15875162966

Committed 25 Jun 2025 11:31AM UTC coverage: 95.929% (-1.3%) from 97.22%

Build # 15875162966

Build Type

push

github

Committed by

olvlvl

Commit Message

Add LanguageId and LocaleId

Coverage Stats

204 of 227 new or added lines in 18 files covered. (89.87%)

1296 of 1351 relevant lines covered (95.93%)

36.59 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

92.45

/src/Core/UnicodeLocaleExtensions.php

<?php

namespace ICanBoogie\CLDR\Core;

use ICanBoogie\CLDR\BCP47\BCP47Data;
use ICanBoogie\CLDR\Supplemental\Territory\TerritoryData;

final class UnicodeLocaleExtensions
{
    private const MAP_PREFIX_TO_PROPERTY = [
        'ca' => 'calendar',
        'cf' => 'currency_format',
        'co' => 'collation',
        'cu' => 'currency',
        'dx' => 'dictionary_break_script_exclusions',
        'em' => 'emoji',
        'fw' => 'first_day_of_week',
        'hc' => 'hour_cycle',
        'lb' => 'line_break_style',
        'lw' => 'line_break_word_handling',
        'ms' => 'measurement_system',
        'mu' => 'measurement_unit_override',
        'nu' => 'numbering_system',
        'rg' => 'region_override',
        'sd' => 'regional_subdivision',
        'ss' => 'sentence_break_suppressions',
        'tz' => 'time_zone',
        'va' => 'common_variant',
    ];

    private const PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS = 'dx';
    private const PREFIX_FOR_REGION_OVERRIDE = 'rg';
    private const PREFIX_FOR_REGIONAL_SUBDIVISION = 'sd';

    private const REGEXP_EXTENSION = '/([a-z]{2})-((?:(?!-[a-z]{2}-).)+)/';

    /**
     * @param string $unicode_ext
     *     A Unicode Locale Extension (-u-) format (defined in BCP 47).
     *
     * @link https://unicode.org/reports/tr35/tr35.html#u_Extension
     * @link https://unicode.org/reports/tr35/tr35.html#Key_And_Type_Definitions_
     */
    public static function parse(string $unicode_ext): self
    {
        if (str_starts_with($unicode_ext, 'u-')) {
            $unicode_ext = substr($unicode_ext, 2);
        }

        preg_match_all(self::REGEXP_EXTENSION, $unicode_ext, $matches, PREG_SET_ORDER);

        $properties = [];

        foreach ($matches as [, $prefix, $value]) {
            $property = self::MAP_PREFIX_TO_PROPERTY[$prefix]
                ?? throw new InvalidUnicodeLocaleExtensions(
                    "Invalid extension prefix '$prefix' in '$unicode_ext'",
                );

            $value = match ($prefix) {
                self::PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS => explode('-', $value),
                self::PREFIX_FOR_REGION_OVERRIDE => self::parse_region_override($value),
                self::PREFIX_FOR_REGIONAL_SUBDIVISION => self::parse_regional_subdivision($value),
                default => $value,
            };

            $properties[$property] = $value;
        }

        return new self(...$properties);
    }

    private static function parse_region_override(string $value): string
    {
        $region = strtoupper(rtrim($value, 'z'));

        return $region;
    }

    private static function parse_regional_subdivision(string $value): string
    {
        throw new \RuntimeException("prefix not implemented yet: sd");
    }

    /**
     * @param string[] $dictionary_break_script_exclusions
     */
    public function __construct(
        public readonly ?string $calendar = null,
        public readonly ?string $currency_format = null,
        public readonly ?string $collation = null,
        public readonly ?string $currency = null,
        public readonly array $dictionary_break_script_exclusions = [],
        public readonly ?string $emoji = null,
        public readonly ?string $first_day_of_week = null,
        public readonly ?string $hour_cycle = null,
        public readonly ?string $line_break_style = null,
        public readonly ?string $line_break_word_handling = null,
        public readonly ?string $measurement_system = null,
        public readonly ?string $measurement_unit_override = null,
        public readonly ?string $numbering_system = null,
        public readonly ?string $region_override = null,
        public readonly ?string $regional_subdivision = null,
        public readonly ?string $sentence_break_suppressions = null,
        public readonly ?string $time_zone = null,
        public readonly ?string $common_variant = null,
    ) {
        $this->validate();
    }

    public function validate(): void
    {
        $err = [];

        foreach ($this->to_array() as $prefix => $value) {
            if ($value === null) {
                continue;
            }

            if ($prefix === self::PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS) {
                // TODO: validate scripts
                continue;
            }

            if ($prefix === self::PREFIX_FOR_REGION_OVERRIDE) {
                if (!in_array($value, TerritoryData::CODES)) {
                    $err[] = "$prefix: $value";
                }

                continue;
            }

            if (!in_array($value, BCP47Data::U_MAPPING[$prefix])) {
                $err[] = "$prefix: $value";
            }
        }

        if ($err) {
            throw new InvalidUnicodeLocaleExtensions("Invalid unicode extension: " . implode("; ", $err));
        }
    }

    /**
     * @return array<string, ?string>
     */
    public function to_array(): array
    {
        return [

            'ca' => $this->calendar,
            'cf' => $this->currency_format,
            'co' => $this->collation,
            'cu' => $this->currency,
            'dx' => $this->dictionary_break_script_exclusions,
            'em' => $this->emoji,
            'fw' => $this->first_day_of_week,
            'hc' => $this->hour_cycle,
            'lb' => $this->line_break_style,
            'lw' => $this->line_break_word_handling,
            'ms' => $this->measurement_system,
            'mu' => $this->measurement_unit_override,
            'nu' => $this->numbering_system,
            'rg' => $this->region_override,
            'sd' => $this->regional_subdivision,
            'ss' => $this->sentence_break_suppressions,
            'tz' => $this->time_zone,
            'va' => $this->common_variant,

        ];
    }
}

1	<?php
2
3	namespace ICanBoogie\CLDR\Core;
4
5	use ICanBoogie\CLDR\BCP47\BCP47Data;
6	use ICanBoogie\CLDR\Supplemental\Territory\TerritoryData;
7
8	final class UnicodeLocaleExtensions
9	{
10	private const MAP_PREFIX_TO_PROPERTY = [
11	'ca' => 'calendar',
12	'cf' => 'currency_format',
13	'co' => 'collation',
14	'cu' => 'currency',
15	'dx' => 'dictionary_break_script_exclusions',
16	'em' => 'emoji',
17	'fw' => 'first_day_of_week',
18	'hc' => 'hour_cycle',
19	'lb' => 'line_break_style',
20	'lw' => 'line_break_word_handling',
21	'ms' => 'measurement_system',
22	'mu' => 'measurement_unit_override',
23	'nu' => 'numbering_system',
24	'rg' => 'region_override',
25	'sd' => 'regional_subdivision',
26	'ss' => 'sentence_break_suppressions',
27	'tz' => 'time_zone',
28	'va' => 'common_variant',
29	];
30
31	private const PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS = 'dx';
32	private const PREFIX_FOR_REGION_OVERRIDE = 'rg';
33	private const PREFIX_FOR_REGIONAL_SUBDIVISION = 'sd';
34
35	private const REGEXP_EXTENSION = '/([a-z]{2})-((?:(?!-[a-z]{2}-).)+)/';
36
37	/**
38	* @param string $unicode_ext
39	* A Unicode Locale Extension (-u-) format (defined in BCP 47).
40	*
41	* @link https://unicode.org/reports/tr35/tr35.html#u_Extension
42	* @link https://unicode.org/reports/tr35/tr35.html#Key_And_Type_Definitions_
43	*/
44	public static function parse(string $unicode_ext): self
45	{
46	if (str_starts_with($unicode_ext, 'u-')) {	9✔
47	$unicode_ext = substr($unicode_ext, 2);	1✔
48	}
49
50	preg_match_all(self::REGEXP_EXTENSION, $unicode_ext, $matches, PREG_SET_ORDER);	9✔
51
52	$properties = [];	9✔
53
54	foreach ($matches as [, $prefix, $value]) {	9✔
55	$property = self::MAP_PREFIX_TO_PROPERTY[$prefix]	9✔
56	?? throw new InvalidUnicodeLocaleExtensions(	9✔
57	"Invalid extension prefix '$prefix' in '$unicode_ext'",	9✔
58	);	9✔
59
60	$value = match ($prefix) {	9✔
61	self::PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS => explode('-', $value),
62	self::PREFIX_FOR_REGION_OVERRIDE => self::parse_region_override($value),	2✔
63	self::PREFIX_FOR_REGIONAL_SUBDIVISION => self::parse_regional_subdivision($value),
64	default => $value,	7✔
65	};	9✔
66
67	$properties[$property] = $value;	9✔
68	}
69
70	return new self(...$properties);	9✔
71	}
72
73	private static function parse_region_override(string $value): string
74	{
75	$region = strtoupper(rtrim($value, 'z'));	2✔
76
77	return $region;	2✔
78	}
79
80	private static function parse_regional_subdivision(string $value): string
81	{
NEW 82	throw new \RuntimeException("prefix not implemented yet: sd");	×
83	}
84
85	/**
86	* @param string[] $dictionary_break_script_exclusions
87	*/
88	public function __construct(
89	public readonly ?string $calendar = null,
90	public readonly ?string $currency_format = null,
91	public readonly ?string $collation = null,
92	public readonly ?string $currency = null,
93	public readonly array $dictionary_break_script_exclusions = [],
94	public readonly ?string $emoji = null,
95	public readonly ?string $first_day_of_week = null,
96	public readonly ?string $hour_cycle = null,
97	public readonly ?string $line_break_style = null,
98	public readonly ?string $line_break_word_handling = null,
99	public readonly ?string $measurement_system = null,
100	public readonly ?string $measurement_unit_override = null,
101	public readonly ?string $numbering_system = null,
102	public readonly ?string $region_override = null,
103	public readonly ?string $regional_subdivision = null,
104	public readonly ?string $sentence_break_suppressions = null,
105	public readonly ?string $time_zone = null,
106	public readonly ?string $common_variant = null,
107	) {
108	$this->validate();	28✔
109	}
110
111	public function validate(): void
112	{
113	$err = [];	28✔
114
115	foreach ($this->to_array() as $prefix => $value) {	28✔
116	if ($value === null) {	28✔
117	continue;	28✔
118	}
119
120	if ($prefix === self::PREFIX_FOR_DICTIONARY_BREAK_SCRIPT_EXCLUSIONS) {	28✔
121	// TODO: validate scripts
122	continue;	28✔
123	}
124
125	if ($prefix === self::PREFIX_FOR_REGION_OVERRIDE) {	9✔
126	if (!in_array($value, TerritoryData::CODES)) {	2✔
NEW 127	$err[] = "$prefix: $value";	×
128	}
129
130	continue;	2✔
131	}
132
133	if (!in_array($value, BCP47Data::U_MAPPING[$prefix])) {	7✔
NEW 134	$err[] = "$prefix: $value";	×
135	}
136	}
137
138	if ($err) {	28✔
NEW 139	throw new InvalidUnicodeLocaleExtensions("Invalid unicode extension: " . implode("; ", $err));	×
140	}
141	}
142
143	/**
144	* @return array<string, ?string>
145	*/
146	public function to_array(): array
147	{
148	return [	28✔
149
150	'ca' => $this->calendar,	28✔
151	'cf' => $this->currency_format,	28✔
152	'co' => $this->collation,	28✔
153	'cu' => $this->currency,	28✔
154	'dx' => $this->dictionary_break_script_exclusions,	28✔
155	'em' => $this->emoji,	28✔
156	'fw' => $this->first_day_of_week,	28✔
157	'hc' => $this->hour_cycle,	28✔
158	'lb' => $this->line_break_style,	28✔
159	'lw' => $this->line_break_word_handling,	28✔
160	'ms' => $this->measurement_system,	28✔
161	'mu' => $this->measurement_unit_override,	28✔
162	'nu' => $this->numbering_system,	28✔
163	'rg' => $this->region_override,	28✔
164	'sd' => $this->regional_subdivision,	28✔
165	'ss' => $this->sentence_break_suppressions,	28✔
166	'tz' => $this->time_zone,	28✔
167	'va' => $this->common_variant,	28✔
168
169	];	28✔
170	}
171	}

ICanBoogie / CLDR / 15875162966

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous