From 637f78503666f3551c199ad5b264b03d7d61bcd0 Mon Sep 17 00:00:00 2001
From: Ines Montani <ines@ines.io>
Date: Thu, 5 Jan 2017 16:25:38 +0100
Subject: [PATCH] Add general sanity tests for all tokenizers

---
 spacy/tests/tokenizer/conftest.py       | 23 +++++++
 spacy/tests/tokenizer/test_tokenizer.py | 80 +++++++++++++++++++++++++
 2 files changed, 103 insertions(+)
 create mode 100644 spacy/tests/tokenizer/conftest.py
 create mode 100644 spacy/tests/tokenizer/test_tokenizer.py

diff --git a/spacy/tests/tokenizer/conftest.py b/spacy/tests/tokenizer/conftest.py
new file mode 100644
index 000000000..c8e340208
--- /dev/null
+++ b/spacy/tests/tokenizer/conftest.py
@@ -0,0 +1,23 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+import pytest
+
+from ...en import English
+from ...de import German
+from ...es import Spanish
+from ...it import Italian
+from ...fr import French
+from ...pt import Portuguese
+from ...nl import Dutch
+from ...sv import Swedish
+from ...hu import Hungarian
+
+
+LANGUAGES = [English, German, Spanish, Italian, French, Dutch, Swedish, Hungarian]
+
+
+@pytest.fixture(params=LANGUAGES)
+def tokenizer(request):
+    lang = request.param
+    return lang.Defaults.create_tokenizer()
diff --git a/spacy/tests/tokenizer/test_tokenizer.py b/spacy/tests/tokenizer/test_tokenizer.py
new file mode 100644
index 000000000..49bfdcb26
--- /dev/null
+++ b/spacy/tests/tokenizer/test_tokenizer.py
@@ -0,0 +1,80 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+import pytest
+
+
+def test_tokenizer_handles_no_word(tokenizer):
+    tokens = tokenizer("")
+    assert len(tokens) == 0
+
+
+@pytest.mark.parametrize('text', ["lorem"])
+def test_tokenizer_handles_single_word(tokenizer, text):
+    tokens = tokenizer(text)
+    assert tokens[0].text == text
+
+
+@pytest.mark.parametrize('text', ["lorem ipsum"])
+def test_tokenizer_handles_two_words(tokenizer, text):
+    tokens = tokenizer(text)
+    assert len(tokens) == 2
+    assert tokens[0].text != tokens[1].text
+
+
+@pytest.mark.parametrize('text', ["lorem  ipsum"])
+def test_tokenizer_splits_double_space(tokenizer, text):
+    tokens = tokenizer(text)
+    assert len(tokens) == 3
+    assert tokens[1].text == " "
+
+
+@pytest.mark.parametrize('text', ["lorem\nipsum"])
+def test_tokenizer_splits_newline(tokenizer, text):
+    tokens = tokenizer(text)
+    assert len(tokens) == 3
+    assert tokens[1].text == "\n"
+
+
+def test_tokenizer_handles_punct(tokenizer):
+    text = "Lorem, ipsum."
+    tokens = tokenizer(text)
+    assert len(tokens) == 4
+    assert tokens[0].text == "Lorem"
+    assert tokens[1].text == ","
+    assert tokens[2].text == "ipsum"
+    assert tokens[1].text != "Lorem"
+
+
+def test_tokenizer_handles_digits(tokenizer):
+    exceptions = ["hu"]
+    text = "Lorem ipsum: 1984."
+    tokens = tokenizer(text)
+
+    if tokens[0].lang_ not in exceptions:
+        assert len(tokens) == 5
+        assert tokens[0].text == "Lorem"
+        assert tokens[3].text == "1984"
+
+
+def test_tokenizer_handles_long_text(tokenizer):
+    text = """Lorem ipsum dolor sit amet, consectetur adipiscing elit
+
+Cras egestas orci non porttitor maximus.
+Maecenas quis odio id dolor rhoncus dignissim. Curabitur sed velit at orci ultrices sagittis. Nulla commodo euismod arcu eget vulputate.
+
+Phasellus tincidunt, augue quis porta finibus, massa sapien consectetur augue, non lacinia enim nibh eget ipsum. Vestibulum in bibendum mauris.
+
+"Nullam porta fringilla enim, a dictum orci consequat in." Mauris nec malesuada justo."""
+
+    tokens = tokenizer(text)
+    assert len(tokens) > 5
+
+
+def test_tokenizer_suspected_freeing_strings(tokenizer):
+    text1 = "Lorem dolor sit amet, consectetur adipiscing elit."
+    text2 = "Lorem ipsum dolor sit amet, consectetur adipiscing elit."
+    tokens1 = tokenizer(text1)
+    tokens2 = tokenizer(text2)
+    assert tokens1[0].text == "Lorem"
+    assert tokens2[0].text == "Lorem"