Modernise tests for merging spans and don't depend on models

spacy/tests/spans/test_merge.py

@@ -1,89 +1,117 @@
+# coding: utf-8
 from __future__ import unicode_literals
-from spacy.attrs import HEAD
+
+from ..util import get_doc
+
 import pytest
-import numpy
 
 
-def test_merge_tokens(EN):
-    tokens = EN(u'Los Angeles start.')
-    tokens.from_array([HEAD], numpy.asarray([[1, 1, 0, -1]], dtype='int32').T)
-    assert len(tokens) == 4
-    assert tokens[0].head.orth_ == 'Angeles'
-    assert tokens[1].head.orth_ == 'start'
-    tokens.merge(0, len('Los Angeles'), 'NNP', 'Los Angeles', 'GPE')
-    assert len(tokens) == 3
-    assert tokens[0].orth_ == 'Los Angeles'
-    assert tokens[0].head.orth_ == 'start'
+def test_spans_merge_tokens(en_tokenizer):
+    text = "Los Angeles start."
+    heads = [1, 1, 0, -1]
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads)
+    assert len(doc) == 4
+    assert doc[0].head.text == 'Angeles'
+    assert doc[1].head.text == 'start'
+    doc.merge(0, len('Los Angeles'), 'NNP', 'Los Angeles', 'GPE')
+    assert len(doc) == 3
+    assert doc[0].text == 'Los Angeles'
+    assert doc[0].head.text == 'start'
 
 
-@pytest.mark.models
-def test_merge_heads(EN):
-    tokens = EN(u'I found a pilates class near work.')
-    assert len(tokens) == 8
-    tokens.merge(tokens[3].idx, tokens[4].idx + len(tokens[4]), tokens[4].tag_,
-                 'pilates class', 'O')
-    assert len(tokens) == 7
-    assert tokens[0].head.i == 1
-    assert tokens[1].head.i == 1
-    assert tokens[2].head.i == 3
-    assert tokens[3].head.i == 1
-    assert tokens[4].head.i in [1, 3]
-    assert tokens[5].head.i == 4
+def test_spans_merge_heads(en_tokenizer):
+    text = "I found a pilates class near work."
+    heads = [1, 0, 2, 1, -3, -1, -1, -6]
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads)
+
+    assert len(doc) == 8
+    doc.merge(doc[3].idx, doc[4].idx + len(doc[4]), doc[4].tag_, 'pilates class', 'O')
+    assert len(doc) == 7
+    assert doc[0].head.i == 1
+    assert doc[1].head.i == 1
+    assert doc[2].head.i == 3
+    assert doc[3].head.i == 1
+    assert doc[4].head.i in [1, 3]
+    assert doc[5].head.i == 4
 
 
+def test_span_np_merges(en_tokenizer):
+    text = "displaCy is a parse tool built with Javascript"
+    heads = [1, 0, 2, 1, -3, -1, -1, -1]
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads)
 
-@pytest.mark.models
-def test_np_merges(EN):
-    text = u'displaCy is a parse tool built with Javascript'
-    tokens = EN(text)
-    assert tokens[4].head.i == 1
-    tokens.merge(tokens[2].idx, tokens[4].idx + len(tokens[4]), u'NP', u'tool', u'O')
-    assert tokens[2].head.i == 1
-    tokens = EN('displaCy is a lightweight and modern dependency parse tree visualization tool built with CSS3 and JavaScript.')
-    
-    ents = [(e[0].idx, e[-1].idx + len(e[-1]), e.label_, e.lemma_)
-            for e in tokens.ents]
+    assert doc[4].head.i == 1
+    doc.merge(doc[2].idx, doc[4].idx + len(doc[4]), 'NP', 'tool', 'O')
+    assert doc[2].head.i == 1
+
+    text = "displaCy is a lightweight and modern dependency parse tree visualization tool built with CSS3 and JavaScript."
+    heads = [1, 0, 8, 3, -1, -2, 4, 3, 1, 1, -9, -1, -1, -1, -1, -2, -15]
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads)
+
+    ents = [(e[0].idx, e[-1].idx + len(e[-1]), e.label_, e.lemma_) for e in doc.ents]
     for start, end, label, lemma in ents:
-        merged = tokens.merge(start, end, label, lemma, label)
-        assert merged != None, (start, end, label, lemma) 
+        merged = doc.merge(start, end, label, lemma, label)
+        assert merged != None, (start, end, label, lemma)
 
 
-    tokens = EN(u'One test with entities like New York City so the ents list is not void')
+    text = "One test with entities like New York City so the ents list is not void"
+    heads = [1, 11, -1, -1, -1, 1, 1, -3, 4, 2, 1, 1, 0, -1, -2]
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads)
 
-    for span in tokens.ents:
-        merged = span.merge()
-        assert merged != None, (span.start, span.end, span.label_, span.lemma_) 
+    for span in doc.ents:
+        merged = doc.merge()
+        assert merged != None, (span.start, span.end, span.label_, span.lemma_)
 
-@pytest.mark.models
-def test_entity_merge(EN):
-    tokens = EN(u'Stewart Lee is a stand up comedian who lives in England and loves Joe Pasquale.\n')
-    assert(len(tokens) == 17)
-    for ent in tokens.ents:
+
+def test_spans_entity_merge(en_tokenizer):
+    text = "Stewart Lee is a stand up comedian who lives in England and loves Joe Pasquale.\n"
+    heads = [1, 1, 0, 1, 2, -1, -4, 1, -2, -1, -1, -3, -10, 1, -2, -13, -1]
+    tags = ['NNP', 'NNP', 'VBZ', 'DT', 'VB', 'RP', 'NN', 'WP', 'VBZ', 'IN', 'NNP', 'CC', 'VBZ', 'NNP', 'NNP', '.', 'SP']
+    ents = [('Stewart Lee', 'PERSON', 0, 2), ('England', 'GPE', 10, 11), ('Joe Pasquale', 'PERSON', 13, 15)]
+
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads, tags=tags, ents=ents)
+    assert len(doc) == 17
+    for ent in doc.ents:
         label, lemma, type_ = (ent.root.tag_, ent.root.lemma_, max(w.ent_type_ for w in ent))
         ent.merge(label, lemma, type_)
     # check looping is ok
-    assert(len(tokens) == 15)
+    assert len(doc) == 15
 
 
-@pytest.mark.models
-def test_sentence_update_after_merge(EN):
-    tokens = EN(u'Stewart Lee is a stand up comedian. He lives in England and loves Joe Pasquale.')
-    sent1, sent2 = list(tokens.sents)
+def test_spans_sentence_update_after_merge(en_tokenizer):
+    text = "Stewart Lee is a stand up comedian. He lives in England and loves Joe Pasquale."
+    heads = [1, 1, 0, 1, 2, -1, -4, -5, 1, 0, -1, -1, -3, -4, 1, -2, -7]
+    deps = ['compound', 'nsubj', 'ROOT', 'det', 'amod', 'prt', 'attr',
+            'punct', 'nsubj', 'ROOT', 'prep', 'pobj', 'cc', 'conj',
+            'compound', 'dobj', 'punct']
+
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads, deps=deps)
+    sent1, sent2 = list(doc.sents)
     init_len = len(sent1)
     init_len2 = len(sent2)
-    merge_me = tokens[0:2]
-    merge_me.merge(u'none', u'none', u'none')
-    merge_me2 = tokens[-2:]
-    merge_me2.merge(u'none', u'none', u'none')
-    assert(len(sent1) == init_len - 1)
-    assert(len(sent2) == init_len2 - 1)
+    doc[0:2].merge('none', 'none', 'none')
+    doc[-2:].merge('none', 'none', 'none')
+    assert len(sent1) == init_len - 1
+    assert len(sent2) == init_len2 - 1
 
 
-@pytest.mark.models
-def test_subtree_size_check(EN):
-    tokens = EN(u'Stewart Lee is a stand up comedian who lives in England and loves Joe Pasquale')
-    sent1 = list(tokens.sents)[0]
+def test_spans_subtree_size_check(en_tokenizer):
+    text = "Stewart Lee is a stand up comedian who lives in England and loves Joe Pasquale"
+    heads = [1, 1, 0, 1, 2, -1, -4, 1, -2, -1, -1, -3, -10, 1, -2]
+    deps = ['compound', 'nsubj', 'ROOT', 'det', 'amod', 'prt', 'attr',
+            'nsubj', 'relcl', 'prep', 'pobj', 'cc', 'conj', 'compound',
+            'dobj']
+
+    tokens = en_tokenizer(text)
+    doc = get_doc(tokens.vocab, [t.text for t in tokens], heads=heads, deps=deps)
+    sent1 = list(doc.sents)[0]
     init_len = len(list(sent1.root.subtree))
-    merge_me = tokens[0:2]
-    merge_me.merge(u'none', u'none', u'none')
-    assert(len(list(sent1.root.subtree)) == init_len - 1)
+    doc[0:2].merge('none', 'none', 'none')
+    assert len(list(sent1.root.subtree)) == init_len - 1