Modernise parser tests and don't depend on models

spacy/tests/parser/test_parse.py

@@ -1,102 +1,101 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
+from ..util import get_doc, apply_transition_sequence
+
 import pytest
 
 
-@pytest.mark.models
+def test_parser_root(en_tokenizer):
-def test_root(EN):
+    text = "i don't have other assistance"
-    tokens = EN(u"i don't have other assistance")
+    heads = [3, 2, 1, 0, 1, -2]
-    for t in tokens:
+    deps = ['nsubj', 'aux', 'neg', 'ROOT', 'amod', 'dobj']
-        assert t.dep != 0, t.orth_
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads, deps=deps)
+    for t in doc:
+        assert t.dep != 0, t.text
 
 
-@pytest.mark.models
+@pytest.mark.parametrize('text', ["Hello"])
-def test_one_word_sentence(EN):
+def test_parser_parse_one_word_sentence(en_tokenizer, en_parser, text):
-    # one word sentence
+    tokens = en_tokenizer(text)
-    doc = EN.tokenizer.tokens_from_list(['Hello'])
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=[0], deps=['ROOT'])
-    EN.tagger(doc)
+
     assert len(doc) == 1
-    with EN.parser.step_through(doc) as _:
+    with en_parser.step_through(doc) as _:
         pass
     assert doc[0].dep != 0
 
 
-def apply_transition_sequence(model, doc, sequence):
+def test_parser_arc_eager_finalize_state(en_tokenizer, en_parser):
-    for action_name in sequence:
+    text = "a b c d e"
-        if '-' in action_name:
-            move, label = action_name.split('-')
-            model.parser.add_label(label)
-    with model.parser.step_through(doc) as stepwise:
-        for transition in sequence:
-            stepwise.transition(transition)
 
-
-@pytest.mark.models
-def test_arc_eager_finalize_state(EN):
     # right branching
-    example = EN.tokenizer.tokens_from_list(u"a b c d e".split(' '))
+    transition = ['R-nsubj', 'D', 'R-nsubj', 'R-nsubj', 'D', 'R-ROOT']
-    apply_transition_sequence(EN, example, ['R-nsubj','D','R-nsubj','R-nsubj','D','R-ROOT'])
+    tokens = en_tokenizer(text)
+    apply_transition_sequence(en_parser, tokens, transition)
 
-    assert example[0].n_lefts == 0
+    assert tokens[0].n_lefts == 0
-    assert example[0].n_rights == 2
+    assert tokens[0].n_rights == 2
-    assert example[0].left_edge.i == 0
+    assert tokens[0].left_edge.i == 0
-    assert example[0].right_edge.i == 4
+    assert tokens[0].right_edge.i == 4
-    assert example[0].head.i == 0
+    assert tokens[0].head.i == 0
 
-    assert example[1].n_lefts == 0
+    assert tokens[1].n_lefts == 0
-    assert example[1].n_rights == 0
+    assert tokens[1].n_rights == 0
-    assert example[1].left_edge.i == 1
+    assert tokens[1].left_edge.i == 1
-    assert example[1].right_edge.i == 1
+    assert tokens[1].right_edge.i == 1
-    assert example[1].head.i == 0
+    assert tokens[1].head.i == 0
 
-    assert example[2].n_lefts == 0
+    assert tokens[2].n_lefts == 0
-    assert example[2].n_rights == 2
+    assert tokens[2].n_rights == 2
-    assert example[2].left_edge.i == 2
+    assert tokens[2].left_edge.i == 2
-    assert example[2].right_edge.i == 4
+    assert tokens[2].right_edge.i == 4
-    assert example[2].head.i == 0
+    assert tokens[2].head.i == 0
 
-    assert example[3].n_lefts == 0
+    assert tokens[3].n_lefts == 0
-    assert example[3].n_rights == 0
+    assert tokens[3].n_rights == 0
-    assert example[3].left_edge.i == 3
+    assert tokens[3].left_edge.i == 3
-    assert example[3].right_edge.i == 3
+    assert tokens[3].right_edge.i == 3
-    assert example[3].head.i == 2
+    assert tokens[3].head.i == 2
 
-    assert example[4].n_lefts == 0
+    assert tokens[4].n_lefts == 0
-    assert example[4].n_rights == 0
+    assert tokens[4].n_rights == 0
-    assert example[4].left_edge.i == 4
+    assert tokens[4].left_edge.i == 4
-    assert example[4].right_edge.i == 4
+    assert tokens[4].right_edge.i == 4
-    assert example[4].head.i == 2
+    assert tokens[4].head.i == 2
 
     # left branching
-    example = EN.tokenizer.tokens_from_list(u"a b c d e".split(' '))
+    transition = ['S', 'S', 'S', 'L-nsubj','L-nsubj','L-nsubj', 'L-nsubj']
-    apply_transition_sequence(EN, example, ['S', 'S', 'S', 'L-nsubj','L-nsubj','L-nsubj', 'L-nsubj'])
+    tokens = en_tokenizer(text)
+    apply_transition_sequence(en_parser, tokens, transition)
 
-    assert example[0].n_lefts == 0
+    assert tokens[0].n_lefts == 0
-    assert example[0].n_rights == 0
+    assert tokens[0].n_rights == 0
-    assert example[0].left_edge.i == 0
+    assert tokens[0].left_edge.i == 0
-    assert example[0].right_edge.i == 0
+    assert tokens[0].right_edge.i == 0
-    assert example[0].head.i == 4
+    assert tokens[0].head.i == 4
 
-    assert example[1].n_lefts == 0
+    assert tokens[1].n_lefts == 0
-    assert example[1].n_rights == 0
+    assert tokens[1].n_rights == 0
-    assert example[1].left_edge.i == 1
+    assert tokens[1].left_edge.i == 1
-    assert example[1].right_edge.i == 1
+    assert tokens[1].right_edge.i == 1
-    assert example[1].head.i == 4
+    assert tokens[1].head.i == 4
 
-    assert example[2].n_lefts == 0
+    assert tokens[2].n_lefts == 0
-    assert example[2].n_rights == 0
+    assert tokens[2].n_rights == 0
-    assert example[2].left_edge.i == 2
+    assert tokens[2].left_edge.i == 2
-    assert example[2].right_edge.i == 2
+    assert tokens[2].right_edge.i == 2
-    assert example[2].head.i == 4
+    assert tokens[2].head.i == 4
 
-    assert example[3].n_lefts == 0
+    assert tokens[3].n_lefts == 0
-    assert example[3].n_rights == 0
+    assert tokens[3].n_rights == 0
-    assert example[3].left_edge.i == 3
+    assert tokens[3].left_edge.i == 3
-    assert example[3].right_edge.i == 3
+    assert tokens[3].right_edge.i == 3
-    assert example[3].head.i == 4
+    assert tokens[3].head.i == 4
 
-    assert example[4].n_lefts == 4
+    assert tokens[4].n_lefts == 4
-    assert example[4].n_rights == 0
+    assert tokens[4].n_rights == 0
-    assert example[4].left_edge.i == 0
+    assert tokens[4].left_edge.i == 0
-    assert example[4].right_edge.i == 4
+    assert tokens[4].right_edge.i == 4
-    assert example[4].head.i == 4
+    assert tokens[4].head.i == 4