I18n & L10n - 010PHP

These are the slides as presented at 010PHP on 14 Apr 2016

Transcript

1. Internationalis(z)ation and Localis(z)ation: The Basics® 010PHP 2016 - 04 - 14 Camilo Sperberg / @unreal4u
2. Main index 2 ‣Definitions and requirements ‣The \Locale object ‣What constitutes a locale ‣Encodings ‣Database ‣Other considerations
3. What is internationalisation? Internationalisation (i18n): Preparing your application to be localized (Pro-tip: How? Watch this talk!) Localisation (L10n): Translating, adding icons and other things of certain zone 3
4. …. …. Most discussed thing in RFC's 4 Bilbo Baggins 1290 - ? RFC 3066 Jan. 2001 RFC 4646 Sep. 2006 RFC 4647 Sep. 2006 Bungo Baggins 1246 - 1326 RFC 5646 Sep. 2009 Balbo Baggins 1167 - 1258 Mungo Baggins 1207 - 1300 RFC 1766 Mar. 1995 Son of Son of Son of Replaces Replaces Replaces Belladonna Took Laura Grubb Berylla Boffin 1172 - 1265 LOTR RFC i18n
5. i18n and PHP: recommended extensions 5 ‣ For gettext: php-gettext ‣ Needs gettext ‣ https://www.gnu.org/software/gettext/ ‣ For intl: php-intl ‣ Needs ICU4C ‣ http://site.icu-project.org/ ‣ mb_* functions: multibyte support
6. L10n: Definition "A set of parameters that defines the user's language, country and any special variant preferences that the user wants to see in their user interface" Wikipedia AKA: rules for a specific region ‣ Current code standard: RFC 4646 ‣ Language in lowercase (ISO 639-1), hyphen, region in uppercase (ISO 3166-1 alpha-2) 6
7. L10n: Code standards nl-NL Dutch as SUISHE*® in the Netherlands nl-BE Dutch as SUISHE*® in Belgium 7 * Spoken, Used, Interpreted, Seen, Heard, Etc
8. L10n: Code standards 8 * Spoken, Used, Interpreted, Seen, Heard, Etc pt-BR Portuguese as SUISHE*® in Brazil pt-PT Portuguese as SUISHE*® in Portugal
9. L10n: Code standards 9 * Spoken, Used, Interpreted, Seen, Heard, Etc es-ES, es-CL, es-AR, es-PE, es-* Spanish as SUISHE*® in Spain, Chile, Argentina, Peru, etc.
10. L10n: Code standards de-CH-1901 German as used in Switzerland using the 1901 variant 10
11. L10n: Code standards hy-Latn-IT-arevela Eastern Armenian written in Latin script, as used in Italy 11
12. L10n: Determining which to use Introducing the \Locale object! 12 Best practice tips: ‣ Get it from your $_GET or $_POST ‣ Get it from the headers ‣ Get it based on ip
13. i18n: Detecting locale 13 public function getLocaleFromClient() {
 $this->locale = $this->getLocaleFromGetRequest();
 if (empty($this->locale)) {
 $this->locale = $this->getLocaleFromHeaders();
 if (empty($this->locale)) {
 $this->locale = $this->getLocaleFromIP();
 }
 }
 
 return $this->locale;
 }
14. i18n: Detecting locale 14 public function getLocaleFromHeaders() {
 $this->locale = '';
 
 if (isset($_SERVER['HTTP_ACCEPT_LANGUAGE'])) {
 $preferredLocale = \Locale::acceptFromHttp($_SERVER['HTTP_ACCEPT_LANGUAGE']);
 $this->locale = $this->_checkLocale($preferredLocale);
 }
 
 return $this->locale;
 }
15. \Locale: cool functions! 15 Use To… ‣ \Locale::getDisplayLanguage() Create a list of languages ‣ \Locale::getDisplayRegion() Display the country where the locale is used ‣ \Locale::acceptFromHttp() Parse browser Accept header ‣ \Locale::setDefault() Set the default localization to use
16. L10n: what to look out for? ‣ Warning! Locale is not just a translation ‣ What's normal for you can be strange to others 16
17. Images and gestures ‣ Pointing fingers can be offensive in Arabic countries 17 ✌ ‣ The "peace" sign is offensive in Australia, Ireland, New Zealand, South Africa and the United Kingdom
18. Translations - Semantics (L10n) 18 echo $numberResults.' results found within a '.$range.'km range'; If translated directly into Spanish, it will sound like this: ➡ "Found 123 results within range 5km"
19. Translations - Semantics (i18n) 19 ‣Solvable by implementing printf() printf( '%1$d results found within a %2$d km range', $numberResults, $range ); ‣Translator decides where to print out variables
20. Translations - Semantics (L10n) 20 es-ES: Carro Coche es-CL: Auto
21. This is a “carro” or “coche” in es-CL:
22. Translations - Semantics (L10n) 22 British American Holiday Vacation Football Soccer American Football Football Flat Apartment Garden Yard Rubbish Garbage / Trash
23. Translations - Semantics 23 Smart Phones Mobile phones Barcode Scanners Control Remotes See Printers Interactive question! Suppose an electronics shop: what's wrong with the following list?
24. Translations - Semantics 24 Captain here: "Watch" was translated as the verb instead of the noun
26. Translations - Plural forms (L10n) 26 English Polish 0 Apples Jabłek 1 Apple Jabłko 2 .. 4 Apples Jabłka 5 .. 21 Apples Jabłek 22 .. 24 Apples Jabłka 25 .. 31 Apples Jabłek More complex cases do exist! ‣ Slovenian: 4 plural forms There are also cases with 1 plural form ‣ Japanese ‣ Vietnamese
27. Translations - Plural forms (i18n) 27 Gettext! ‣ Supports plural forms ‣ Is cached in RAM (pros and cons) ‣ Very easy to edit (poEdit) ‣ Can be separated into modules ‣ Produces compiled language files
28. Translations - Plural forms (i18n) ‣ \MessageFormatter can also help ‣ Can do pretty amazing stuff ‣ I personally don't have experience with it 28 $fmt = new MessageFormatter(
 'en_GB',
 'Peter has {0, plural, =0{no cat} =1{a cat} other{# cats}}'
 );
 echo $fmt->format(array(0));
 $fmt = new MessageFormatter(
 'nl_NL',
 'Peter heeft {0, plural, =0{geen kat} =1{een kat} other{# katten}}'
 );
 echo $fmt->format(array(0));
 
 // Outputs:
 // Peter has no cat
 // Peter heeft geen kat
29. Number formatting - L10n ‣ Numbers have a lot of different types of annotations ‣ Corollary: nobody really knows well how a number should be formatted 29 Interactive question! How do you format the following negative number, in Euros, here in the Netherlands? 1234,57
30. Number formatting - L10n 30 nl-NL fr-FR pt-BR hi-IN ps-AR "-1.234,57" "-1 234,57" "-1.234,57" "-१,२३४.५७" "-۱٬۲۳۴٫۵۷" "€ 1.234,57-" "-1 234,57 €" "(€1.234,57)" "-€ १,२३४.५७" "-۱٬۲۳۴٫۵۷ €" "25%" "25 %" "25%" "२५%" "۲۵٪"
31. Number formatting: i18n 31 $locales = ['nl-NL', 'fr-FR', 'pt-BR', 'hi-IN', 'ps-AR',];
 
 foreach ($locales as $myLocale) {
 $numberFormatter = new \NumberFormatter($myLocale, \NumberFormatter::DECIMAL);
 $percentFormatter = new \NumberFormatter($myLocale, \NumberFormatter::PERCENT);
 $currencyFormatter = new \NumberFormatter($myLocale, \NumberFormatter::CURRENCY);
 printf('Locale: %s'.PHP_EOL, $myLocale);
 printf('[DEC]-1.234,57: "%s" :: ', $numberFormatter->format(-1234.57));
 printf('[PER]25%%: "%s" :: ', $percentFormatter->format(0.25));
 printf('[CUR]1.234,57-: "%s"'.PHP_EOL, $currencyFormatter->formatCurrency(-1234.57, 'EUR')); // Last argument can also be \NumberFormatter::CURRENCY to print in CURRENCY of loaded locale
 }
32. Number formatting: i18n 32 $locales = ['nl-NL', 'fr-FR', 'pt-BR', 'hi-IN', 'ps-AR',];
 
 foreach ($locales as $myLocale) {
 $numberFormatter = new \NumberFormatter($myLocale, \NumberFormatter::DECIMAL);
 $percentFormatter = new \NumberFormatter($myLocale, \NumberFormatter::PERCENT);
 $currencyFormatter = new \NumberFormatter($myLocale, \NumberFormatter::CURRENCY);
 printf('Locale: %s'.PHP_EOL, $myLocale);
 printf('[DEC]-1.234,57: "%s" :: ', $numberFormatter->format(-1234.57));
 printf('[PER]25%%: "%s" :: ', $percentFormatter->format(0.25));
 printf('[CUR]1.234,57-: "%s"'.PHP_EOL, $currencyFormatter->formatCurrency(-1234.57, 'EUR')); // Last argument can also be \NumberFormatter::CURRENCY to print in CURRENCY of loaded locale
 }
33. 33 unreal4u-MBP:localization unreal4u$ php numbers.php Locale: nl-NL [DEC]-1.234,57: "-1.234,57" :: [PER]25%: "25%" :: [CUR]1.234,57-: "€ 1.234,57-" All this in Dutch AKA Nederlands -------------------------------------------------------------------------------- Locale: fr-FR [DEC]-1.234,57: "-1 234,57" :: [PER]25%: "25 %" :: [CUR]1.234,57-: "-1 234,57 €" All this in French AKA français -------------------------------------------------------------------------------- Locale: pt-BR [DEC]-1.234,57: "-1.234,57" :: [PER]25%: "25%" :: [CUR]1.234,57-: "(€1.234,57)" All this in Portuguese AKA português -------------------------------------------------------------------------------- Locale: hi-IN [DEC]-1.234,57: "-१,२३४.५७" :: [PER]25%: "२५%" :: [CUR]1.234,57-: "-€ १,२३४.५७" All this in Hindi AKA िहन्दी -------------------------------------------------------------------------------- Locale: ps-AR [DEC]-1.234,57: "-١،٢٣۴٫۵٧" :: [PER]25%: "٢۵٪" :: [CUR]1.234,57-: "-١،٢٣۴٫۵٧ €" All this in Pashto AKA ﻮﺘ,ﭘ
34. Some \NumberFormatter problems ‣ \NumberFormatter::DURATION isn't implemented in much locales ‣ echo $fmt->format(12345) -> 3 hours, 25 minutes, 45 seconds ‣ "Easy" to implement using getPattern() and setPattern() ‣ Documentation exists, but is not optimal 34
35. Date and time formatting - L10n ‣ Dates have 3 different annotations ‣ YYYY-MM-DD (1.660M) ‣ DD-MM-YYYY (4.810M) ‣ MM-DD-YYYY (320M) ‣ "It's complicated" (457M) ‣ Contrary to numbers, everybody knows how a date is formatted 35 Interactive question! What is the value of the following date in The Netherlands? 05-03-13
36. Date and time formatting - L10n ➡ YMD (1660) ➡ YMD and DMY (287) ➡ DMY (3295) ➡ DMY and MDY (130) ➡ MDY (320) ➡ YMD and DMY and MDY (40) 36 https://en.wikipedia.org/wiki/Date_format_by_country
37. Date and time formatting: i18n ‣ Use PHP's \*Date* related classes, like ALWAYS! ‣ Incredibly versatile yet powerful functions ‣ Specially in combination with locales ‣ Always work in UTC, let the \*Date* classes do the rest 37
38. Date and time formatting: i18n 38 nl-NL fr-FR hi-IN ps-AR Short "23-05-15" "23/05/15" "२३-५-१५" "۲۰۱۵/۵/۲۳" Medium "23 mei 2015" "23 mai 2015" "२३-०५-२०१५" "۲۳ ۲۰۱۵ یم" With time "23 mei 2015 01:34:09" "23 mai 2015 01:34:09" "२३-०५-२०१५ १:३४:०९ पूवार्ह्न" "۲۳ ۱:۳۴:۰۹ ۲۰۱۵ یم"
39. Date and time formatting: i18n 39 $locales = ['nl-NL', 'fr-FR', 'pt-BR', 'hi-IN', 'ps-AR',];
 
 $printDate = new \DateTime('23-05-2015 01:34:09', new \DateTimeZone('UTC'));
 
 foreach ($locales as $myLocale) {
 $dateObject = \intlDateFormatter::create( $myLocale, \intlDateFormatter::MEDIUM, \intlDateFormatter::SHORT );
 printf(
 'Locale: %s, Short: "%s", Medium "%s"'.PHP_EOL, 
 $myLocale, 
 $dateObject->format($printDate), 
 );
 }
40. i18n/L10n and OS ‣ Variety in L10n is almost infinite ‣ Automatic in i18n and L10n is better ‣ Operating system plays an important role ‣ Why reinvent a very very very complicated wheel if it already exists? 40
42. Timezones - L10n 42 Interactive question! What time is it now in Seoul (ko-KR)?
43. Ask my timebot! https://telegram.me/TheTimeBot 43 Disclaimer: Feel free to use it, but please do only provide perfect input https://github.com/unreal4u/tg-timebot
44. Timezones - Some data ‣ PHP has full support for timezones ‣ 39 (40?) official timezones ‣ Multiple timezones in one locale ‣ nl-NL: Europe/Amsterdam ‣ es-CL: America/Santiago and Easter/Pacific ‣ en-US: Has 4 timezones (Plus Alaska, Samoa, Hawaii and Chamorro) ‣ ru-RU: Has 8 timezones 44
45. Timezones: i18n Caution! Calls to ICU library can get pretty expensive! ‣ With a known locale, get all timezones ‣ If there's only one, instantiate \DateTimeZone ‣ More than 1? Get precise timezoneId and DST settings (cache them!) ‣ Now calculate the offset of a timezone for the view 45
46. Timezones: i18n - Check validity 46 public function isValidTimeZone($timeZoneName='') {
 if (!is_string($timeZoneName)) {
 $timeZoneName = '';
 }
 
 try {
 new \DateTimeZone($timeZoneName);
 return true;
 } catch (\Exception $e) {
 return false;
 }
 }
47. Timezones: i18n - Get timezone candidates 47 /** * $region is defined as \Locale::getRegion($currentLocale) */ private function _setTimezoneCandidates($region='') {
 if (!empty($region)) {
 $this->_timezoneCandidates = \DateTimeZone::listIdentifiers( \DateTimeZone::PER_COUNTRY, $region ); if (!empty($this->_timezoneCandidates) && count($this->_timezoneCandidates) == 1) {
 $this->setTimezone($this->_timezoneCandidates[0]);
 }
 }
 }
48. Timezones: i18n - Set timezone 48 public function setTimezone($timeZoneName='UTC') {
 if (!$this->isValidTimeZone($timeZoneName)) {
 $timeZoneName = 'UTC';
 }
 
 $this->timezone = new \DateTimeZone($timeZoneName);
 $this->timezoneId = $this->timezone->getName();
 $transitions = $this->timezone->getTransitions();
 $this->timezoneInDST = $transitions[0]['isdst'];
 
 return $this->timezoneId;
 }
49. Timezones: i18n - Display 49 $idf = \intlDateFormatter::create(
 'ko-KR', // $this->_currentLocale
 \IntlDateFormatter::MEDIUM,
 \IntlDateFormatter::MEDIUM,
 $this->timeZoneId // Asia/Seoul
 );
 $theDate = new \DateTime( '23-05-2015 21:34:09', new \DateTimeZone(‘UTC') ); echo $idf->format($theDate);
50. What's the time in Seoul then? Result? UTC 23-05-2015 21:34:09 is 2015. 5. 24. য়੹ 6:34:09 in ko-KR (Offset: +9 hours) 50
51. Encoding and charsets - L10n ‣ Difficult, often misunderstood subject ‣ Difficult to debug ‣ First step of debugging is knowing what encoding you are working with ‣ Convert to an appropriate charset with iconv() 51
52. Encoding in PHP ‣ Internal work always in UTF-8, EVERYWHERE ‣ Include some basic stuff so that PHP also knows that it has to work in UTF-8 ‣ Don't forget to send the browser information as well 52 mb_internal_encoding('UTF-8'); header('Content-type: %s; charset=UTF-8'); ‣ Lots of small things to consider, but can vary on each case
53. Encoding in PHP: mails Caution with the imap extension! Has some problems with UTF-7 Always encode "To" (BC, BCC) and "Subject" fields 53 Code Output "=?utf-8?B?5L2p5ae/?= <my@name.com.tw>" ֫঵ "=?iso-8859-1?Q?B=F8lla?=, med =?iso-8859-1?Q?=F8l?= i baggen <my@name.com>" Bølla , med øl i baggen "=?utf-7?Q?Petra_M+APw-ller?=" Petra Müller
54. Encoding in PHP: mails ‣ Buggy functions ‣ imap_rfc822_parse_adrlist() ‣ imap_mime_header_decode() ‣ Others? ‣ Check out https://github.com/unreal4u/string-operations/ for replacement functions 54
55. Databases and L10n / i18n
56. But before we begin…
58. Database and encodings/charsets CHARSET 58 COLLATION
59. Practical use of charset md5/SHA1-like strings should be ASCII-encoded (Why? It helps the db engine to predict better its memory assignment) 59 Interactive question! What charset should be used to save the following string? f5d39e997c5d7e4e2a3ef49973f61fb2
60. Practical use of charset CREATE TABLE `t1` (
 `md5HashCalculation` CHAR(32) CHARSET ASCII COLLATE ascii_bin
 ); 60
61. Differences between TEXT and [VAR]CHAR ‣ [VAR]CHAR(255) holds up to 255 characters ‣ TINYTEXT can hold up to 255 bytes ‣ UTF-8 characters can take up to 5 (or more) bytes 61
62. Indexes and charsets When working with Unicode characters, performance can be indirectly and negatively impacted ‣ Too big (and complex) of a topic for now ‣ Use EXPLAIN to understand underlying decisions of MySQL (in some cases) ‣ Don't bother in micro-optimization either 62
63. COLLATION ‣ Used to order data in a "natural" way ‣ Different languages have different rules 63 CREATE TABLE `spanishCollation` (
 `name01` VARCHAR(15) COLLATE utf8_spanish_ci,
 `name02` VARCHAR(15) COLLATE utf8_spanish2_ci
 ) DEFAULT CHARSET utf8;
64. Some notes on Collation "*_ci" stands for case-insensitive Watch out with utf8_general_ci and utf8_unicode_ci! ➡ utf8_general_ci has some problems with Hebrew and some cyrillic characters ➡ It's generally faster (7~12%) ➡ But utf8_unicode_ci is more compatible 64
65. Collation and performance ‣ Performance penalty: order in another collation ‣ It will have to do a filesort ‣ Which is MySQL's way of saying "quicksort" ‣ [Partial] keys can help avoid this quicksort operation 65
66. General database localization ‣ Not recommended: translation on database level ‣ If absolutely needed, investigate EAV model ‣ PRO: Quick, simple and cheap ‣ CON: Queries may become complex 66
67. Names and addresses ‣ UTF-8 does NOT cover all cases! ‣ Best way to save information is to save it in binary format: 67 CREATE TABLE `thaPeople` (
 `name` MEDIUMBLOB NULL DEFAULT NULL,
 `address` MEDIUMBLOB NULL DEFAULT NULL
 ); ‣ However this is a very extreme case
68. Your own L10n database ‣ Does the locale use the metric or imperial system (either British or American)? ‣ What type of rounding is used in that locale? ‣ Optional: custom number and currency pattern to overwrite any default rules ‣ The preferred timezone (user based, not L10n based) ‣ Direction of text 68
69. Fonts ‣ Easily overseen, yet very important ‣ Web-safe fonts are generally safe to use ‣ Don't forget to test multibyte characters ‣ 2 bytes: ñÖÑú - ӬģĽ ‣ 3 bytes: 佸ਁ - —၍₶ ‣ 4+bytes: 韴韵韶 - ‣ Example: Mamá vive en Föllinge en el bosque del Ñañdú.¿Enredado? ¡Deberías! (SimSun-ExtB) 69
70. JavaScript considerations ‣ Always use native Date() object ‣ Has support for timezones ‣ No native support for i18n on Javascript ‣ http://i18next.com is able to save the day! 70
71. 71 Finally: Who am I? Want to know more? My name is Camilo Sperberg Tweet me @unreal4u Email me@unreal4u.com
72. Finally: Who am I? ‣ Blog: http://blog.unreal4u.com/ (Spanish) ‣ Leave comments on https://legacy.joind.in/17727 ‣ Slides will be ready to be downloaded on: ‣ http://unreal4u.com/talks/ ‣ https://speakerdeck.com/unreal4u 72
73. Thanks! 73
74. Nice reads and more information ‣ https://github.com/triplepoint/php-units-of-measure ‣ https://github.com/unreal4u/localization ‣ http://www.w3.org/International/articles/language-tags/ ‣ http://php.net/manual/en/book.intl.php ‣ http://www.sitepoint.com/localizing-php-applications-1/ ‣ http://www.utf8-chartable.de/ ‣ http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/ 74