* Added pyproject.toml
This adds the build requirements metadata to the repo, which can be used
with any build tools that implement PEP 517 and PEP 518 (e.g. pip, tox).
It is no longer necessary to have the build dependencies installed when
installing from source.
* Add python_requires for 2.7, 3.4+
This directive specifies in the build metadata which version of CPython
is supported by this version of spaCy, which pip will take into account
when determining what version to download. This will allow you to safely
drop old versions of Python without `pip install spaCy` breaking for those
versions.
* Add Python 3.7 to the trove classifiers
* Add note that Unidic is required for Japanese
This addresses #3001. -POLM
* Add extras_require for mecab with old version
Related to issue #3018.
* mecab → ja
Co-Authored-By: polm <polm@dampfkraft.com>
* Go back to using requests instead of urllib (closes#2320)
Fewer dependencies are good, but this one was simply causing too many other problems around SSL verification and Python 2/3 compatibility. requests is a popular enough package that it's okay for spaCy to depend on it – and this will hopefully make model downloads less flakey.
* Only download model if not installed (see #1456)
Use #egg=model==version to allow pip to check for existing installations. The download is only started if no installation matching the package/version is found. Fixes a long-standing inconvenience.
* Pass additional options to pip when installing model (resolves#1456)
Treat all additional arguments passed to the download command as pip options to allow user to customise the command. For example:
python -m spacy download en --user
* Add CLI option to enable installing model package dependencies
* Revert "Add CLI option to enable installing model package dependencies"
This reverts commit 9336ffe695.
* Update documentation
This patch takes a step towards #1487 by introducing the
doc.retokenize() context manager, to handle merging spans, and soon
splitting tokens.
The idea is to do merging and splitting like this:
with doc.retokenize() as retokenizer:
for start, end, label in matches:
retokenizer.merge(doc[start : end], attrs={'ent_type': label})
The retokenizer accumulates the merge requests, and applies them
together at the end of the block. This will allow retokenization to be
more efficient, and much less error prone.
A retokenizer.split() function will then be added, to handle splitting a
single token into multiple tokens. These methods take `Span` and `Token`
objects; if the user wants to go directly from offsets, they can append
to the .merges and .splits lists on the retokenizer.
The doc.merge() method's behaviour remains unchanged, so this patch
should be 100% backwards incompatible (modulo bugs). Internally,
doc.merge() fixes up the arguments (to handle the various deprecated styles),
opens the retokenizer, and makes the single merge.
We can later start making deprecation warnings on direct calls to doc.merge(),
to migrate people to use of the retokenize context manager.
