Merge branch 'rethinc2' of ssh://github.com/honnibal/spaCy into rethinc2

This commit is contained in:
Matthew Honnibal 2016-02-05 12:48:28 +01:00
commit 4cf34fc170
4 changed files with 11 additions and 7 deletions

View File

@ -2,7 +2,7 @@ cython
cymem>=1.30,<1.31 cymem>=1.30,<1.31
pathlib pathlib
preshed>=0.46.1,<0.47.0 preshed>=0.46.1,<0.47.0
thinc>=4.2.0,<4.3.0 thinc>=5.0.0,<5.1.0
murmurhash>=0.26,<0.27 murmurhash>=0.26,<0.27
text-unidecode text-unidecode
numpy numpy

View File

@ -79,6 +79,9 @@ if sys.platform.startswith('darwin'):
compile_options['other'].append('-mmacosx-version-min=10.8') compile_options['other'].append('-mmacosx-version-min=10.8')
compile_options['other'].append('-stdlib=libc++') compile_options['other'].append('-stdlib=libc++')
link_options['other'].append('-lc++') link_options['other'].append('-lc++')
else:
compile_options['other'].append('-fopenmp')
link_options['other'].append('-fopenmp')
class build_ext_options: class build_ext_options:

View File

@ -271,20 +271,20 @@ class Language(object):
def pipe(self, texts, tag=True, parse=True, entity=True, n_threads=2, def pipe(self, texts, tag=True, parse=True, entity=True, n_threads=2,
batch_size=1000): batch_size=1000):
stream = self.tokenizer.stream(texts, stream = self.tokenizer.pipe(texts,
n_threads=n_threads, batch_size=batch_size) n_threads=n_threads, batch_size=batch_size)
if self.tagger and tag: if self.tagger and tag:
stream = self.tagger.stream(stream, stream = self.tagger.pipe(stream,
n_threads=n_threads, batch_size=batch_size) n_threads=n_threads, batch_size=batch_size)
if self.matcher and entity: if self.matcher and entity:
stream = self.matcher.stream(stream, stream = self.matcher.pipe(stream,
n_threads=n_threads, batch_size=batch_size) n_threads=n_threads, batch_size=batch_size)
if self.parser and parse: if self.parser and parse:
stream = self.parser.stream(stream, stream = self.parser.pipe(stream,
n_threads=n_threads, batch_size=batch_size) n_threads=n_threads, batch_size=batch_size)
if self.entity and entity: if self.entity and entity:
stream = self.entity.stream(stream, stream = self.entity.pipe(stream,
n_threads=n_threads, batch_size=batch_size) n_threads=1, batch_size=batch_size)
for doc in stream: for doc in stream:
yield doc yield doc

View File

@ -4,6 +4,7 @@ MALT-style dependency parser
""" """
from __future__ import unicode_literals from __future__ import unicode_literals
cimport cython cimport cython
cimport cython.parallel
from cpython.ref cimport PyObject, Py_INCREF, Py_XDECREF from cpython.ref cimport PyObject, Py_INCREF, Py_XDECREF
from cpython.exc cimport PyErr_CheckSignals from cpython.exc cimport PyErr_CheckSignals