Fix dates formatting Y, w and W symbols for week-numbering
#1179
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #1179 +/- ##
==========================================
+ Coverage 91.37% 91.50% +0.13%
==========================================
Files 27 27
Lines 4672 4687 +15
==========================================
+ Hits 4269 4289 +20
+ Misses 403 398 -5
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
babel/dates.py
Outdated
| @@ -1677,6 +1675,33 @@ def get_day_of_year(self, date: datetime.date | None = None) -> int: | |||
| date = self.value | |||
| return (date - date.replace(month=1, day=1)).days + 1 | |||
|
|
|||
| def get_week_of_year(self, value: datetime.datetime | datetime.date | None = None) -> int: | |||
There was a problem hiding this comment.
Should these methods be public or do we want to keep them private? Either way, it might be nice to add a docstring and some examples perhaps?
There was a problem hiding this comment.
I think these methods should be kept private.
If there is a use case that determines week of year/month from a date/datetime instance, we should add them as helper methods to babel/dates.py, not instance methods of DateTimeFormat.
import datetime
import icu
def icu_format(value, format, locale):
if isinstance(locale, str):
locale = icu.Locale(locale)
tz = icu.TimeZone.getGMT()
cal = icu.Calendar.createInstance(tz, locale)
f = icu.SimpleDateFormat(format, locale)
if isinstance(value, datetime.date):
cal.set(value.year, value.month - 1, value.day)
elif isinstance(value, datetime.datetime):
cal.set(value.year, value.month - 1, value.day, value.hour, value.minute, value.second)
else:
raise ValueError(repr(value))
return f.format(cal.getTime())>>> icu_format(datetime.date(2023, 12, 31), "YYYY-'W'ww-e", 'de_DE')
'2023-W52-7'
>>> icu_format(datetime.date(2023, 12, 31), "YYYY-'W'ww-e", 'en_US')
'2024-W01-1'
>>> icu_format(datetime.date(2023, 12, 31), "YYYY-'W'ww-e", 'en_AU')
'2023-W53-7'
I just moved to the Trying to add tests for the week of month, however results of ICU and Babel differ. I think one or both are incorrect. I'm unsure why. >>> icu_format(datetime.date(2003, 1, 31), 'W', 'de_DE')
'5'
>>> icu_format(datetime.date(2003, 2, 1), 'W', 'de_DE')
'0' # <= I think it should be '1'
>>> format_date(parse_date('2003-01-31'), format='W', locale='de_DE')
'5'
>>> format_date(parse_date('2003-02-01'), format='W', locale='de_DE')
'5' # <= I think it should be '1' |
>>> format_date(parse_date('2003-02-01'), format='W', locale='de_DE')
'5' # <= I think it should be '1'Maybe |
7ccfdf7 to
d5f31af
Compare
The lack of the coverage is for
|
…_week_of_month` and add simple docstring
Reconsidering with week-of-month calculation in Week Data section in TR#35, it seems that ICU is correct. I fixed week-of-month calculation in Babel in the additional commit of this pullreq. |
Y and w symbols for week-numberingY, w and W symbols for week-numbering
Fix #1133.
I confirmed the added test cases for week-numbering are correct using PyICU library.