From 168fdc61cd717767851cf8efded9cd6e8a939fba Mon Sep 17 00:00:00 2001
From: Bernie Hackett <bernie@10gen.com>
Date: Fri, 14 Dec 2012 13:21:32 -0800
Subject: [PATCH] Fix accuracy issues in datetime decoding PYTHON-392

---
 bson/__init__.py    | 10 +++++++---
 bson/_cbsonmodule.c | 33 +++++++++++++++++++++++++++++++--
 test/test_bson.py   |  8 ++++----
 3 files changed, 42 insertions(+), 9 deletions(-)

diff --git a/bson/__init__.py b/bson/__init__.py
index 4f9cae710..02c8452bb 100644
--- a/bson/__init__.py
+++ b/bson/__init__.py
@@ -217,11 +217,15 @@ def _get_boolean(data, position, as_class, tz_aware, uuid_subtype):
 
 
 def _get_date(data, position, as_class, tz_aware, uuid_subtype):
-    seconds = float(struct.unpack("<q", data[position:position + 8])[0]) / 1000.0
+    millis = struct.unpack("<q", data[position:position + 8])[0]
+    diff = millis % 1000
+    seconds = (millis - diff) / 1000
     position += 8
     if tz_aware:
-        return EPOCH_AWARE + datetime.timedelta(seconds=seconds), position
-    return EPOCH_NAIVE + datetime.timedelta(seconds=seconds), position
+        dt = EPOCH_AWARE + datetime.timedelta(seconds=seconds)
+    else:
+        dt = EPOCH_NAIVE + datetime.timedelta(seconds=seconds)
+    return dt.replace(microsecond=diff * 1000), position
 
 
 def _get_code(data, position, as_class, tz_aware, uuid_subtype):
diff --git a/bson/_cbsonmodule.c b/bson/_cbsonmodule.c
index 390e92bb8..0e50f1d33 100644
--- a/bson/_cbsonmodule.c
+++ b/bson/_cbsonmodule.c
@@ -117,8 +117,37 @@ static int _write_element_to_buffer(PyObject* self, buffer_t buffer, int type_by
 
 /* Date stuff */
 static PyObject* datetime_from_millis(long long millis) {
-    int microseconds = (millis % 1000) * 1000;
-    Time64_T seconds = millis / 1000;
+    /* To encode a datetime instance like datetime(9999, 12, 31, 23, 59, 59, 999999)
+     * we follow these steps:
+     * 1. Calculate a timestamp in seconds:       253402300799
+     * 2. Multiply that by 1000:                  253402300799000
+     * 3. Add in microseconds divided by 1000     253402300799999
+     *
+     * (Note: BSON doesn't support microsecond accuracy, hence the rounding.)
+     *
+     * To decode we could do:
+     * 1. Get seconds: timestamp / 1000:          253402300799
+     * 2. Get micros: (timestamp % 1000) * 1000:  999000
+     * Resulting in datetime(9999, 12, 31, 23, 59, 59, 999000) -- the expected result
+     *
+     * Now what if the we encode (1, 1, 1, 1, 1, 1, 111111)?
+     * 1. and 2. gives:                           -62135593139000
+     * 3. Gives us:                               -62135593138889
+     *
+     * Now decode:
+     * 1. Gives us:                               -62135593138
+     * 2. Gives us:                               -889000
+     * Resulting in datetime(1, 1, 1, 1, 1, 2, 15888216) -- an invalid result
+     *
+     * If instead to decode we do:
+     * diff = ((millis % 1000) + 1000) % 1000:    111
+     * seconds = (millis - diff) / 1000:          -62135593139
+     * micros = diff * 1000                       111000
+     * Resulting in datetime(1, 1, 1, 1, 1, 1, 111000) -- the expected result
+     */
+    int diff = (int)(((millis % 1000) + 1000) % 1000);
+    int microseconds = diff * 1000;
+    Time64_T seconds = (millis - diff) / 1000;
     struct TM timeinfo;
     gmtime64_r(&seconds, &timeinfo);
 
diff --git a/test/test_bson.py b/test/test_bson.py
index c4612bb1e..2d93b89a3 100644
--- a/test/test_bson.py
+++ b/test/test_bson.py
@@ -224,20 +224,20 @@ class TestBSON(unittest.TestCase):
 
     def test_datetime_encode_decode(self):
         # Negative timestamps
-        dt1 = datetime.datetime(1, 1, 1, 1, 1, 1)
+        dt1 = datetime.datetime(1, 1, 1, 1, 1, 1, 111000)
         dt2 = BSON.encode({"date": dt1}).decode()["date"]
         self.assertEqual(dt1, dt2)
 
-        dt1 = datetime.datetime(1959, 6, 25, 12, 16, 59)
+        dt1 = datetime.datetime(1959, 6, 25, 12, 16, 59, 999000)
         dt2 = BSON.encode({"date": dt1}).decode()["date"]
         self.assertEqual(dt1, dt2)
 
         # Positive timestamps
-        dt1 = datetime.datetime(9999, 12, 31, 23, 59, 59)
+        dt1 = datetime.datetime(9999, 12, 31, 23, 59, 59, 999000)
         dt2 = BSON.encode({"date": dt1}).decode()["date"]
         self.assertEqual(dt1, dt2)
 
-        dt1 = datetime.datetime(2011, 6, 14, 10, 47, 53)
+        dt1 = datetime.datetime(2011, 6, 14, 10, 47, 53, 444000)
         dt2 = BSON.encode({"date": dt1}).decode()["date"]
         self.assertEqual(dt1, dt2)