Coverage for app/backend/src/couchers/i18n/localize.py: 92%
72 statements
« prev ^ index » next coverage.py v7.14.3, created at 2026-06-28 21:22 +0000
1"""
2Defines low-level localization functions for strings, dates, etc.
3Most code should use the higher-level couchers.i18n.LocalizationContext object.
4"""
6import re
7from collections.abc import Mapping, Sequence
8from datetime import date, datetime, time, tzinfo
9from typing import cast
11import babel
12import phonenumbers
13from babel.dates import get_datetime_format, get_timezone_name, match_skeleton, parse_pattern
14from babel.lists import format_list
16from couchers.i18n.locales import DEFAULT_LOCALE, get_main_i18next
17from couchers.resources import get_region_code_iso3166_alpha3_to_alpha2
20def localize_string(lang: str | None, key: str, *, substitutions: Mapping[str, str | int] | None = None) -> str:
21 """
22 Retrieves a translated string and performs substitutions.
24 Args:
25 lang: Language code (e.g., "en", "pt-BR"). If None, defaults to the default fallback language ("en")
26 key: The key for the string to be looked up.
27 substitutions: Dictionary of variable substitutions for the string (e.g., {"hours": 24})
29 Returns:
30 The translated string with substitutions applied
31 """
32 return get_main_i18next().localize(key, lang or DEFAULT_LOCALE, substitutions)
35def localize_list(items: Sequence[str], locale: babel.Locale) -> str:
36 return format_list(items, locale=locale)
39def try_localize_language_name_from_iso639(code: str, locale: babel.Locale, standalone: bool = False) -> str | None:
40 """
41 Attempts to localize the name of a language expressed as an ISO639 code.
43 Args:
44 code: The ISO639 language code.
45 locale: The locale to render the language name in.
46 standalone: The result won't be part of a larger sentence and should be capitalized if the language has capitals.
48 Returns:
49 The localized name, or None if no localized name is available.
50 """
51 try:
52 name = babel.Locale.parse(code).get_language_name(locale)
53 if name is None: 53 ↛ 54line 53 didn't jump to line 54 because the condition on line 53 was never true
54 return None
55 if standalone:
56 # The Unicode CLDR returns a casing that allows embedding in a larger sentence, e.g. "español".
57 # If we're displaying the language name on its own, capitalize its first letter if applicable.
58 # An LLM prompt revealed that this holds for all major languages.
59 # It is a no-op for scripts that don't have capital letters.
60 name = name[:1].title() + name[1:]
61 return name
62 except (ValueError, babel.UnknownLocaleError):
63 return None
66def try_localize_region_name_from_iso3166(code: str, locale: babel.Locale) -> str | None:
67 """
68 Gets a region name specified as an ISO3166 alpha2 or alpha3 code, localized in the given locale.
69 """
70 # The Unicode CLDR uses alpha2 codes as keys (all alpha3 codes have a corresponding alpha2 code)
71 code = get_region_code_iso3166_alpha3_to_alpha2().get(code, code)
72 region_name: str | None = locale.territories.get(code, None)
73 return region_name
76def localize_date(
77 value: date, locale: babel.Locale, *, abbrev: bool = False, with_year: bool = True, with_day_of_week: bool = False
78) -> str:
79 """Formats a time- and timezone-agnostic date for the given locale."""
80 pattern = _get_cldr_date_pattern(locale, abbrev=abbrev, with_year=with_year, with_day_of_week=with_day_of_week)
81 return parse_pattern(pattern).apply(value, locale)
84def localize_time(value: time, locale: babel.Locale, *, with_seconds: bool = False) -> str:
85 """Formats a date- and timezone-agnostic time for the given locale."""
86 pattern = _get_cldr_time_pattern(locale, with_seconds=with_seconds)
87 return parse_pattern(pattern).apply(value, locale)
90def localize_datetime(
91 value: datetime,
92 locale: babel.Locale,
93 *,
94 abbrev: bool = False,
95 with_year: bool = True,
96 with_day_of_week: bool = False,
97 with_seconds: bool = False,
98) -> str:
99 """Formats a date and time for the given locale."""
100 # A timezone-unaware datetime is almost certainly a bug, so we don't support it.
101 assert value.tzinfo is not None, "Cannot localize a timezone-unaware datetime."
103 pattern = _combine_cldr_date_time_patterns(
104 locale,
105 _get_cldr_date_pattern(locale, abbrev=abbrev, with_year=with_year, with_day_of_week=with_day_of_week),
106 _get_cldr_time_pattern(locale, with_seconds=with_seconds),
107 )
108 return parse_pattern(pattern).apply(value, locale)
111def _get_cldr_date_pattern(
112 locale: babel.Locale, *, abbrev: bool = False, with_year: bool = True, with_day_of_week: bool = False
113) -> str:
114 # First build a Unicode CLDR datetime pattern skeleton, which is locale and order-agnostic,
115 # and only indicates the components we're interested in formatting.
116 # This is similar to Intl.DateTimeFormat in Javascript.
117 # See https://cldr.unicode.org/translation/date-time/date-time-symbols.
118 requested_skeleton = ""
120 if with_year:
121 requested_skeleton += "y"
122 requested_skeleton += "MMM" if abbrev else "MMMM"
123 requested_skeleton += "d"
124 if with_day_of_week:
125 requested_skeleton += "EEE" if abbrev else "EEEE"
127 # Next, match that skeleton to a similar locale-supported skeleton,
128 # which allows us to lower it to a datetime pattern (locale and order-specific).
129 matched_skeleton = match_skeleton(requested_skeleton, options=locale.datetime_skeletons)
130 if not matched_skeleton: 130 ↛ 131line 130 didn't jump to line 131 because the condition on line 130 was never true
131 raise ValueError(f"Locale {locale.english_name} has no matching datetime skeleton for '{requested_skeleton}'")
133 pattern: str = locale.datetime_skeletons[matched_skeleton].pattern
135 # By CLDR rules, skeleton matching might return a pattern with abbreviations where
136 # we asked for non-abbreviated forms, in which case we can update the returned pattern.
137 if not abbrev:
138 # Abbreviated to non-abbreviated month (MMM = abbreviated)
139 pattern = re.sub(r"(?<!M)MMM(?!M)", "MMMM", pattern)
140 if with_day_of_week:
141 # Abbreviated to non-abbreviated day of week (E = EEE = abbreviated)
142 pattern = re.sub(r"(?<!E)E{1,3}(?!E)", "EEEE", pattern)
144 return pattern
147def _get_cldr_time_pattern(locale: babel.Locale, *, with_seconds: bool = False) -> str:
148 # Use a reference format pattern to figure out if it's using 24h clock
149 reference_time_pattern: str = locale.time_formats["medium"].pattern
151 # Remove literals like 'of'
152 reference_time_pattern = re.sub("'[^']*'", "", reference_time_pattern)
154 # Extract only the hours, minutes and am/pm patterns.
155 requested_skeleton = re.sub("[^hHkKma]+", "", reference_time_pattern)
156 if with_seconds:
157 requested_skeleton += "ss"
159 # Next, match that skeleton to a similar locale-supported skeleton,
160 # which allows us to lower it to a datetime pattern (locale and order-specific).
161 matched_skeleton = match_skeleton(requested_skeleton, options=locale.datetime_skeletons)
162 if not matched_skeleton: 162 ↛ 163line 162 didn't jump to line 163 because the condition on line 162 was never true
163 raise ValueError(f"Locale {locale.english_name} has no matching datetime skeleton for '{requested_skeleton}'")
165 return cast(str, locale.datetime_skeletons[matched_skeleton].pattern) # "pattern" is Any-typed
168def _combine_cldr_date_time_patterns(locale: babel.Locale, date_pattern: str, time_pattern: str) -> str:
169 # get_datetime_format's return value is statically mistyped
170 combining_format = cast(str, get_datetime_format(locale=locale))
172 # CLDR defines {0} to be the time and {1} to be the date
173 return combining_format.replace("{1}", date_pattern).replace("{0}", time_pattern)
176def localize_timezone(timezone: tzinfo, locale: babel.Locale, *, short: bool = False) -> str:
177 return get_timezone_name(timezone, width="short" if short else "long", locale=locale)
180def format_phone_number(value: str) -> str:
181 """Formats a phone number from E.164 format to the international format."""
182 return phonenumbers.format_number(phonenumbers.parse(value), phonenumbers.PhoneNumberFormat.INTERNATIONAL)