Textblob

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

Features:

- Add `Word.stem` and `WordList.stem` methods (:issue:`145`). Thanks :user:`nitkul`.

Bug fixes:

- Fix translation and language detection (:issue:`137`). Thanks :user:`EpicJhon` for the fix.

Changes:

- *Backwards-incompatible*: Remove Python 2.6 and 3.3 support.

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

Bug fixes:

- Fix translation and language detection (:issue:`115`, :issue:`117`, :issue:`119`). Thanks :user:`AdrianLC` and :user:`jschnurr` for the fix. Thanks :user:`AdrianLC`, :user:`edgaralts`, and :user:`pouya-cognitiv` for reporting.

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

Changes:

- Compatible with nltk>=3.1. NLTK versions < 3.1 are no longer supported.
- Change default tagger to NLTKTagger (uses NLTK's averaged perceptron tagger).
- Tested on Python 3.5.

Bug fixes:

- Fix singularization of a number of words. Thanks :user:`jonmcoe`.
- Fix spelling correction when nltk>=3.1 is installed (:issue:`99`). Thanks :user:`shubham12101` for reporting.

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

Changes:

- Unchanged text is now considered a translation error. Raises ``NotTranslated`` (:issue:`76`). Thanks :user:`jschnurr`.

Bug fixes:

- ``Translator.translate`` will detect language of input text by default (:issue:`85`). Thanks again :user:`jschnurr`.
- Fix matching of tagged phrases with CFG in ``ConllExtractor``. Thanks :user:`lragnarsson`.
- Fix inflection of a few irregular English nouns. Thanks :user:`jonmcoe`.

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

Bug fixes:

- Fix ``DecisionTreeClassifier.pprint`` for compatibility with nltk>=3.0.2.
- Translation no longer adds erroneous whitespace around punctuation characters (:issue:`83`). Thanks :user:`AdrianLC` for reporting and thanks :user:`jschnurr` for the patch.

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

- TextBlob now depends on NLTK 3. The vendorized version of NLTK has been removed.
- Fix bug that raised a `SyntaxError` when translating text with non-ascii characters on Python 3.
- Fix bug that showed "double-escaped" unicode characters in translator output (issue 56). Thanks Evan Dempsey.
- *Backwards-incompatible*: Completely remove ``import text.blob``. You should ``import textblob`` instead.
- *Backwards-incompatible*: Completely remove ``PerceptronTagger``. Install ``textblob-aptagger`` instead.
- *Backwards-incompatible*: Rename ``TextBlobException`` to ``TextBlobError`` and ``MissingCorpusException`` to ``MissingCorpusError``.
- *Backwards-incompatible*: ``Format`` classes are passed a file object rather than a file path.
- *Backwards-incompatible*: If training a classifier with data from a file, you must pass a file object (rather than a file path).
- Updated English sentiment corpus.
- Add ``feature_extractor`` parameter to ``NaiveBayesAnalyzer``.
- Add ``textblob.formats.get_registry()`` and ``textblob.formats.register()`` which allows users to register custom data source formats.
- Change ``BaseClassifier.detect`` from a ``staticmethod`` to a ``classmethod``.
- Improved docs.
- Tested on Python 3.4.