Merge branch 'tokenization_w_exception_patterns' of https://github.com/oroszgy/spaCy.hu into oroszgy-tokenization_w_exception_patterns

spacy/language.py

@@ -67,6 +67,8 @@ class BaseDefaults(object):
     @classmethod
     def create_tokenizer(cls, nlp=None):
         rules = cls.tokenizer_exceptions
+        if cls.token_match:
+            token_match = cls.token_match
         if cls.prefixes:
             prefix_search  = util.compile_prefix_regex(cls.prefixes).search
         else:
@@ -82,7 +84,7 @@ class BaseDefaults(object):
         vocab = nlp.vocab if nlp is not None else cls.create_vocab(nlp)
         return Tokenizer(vocab, rules=rules,
                          prefix_search=prefix_search, suffix_search=suffix_search,
-                         infix_finditer=infix_finditer)
+                         infix_finditer=infix_finditer, token_match=token_match)
 
     @classmethod
     def create_tagger(cls, nlp=None):
@@ -142,6 +144,8 @@ class BaseDefaults(object):
             pipeline.append(nlp.entity)
         return pipeline
 
+    token_match = language_data.TOKEN_MATCH
+
     prefixes = tuple(language_data.TOKENIZER_PREFIXES)
 
     suffixes = tuple(language_data.TOKENIZER_SUFFIXES)

spacy/language_data/__init__.py

@@ -3,3 +3,4 @@ from .punctuation import *
 from .tag_map import *
 from .entity_rules import *
 from .util import *
+from .tokenizer_exceptions import *

spacy/language_data/tokenizer_exceptions.py

@@ -0,0 +1,11 @@
+from __future__ import unicode_literals
+
+import re
+
+_URL_PATTERN = r'''
+^((([A-Za-z]{3,9}:(?:\/\/)?)(?:[-;:&=\+\$,\w]+@)?[A-Za-z0-9.-]+|(?:www.|[-;:&=\+\$,\w]+@)[A-Za-z0-9.-]+)((?:\/[\+~%\/.\w\-_]*)?\??(?:[-\+=&;%@.\w_]*)#?(?:[\w]*))?)$
+'''.strip()
+
+TOKEN_MATCH = re.compile(_URL_PATTERN).match
+
+__all__ = ['TOKEN_MATCH']

spacy/tests/tokenizer/test_urls.py

@@ -0,0 +1,77 @@
+from __future__ import unicode_literals
+
+import pytest
+
+URLS = [
+    u"http://www.nytimes.com/2016/04/20/us/politics/new-york-primary-preview.html?hp&action=click&pgtype=Homepage&clickSource=story-heading&module=a-lede-package-region&region=top-news&WT.nav=top-news&_r=0",
+    u"www.google.com?q=google",
+    u"google.com",
+    u"www.red-stars.com",
+    pytest.mark.xfail(u"red-stars.com"),
+    u"http://foo.com/blah_(wikipedia)#cite-1",
+    u"http://www.example.com/wpstyle/?bar=baz&inga=42&quux",
+    u"mailto:foo.bar@baz.com",
+    u"mailto:foo-bar@baz-co.com"
+]
+
+# Punctuation we want to check is split away before the URL
+PREFIXES = [
+    "(", '"', "...", ":", "<", ">", ")"
+]
+
+# Punctuation we want to check is split away after the URL
+SUFFIXES = [
+    "(", '"', "...", ":", "<", ">"]
+
+@pytest.mark.parametrize("text", URLS)
+def test_simple_url(en_tokenizer, text):
+    tokens = en_tokenizer(text)
+    assert tokens[0].orth_ == text
+    assert len(tokens) == 1
+
+
+@pytest.mark.parametrize("prefix", PREFIXES)
+@pytest.mark.parametrize("url", URLS)
+def test_prefixed_url(en_tokenizer, prefix, url):
+    tokens = en_tokenizer(prefix + url)
+    assert tokens[0].text == prefix
+    assert tokens[1].text == url
+    assert len(tokens) == 2
+    
+@pytest.mark.parametrize("suffix", SUFFIXES)
+@pytest.mark.parametrize("url", URLS)
+def test_prefixed_url(en_tokenizer, suffix, url):
+    tokens = en_tokenizer(url + suffix)
+    assert tokens[1].text == suffix
+    assert tokens[0].text == url
+    assert len(tokens) == 2
+    
+@pytest.mark.parametrize("prefix", PREFIXES)
+@pytest.mark.parametrize("suffix", SUFFIXES)
+@pytest.mark.parametrize("url", URLS)
+def test_surround_url(en_tokenizer, prefix, suffix, url):
+    tokens = en_tokenizer(prefix + url + suffix)
+    assert tokens[0].text == prefix
+    assert tokens[1].text == url
+    assert tokens[2].text == suffix
+    assert len(tokens) == 3
+    
+@pytest.mark.parametrize("prefix1", PREFIXES)
+@pytest.mark.parametrize("prefix2", PREFIXES)
+@pytest.mark.parametrize("url", URLS)
+def test_two_prefix_url(en_tokenizer, prefix1, prefix2, url):
+    tokens = en_tokenizer(prefix1 + prefix2 + url)
+    assert tokens[0].text == prefix1
+    assert tokens[1].text == prefix2
+    assert tokens[2].text == url
+    assert len(tokens) == 3
+    
+@pytest.mark.parametrize("suffix1", SUFFIXES)
+@pytest.mark.parametrize("suffix2", SUFFIXES)
+@pytest.mark.parametrize("url", URLS)
+def test_two_prefix_url(en_tokenizer, suffix1, suffix2, url):
+    tokens = en_tokenizer(url + suffix1 + suffix2)
+    assert tokens[0].text == url
+    assert tokens[1].text == suffix1
+    assert tokens[2].text == suffix2
+    assert len(tokens) == 3

spacy/tokenizer.pxd

@@ -16,6 +16,7 @@ cdef class Tokenizer:
     cdef PreshMap _specials
     cpdef readonly Vocab vocab
 
+    cdef public object token_match
     cdef public object prefix_search
     cdef public object suffix_search
     cdef public object infix_finditer

spacy/tokenizer.pyx

@@ -29,7 +29,7 @@ cdef class Tokenizer:
     """Segment text, and create Doc objects with the discovered segment boundaries."""
     @classmethod
     def load(cls, path, Vocab vocab, rules=None, prefix_search=None, suffix_search=None,
-             infix_finditer=None):
+             infix_finditer=None, token_match=None):
         '''Load a Tokenizer, reading unsupplied components from the path.
         
         Arguments:
@@ -39,6 +39,8 @@ cdef class Tokenizer:
                 A storage container for lexical types.
             rules (dict):
                 Exceptions and special-cases for the tokenizer.
+            token_match:
+                A boolean function matching strings that becomes tokens.
             prefix_search:
                 Signature of re.compile(string).search
             suffix_search:
@@ -65,10 +67,9 @@ cdef class Tokenizer:
             with (path / 'tokenizer' / 'infix.txt').open() as file_:
                 entries = file_.read().split('\n')
             infix_finditer = util.compile_infix_regex(entries).finditer
-        return cls(vocab, rules, prefix_search, suffix_search, infix_finditer)
+        return cls(vocab, rules, prefix_search, suffix_search, infix_finditer, token_match)
 
-
+    def __init__(self, Vocab vocab, rules, prefix_search, suffix_search, infix_finditer, token_match=None):
-    def __init__(self, Vocab vocab, rules, prefix_search, suffix_search, infix_finditer):
         '''Create a Tokenizer, to create Doc objects given unicode text.
         
         Arguments:
@@ -85,10 +86,13 @@ cdef class Tokenizer:
             infix_finditer:
                 A function matching the signature of re.compile(string).finditer
                 to find infixes.
+            token_match:
+                A boolean function matching strings that becomes tokens.
         '''
         self.mem = Pool()
         self._cache = PreshMap()
         self._specials = PreshMap()
+        self.token_match = token_match
         self.prefix_search = prefix_search
         self.suffix_search = suffix_search
         self.infix_finditer = infix_finditer
@@ -100,9 +104,10 @@ cdef class Tokenizer:
     def __reduce__(self):
         args = (self.vocab,
                 self._rules,
-                self._prefix_re, 
+                self._prefix_re,
-                self._suffix_re, 
+                self._suffix_re,
-                self._infix_re)
+                self._infix_re,
+                self.token_match)
 
         return (self.__class__, args, None, None)
     
@@ -202,9 +207,12 @@ cdef class Tokenizer:
         cdef vector[LexemeC*] suffixes
         cdef int orig_size
         orig_size = tokens.length
-        span = self._split_affixes(tokens.mem, span, &prefixes, &suffixes)
+        if self.token_match and self.token_match(span):
-        self._attach_tokens(tokens, span, &prefixes, &suffixes)
+            tokens.push_back(self.vocab.get(tokens.mem, span), False)
-        self._save_cached(&tokens.c[orig_size], orig_key, tokens.length - orig_size)
+        else:
+            span = self._split_affixes(tokens.mem, span, &prefixes, &suffixes)
+            self._attach_tokens(tokens, span, &prefixes, &suffixes)
+            self._save_cached(&tokens.c[orig_size], orig_key, tokens.length - orig_size)
 
     cdef unicode _split_affixes(self, Pool mem, unicode string,
                                 vector[const LexemeC*] *prefixes,