* Reimplement matching in Cython, instead of Python.

spacy/matcher.pyx

@@ -1,52 +1,100 @@
-class MatchState(object):
+from .typedefs cimport attr_t
-    def __init__(self, token_spec, ext):
+from .attrs cimport attr_id_t
-        self.token_spec = token_spec
+from .structs cimport TokenC
-        self.ext = ext
-        self.is_final = False
 
-    def match(self, token):
+from cymem.cymem cimport Pool
-        for attr, value in self.token_spec:
+from libcpp.vector cimport vector
-            if getattr(token, attr) != value:
+
+from .attrs cimport LENGTH, ENT_TYPE
+from .tokens.doc cimport get_token_attr
+from .tokens.doc cimport Doc
+from .vocab cimport Vocab
+
+
+cdef struct AttrValue:
+    attr_id_t attr
+    attr_t value
+
+
+cdef struct Pattern:
+    AttrValue* spec
+    int length
+
+
+cdef Pattern* init_pattern(Pool mem, object token_specs, attr_t entity_type) except NULL:
+    pattern = <Pattern*>mem.alloc(len(token_specs) + 1, sizeof(Pattern))
+    cdef int i
+    for i, spec in enumerate(token_specs):
+        pattern[i].spec = <AttrValue*>mem.alloc(len(spec), sizeof(AttrValue))
+        pattern[i].length = len(spec)
+        for j, (attr, value) in enumerate(spec):
+            pattern[i].spec[j].attr = attr
+            pattern[i].spec[j].value = value
+    i = len(token_specs)
+    pattern[i].spec = <AttrValue*>mem.alloc(1, sizeof(AttrValue))
+    pattern[i].spec[0].attr = ENT_TYPE
+    pattern[i].spec[0].value = entity_type
+    pattern[i].spec[1].attr = LENGTH
+    pattern[i].spec[1].value = len(token_specs)
+    pattern[i].length = 0
+    return pattern
+
+
+cdef int match(const Pattern* pattern, const TokenC* token) except -1:
+    cdef int i
+    for i in range(pattern.length):
+        if get_token_attr(token, pattern.spec[i].attr) != pattern.spec[i].value:
             return False
-        else:
     return True
 
-    def __repr__(self):
+
-        return '<spec %s>' % (self.token_spec)
+cdef int is_final(const Pattern* pattern) except -1:
+    return (pattern + 1).length == 0
 
 
-class EndState(object):
+cdef object get_entity(const Pattern* pattern, const TokenC* tokens, int i):
-    def __init__(self, entity_type, length):
+    pattern += 1
-        self.entity_type = entity_type
+    i += 1
-        self.length = length
+    return (pattern.spec[0].value, i - pattern.spec[1].value, i)
-        self.is_final = True
-
-    def __call__(self, token):
-        return (self.entity_type, ((token.i+1) - self.length), token.i+1)
-
-    def __repr__(self):
-        return '<end %s>' % (self.entity_type)
 
 
-class Matcher(object):
+cdef class Matcher:
+    cdef Pool mem
+    cdef Pattern** patterns
+    cdef readonly int n_patterns
+
     def __init__(self, patterns):
-        self.start_states = []
+        self.mem = Pool()
-        for token_specs, entity_type in patterns:
+        self.patterns = <Pattern**>self.mem.alloc(len(patterns), sizeof(Pattern*))
-            state = EndState(entity_type, len(token_specs))
+        for i, (token_specs, entity_type) in enumerate(patterns):
-            for spec in reversed(token_specs):
+            self.patterns[i] = init_pattern(self.mem, token_specs, entity_type)
-                state = MatchState(spec, state)
+        self.n_patterns = len(patterns)
-            self.start_states.append(state)
 
-    def __call__(self, tokens):
+    def __call__(self, Doc doc):
-        queue = list(self.start_states)
+        cdef vector[Pattern*] partials
+        cdef int n_partials = 0
+        cdef int q = 0
+        cdef int i, token_i
+        cdef const TokenC* token
+        cdef Pattern* state
         matches = []
-        for token in tokens:
+        for token_i in range(doc.length):
-            next_queue = list(self.start_states)
+            token = &doc.data[token_i]
-            for pattern in queue:
+            q = 0
-                if pattern.match(token):
+            for i in range(partials.size()):
-                    if pattern.ext.is_final:
+                state = partials.at(i)
-                        matches.append(pattern.ext(token))
+                if match(state, token):
+                    if is_final(state):
+                        matches.append(get_entity(state, token, token_i))
                     else:
-                        next_queue.append(pattern.ext)
+                        partials[q] = state + 1
-            queue = next_queue
+                        q += 1
+            partials.resize(q)
+            for i in range(self.n_patterns):
+                state = self.patterns[i]
+                if match(state, token):
+                    if is_final(state):
+                        matches.append(get_entity(state, token, token_i))
+                    else:
+                        partials.push_back(state + 1)
         return matches

tests/test_matcher.py

@@ -1,52 +1,51 @@
 from __future__ import unicode_literals
 import pytest
 
+from spacy.strings import StringStore
 from spacy.matcher import *
-
+from spacy.attrs import ORTH
-
+from spacy.tokens.doc import Doc
-class MockToken(object):
+from spacy.vocab import Vocab
-    def __init__(self, i, string):
-        self.i = i
-        self.orth_ = string
-
-
-def make_tokens(string):
-    return [MockToken(i, s) for i, s in enumerate(string.split())]
 
 
 @pytest.fixture
-def matcher():
+def matcher(EN):
     specs = []
     for string in ['JavaScript', 'Google Now', 'Java']:
-        spec = tuple([[('orth_', orth)] for orth in string.split()])
+        spec = []
-        specs.append((spec, 'product'))
+        for orth_ in string.split():
+            spec.append([(ORTH, EN.vocab.strings[orth_])])
+        specs.append((spec, EN.vocab.strings['product']))
     return Matcher(specs)
 
 
 def test_compile(matcher):
-    assert len(matcher.start_states) == 3
+    assert matcher.n_patterns == 3
 
-
+def test_no_match(matcher, EN):
-def test_no_match(matcher):
+    tokens = EN('I like cheese')
-    tokens = make_tokens('I like cheese')
     assert matcher(tokens) == []
 
 
-def test_match_start(matcher):
+def test_match_start(matcher, EN):
-    tokens = make_tokens('JavaScript is good')
+    tokens = EN('JavaScript is good')
-    assert matcher(tokens) == [('product', 0, 1)]
+    assert matcher(tokens) == [(EN.vocab.strings['product'], 0, 1)]
 
 
-def test_match_end(matcher):
+def test_match_end(matcher, EN):
-    tokens = make_tokens('I like Java')
+    tokens = EN('I like Java')
-    assert matcher(tokens) == [('product', 2, 3)]
+    assert matcher(tokens) == [(EN.vocab.strings['product'], 2, 3)]
 
 
-def test_match_middle(matcher):
+def test_match_middle(matcher, EN):
-    tokens = make_tokens('I like Google Now best')
+    tokens = EN('I like Google Now best')
-    assert matcher(tokens) == [('product', 2, 4)]
+    assert matcher(tokens) == [(EN.vocab.strings['product'], 2, 4)]
 
 
-def test_match_multi(matcher):
+def test_match_multi(matcher, EN):
-    tokens = make_tokens('I like Google Now and Java best')
+    tokens = EN('I like Google Now and Java best')
-    assert matcher(tokens) == [('product', 2, 4), ('product', 5, 6)]
+    assert matcher(tokens) == [(EN.vocab.strings['product'], 2, 4),
+                               (EN.vocab.strings['product'], 5, 6)]
+
+def test_dummy():
+    pass