From a4a74e22e251b03a76e0a1173392e9fee1293096 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ivan=20Koles=C3=A1r?= <ikol@statoil.com>
Date: Tue, 5 Feb 2019 15:19:35 +0100
Subject: [PATCH] Make windiris source dates optional

---
 camille/source/windiris.py    | 46 ++++++++++++++++++++---------------
 tests/source/test_windiris.py | 32 +++++++++++++++++-------
 2 files changed, 50 insertions(+), 28 deletions(-)

diff --git a/camille/source/windiris.py b/camille/source/windiris.py
index 0954ba62..3a468b82 100644
--- a/camille/source/windiris.py
+++ b/camille/source/windiris.py
@@ -5,6 +5,7 @@
 import re
 import pytz
 
+
 def _to_string(x):
     x = re.sub('[^A-Za-z0-9,.]', '', str(x))
     return '(' + x + ')'
@@ -18,25 +19,31 @@ def _sqlite(start_date,
             los_id=None,
             distance=None,
             status=None):
+
+    query_params = []
+    if start_date is not None:
+        query_params.append(' Timestamp >= "{}" '
+                            .format(str(start_date.astimezone(pytz.utc)
+                                        .replace(tzinfo=None))))
+    if end_date is not None:
+        query_params.append(' Timestamp < "{}" '
+                            .format(str(end_date.astimezone(pytz.utc)
+                                        .replace(tzinfo=None))))
+    if los_id is not None:
+        query_params.append(' "LOS Index" IN {} '.format(_to_string(los_id)))
+    if distance is not None:
+        query_params.append(' Distance IN {} '.format(_to_string(distance)))
+    if status is not None:
+        query_params.append(' "RWS Status" IN {} '.format(_to_string(status)))
+
     query = (
-        'SELECT * FROM ' + installation #nosec
-        + ' WHERE Timestamp >= :start AND Timestamp < :end'
-        + ( '' if los_id is None
-               else ' AND "LOS Index" IN {}'.format(_to_string(los_id)) )
-        + ( '' if distance is None
-               else ' AND Distance IN {}'.format(_to_string(distance)) )
-        + ( '' if status is None
-               else ' AND "RWS Status" IN {}'.format(_to_string(status)) )
+        'SELECT * FROM ' + _to_string(installation) #nosec
+        + ('' if len(query_params) == 0
+           else ' WHERE ' + 'AND'.join(query_params))
         + ';'
     )
 
     df = pd.read_sql_query(query, connection,
-                           params={
-                                'start': str(start_date.astimezone(pytz.utc)
-                                             .replace(tzinfo=None)),
-                                'end': str(end_date.astimezone(pytz.utc)
-                                           .replace(tzinfo=None))
-                           },
                            index_col='Timestamp',
                            parse_dates={
                                 'Timestamp': {'utc': True}
@@ -107,15 +114,12 @@ def windiris(root, tzinfo=pytz.utc):
         raise ValueError('{} is not a directory'.format(root))
 
     def windiris_internal(installation,
-                          start_date,
-                          end_date,
+                          start_date=None,
+                          end_date=None,
                           los_id=None,
                           distance=None,
                           status=None):
 
-        if start_date.tzinfo is None or end_date.tzinfo is None:
-            raise ValueError('dates must be timezone aware')
-
         f = os.path.join(root, installation, installation + '_rtd.db' )
 
         if not os.path.isfile(f):
@@ -123,6 +127,10 @@ def windiris_internal(installation,
 
         conn = sqlite3.connect(f)
 
+        if (start_date is not None and start_date.tzinfo is None) or \
+                (end_date is not None and end_date.tzinfo is None):
+            raise ValueError('dates must be timezone aware')
+
         return _sqlite(start_date,
                        end_date,
                        conn,
diff --git a/tests/source/test_windiris.py b/tests/source/test_windiris.py
index 55c80bd5..4c76f023 100644
--- a/tests/source/test_windiris.py
+++ b/tests/source/test_windiris.py
@@ -6,18 +6,17 @@
 
 wi = windiris('tests/test_data/windiris')
 
-def test_load_all_data():
-    s = datetime(2017, 12, 17, tzinfo=utc)
-    e = datetime(2018, 10, 23, tzinfo=utc)
+all_radial_windspeed = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+                        2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+                        3, 3, 3, 3, 3, 3]
+db_start_datetime = datetime(2017, 12, 17, tzinfo=utc)
+db_end_datetime = datetime(2018, 10, 23, tzinfo=utc)
 
-    df = wi('inst2', s, e)
+def test_load_all_data():
+    df = wi('inst2', db_start_datetime, db_end_datetime)
 
     assert df.shape[0] == 27
-    assert (
-                df.radial_windspeed == [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-                                         2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
-                                         3, 3, 3, 3, 3, 3 ]
-           ).all()
+    assert (df.radial_windspeed == all_radial_windspeed).all()
 
 
 def test_load_one_day():
@@ -138,3 +137,18 @@ def test_no_time_zone():
     with pytest.raises(ValueError) as exc:
         wi('inst2', ntz_date, tz_date)
     assert 'timezone aware' in str(exc)
+
+def test_load_all_data_by_no_date():
+    df_no_dates = wi('inst2')
+
+    assert (df_no_dates.radial_windspeed == all_radial_windspeed).all()
+
+def test_load_data_without_start_date():
+    df_no_start_date = wi('inst2', end_date=db_end_datetime)
+
+    assert (df_no_start_date.radial_windspeed == all_radial_windspeed).all()
+
+def test_load_data_without_end_date():
+    df_no_end_date = wi('inst2', start_date=db_start_datetime)
+
+    assert (df_no_end_date.radial_windspeed == all_radial_windspeed).all()