From 8573099c4be03e134e22d61eb3b5e2c65dfc78ae Mon Sep 17 00:00:00 2001
From: Bernie Hackett <bernie.hackett@gmail.com>
Date: Sun, 24 Jun 2018 20:11:35 -0700
Subject: [PATCH] Move time consuming utf8 tests out of the main suite

---
 test/test_bson.py         | 63 --------------------------------
 test/unicode/test_utf8.py | 76 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 76 insertions(+), 63 deletions(-)
 create mode 100644 test/unicode/test_utf8.py

diff --git a/test/test_bson.py b/test/test_bson.py
index e4b6a7491..b5291c5ad 100644
--- a/test/test_bson.py
+++ b/test/test_bson.py
@@ -650,69 +650,6 @@ class TestBSON(unittest.TestCase):
             z = {iso8859_bytes: "hello"}
             self.assertRaises(InvalidStringData, BSON.encode, z)
 
-    # Verify that python and bson have the same understanding of
-    # legal utf-8 if the first byte is 0xf4 (244)
-    def _assert_same_utf8_validation(self, data):
-        try:
-            data.decode('utf-8')
-            py_is_legal = True
-        except UnicodeDecodeError:
-            py_is_legal = False
-
-        try:
-            BSON.encode({'x': data})
-            bson_is_legal = True
-        except InvalidStringData:
-            bson_is_legal = False
-
-        self.assertEqual(py_is_legal, bson_is_legal, data)
-
-    @unittest.skipIf(PY3, "python3 has strong separation between bytes/unicode")
-    def test_legal_utf8_full_coverage(self):
-        # this tests takes 400 seconds. Which is too long to run each time.
-        # However it is the only one which covers all possible bit combinations
-        # in the 244 space.
-        b1 = chr(0xf4)
-
-        for b2 in map(chr, range(255)):
-            m2 = b1 + b2
-            self._assert_same_utf8_validation(m2)
-
-            for b3 in map(chr, range(255)):
-                m3 = m2 + b3
-                self._assert_same_utf8_validation(m3)
-
-                for b4 in map(chr, range(255)):
-                    m4 = m3 + b4
-                    self._assert_same_utf8_validation(m4)
-
-    # In python3:
-    #  - 'bytes' are not checked with isLegalutf
-    #  - 'unicode' I cannot create unicode objects with invalid utf8, since it
-    #    would result in non valid code-points.
-    @unittest.skipIf(PY3, "python3 has strong separation between bytes/unicode")
-    def test_legal_utf8_few_samples(self):
-        good_samples = [
-            '\xf4\x80\x80\x80',
-            '\xf4\x8a\x80\x80',
-            '\xf4\x8e\x80\x80',
-            '\xf4\x81\x80\x80',
-        ]
-
-        for data in good_samples:
-            self._assert_same_utf8_validation(data)
-
-        bad_samples = [
-            '\xf4\x00\x80\x80',
-            '\xf4\x3a\x80\x80',
-            '\xf4\x7f\x80\x80',
-            '\xf4\x90\x80\x80',
-            '\xf4\xff\x80\x80',
-        ]
-
-        for data in bad_samples:
-            self._assert_same_utf8_validation(data)
-
     def test_null_character(self):
         doc = {"a": "\x00"}
         self.assertEqual(doc, BSON.encode(doc).decode())
diff --git a/test/unicode/test_utf8.py b/test/unicode/test_utf8.py
new file mode 100644
index 000000000..d66515d0a
--- /dev/null
+++ b/test/unicode/test_utf8.py
@@ -0,0 +1,76 @@
+import sys
+
+sys.path[0:0] = [""]
+
+from bson import BSON
+from bson.errors import InvalidStringData
+from bson.py3compat import PY3
+from test import unittest
+
+class TestUTF8(unittest.TestCase):
+
+    # Verify that python and bson have the same understanding of
+    # legal utf-8 if the first byte is 0xf4 (244)
+    def _assert_same_utf8_validation(self, data):
+        try:
+            data.decode('utf-8')
+            py_is_legal = True 
+        except UnicodeDecodeError:
+            py_is_legal = False
+
+        try:
+            BSON.encode({'x': data})
+            bson_is_legal = True 
+        except InvalidStringData:
+            bson_is_legal = False
+
+        self.assertEqual(py_is_legal, bson_is_legal, data)
+
+    @unittest.skipIf(PY3, "python3 has strong separation between bytes/unicode")
+    def test_legal_utf8_full_coverage(self):
+        # This test takes 400 seconds. Which is too long to run each time.
+        # However it is the only one which covers all possible bit combinations
+        # in the 244 space.
+        b1 = chr(0xf4)
+
+        for b2 in map(chr, range(255)):
+            m2 = b1 + b2 
+            self._assert_same_utf8_validation(m2)
+
+            for b3 in map(chr, range(255)):
+                m3 = m2 + b3 
+                self._assert_same_utf8_validation(m3)
+
+                for b4 in map(chr, range(255)):
+                    m4 = m3 + b4 
+                    self._assert_same_utf8_validation(m4)
+
+    # In python3:
+    #  - 'bytes' are not checked with isLegalutf
+    #  - 'unicode' We cannot create unicode objects with invalid utf8, since it
+    #    would result in non valid code-points.
+    @unittest.skipIf(PY3, "python3 has strong separation between bytes/unicode")
+    def test_legal_utf8_few_samples(self):
+        good_samples = [
+            '\xf4\x80\x80\x80',
+            '\xf4\x8a\x80\x80',
+            '\xf4\x8e\x80\x80',
+            '\xf4\x81\x80\x80',
+        ]
+
+        for data in good_samples:
+            self._assert_same_utf8_validation(data)
+
+        bad_samples = [
+            '\xf4\x00\x80\x80',
+            '\xf4\x3a\x80\x80',
+            '\xf4\x7f\x80\x80',
+            '\xf4\x90\x80\x80',
+            '\xf4\xff\x80\x80',
+        ]
+
+        for data in bad_samples:
+            self._assert_same_utf8_validation(data)
+
+if __name__ == "__main__":
+    unittest.main()