 [ '', 'א', 'ב', 'ג', 'ד', 'ה', 'ו', 'ז', 'ח', 'ט', 'י' ],
1970 [ '', 'י', 'כ', 'ל', 'מ', 'נ', 'ס', 'ע', 'פ', 'צ', 'ק' ],
1971 [ '',
1972 [ 'ק' ],
1973 [ 'ר' ],
1974 [ 'ש' ],
1975 [ 'ת' ],
1976 [ 'ת', 'ק' ],
1977 [ 'ת', 'ר' ],
1978 [ 'ת', 'ש' ],
1979 [ 'ת', 'ת' ],
1980 [ 'ת', 'ת', 'ק' ],
1981 [ 'ת', 'ת', 'ר' ],
1982 ],
1983 [ '', 'א', 'ב', 'ג', 'ד', 'ה', 'ו', 'ז', 'ח', 'ט', 'י' ]
1984 ];
1985
1986 $num = intval( $num );
1987 if ( $num > 9999 || $num <= 0 ) {
1988 return $num;
1989 }
1990
1991 // Round thousands have special notations
1992 if ( $num === 1000 ) {
1993 return "א' אלף";
1994 } elseif ( $num % 1000 === 0 ) {
1995 return $table[0][$num / 1000] . "' אלפים";
1996 }
1997
1998 $letters = [];
1999
2000 for ( $pow10 = 1000, $i = 3; $i >= 0; $pow10 /= 10, $i-- ) {
2001 if ( $num >= $pow10 ) {
2002 if ( $num === 15 || $num === 16 ) {
2003 $letters[] = $table[0][9];
2004 $letters[] = $table[0][$num - 9];
2005 $num = 0;
2006 } else {
2007 $letters = array_merge(
2008 $letters,
2009 (array)$table[$i][intval( $num / $pow10 )]
2010 );
2011
2012 if ( $pow10 === 1000 ) {
2013 $letters[] = "'";
2014 }
2015 }
2016 }
2017
2018 $num = $num % $pow10;
2019 }
2020
2021 $preTransformLength = count( $letters );
2022 if ( $preTransformLength === 1 ) {
2023 // Add geresh (single quote) to one-letter numbers
2024 $letters[] = "'";
2025 } else {
2026 $lastIndex = $preTransformLength - 1;
2027 $letters[$lastIndex] = str_replace(
2028 [ 'כ', 'מ', 'נ', 'פ', 'צ' ],
2029 [ 'ך', 'ם', 'ן', 'ף', 'ץ' ],
2030 $letters[$lastIndex]
2031 );
2032
2033 // Add gershayim (double quote) to multiple-letter numbers,
2034 // but exclude numbers with only one letter after the thousands
2035 // (1001-1009, 1020, 1030, 2001-2009, etc.)
2036 if ( $letters[1] === "'" && $preTransformLength === 3 ) {
2037 $letters[] = "'";
2038 } else {
2039 array_splice( $letters, -1, 0, '"' );
2040 }
2041 }
2042
2043 return implode( $letters );
2044 }
2045
2046 /**
2047 * Used by date() and time() to adjust the time output.
2048 *
2049 * @param string $ts The time in date('YmdHis') format
2050 * @param mixed $tz Adjust the time by this amount (default false, mean we
2051 * get user timecorrection setting)
2052 * @return int
2053 */
2054 public function userAdjust( $ts, $tz = false ) {
2055 global $wgUser, $wgLocalTZoffset;
2056
2057 if ( $tz === false ) {
2058 $tz = $wgUser->getOption( 'timecorrection' );
2059 }
2060
2061 $data = explode( '|', $tz, 3 );
2062
2063 if ( $data[0] == 'ZoneInfo' ) {
2064 try {
2065 $userTZ = new DateTimeZone( $data[2] );
2066 $date = new DateTime( $ts, new DateTimeZone( 'UTC' ) );
2067 $date->setTimezone( $userTZ );
2068 return $date->format( 'YmdHis' );
2069 } catch ( Exception $e ) {
2070 // Unrecognized timezone, default to 'Offset' with the stored offset.
2071 $data[0] = 'Offset';
2072 }
2073 }
2074
2075 if ( $data[0] == 'System' || $tz == '' ) {
2076 # Global offset in minutes.
2077 $minDiff = $wgLocalTZoffset;
2078 } elseif ( $data[0] == 'Offset' ) {
2079 $minDiff = intval( $data[1] );
2080 } else {
2081 $data = explode( ':', $tz );
2082 if ( count( $data ) == 2 ) {
2083 $data[0] = intval( $data[0] );
2084 $data[1] = intval( $data[1] );
2085 $minDiff = abs( $data[0] ) * 60 + $data[1];
2086 if ( $data[0] < 0 ) {
2087 $minDiff = -$minDiff;
2088 }
2089 } else {
2090 $minDiff = intval( $data[0] ) * 60;
2091 }
2092 }
2093
2094 # No difference ? Return time unchanged
2095 if ( $minDiff == 0 ) {
2096 return $ts;
2097 }
2098
2099 Wikimedia\suppressWarnings(); // E_STRICT system time bitching
2100 # Generate an adjusted date; take advantage of the fact that mktime
2101 # will normalize out-of-range values so we don't have to split $minDiff
2102 # into hours and minutes.
2103 $t = mktime( (
2104 (int)substr( $ts, 8, 2 ) ), # Hours
2105 (int)substr( $ts, 10, 2 ) + $minDiff, # Minutes
2106 (int)substr( $ts, 12, 2 ), # Seconds
2107 (int)substr( $ts, 4, 2 ), # Month
2108 (int)substr( $ts, 6, 2 ), # Day
2109 (int)substr( $ts, 0, 4 ) ); # Year
2110
2111 $date = date( 'YmdHis', $t );
2112 Wikimedia\restoreWarnings();
2113
2114 return $date;
2115 }
2116
2117 /**
2118 * This is meant to be used by time(), date(), and timeanddate() to get
2119 * the date preference they're supposed to use, it should be used in
2120 * all children.
2121 *
2122 * function timeanddate([...], $format = true) {
2123 * $datePreference = $this->dateFormat($format);
2124 * [...]
2125 * }
2126 *
2127 * @param int|string|bool $usePrefs If true, the user's preference is used
2128 * if false, the site/language default is used
2129 * if int/string, assumed to be a format.
2130 * @return string
2131 */
2132 function dateFormat( $usePrefs = true ) {
2133 global $wgUser;
2134
2135 if ( is_bool( $usePrefs ) ) {
2136 if ( $usePrefs ) {
2137 $datePreference = $wgUser->getDatePreference();
2138 } else {
2139 $datePreference = (string)User::getDefaultOption( 'date' );
2140 }
2141 } else {
2142 $datePreference = (string)$usePrefs;
2143 }
2144
2145 // return int
2146 if ( $datePreference == '' ) {
2147 return 'default';
2148 }
2149
2150 return $datePreference;
2151 }
2152
2153 /**
2154 * Get a format string for a given type and preference
2155 * @param string $type May be 'date', 'time', 'both', or 'pretty'.
2156 * @param string $pref The format name as it appears in Messages*.php under
2157 * $datePreferences.
2158 *
2159 * @since 1.22 New type 'pretty' that provides a more readable timestamp format
2160 *
2161 * @return string
2162 */
2163 function getDateFormatString( $type, $pref ) {
2164 $wasDefault = false;
2165 if ( $pref == 'default' ) {
2166 $wasDefault = true;
2167 $pref = $this->getDefaultDateFormat();
2168 }
2169
2170 if ( !isset( $this->dateFormatStrings[$type][$pref] ) ) {
2171 $df =
2172 $this->localisationCache->getSubitem( $this->mCode, 'dateFormats', "$pref $type" );
2173
2174 if ( $type === 'pretty' && $df === null ) {
2175 $df = $this->getDateFormatString( 'date', $pref );
2176 }
2177
2178 if ( !$wasDefault && $df === null ) {
2179 $pref = $this->getDefaultDateFormat();
2180 $df = $this->getLocalisationCache()
2181 ->getSubitem( $this->mCode, 'dateFormats', "$pref $type" );
2182 }
2183
2184 $this->dateFormatStrings[$type][$pref] = $df;
2185 }
2186 return $this->dateFormatStrings[$type][$pref];
2187 }
2188
2189 /**
2190 * @param string $ts The time format which needs to be turned into a
2191 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2192 * @param bool $adj Whether to adjust the time output according to the
2193 * user configured offset ($timecorrection)
2194 * @param mixed $format True to use user's date format preference
2195 * @param string|bool $timecorrection The time offset as returned by
2196 * validateTimeZone() in Special:Preferences
2197 * @return string
2198 */
2199 public function date( $ts, $adj = false, $format = true, $timecorrection = false ) {
2200 $ts = wfTimestamp( TS_MW, $ts );
2201 if ( $adj ) {
2202 $ts = $this->userAdjust( $ts, $timecorrection );
2203 }
2204 $df = $this->getDateFormatString( 'date', $this->dateFormat( $format ) );
2205 return $this->sprintfDate( $df, $ts );
2206 }
2207
2208 /**
2209 * @param string $ts The time format which needs to be turned into a
2210 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2211 * @param bool $adj Whether to adjust the time output according to the
2212 * user configured offset ($timecorrection)
2213 * @param mixed $format True to use user's date format preference
2214 * @param string|bool $timecorrection The time offset as returned by
2215 * validateTimeZone() in Special:Preferences
2216 * @return string
2217 */
2218 public function time( $ts, $adj = false, $format = true, $timecorrection = false ) {
2219 $ts = wfTimestamp( TS_MW, $ts );
2220 if ( $adj ) {
2221 $ts = $this->userAdjust( $ts, $timecorrection );
2222 }
2223 $df = $this->getDateFormatString( 'time', $this->dateFormat( $format ) );
2224 return $this->sprintfDate( $df, $ts );
2225 }
2226
2227 /**
2228 * @param string $ts The time format which needs to be turned into a
2229 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2230 * @param bool $adj Whether to adjust the time output according to the
2231 * user configured offset ($timecorrection)
2232 * @param mixed $format What format to return, if it's false output the
2233 * default one (default true)
2234 * @param string|bool $timecorrection The time offset as returned by
2235 * validateTimeZone() in Special:Preferences
2236 * @return string
2237 */
2238 public function timeanddate( $ts, $adj = false, $format = true, $timecorrection = false ) {
2239 $ts = wfTimestamp( TS_MW, $ts );
2240 if ( $adj ) {
2241 $ts = $this->userAdjust( $ts, $timecorrection );
2242 }
2243 $df = $this->getDateFormatString( 'both', $this->dateFormat( $format ) );
2244 return $this->sprintfDate( $df, $ts );
2245 }
2246
2247 /**
2248 * Takes a number of seconds and turns it into a text using values such as hours and minutes.
2249 *
2250 * @since 1.20
2251 *
2252 * @param int $seconds The amount of seconds.
2253 * @param array $chosenIntervals The intervals to enable.
2254 *
2255 * @return string
2256 */
2257 public function formatDuration( $seconds, array $chosenIntervals = [] ) {
2258 $intervals = $this->getDurationIntervals( $seconds, $chosenIntervals );
2259
2260 $segments = [];
2261
2262 foreach ( $intervals as $intervalName => $intervalValue ) {
2263 // Messages: duration-seconds, duration-minutes, duration-hours, duration-days, duration-weeks,
2264 // duration-years, duration-decades, duration-centuries, duration-millennia
2265 $message = wfMessage( 'duration-' . $intervalName )->numParams( $intervalValue );
2266 $segments[] = $message->inLanguage( $this )->escaped();
2267 }
2268
2269 return $this->listToText( $segments );
2270 }
2271
2272 /**
2273 * Takes a number of seconds and returns an array with a set of corresponding intervals.
2274 * For example 65 will be turned into [ minutes => 1, seconds => 5 ].
2275 *
2276 * @since 1.20
2277 *
2278 * @param int $seconds The amount of seconds.
2279 * @param array $chosenIntervals The intervals to enable.
2280 *
2281 * @return array
2282 */
2283 public function getDurationIntervals( $seconds, array $chosenIntervals = [] ) {
2284 if ( empty( $chosenIntervals ) ) {
2285 $chosenIntervals = [
2286 'millennia',
2287 'centuries',
2288 'decades',
2289 'years',
2290 'days',
2291 'hours',
2292 'minutes',
2293 'seconds'
2294 ];
2295 }
2296
2297 $intervals = array_intersect_key( self::$durationIntervals, array_flip( $chosenIntervals ) );
2298 $sortedNames = array_keys( $intervals );
2299 $smallestInterval = array_pop( $sortedNames );
2300
2301 $segments = [];
2302
2303 foreach ( $intervals as $name => $length ) {
2304 $value = floor( $seconds / $length );
2305
2306 if ( $value > 0 || ( $name == $smallestInterval && empty( $segments ) ) ) {
2307 $seconds -= $value * $length;
2308 $segments[$name] = $value;
2309 }
2310 }
2311
2312 return $segments;
2313 }
2314
2315 /**
2316 * Internal helper function for userDate(), userTime() and userTimeAndDate()
2317 *
2318 * @param string $type Can be 'date', 'time' or 'both'
2319 * @param string $ts The time format which needs to be turned into a
2320 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2321 * @param User $user User object used to get preferences for timezone and format
2322 * @param array $options Array, can contain the following keys:
2323 * - 'timecorrection': time correction, can have the following values:
2324 * - true: use user's preference
2325 * - false: don't use time correction
2326 * - int: value of time correction in minutes
2327 * - 'format': format to use, can have the following values:
2328 * - true: use user's preference
2329 * - false: use default preference
2330 * - string: format to use
2331 * @since 1.19
2332 * @return string
2333 */
2334 private function internalUserTimeAndDate( $type, $ts, User $user, array $options ) {
2335 $ts = wfTimestamp( TS_MW, $ts );
2336 $options += [ 'timecorrection' => true, 'format' => true ];
2337 if ( $options['timecorrection'] !== false ) {
2338 if ( $options['timecorrection'] === true ) {
2339 $offset = $user->getOption( 'timecorrection' );
2340 } else {
2341 $offset = $options['timecorrection'];
2342 }
2343 $ts = $this->userAdjust( $ts, $offset );
2344 }
2345 if ( $options['format'] === true ) {
2346 $format = $user->getDatePreference();
2347 } else {
2348 $format = $options['format'];
2349 }
2350 $df = $this->getDateFormatString( $type, $this->dateFormat( $format ) );
2351 return $this->sprintfDate( $df, $ts );
2352 }
2353
2354 /**
2355 * Get the formatted date for the given timestamp and formatted for
2356 * the given user.
2357 *
2358 * @param mixed $ts Mixed: the time format which needs to be turned into a
2359 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2360 * @param User $user User object used to get preferences for timezone and format
2361 * @param array $options Array, can contain the following keys:
2362 * - 'timecorrection': time correction, can have the following values:
2363 * - true: use user's preference
2364 * - false: don't use time correction
2365 * - int: value of time correction in minutes
2366 * - 'format': format to use, can have the following values:
2367 * - true: use user's preference
2368 * - false: use default preference
2369 * - string: format to use
2370 * @since 1.19
2371 * @return string
2372 */
2373 public function userDate( $ts, User $user, array $options = [] ) {
2374 return $this->internalUserTimeAndDate( 'date', $ts, $user, $options );
2375 }
2376
2377 /**
2378 * Get the formatted time for the given timestamp and formatted for
2379 * the given user.
2380 *
2381 * @param mixed $ts The time format which needs to be turned into a
2382 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2383 * @param User $user User object used to get preferences for timezone and format
2384 * @param array $options Array, can contain the following keys:
2385 * - 'timecorrection': time correction, can have the following values:
2386 * - true: use user's preference
2387 * - false: don't use time correction
2388 * - int: value of time correction in minutes
2389 * - 'format': format to use, can have the following values:
2390 * - true: use user's preference
2391 * - false: use default preference
2392 * - string: format to use
2393 * @since 1.19
2394 * @return string
2395 */
2396 public function userTime( $ts, User $user, array $options = [] ) {
2397 return $this->internalUserTimeAndDate( 'time', $ts, $user, $options );
2398 }
2399
2400 /**
2401 * Get the formatted date and time for the given timestamp and formatted for
2402 * the given user.
2403 *
2404 * @param mixed $ts The time format which needs to be turned into a
2405 * date('YmdHis') format with wfTimestamp(TS_MW,$ts)
2406 * @param User $user User object used to get preferences for timezone and format
2407 * @param array $options Array, can contain the following keys:
2408 * - 'timecorrection': time correction, can have the following values:
2409 * - true: use user's preference
2410 * - false: don't use time correction
2411 * - int: value of time correction in minutes
2412 * - 'format': format to use, can have the following values:
2413 * - true: use user's preference
2414 * - false: use default preference
2415 * - string: format to use
2416 * @since 1.19
2417 * @return string
2418 */
2419 public function userTimeAndDate( $ts, User $user, array $options = [] ) {
2420 return $this->internalUserTimeAndDate( 'both', $ts, $user, $options );
2421 }
2422
2423 /**
2424 * Get the timestamp in a human-friendly relative format, e.g., "3 days ago".
2425 *
2426 * Determine the difference between the timestamp and the current time, and
2427 * generate a readable timestamp by returning "<N> <units> ago", where the
2428 * largest possible unit is used.
2429 *
2430 * @since 1.26 (Prior to 1.26 method existed but was not meant to be used directly)
2431 *
2432 * @param MWTimestamp $time
2433 * @param MWTimestamp|null $relativeTo The base timestamp to compare to (defaults to now)
2434 * @param User|null $user User the timestamp is being generated for
2435 * (or null to use main context's user)
2436 * @return string Formatted timestamp
2437 */
2438 public function getHumanTimestamp(
2439 MWTimestamp $time, MWTimestamp $relativeTo = null, User $user = null
2440 ) {
2441 if ( $relativeTo === null ) {
2442 $relativeTo = new MWTimestamp();
2443 }
2444 if ( $user === null ) {
2445 $user = RequestContext::getMain()->getUser();
2446 }
2447
2448 // Adjust for the user's timezone.
2449 $offsetThis = $time->offsetForUser( $user );
2450 $offsetRel = $relativeTo->offsetForUser( $user );
2451
2452 $ts = '';
2453 if ( Hooks::run( 'GetHumanTimestamp', [ &$ts, $time, $relativeTo, $user, $this ] ) ) {
2454 $ts = $this->getHumanTimestampInternal( $time, $relativeTo, $user );
2455 }
2456
2457 // Reset the timezone on the objects.
2458 $time->timestamp->sub( $offsetThis );
2459 $relativeTo->timestamp->sub( $offsetRel );
2460
2461 return $ts;
2462 }
2463
2464 /**
2465 * Convert an MWTimestamp into a pretty human-readable timestamp using
2466 * the given user preferences and relative base time.
2467 *
2468 * @see Language::getHumanTimestamp
2469 * @param MWTimestamp $ts Timestamp to prettify
2470 * @param MWTimestamp $relativeTo Base timestamp
2471 * @param User $user User preferences to use
2472 * @return string Human timestamp
2473 * @since 1.26
2474 */
2475 private function getHumanTimestampInternal(
2476 MWTimestamp $ts, MWTimestamp $relativeTo, User $user
2477 ) {
2478 $diff = $ts->diff( $relativeTo );
2479 $diffDay = (bool)( (int)$ts->timestamp->format( 'w' ) -
2480 (int)$relativeTo->timestamp->format( 'w' ) );
2481 $days = $diff->days ?: (int)$diffDay;
2482 if ( $diff->invert || $days > 5
2483 && $ts->timestamp->format( 'Y' ) !== $relativeTo->timestamp->format( 'Y' )
2484 ) {
2485 // Timestamps are in different years: use full timestamp
2486 // Also do full timestamp for future dates
2487 /**
2488 * @todo FIXME: Add better handling of future timestamps.
2489 */
2490 $format = $this->getDateFormatString( 'both', $user->getDatePreference() ?: 'default' );
2491 $ts = $this->sprintfDate( $format, $ts->getTimestamp( TS_MW ) );
2492 } elseif ( $days > 5 ) {
2493 // Timestamps are in same year, but more than 5 days ago: show day and month only.
2494 $format = $this->getDateFormatString( 'pretty', $user->getDatePreference() ?: 'default' );
2495 $ts = $this->sprintfDate( $format, $ts->getTimestamp( TS_MW ) );
2496 } elseif ( $days > 1 ) {
2497 // Timestamp within the past week: show the day of the week and time
2498 $format = $this->getDateFormatString( 'time', $user->getDatePreference() ?: 'default' );
2499 $weekday = self::$mWeekdayMsgs[$ts->timestamp->format( 'w' )];
2500 // Messages:
2501 // sunday-at, monday-at, tuesday-at, wednesday-at, thursday-at, friday-at, saturday-at
2502 $ts = wfMessage( "$weekday-at" )
2503 ->inLanguage( $this )
2504 ->params( $this->sprintfDate( $format, $ts->getTimestamp( TS_MW ) ) )
2505 ->text();
2506 } elseif ( $days == 1 ) {
2507 // Timestamp was yesterday: say 'yesterday' and the time.
2508 $format = $this->getDateFormatString( 'time', $user->getDatePreference() ?: 'default' );
2509 $ts = wfMessage( 'yesterday-at' )
2510 ->inLanguage( $this )
2511 ->params( $this->sprintfDate( $format, $ts->getTimestamp( TS_MW ) ) )
2512 ->text();
2513 } elseif ( $diff->h > 1 || $diff->h == 1 && $diff->i > 30 ) {
2514 // Timestamp was today, but more than 90 minutes ago: say 'today' and the time.
2515 $format = $this->getDateFormatString( 'time', $user->getDatePreference() ?: 'default' );
2516 $ts = wfMessage( 'today-at' )
2517 ->inLanguage( $this )
2518 ->params( $this->sprintfDate( $format, $ts->getTimestamp( TS_MW ) ) )
2519 ->text();
2520
2521 // From here on in, the timestamp was soon enough ago so that we can simply say
2522 // XX units ago, e.g., "2 hours ago" or "5 minutes ago"
2523 } elseif ( $diff->h == 1 ) {
2524 // Less than 90 minutes, but more than an hour ago.
2525 $ts = wfMessage( 'hours-ago' )->inLanguage( $this )->numParams( 1 )->text();
2526 } elseif ( $diff->i >= 1 ) {
2527 // A few minutes ago.
2528 $ts = wfMessage( 'minutes-ago' )->inLanguage( $this )->numParams( $diff->i )->text();
2529 } elseif ( $diff->s >= 30 ) {
2530 // Less than a minute, but more than 30 sec ago.
2531 $ts = wfMessage( 'seconds-ago' )->inLanguage( $this )->numParams( $diff->s )->text();
2532 } else {
2533 // Less than 30 seconds ago.
2534 $ts = wfMessage( 'just-now' )->text();
2535 }
2536
2537 return $ts;
2538 }
2539
2540 /**
2541 * @param string $key
2542 * @return string|null
2543 */
2544 public function getMessage( $key ) {
2545 return $this->localisationCache->getSubitem( $this->mCode, 'messages', $key );
2546 }
2547
2548 /**
2549 * @return array
2550 */
2551 function getAllMessages() {
2552 return $this->localisationCache->getItem( $this->mCode, 'messages' );
2553 }
2554
2555 /**
2556 * @param string $in
2557 * @param string $out
2558 * @param string $string
2559 * @return string
2560 */
2561 public function iconv( $in, $out, $string ) {
2562 # Even with //IGNORE iconv can whine about illegal characters in
2563 # *input* string. We just ignore those too.
2564 # REF: https://bugs.php.net/bug.php?id=37166
2565 # REF: https://phabricator.wikimedia.org/T18885
2566 Wikimedia\suppressWarnings();
2567 $text = iconv( $in, $out . '//IGNORE', $string );
2568 Wikimedia\restoreWarnings();
2569 return $text;
2570 }
2571
2572 // callback functions for ucwords(), ucwordbreaks()
2573
2574 /**
2575 * @param array $matches
2576 * @return mixed|string
2577 */
2578 function ucwordbreaksCallbackAscii( $matches ) {
2579 return $this->ucfirst( $matches[1] );
2580 }
2581
2582 /**
2583 * @param array $matches
2584 * @return string
2585 */
2586 function ucwordbreaksCallbackMB( $matches ) {
2587 return mb_strtoupper( $matches[0] );
2588 }
2589
2590 /**
2591 * @param array $matches
2592 * @return string
2593 */
2594 function ucwordsCallbackMB( $matches ) {
2595 return mb_strtoupper( $matches[0] );
2596 }
2597
2598 /**
2599 * Make a string's first character uppercase
2600 *
2601 * @param string $str
2602 *
2603 * @return string
2604 */
2605 public function ucfirst( $str ) {
2606 $o = ord( $str );
2607 if ( $o < 96 ) { // if already uppercase...
2608 return $str;
2609 } elseif ( $o < 128 ) {
2610 return ucfirst( $str ); // use PHP's ucfirst()
2611 } else {
2612 // fall back to more complex logic in case of multibyte strings
2613 return $this->uc( $str, true );
2614 }
2615 }
2616
2617 /**
2618 * Convert a string to uppercase
2619 *
2620 * @param string $str
2621 * @param bool $first
2622 *
2623 * @return string
2624 */
2625 public function uc( $str, $first = false ) {
2626 if ( $first ) {
2627 if ( $this->isMultibyte( $str ) ) {
2628 return $this->mbUpperChar( mb_substr( $str, 0, 1 ) ) . mb_substr( $str, 1 );
2629 } else {
2630 return ucfirst( $str );
2631 }
2632 } else {
2633 return $this->isMultibyte( $str ) ? mb_strtoupper( $str ) : strtoupper( $str );
2634 }
2635 }
2636
2637 /**
2638 * Convert character to uppercase, allowing overrides of the default mb_upper
2639 * behaviour, which is buggy in many ways. Having a conversion table can be
2640 * useful during transitions between PHP versions where unicode changes happen.
2641 * This can make some resources unreachable on-wiki, see discussion at T219279.
2642 * Providing such a conversion table can allow to manage the transition period.
2643 *
2644 * @since 1.34
2645 *
2646 * @param string $char
2647 *
2648 * @return string
2649 */
2650 protected function mbUpperChar( $char ) {
2651 global $wgOverrideUcfirstCharacters;
2652 if ( array_key_exists( $char, $wgOverrideUcfirstCharacters ) ) {
2653 return $wgOverrideUcfirstCharacters[$char];
2654 } else {
2655 return mb_strtoupper( $char );
2656 }
2657 }
2658
2659 /**
2660 * @param string $str
2661 * @return mixed|string
2662 */
2663 function lcfirst( $str ) {
2664 $o = ord( $str );
2665 if ( !$o ) {
2666 return strval( $str );
2667 } elseif ( $o >= 128 ) {
2668 return $this->lc( $str, true );
2669 } elseif ( $o > 96 ) {
2670 return $str;
2671 } else {
2672 $str[0] = strtolower( $str[0] );
2673 return $str;
2674 }
2675 }
2676
2677 /**
2678 * @param string $str
2679 * @param bool $first
2680 * @return mixed|string
2681 */
2682 function lc( $str, $first = false ) {
2683 if ( $first ) {
2684 if ( $this->isMultibyte( $str ) ) {
2685 return mb_strtolower( mb_substr( $str, 0, 1 ) ) . mb_substr( $str, 1 );
2686 } else {
2687 return strtolower( substr( $str, 0, 1 ) ) . substr( $str, 1 );
2688 }
2689 } else {
2690 return $this->isMultibyte( $str ) ? mb_strtolower( $str ) : strtolower( $str );
2691 }
2692 }
2693
2694 /**
2695 * @param string $str
2696 * @return bool
2697 */
2698 function isMultibyte( $str ) {
2699 return strlen( $str ) !== mb_strlen( $str );
2700 }
2701
2702 /**
2703 * @param string $str
2704 * @return mixed|string
2705 */
2706 function ucwords( $str ) {
2707 if ( $this->isMultibyte( $str ) ) {
2708 $str = $this->lc( $str );
2709
2710 // regexp to find first letter in each word (i.e. after each space)
2711 $replaceRegexp = "/^([a-z]|[\\xc0-\\xff][\\x80-\\xbf]*)| ([a-z]|[\\xc0-\\xff][\\x80-\\xbf]*)/";
2712
2713 // function to use to capitalize a single char
2714 return preg_replace_callback(
2715 $replaceRegexp,
2716 [ $this, 'ucwordsCallbackMB' ],
2717 $str
2718 );
2719 } else {
2720 return ucwords( strtolower( $str ) );
2721 }
2722 }
2723
2724 /**
2725 * capitalize words at word breaks
2726 *
2727 * @param string $str
2728 * @return mixed
2729 */
2730 function ucwordbreaks( $str ) {
2731 if ( $this->isMultibyte( $str ) ) {
2732 $str = $this->lc( $str );
2733
2734 // since \b doesn't work for UTF-8, we explicitely define word break chars
2735 $breaks = "[ \-\(\)\}\{\.,\?!]";
2736
2737 // find first letter after word break
2738 $replaceRegexp = "/^([a-z]|[\\xc0-\\xff][\\x80-\\xbf]*)|" .
2739 "$breaks([a-z]|[\\xc0-\\xff][\\x80-\\xbf]*)/";
2740
2741 return preg_replace_callback(
2742 $replaceRegexp,
2743 [ $this, 'ucwordbreaksCallbackMB' ],
2744 $str
2745 );
2746 } else {
2747 return preg_replace_callback(
2748 '/\b([\w\x80-\xff]+)\b/',
2749 [ $this, 'ucwordbreaksCallbackAscii' ],
2750 $str
2751 );
2752 }
2753 }
2754
2755 /**
2756 * Return a case-folded representation of $s
2757 *
2758 * This is a representation such that caseFold($s1)==caseFold($s2) if $s1
2759 * and $s2 are the same except for the case of their characters. It is not
2760 * necessary for the value returned to make sense when displayed.
2761 *
2762 * Do *not* perform any other normalisation in this function. If a caller
2763 * uses this function when it should be using a more general normalisation
2764 * function, then fix the caller.
2765 *
2766 * @param string $s
2767 *
2768 * @return string
2769 */
2770 function caseFold( $s ) {
2771 return $this->uc( $s );
2772 }
2773
2774 /**
2775 * TODO: $s is not always a string per T218883
2776 * @param string $s
2777 * @return string
2778 */
2779 function checkTitleEncoding( $s ) {
2780 if ( is_array( $s ) ) {
2781 throw new MWException( 'Given array to checkTitleEncoding.' );
2782 }
2783 if ( StringUtils::isUtf8( $s ) ) {
2784 return $s;
2785 }
2786
2787 return $this->iconv( $this->fallback8bitEncoding(), 'utf-8', $s );
2788 }
2789
2790 /**
2791 * @return string
2792 */
2793 function fallback8bitEncoding() {
2794 return $this->localisationCache->getItem( $this->mCode, 'fallback8bitEncoding' );
2795 }
2796
2797 /**
2798 * Most writing systems use whitespace to break up words.
2799 * Some languages such as Chinese don't conventionally do this,
2800 * which requires special handling when breaking up words for
2801 * searching etc.
2802 *
2803 * @return bool
2804 */
2805 function hasWordBreaks() {
2806 return true;
2807 }
2808
2809 /**
2810 * Some languages such as Chinese require word segmentation,
2811 * Specify such segmentation when overridden in derived class.
2812 *
2813 * @param string $string
2814 * @return string
2815 */
2816 function segmentByWord( $string ) {
2817 return $string;
2818 }
2819
2820 /**
2821 * Some languages have special punctuation need to be normalized.
2822 * Make such changes here.
2823 *
2824 * @param string $string
2825 * @return string
2826 */
2827 function normalizeForSearch( $string ) {
2828 return self::convertDoubleWidth( $string );
2829 }
2830
2831 /**
2832 * convert double-width roman characters to single-width.
2833 * range: ff00-ff5f ~= 0020-007f
2834 *
2835 * @param string $string
2836 *
2837 * @return string
2838 */
2839 protected static function convertDoubleWidth( $string ) {
2840 static $full = null;
2841 static $half = null;
2842
2843 if ( $full === null ) {
2844 $fullWidth = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
2845 $halfWidth = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
2846 $full = str_split( $fullWidth, 3 );
2847 $half = str_split( $halfWidth );
2848 }
2849
2850 $string = str_replace( $full, $half, $string );
2851 return $string;
2852 }
2853
2854 /**
2855 * @param string $string
2856 * @param string $pattern
2857 * @return string
2858 */
2859 protected static function insertSpace( $string, $pattern ) {
2860 $string = preg_replace( $pattern, " $1 ", $string );
2861 $string = preg_replace( '/ +/', ' ', $string );
2862 return $string;
2863 }
2864
2865 /**
2866 * @param string[] $termsArray
2867 * @return string[]
2868 */
2869 function convertForSearchResult( $termsArray ) {
2870 # some languages, e.g. Chinese, need to do a conversion
2871 # in order for search results to be displayed correctly
2872 return $termsArray;
2873 }
2874
2875 /**
2876 * Get the first character of a string.
2877 *
2878 * @param string $s
2879 * @return string
2880 */
2881 function firstChar( $s ) {
2882 $matches = [];
2883 preg_match(
2884 '/^([\x00-\x7f]|[\xc0-\xdf][\x80-\xbf]|' .
2885 '[\xe0-\xef][\x80-\xbf]{2}|[\xf0-\xf7][\x80-\xbf]{3})/',
2886 $s,
2887 $matches
2888 );
2889
2890 if ( isset( $matches[1] ) ) {
2891 if ( strlen( $matches[1] ) != 3