mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-11 12:18:04 +03:00
143 lines
6.5 KiB
ReStructuredText
143 lines
6.5 KiB
ReStructuredText
===
|
|
API
|
|
===
|
|
|
|
.. warning:: The documentation is currently being rewritten. I started out
|
|
using Sphinx, but I've found it too limiting.
|
|
|
|
For now, the docs here are incomplete and may even tell you lies (please
|
|
report the lies).
|
|
|
|
.. py:currentmodule:: spacy
|
|
|
|
.. class:: en.English(self, data_dir=join(dirname(__file__, 'data')))
|
|
:noindex:
|
|
|
|
.. method:: __call__(self, unicode text, tag=True, parse=False) --> Tokens
|
|
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
| Attribute | Type | Attr API | NoteS |
|
|
+===========+========================================+=============+==========================+
|
|
| strings | :py:class:`strings.StringStore` | __getitem__ | string <-> int mapping |
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
| vocab | :py:class:`vocab.Vocab` | __getitem__ | Look up Lexeme object |
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
| tokenizer | :py:class:`tokenizer.Tokenizer` | __call__ | Get Tokens given unicode |
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
| tagger | :py:class:`en.pos.EnPosTagger` | __call__ | Set POS tags on Tokens |
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
| parser | :py:class:`syntax.parser.GreedyParser` | __call__ | Set parse on Tokens |
|
|
+-----------+----------------------------------------+-------------+--------------------------+
|
|
|
|
.. py:class:: spacy.tokens.Tokens(self, vocab: Vocab, string_length=0)
|
|
|
|
.. py:method:: __getitem__(self, i) --> Token
|
|
|
|
.. py:method:: __iter__(self) --> Iterator[Token]
|
|
|
|
.. py:method:: __len__(self) --> int
|
|
|
|
.. py:method:: to_array(self, attr_ids: List[int]) --> numpy.ndarray[ndim=2, dtype=int32]
|
|
|
|
.. py:method:: count_by(self, attr_id: int) --> Dict[int, int]
|
|
|
|
+---------------+-------------+-------------+
|
|
| Attribute | Type | Useful |
|
|
+===============+=============+=============+
|
|
| vocab | Vocab | __getitem__ |
|
|
+---------------+-------------+-------------+
|
|
| vocab.strings | StringStore | __getitem__ |
|
|
+---------------+-------------+-------------+
|
|
|
|
.. py:class:: spacy.tokens.Token(self, parent: Tokens, i: int)
|
|
|
|
.. py:method:: __unicode__(self) --> unicode
|
|
|
|
.. py:method:: __len__(self) --> int
|
|
|
|
.. py:method:: nbor(self, i=1) --> Token
|
|
|
|
.. py:method:: child(self, i=1) --> Token
|
|
|
|
.. py:method:: sibling(self, i=1) --> Token
|
|
|
|
.. py:attribute:: head: Token
|
|
|
|
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| Attribute | Type | Attribute | Type | Attribute | Type |
|
|
+===========+======+===========+=========+===========+====================================+
|
|
| orth | int | orth\_ | unicode | idx | int |
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| lemma | int | lemma\_ | unicode | cluster | int |
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| lower | int | lower\_ | unicode | length | int |
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| norm | int | norm\_ | unicode | prob | float |
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| shape | int | shape\_ | unicode | repvec | ndarray(shape=(300,), dtype=float) |
|
|
+-----------+------+-----------+---------+-----------+------------------------------------+
|
|
| prefix | int | prefix\_ | unicode | |
|
|
+-----------+------+-----------+---------+------------------------------------------------+
|
|
| suffix | int | suffix\_ | unicode | |
|
|
+-----------+------+-----------+---------+------------------------------------------------+
|
|
| pos | int | pos\_ | unicode | |
|
|
+-----------+------+-----------+---------+------------------------------------------------+
|
|
| tag | int | tag\_ | unicode | |
|
|
+-----------+------+-----------+---------+------------------------------------------------+
|
|
| dep | int | dep\_ | unicode | |
|
|
+-----------+------+-----------+---------+------------------------------------------------+
|
|
|
|
|
|
.. py:class:: spacy.vocab.Vocab(self, data_dir=None, lex_props_getter=None)
|
|
|
|
.. py:method:: __len__(self) --> int
|
|
|
|
.. py:method:: __getitem__(self, id: int) --> unicode
|
|
|
|
.. py:method:: __getitem__(self, string: unicode) --> int
|
|
|
|
.. py:method:: __setitem__(self, py_str: unicode, props: Dict[str, int[float]) --> None
|
|
|
|
.. py:method:: dump(self, loc: unicode) --> None
|
|
|
|
.. py:method:: load_lexemes(self, loc: unicode) --> None
|
|
|
|
.. py:method:: load_vectors(self, loc: unicode) --> None
|
|
|
|
.. py:class:: spacy.strings.StringStore(self)
|
|
|
|
.. py:method:: __len__(self) --> int
|
|
|
|
.. py:method:: __getitem__(self, id: int) --> unicode
|
|
|
|
.. py:method:: __getitem__(self, string: bytes) --> id
|
|
|
|
.. py:method:: __getitem__(self, string: unicode) --> id
|
|
|
|
.. py:method:: dump(self, loc: unicode) --> None
|
|
|
|
.. py:method:: load(self, loc: unicode) --> None
|
|
|
|
.. py:class:: spacy.tokenizer.Tokenizer(self, Vocab vocab, rules, prefix_re, suffix_re, infix_re, pos_tags, tag_names)
|
|
|
|
.. py:method:: tokens_from_list(self, List[unicode]) --> spacy.tokens.Tokens
|
|
|
|
.. py:method:: __call__(self, string: unicode) --> spacy.tokens.Tokens)
|
|
|
|
.. py:attribute:: vocab: spacy.vocab.Vocab
|
|
|
|
.. py:class:: spacy.en.pos.EnPosTagger(self, strings: spacy.strings.StringStore, data_dir: unicode)
|
|
|
|
.. py:method:: __call__(self, tokens: spacy.tokens.Tokens)
|
|
|
|
.. py:method:: train(self, tokens: spacy.tokens.Tokens, List[int] golds) --> int
|
|
|
|
.. py:method:: load_morph_exceptions(self, exc: Dict[unicode, Dict])
|
|
|
|
.. py:class:: GreedyParser(self, model_dir: unicode)
|
|
|
|
.. py:method:: __call__(self, tokens: spacy.tokens.Tokens) --> None
|
|
|
|
.. py:method:: train(self, spacy.tokens.Tokens) --> None
|