Merge remote-tracking branch 'refs/remotes/honnibal/master'

.appveyor.yml

@@ -12,9 +12,9 @@ environment:
       PYTHON_VERSION: "2.7.10"
       PYTHON_ARCH: "64"
 
-    #- PYTHON: "C:\\Python27.10-x32"
-    #  PYTHON_VERSION: "2.7.10"
-    #  PYTHON_ARCH: "32"
+    - PYTHON: "C:\\Python27.10-x32"
+      PYTHON_VERSION: "2.7.10"
+      PYTHON_ARCH: "32"
 
     # The lastest Python 3.4.
     - PYTHON: "C:\\Python34-x64" 
@@ -38,10 +38,11 @@ install:
   - "SET PYTHONPATH=%CD%;%PYTHONPATH%"
 
   # Filesystem root
-  - ps: "ls \"C:/\""
+  #- ps: "ls \"C:/\""
+  #- SET
 
   # Installed SDKs
-  - ps: "ls \"C:/Program Files/Microsoft SDKs/Windows\""
+  #- ps: "ls \"C:/Program Files/Microsoft SDKs/Windows\""
 
   # Checking stdint.h
   #- ps: "ls \"C:/projects/spacy/include/\""

bin/init_model.py

@@ -191,7 +191,8 @@ def setup_vocab(get_lex_attr, tag_map, src_dir, dst_dir):
         else:
             lexeme.cluster = 0
     vocab.dump(str(dst_dir / 'lexemes.bin'))
-    vocab.strings.dump(str(dst_dir / 'strings.txt'))
+    with (dst_dir / 'strings.json').open('w') as file_:
+        vocab.strings.dump(file_)
     with (dst_dir / 'oov_prob').open('w') as file_:
         file_.write('%f' % oov_prob)
 

fabfile.py

@@ -68,7 +68,7 @@ def publish(version):
         local('git push origin %s' % version)
         local('python setup.py sdist')
         local('python setup.py register')
-        local('twine upload dist/%s.tar.gz' % version)
+        local('twine upload dist/spacy-%s.tar.gz' % version)
 
 
 def env(lang="python2.7"):

setup.py

@@ -129,11 +129,13 @@ def cython_setup(mod_names, language, includes):
         version=VERSION,
         url="http://honnibal.github.io/spaCy/",
         package_data={"spacy": ["*.pxd"],
+                      "spacy.tokens": ["*.pxd"],
+                      "spacy.serialize": ["*.pxd"],
                       "spacy.en": ["*.pxd", "data/pos/*",
                                    "data/wordnet/*", "data/tokenizer/*",
                                    "data/vocab/tag_map.json",
                                    "data/vocab/lexemes.bin",
-                                   "data/vocab/strings.txt"],
+                                   "data/vocab/strings.json"],
                       "spacy.syntax": ["*.pxd"]},
         ext_modules=exts,
         cmdclass={'build_ext': build_ext_cython_subclass},
@@ -175,7 +177,7 @@ def run_setup(exts):
     headers_workaround.install_headers('numpy')
 
 
-VERSION = '0.96'
+VERSION = '0.97'
 def main(modules, is_pypy):
     language = "cpp"
     includes = ['.', path.join(sys.prefix, 'include')]

spacy/en/download.py

@@ -1,5 +1,4 @@
 from __future__ import print_function
-from os import path
 import sys
 import os
 import tarfile
@@ -8,52 +7,58 @@ import plac
 
 from . import uget
 
+
+try:
+    FileExistsError
+except NameError:
+    FileExistsError = Exception
+
+
 # TODO: Read this from the same source as the setup
-VERSION = '0.9.5'
+VERSION = '0.9.6'
 
 AWS_STORE = 'https://s3-us-west-1.amazonaws.com/media.spacynlp.com'
 
 ALL_DATA_DIR_URL = '%s/en_data_all-%s.tgz' % (AWS_STORE, VERSION)
 
-DEST_DIR = path.join(path.dirname(path.abspath(__file__)), 'data')
+DEST_DIR = os.path.dirname(os.path.abspath(__file__))
 
 
-def download_file(url, dest_dir):
-    return uget.download(url, dest_dir, console=sys.stdout)
+def download_file(url, download_path):
+    return uget.download(url, download_path, console=sys.stdout)
 
 
-def install_data(url, dest_dir):
-    filename = download_file(url, dest_dir)
-    t = tarfile.open(filename)
-    t.extractall(dest_dir)
+def install_data(url, extract_path, download_path):
+    try:
+        os.makedirs(extract_path)
+    except FileExistsError:
+        pass
 
-
-def install_parser_model(url, dest_dir):
-    filename = download_file(url, dest_dir)
-    t = tarfile.open(filename, mode=":gz")
-    t.extractall(dest_dir)
-
-
-def install_dep_vectors(url, dest_dir):
-    download_file(url, dest_dir)
+    tmp = download_file(url, download_path)
+    assert tmp == download_path
+    t = tarfile.open(download_path)
+    t.extractall(extract_path)
 
 
 @plac.annotations(
     force=("Force overwrite", "flag", "f", bool),
 )
 def main(data_size='all', force=False):
-    if data_size == 'all':
-        data_url = ALL_DATA_DIR_URL
-    elif data_size == 'small':
-        data_url = SM_DATA_DIR_URL
+    filename = ALL_DATA_DIR_URL.rsplit('/', 1)[1]
+    download_path = os.path.join(DEST_DIR, filename)
+    data_path = os.path.join(DEST_DIR, 'data')
 
-    if force and path.exists(DEST_DIR):
-        shutil.rmtree(DEST_DIR)
+    if force and os.path.exists(download_path):
+        os.unlink(download_path)
 
-    if not os.path.exists(DEST_DIR):
-        os.makedirs(DEST_DIR)
+    if force and os.path.exists(data_path):
+        shutil.rmtree(data_path)
 
-    install_data(data_url, DEST_DIR)
+    if os.path.exists(data_path):
+        print('data already installed at %s, overwrite with --force' % DEST_DIR)
+        sys.exit(1)
+
+    install_data(ALL_DATA_DIR_URL, DEST_DIR, download_path)
 
 
 if __name__ == '__main__':

spacy/language.py

@@ -1,5 +1,6 @@
 from os import path
 from warnings import warn
+import io
 
 try:
     import ujson as json
@@ -247,7 +248,10 @@ class Language(object):
         self.parser.model.end_training(path.join(data_dir, 'deps', 'model'))
         self.entity.model.end_training(path.join(data_dir, 'ner', 'model'))
         self.tagger.model.end_training(path.join(data_dir, 'pos', 'model'))
-        self.vocab.strings.dump(path.join(data_dir, 'vocab', 'strings.txt'))
+
+        strings_loc = path.join(data_dir, 'vocab', 'strings.json')
+        with io.open(strings_loc, 'w', encoding='utf8') as file_:
+            self.vocab.strings.dump(file_)
 
         with open(path.join(data_dir, 'vocab', 'serializer.json'), 'w') as file_:
             file_.write(

spacy/strings.pyx

@@ -12,8 +12,15 @@ from libc.stdint cimport int64_t
 
 from .typedefs cimport hash_t, attr_t
 
+try:
+    import codecs as io
+except ImportError:
+    import io
 
-SEPARATOR = '\n|-SEP-|\n'
+try:
+    import ujson as json
+except ImportError:
+    import json
 
 
 cpdef hash_t hash_string(unicode string) except 0:
@@ -114,7 +121,11 @@ cdef class StringStore:
     def __iter__(self):
         cdef int i
         for i in range(self.size):
-            yield self[i]
+            if i == 0:
+                yield u''
+            else:
+                utf8str = &self.c[i]
+                yield _decode(utf8str)
 
     def __reduce__(self):
         strings = [""]
@@ -138,26 +149,20 @@ cdef class StringStore:
         self.size += 1
         return &self.c[self.size-1]
 
-    def dump(self, loc):
-        cdef Utf8Str* string
-        cdef unicode py_string
-        cdef int i
-        with codecs.open(loc, 'w', 'utf8') as file_:
-            for i in range(1, self.size):
-                string = &self.c[i]
-                py_string = _decode(string)
-                file_.write(py_string)
-                if (i+1) != self.size:
-                    file_.write(SEPARATOR)
+    def dump(self, file_):
+        string_data = json.dumps([s for s in self])
+        if not isinstance(string_data, unicode):
+            string_data = string_data.decode('utf8')
+        file_.write(string_data)
 
-    def load(self, loc):
-        with codecs.open(loc, 'r', 'utf8') as file_:
-            strings = file_.read().split(SEPARATOR)
+    def load(self, file_):
+        strings = json.load(file_)
         if strings == ['']:
             return None
         cdef unicode string
         cdef bytes byte_string
         for string in strings: 
+            if string:
                 byte_string = string.encode('utf8')
                 self.intern(byte_string, len(byte_string))
 

spacy/tokens/doc.pyx

@@ -120,6 +120,9 @@ cdef class Doc:
     def __str__(self):
         return u''.join([t.string for t in self])
 
+    def __repr__(self):
+        return u''.join([t.string for t in self])
+
     def similarity(self, other):
         if self.vector_norm == 0 or other.vector_norm == 0:
             return 0.0

spacy/tokens/spans.pyx

@@ -46,6 +46,12 @@ cdef class Span:
             return 0
         return self.end - self.start
 
+    def __repr__(self):
+        text = self.text_with_ws
+        if self[-1].whitespace_:
+            text = text[:-1]
+        return text
+
     def __getitem__(self, object i):
         if isinstance(i, slice):
             start, end = normalize_slice(len(self), i.start, i.stop, i.step)

spacy/tokens/token.pyx

@@ -43,6 +43,9 @@ cdef class Token:
     def __str__(self):
         return self.string
 
+    def __repr__(self):
+        return self.string
+
     cpdef bint check_flag(self, attr_id_t flag_id) except -1:
         return Lexeme.c_check_flag(self.c.lex, flag_id)
 

spacy/vocab.pyx

@@ -62,7 +62,9 @@ cdef class Vocab:
         cdef Vocab self = cls(get_lex_attr=get_lex_attr, tag_map=tag_map,
                               lemmatizer=lemmatizer, serializer_freqs=serializer_freqs)
 
-        self.load_lexemes(path.join(data_dir, 'strings.txt'), path.join(data_dir, 'lexemes.bin'))
+        with io.open(path.join(data_dir, 'strings.json'), 'r', encoding='utf8') as file_:
+            self.strings.load(file_)
+        self.load_lexemes(path.join(data_dir, 'lexemes.bin'))
         if path.exists(path.join(data_dir, 'vec.bin')):
             self.vectors_length = self.load_vectors_from_bin_loc(path.join(data_dir, 'vec.bin'))
         return self
@@ -106,11 +108,12 @@ cdef class Vocab:
         # TODO: Dump vectors
         tmp_dir = tempfile.mkdtemp()
         lex_loc = path.join(tmp_dir, 'lexemes.bin')
-        str_loc = path.join(tmp_dir, 'strings.txt')
+        str_loc = path.join(tmp_dir, 'strings.json')
         vec_loc = path.join(self.data_dir, 'vec.bin') if self.data_dir is not None else None
 
         self.dump(lex_loc)
-        self.strings.dump(str_loc)
+        with io.open(str_loc, 'w', encoding='utf8') as file_:
+            self.strings.dump(file_)
         
         state = (str_loc, lex_loc, vec_loc, self.morphology, self.get_lex_attr,
                  self.serializer_freqs, self.data_dir)
@@ -250,8 +253,7 @@ cdef class Vocab:
             fp.write_from(&lexeme.l2_norm, sizeof(lexeme.l2_norm), 1)
         fp.close()
 
-    def load_lexemes(self, strings_loc, loc):
-        self.strings.load(strings_loc)
+    def load_lexemes(self, loc):
         if not path.exists(loc):
             raise IOError('LexemeCs file not found at %s' % loc)
         fp = CFile(loc, 'rb')
@@ -369,7 +371,9 @@ def unpickle_vocab(strings_loc, lex_loc, vec_loc, morphology, get_lex_attr,
     vocab.data_dir = data_dir
     vocab.serializer_freqs = serializer_freqs
 
-    vocab.load_lexemes(strings_loc, lex_loc)
+    with io.open(strings_loc, 'r', encoding='utf8') as file_:
+        vocab.strings.load(file_)
+    vocab.load_lexemes(lex_loc)
     if vec_loc is not None:
         vocab.load_vectors_from_bin_loc(vec_loc)
     return vocab

tests/vocab/test_intern.py

@@ -1,12 +1,13 @@
 # -*- coding: utf8 -*-
 from __future__ import unicode_literals
 import pickle
-import io
 
 from spacy.strings import StringStore
 
 import pytest
 
+import io
+
 
 @pytest.fixture
 def sstore():
@@ -92,4 +93,12 @@ def test_pickle_string_store(sstore):
     assert loaded[hello_id] == u'Hi'
 
 
-
+def test_dump_load(sstore):
+    id_ = sstore[u'qqqqq']
+    loc = '/tmp/sstore.json'
+    with io.open(loc, 'w', encoding='utf8') as file_:
+        sstore.dump(file_)
+    new_store = StringStore()
+    with io.open(loc, 'r', encoding='utf8') as file_:
+        new_store.load(file_)
+    assert new_store[id_] == u'qqqqq'