polib

• Fixed issue #70 (occurrences parsing for windows pathes)
• Fixed issue #71 (tcomment and flags not saved for obsolete entries)
• Fixed issue #72 (wrong metadata ordering)
• Fixed issue #73 (can't always unpickle POFile)
• Use natural sorting for additional headers (metadata)
• Fixed typos in various documents

• Fixed bad parsing of indented msgstr_plural
• Fixed ordering of "Language" metadata entry
• Removed space after "#" in header if comment line is empty (like gettext tools)
• Fixed typos / grammar errors (thanks Jakub Wilk)
• Take into account msgid_plural if needed when comparing entries (thanks Leonardo Constantino Oliveira)
• Fixed issue #63 (str() on a bytes instance when using python3) (thanks Jakub Wilk)

• Wheel support
• Add missing 'Language' and 'Plural-Forms' to metadata ordering
• More accurate float operation for POFile.percent_translated()

• Fixed issue #59: tokens variable referenced before assignment
• Implemented feature request #56: line number information in PO entries
• Fixed issue #61: polib does not handle previous msgid on multilines properly

• Fixed issue #43: improved check that determine if polib is dealing with a filepath or unicode content
• Fixed issue #44: polib now checks MO files revision number and throws an error if the number is unexpected
• Fixed issue #45: parse properly mo files with no header entry
• Fixed issue #47: added flags attribute for MOEntry to be consistent with POEntry
• Fixed issue #49: use integers rather than strings for msgstr_plural keys
• Fixed issue #51: if a PO file ends with a comment, polib adds a spurious empty entry at the end
• Fixed issue #52: bad magic number written on big endian platforms
• Fixed issue #53: added a hash() method to POEntry and MOEntry classes
• Fixed issue #54: use lowercase for state identifiers. This fixes issues with certain locales and string.lower()
• Fixed issue #58: use io.open() instead of codecs.open() because the latter doesn't handle very well universal line endings
• Make sure the mo file is closed at garbage collection, this prevents warnings on unclosed file when running tests with python >= 3.2
• Better way to test endianness
• polib download URL is now on Pypi

• Fixed issue #38: POFile.append() raised a duplicate exception when you tried to add a new entry with the same msgid and a different msgctxt (only when check_for_duplicates option is set to True)
• Fixed issue #39: Added init.py file for convenience
• Fixed issue #41: UnicodeDecodeError when running setup.py build on python3 with C locale
• polib is now fully PEP8 compliant
• Small improvements: remove unused "typ" var (thanks Rodrigo Silva), improved Makefile, Make sure _BaseFile.contains returns a boolean value

• allow empty comments, flags or occurrences lines

• speed up POFile.merge method (thanks @encukou)
• allow comments starting with two '#' characters (thanks @goibhniu)

Yeah... after nearly 6 years, polib reaches the stable state :) Changes and fixes in this release :

• polib.pofile and polib.mofile functions can now return a custom class (thanks Craig Blaszczyk)
• polib now can find the metadata entry no matter where it is located (thanks François Poirotte)
• fixed issue #28 (IOError on reading obsolete "previous msgid" entries) (thanks James Ni)

This version adds support for python 3 (thanks to Vinay Sajip). polib now supports out-of-the-box any version of python ranging from 2.4 to latest 3.X version. polib is now 5 years old ;) so the 0.7.X branch will be the last before the 1.X stable branch.

• Better api, autodetected_encoding is no longer required to explicitly set the encoding (fixes issue #23),
• Fixed issue #24 Support indented PO files (thanks to François Poirotte).

• Fixed issue #19 (Disappearing newline characters due to textwrap module),
• ensure wrapping works as expected.

• Backported textwrap.TextWrapper._wrap_chunks that has support for the drop_whitespace parameter added in Python 2.6 (Fixes #18: broken compatibility with python 2.5, thanks @jezdez).

• fixed regression that prevented POFile initialization from data to work (issue #17).

• polib is now fully documented <http://polib.readthedocs.org>_,
• switched from doctests to unit tests to keep the polib.py file clean,
• fixed issue #7 (wrapping issues, thanks @jezdez),
• added a eq method to _BaseFile (thanks @kost BebiX),
• handle msgctxt correctly when compiling mo files,
• compiled mo files are now exactly the same as those compiled by msgfmt without using hash tables.

• Removed multiline handling code, it was a mess and was the source of potential bugs like issue #11,
• Fixed typo in README and CHANGELOG, fixes issue #13.

• fixed an issue with detect_encoding(), in some cases it could return an invalid charset.

• correctly unescape lines containing both \\n and \n (thanks to Martin Geisler),
• fixed issue #6: str() methods are returning unicode instead of str,
• fixed issue #8: POFile.merge error when an entry is obsolete in a .po, that this entry reappears in the .pot and that we merge the two,
• added support to instantiate POFile objects using data instead of file path (thanks to Diego Búrigo Zacarão),
• fixed issue #9: POFile.merge drop fuzzy attributes from translations (thanks to Tim Gerundt),
• fixed issue #10: Finding entries with the same msgid and different context (msgctxt).

• fixed issue #1: untranslated_entries() also show fuzzy message,
• write back the fuzzy header if present in the pofile,
• added support for previous msgctxt, previous msgid and previous msgid_plural comments (fixes issue #5),
• better handling of lines wrapping.

• fixed issue #0025: setup.py requires CHANGELOG but it's not present in polib-0.5.0-tar.gz

• fixed issue #0017: UnicodeDecodeError while writing a mo-file,
• fixed issue #0018: implemented support for msgctxt,
• fixed bug when compiling plural msgids/strs,
• API docs are no longer included, hopefully next release will ship with sphinx documentation,
• parse msg plural entries correctly when reading mo files,
• fixed issue #0020 and #0021: added ability to check for duplicate when adding entries to po/mo files, this is optional and not enabled by default because it slows down considerably the library,
• fixed issue #0022: unescaping code is insufficient,
• fixed issue #0023: encoding error when saving mo file as po file (thanks to sebastien.sable for the patch !).

• fixed issue #0007: use the codecs module to open files,
• fixed issue #0014: plural forms are not saved correctly in the mo file (thanks lorenzo.gil.sanchez for the patch),
• fixed issue #0015: no LICENSE file included in tarball,
• removed Version/Date from README,
• added test pot files to MANIFEST.in,
• performance improvement in find() method (thanks Thomas !).

• fixed issue #0006: plural msgstrs were saved unsorted,
• fixed issue #0008: long comment lines broke 'save()' method,
• removed performance shortcuts: they were in fact inefficient, I was mislead by the python profile module, kudos to Thomas for making me realise that,
• fixed issue #0010: wrong polib version number,
• fixed issue #0011: occurrences parsing is now more robust and can handle weird references formats (like in eToys OLPC po files),
• fixed issue #0012: improved merge() method.

• fixed bug #0005: percent_translated divide by 0 on empty po files,
• fixed bug #0004: occurrences that have hyphens are wrapped when they should not,
• changes in how encoding is handled,
• remove deprecation warnings for typo on "occurrences",
• added POEntry.cmp() method to sort entries like gettext does,
• fixed POEntry.transalated(),
• added a merge() method to POFile class, that behaves like the gettext msgmerge utility,
• obsolete entries are now written at the end of the file and with only msgid/msgstr like gettext does,
• fixed some bugs in mo files parsing,
• renamed quote/unquote functions to escape/unescape,
• various cosmetic changes.

• fixed bug #0002: typo on "occurrences",
• fixed bug #0003: mismatch in exception instance names,
• removed deprecation warnings,
• removed unused charset() method in POFile/MOFile objects,
• fixed bug in multibytes string length (added regression tests),
• fixed a bug in detect_encoding(),
• added a find() method to _BaseFile class,
• proper handling of quoting and unquoting,
• proper handling of multiline strings in metadata

• speed improvements,
• polib can now compile mo files,
• unicode support,
• fixed bug #0001: global name 'sorted' is not defined" on python 2.3.

Initial release