Babel Changelog

===============

Version 2.5.1

-------------

Minor Improvements and bugfixes

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

* Use a fixed datetime to avoid test failures (#520) (@narendravardi)

* Parse multi-line __future__ imports better (#519) (@akx)

* Fix validate_currency docstring (#522)

* Allow normalize_locale and exists to handle various unexpected inputs (#523) (@suhojm)

* Make PyInstaller support more robust (#525, #526) (@thijstriemstra, @akx)

Version 2.5.0

-------------

New Features

~~~~~~~~~~~~

* Numbers: Add currency utilities and helpers (#491) (@kdeldycke)

* Support PyInstaller (#500, #505) (@wodo)

Minor Improvements and bugfixes

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

* Dates: Add __str__ to DateTimePattern (#515) (@sfermigier)

* Dates: Fix an invalid string to bytes comparison when parsing TZ files on Py3 (#498) (@rowillia)

* Dates: Formatting zero-padded components of dates is faster (#517) (@akx)

* Documentation: Fix "Good Commits" link in CONTRIBUTING.md (#511) (@naryanacharya6)

* Documentation: Fix link to Python gettext module (#512) (@Linkid)

* Messages: Allow both dash and underscore separated locale identifiers in pofiles (#489, #490) (@akx)

* Messages: Extract Python messages in nested gettext calls (#488) (@sublee)

* Messages: Fix in-place editing of dir list while iterating (#476, #492) (@MarcDufresne)

* Messages: Stabilize sort order (#482) (@xavfernandez)

* Time zones: Honor the no-inherit marker for metazone names (#405) (@akx)

Version 2.4.0

-------------

New Features

~~~~~~~~~~~~

Some of these changes might break your current code and/or tests.

* CLDR: CLDR 29 is now used instead of CLDR 28 (#405) (@akx)

* Messages: Add option 'add_location' for location line formatting (#438, #459) (@rrader, @alxpy)

* Numbers: Allow full control of decimal behavior (#410) (@etanol)

Minor Improvements and bugfixes

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

* Documentation: Improve Date Fields descriptions (#450) (@ldwoolley)

* Documentation: Typo fixes and documentation improvements (#406, #412, #403, #440, #449, #463) (@zyegfryed, @adamchainz, @jwilk, @akx, @roramirez, @abhishekcs10)

* Messages: Default to UTF-8 source encoding instead of ISO-8859-1 (#399) (@asottile)

* Messages: Ensure messages are extracted in the order they were passed in (#424) (@ngrilly)

* Messages: Message extraction for JSX files is improved (#392, #396, #425) (@karloskar, @georgschoelly)

* Messages: PO file reading supports multi-line obsolete units (#429) (@mbirtwell)

* Messages: Python message extractor respects unicode_literals in __future__ (#427) (@sublee)

* Messages: Roundtrip Language headers (#420) (@kruton)

* Messages: units before obsolete units are no longer erroneously marked obsolete (#452) (@mbirtwell)

* Numbers: `parse_pattern` now preserves the full original pattern (#414) (@jtwang)

* Numbers: Fix float conversion in `extract_operands` (#435) (@akx)

* Plurals: Fix plural forms for Czech and Slovak locales (#373) (@ykshatroff)

* Plurals: More plural form fixes based on Mozilla and CLDR references (#431) (@mshenfield)

Internal improvements

~~~~~~~~~~~~~~~~~~~~~

* Local times are constructed correctly in tests (#411) (@etanol)

* Miscellaneous small improvements (#437) (@scop)

* Regex flags are extracted from the regex strings (#462) (@singingwolfboy)

* The PO file reader is now a class and has seen some refactoring (#429, #452) (@mbirtwell)

Version 2.3.4

-------------

(Bugfix release, released on April 22th)

Bugfixes

~~~~~~~~

* CLDR: The lxml library is no longer used for CLDR importing, so it should not cause strange failures either. Thanks to @aronbierbaum for the bug report and @jtwang for the fix. (https://github.com/python-babel/babel/pull/393)

* CLI: Every last single CLI usage regression should now be gone, and both distutils and stand-alone CLIs should work as they have in the past. Thanks to @paxswill and @ajaeger for bug reports. (https://github.com/python-babel/babel/pull/389)

Version 2.3.3

-------------

(Bugfix release, released on April 12th)

Bugfixes

~~~~~~~~

* CLI: Usage regressions that had snuck in between 2.2 and 2.3 should be no more. (https://github.com/python-babel/babel/pull/386) Thanks to @ajaeger, @sebdiem and @jcristovao for bug reports and patches.

Version 2.3.2

-------------

(Bugfix release, released on April 9th)

Bugfixes

~~~~~~~~

* Dates: Period (am/pm) formatting was broken in certain locales (namely zh_TW). Thanks to @jun66j5 for the bug report. (https://github.com/python-babel/babel/issues/378, https://github.com/python-babel/babel/issues/379)

Version 2.3.1

-------------

(Bugfix release because of deployment problems, released on April 8th)

Version 2.3

-----------

(Feature release, released on April 8th)

Internal improvements

~~~~~~~~~~~~~~~~~~~~~

* The CLI frontend and Distutils commands use a shared implementation (https://github.com/python-babel/babel/pull/311)

* PyPy3 is supported (https://github.com/python-babel/babel/pull/343)

Features

~~~~~~~~

* CLDR: Add an API for territory language data (https://github.com/python-babel/babel/pull/315)

* Core: Character order and measurement system data is imported and exposed (https://github.com/python-babel/babel/pull/368)

* Dates: Add an API for time interval formatting (https://github.com/python-babel/babel/pull/316)

* Dates: More pattern formats and lengths are supported (https://github.com/python-babel/babel/pull/347)

* Dates: Period IDs are imported and exposed (https://github.com/python-babel/babel/pull/349)

* Dates: Support for date-time skeleton formats has been added (https://github.com/python-babel/babel/pull/265)

* Dates: Timezone formatting has been improved (https://github.com/python-babel/babel/pull/338)

* Messages: JavaScript extraction now supports dotted names, ES6 template strings and JSX tags (https://github.com/python-babel/babel/pull/332)

* Messages: npgettext is recognized by default (https://github.com/python-babel/babel/pull/341)

* Messages: The CLI learned to accept multiple domains (https://github.com/python-babel/babel/pull/335)

* Messages: The extraction commands now accept filenames in addition to directories (https://github.com/python-babel/babel/pull/324)

* Units: A new API for unit formatting is implemented (https://github.com/python-babel/babel/pull/369)

Bugfixes

~~~~~~~~

* Core: Mixed-case locale IDs work more reliably (https://github.com/python-babel/babel/pull/361)

* Dates: S...S formats work correctly now (https://github.com/python-babel/babel/pull/360)

* Messages: All messages are now sorted correctly if sorting has been specified (https://github.com/python-babel/babel/pull/300)

* Messages: Fix the unexpected behavior caused by catalog header updating (e0e7ef1) (https://github.com/python-babel/babel/pull/320)

* Messages: Gettext operands are now generated correctly (https://github.com/python-babel/babel/pull/295)

* Messages: Message extraction has been taught to detect encodings better (https://github.com/python-babel/babel/pull/274)

Version 2.2

-----------

(Feature release, released on January 2nd 2016)

Bugfixes

~~~~~~~~

* General: Add __hash__ to Locale. (#303) (2aa8074)

* General: Allow files with BOM if they're UTF-8 (#189) (da87edd)

* General: localedata directory is now locale-data (#109) (2d1882e)

* General: odict: Fix pop method (0a9e97e)

* General: Removed uses of datetime.date class from *.dat files (#174) (94f6830)

* Messages: Fix plural selection for Chinese (531f666)

* Messages: Fix typo and add semicolon in plural_forms (5784501)

* Messages: Flatten NullTranslations.files into a list (ad11101)

* Times: FixedOffsetTimezone: fix display of negative offsets (d816803)

Features

~~~~~~~~

* CLDR: Update to CLDR 28 (#292) (9f7f4d0)

* General: Add __copy__ and __deepcopy__ to LazyProxy. (a1cc3f1)

* General: Add official support for Python 3.4 and 3.5

* General: Improve odict performance by making key search O(1) (6822b7f)

* Locale: Add an ordinal_form property to Locale (#270) (b3f3430)

* Locale: Add support for list formatting (37ce4fa, be6e23d)

* Locale: Check inheritance exceptions first (3ef0d6d)

* Messages: Allow file locations without line numbers (#279) (79bc781)

* Messages: Allow passing a callable to `extract()` (#289) (3f58516)

* Messages: Support 'Language' header field of PO files (#76) (3ce842b)

* Messages: Update catalog headers from templates (e0e7ef1)

* Numbers: Properly load and expose currency format types (#201) (df676ab)

* Numbers: Use cdecimal by default when available (b6169be)

* Numbers: Use the CLDR's suggested number of decimals for format_currency (#139) (201ed50)

* Times: Add format_timedelta(format='narrow') support (edc5eb5)

Version 2.1

-----------

(Bugfix/minor feature release, released on September 25th 2015)

- Parse and honour the locale inheritance exceptions

  (https://github.com/python-babel/babel/issues/97)

- Fix Locale.parse using ``global.dat`` incompatible types 

  (https://github.com/python-babel/babel/issues/174)

- Fix display of negative offsets in ``FixedOffsetTimezone``

  (https://github.com/python-babel/babel/issues/214)

- Improved odict performance which is used during localization file

  build, should improve compilation time for large projects

- Add support for "narrow" format for ``format_timedelta``

- Add universal wheel support

- Support 'Language' header field in .PO files 

  (fixes https://github.com/python-babel/babel/issues/76)

- Test suite enhancements (coverage, broken tests fixed, etc)

- Documentation updated

Version 2.0

-----------

(Released on July 27th 2015, codename Second Coming)

- Added support for looking up currencies that belong to a territory

  through the :func:`babel.numbers.get_territory_currencies`

  function.

- Improved Python 3 support.

- Fixed some broken tests for timezone behavior.

- Improved various smaller things for dealing with dates.

Version 1.4

-----------

(bugfix release, release date to be decided)

- Fixed a bug that caused deprecated territory codes not being

  converted properly by the subtag resolving.  This for instance

  showed up when trying to use ``und_UK`` as a language code

  which now properly resolves to ``en_GB``.

- Fixed a bug that made it impossible to import the CLDR data

  from scratch on windows systems.

Version 1.3

-----------

(bugfix release, released on July 29th 2013)

- Fixed a bug in likely-subtag resolving for some common locales.

  This primarily makes ``zh_CN`` work again which was broken

  due to how it was defined in the likely subtags combined with

  our broken resolving.  This fixes :gh:`37`.

- Fixed a bug that caused pybabel to break when writing to stdout

  on Python 3.

- Removed a stray print that was causing issues when writing to

  stdout for message catalogs.

Version 1.2

-----------

(bugfix release, released on July 27th 2013)

- Included all tests in the tarball.  Previously the include

  skipped past recursive folders.

- Changed how tests are invoked and added separate standalone

  test command.  This simplifies testing of the package for

  linux distributors.

Version 1.1

-----------

(bugfix release, released on July 27th 2013)

- added dummy version requirements for pytz so that it installs

  on pip 1.4.

- Included tests in the tarball.

Version 1.0

-----------

(Released on July 26th 2013, codename Revival)

- support python 2.6, 2.7, 3.3+ and pypy - drop all other versions

- use tox for testing on different pythons

- Added support for the locale plural rules defined by the CLDR.

- Added `format_timedelta` function to support localized formatting of

  relative times with strings such as "2 days" or "1 month" (:trac:`126`).

- Fixed negative offset handling of Catalog._set_mime_headers (:trac:`165`).

- Fixed the case where messages containing square brackets would break with

  an unpack error.

- updated to CLDR 23

- Make the CLDR import script work with Python 2.7.

- Fix various typos.

- Sort output of list-locales.

- Make the POT-Creation-Date of the catalog being updated equal to

  POT-Creation-Date of the template used to update (:trac:`148`).

- Use a more explicit error message if no option or argument (command) is

  passed to pybabel (:trac:`81`).

- Keep the PO-Revision-Date if it is not the default value (:trac:`148`).

- Make --no-wrap work by reworking --width's default and mimic xgettext's

  behaviour of always wrapping comments (:trac:`145`).

- Add --project and --version options for commandline (:trac:`173`).

- Add a __ne__() method to the Local class.

- Explicitly sort instead of using sorted() and don't assume ordering

  (Jython compatibility).

- Removed ValueError raising for string formatting message checkers if the

  string does not contain any string formattings (:trac:`150`).

- Fix Serbian plural forms (:trac:`213`).

- Small speed improvement in format_date() (:trac:`216`).

- Fix so frontend.CommandLineInterface.run does not accumulate logging 

  handlers (:trac:`227`, reported with initial patch by dfraser)

- Fix exception if environment contains an invalid locale setting

  (:trac:`200`)

- use cPickle instead of pickle for better performance (:trac:`225`)

- Only use bankers round algorithm as a tie breaker if there are two nearest

  numbers, round as usual if there is only one nearest number (:trac:`267`,

  patch by Martin)

- Allow disabling cache behaviour in LazyProxy (:trac:`208`, initial patch

  from Pedro Algarvio)

- Support for context-aware methods during message extraction (:trac:`229`,

  patch from David Rios)

- "init" and "update" commands support "--no-wrap" option (:trac:`289`)

- fix formatting of fraction in format_decimal() if the input value is a float

  with more than 7 significant digits (:trac:`183`)

- fix format_date() with datetime parameter (:trac:`282`, patch from Xavier

  Morel)

- fix format_decimal() with small Decimal values (:trac:`214`, patch from

  George Lund)

- fix handling of messages containing '\\n' (:trac:`198`)

- handle irregular multi-line msgstr (no "" as first line) gracefully

  (:trac:`171`)

- parse_decimal() now returns Decimals not floats, API change (:trac:`178`)

- no warnings when running setup.py without installed setuptools (:trac:`262`)

- modified Locale.__eq__ method so Locales are only equal if all of their

  attributes (language, territory, script, variant) are equal

- resort to hard-coded message extractors/checkers if pkg_resources is

  installed but no egg-info was found (:trac:`230`)

- format_time() and format_datetime() now accept also floats (:trac:`242`)

- add babel.support.NullTranslations class similar to gettext.NullTranslations

  but with all of Babel's new gettext methods (:trac:`277`)

- "init" and "update" commands support "--width" option (:trac:`284`)

- fix 'input_dirs' option for setuptools integration (:trac:`232`, initial

  patch by Étienne Bersac)

- ensure .mo file header contains the same information as the source .po file

  (:trac:`199`)

- added support for get_language_name() on the locale objects.

- added support for get_territory_name() on the locale objects.

- added support for get_script_name() on the locale objects.

- added pluralization support for currency names and added a '¤¤¤'

  pattern for currencies that includes the full name.

- depend on pytz now and wrap it nicer.  This gives us improved support

  for things like timezone transitions and an overall nicer API.

- Added support for explicit charset to PO file reading.

- Added experimental Python 3 support.

- Added better support for returning timezone names.

- Don't throw away a Catalog's obsolete messages when updating it.

- Added basic likelySubtag resolving when doing locale parsing and no

  match can be found.

Version 0.9.6

-------------

(released on March 17th 2011)

- Backport r493-494: documentation typo fixes.

- Make the CLDR import script work with Python 2.7.

- Fix various typos.

- Fixed Python 2.3 compatibility (:trac:`146`, :trac:`233`).

- Sort output of list-locales.

- Make the POT-Creation-Date of the catalog being updated equal to

  POT-Creation-Date of the template used to update (:trac:`148`).

- Use a more explicit error message if no option or argument (command) is

  passed to pybabel (:trac:`81`).

- Keep the PO-Revision-Date if it is not the default value (:trac:`148`).

- Make --no-wrap work by reworking --width's default and mimic xgettext's

  behaviour of always wrapping comments (:trac:`145`).

- Fixed negative offset handling of Catalog._set_mime_headers (:trac:`165`).

- Add --project and --version options for commandline (:trac:`173`).

- Add a __ne__() method to the Local class.

- Explicitly sort instead of using sorted() and don't assume ordering

  (Python 2.3 and Jython compatibility).

- Removed ValueError raising for string formatting message checkers if the

  string does not contain any string formattings (:trac:`150`).

- Fix Serbian plural forms (:trac:`213`).

- Small speed improvement in format_date() (:trac:`216`).

- Fix number formatting for locales where CLDR specifies alt or draft 

  items (:trac:`217`)

- Fix bad check in format_time (:trac:`257`, reported with patch and tests by

  jomae)

- Fix so frontend.CommandLineInterface.run does not accumulate logging 

  handlers (:trac:`227`, reported with initial patch by dfraser)

- Fix exception if environment contains an invalid locale setting

  (:trac:`200`)

Version 0.9.5

-------------

(released on April 6th 2010)

- Fixed the case where messages containing square brackets would break with

  an unpack error.

- Backport of r467: Fuzzy matching regarding plurals should *NOT* be checked

  against len(message.id)  because this is always 2, instead, it's should be

  checked against catalog.num_plurals (:trac:`212`).

Version 0.9.4

-------------

(released on August 25th 2008)

- Currency symbol definitions that is defined with choice patterns in the

  CLDR data are no longer imported, so the symbol code will be used instead.

- Fixed quarter support in date formatting.

- Fixed a serious memory leak that was introduces by the support for CLDR

  aliases in 0.9.3 (:trac:`128`).

- Locale modifiers such as "@euro" are now stripped from locale identifiers

  when parsing (:trac:`136`).

- The system locales "C" and "POSIX" are now treated as aliases for

  "en_US_POSIX", for which the CLDR provides the appropriate data. Thanks to

  Manlio Perillo for the suggestion.

- Fixed JavaScript extraction for regular expression literals (:trac:`138`)

  and concatenated strings.

- The `Translation` class in `babel.support` can now manage catalogs with

  different message domains, and exposes the family of `d*gettext` functions

  (:trac:`137`).

Version 0.9.3

-------------

(released on July 9th 2008)

- Fixed invalid message extraction methods causing an UnboundLocalError.

- Extraction method specification can now use a dot instead of the colon to

  separate module and function name (:trac:`105`).

- Fixed message catalog compilation for locales with more than two plural

  forms (:trac:`95`).

- Fixed compilation of message catalogs for locales with more than two plural

  forms where the translations were empty (:trac:`97`).

- The stripping of the comment tags in comments is optional now and

  is done for each line in a comment.

- Added a JavaScript message extractor.

- Updated to CLDR 1.6.

- Fixed timezone calculations when formatting datetime and time values.

- Added a `get_plural` function into the plurals module that returns the

  correct plural forms for a locale as tuple.

- Added support for alias definitions in the CLDR data files, meaning that

  the chance for items missing in certain locales should be greatly reduced

  (:trac:`68`).

Version 0.9.2

-------------

(released on February 4th 2008)

- Fixed catalogs' charset values not being recognized (:trac:`66`).

- Numerous improvements to the default plural forms.

- Fixed fuzzy matching when updating message catalogs (:trac:`82`).

- Fixed bug in catalog updating, that in some cases pulled in translations

  from different catalogs based on the same template.

- Location lines in PO files do no longer get wrapped at hyphens in file

  names (:trac:`79`).

- Fixed division by zero error in catalog compilation on empty catalogs

  (:trac:`60`).

Version 0.9.1

-------------

(released on September 7th 2007)

- Fixed catalog updating when a message is merged that was previously simple

  but now has a plural form, for example by moving from `gettext` to

  `ngettext`, or vice versa.

- Fixed time formatting for 12 am and 12 pm.

- Fixed output encoding of the `pybabel --list-locales` command.

- MO files are now written in binary mode on windows (:trac:`61`).

Version 0.9

-----------

(released on August 20th 2007)

- The `new_catalog` distutils command has been renamed to `init_catalog` for

  consistency with the command-line frontend.

- Added compilation of message catalogs to MO files (:trac:`21`).

- Added updating of message catalogs from POT files (:trac:`22`).

- Support for significant digits in number formatting.

- Apply proper "banker's rounding" in number formatting in a cross-platform

  manner.

- The number formatting functions now also work with numbers represented by

  Python `Decimal` objects (:trac:`53`).

- Added extensible infrastructure for validating translation catalogs.

- Fixed the extractor not filtering out messages that didn't validate against

  the keyword's specification (:trac:`39`).

- Fixed the extractor raising an exception when encountering an empty string

  msgid. It now emits a warning to stderr.

- Numerous Python message extractor fixes: it now handles nested function

  calls within a gettext function call correctly, uses the correct line number

  for multi-line function calls, and other small fixes (tickets :trac:`38` and

  :trac:`39`).

- Improved support for detecting Python string formatting fields in message

  strings (:trac:`57`).

- CLDR upgraded to the 1.5 release.

- Improved timezone formatting.

- Implemented scientific number formatting.

- Added mechanism to lookup locales by alias, for cases where browsers insist

  on including only the language code in the `Accept-Language` header, and

  sometimes even the incorrect language code.

Version 0.8.1

-------------

(released on July 2nd 2007)

- `default_locale()` would fail when the value of the `LANGUAGE` environment

  variable contained multiple language codes separated by colon, as is

  explicitly allowed by the GNU gettext tools. As the `default_locale()`

  function is called at the module level in some modules, this bug would

  completely break importing these modules on systems where `LANGUAGE` is set

  that way.

- The character set specified in PO template files is now respected when

  creating new catalog files based on that template. This allows the use of

  characters outside the ASCII range in POT files (:trac:`17`).

- The default ordering of messages in generated POT files, which is based on

  the order those messages are found when walking the source tree, is no

  longer subject to differences between platforms; directory and file names

  are now always sorted alphabetically.

- The Python message extractor now respects the special encoding comment to be

  able to handle files containing non-ASCII characters (:trac:`23`).

- Added ``N_`` (gettext noop) to the extractor's default keywords.

- Made locale string parsing more robust, and also take the script part into

  account (:trac:`27`).

- Added a function to list all locales for which locale data is available.

- Added a command-line option to the `pybabel` command which prints out all

  available locales (:trac:`24`).

- The name of the command-line script has been changed from just `babel` to

  `pybabel` to avoid a conflict with the OpenBabel project (:trac:`34`).

Version 0.8

-----------

(released on June 20th 2007)

- First public release