added _check_convert_dates including testcase (#65)

* added _check_convert_dates

* added testcase

HISTORY.rst

@@ -14,6 +14,7 @@ UNRELEASED
 * add distinction for Groepering (timeseries vs. extremes) to `ddlpy.locations()` dataframe in https://github.com/deltares/ddlpy/pull/49
 * drop `Tijdstip` column in `ddlpy.measurements()` output dataframe to avoid duplication with time index in https://github.com/deltares/ddlpy/pull/52 and https://github.com/deltares/ddlpy/pull/54
 * add `ddlpy.measurements_amount()` to retrieve the number of available measurements grouped by day/month/year in https://github.com/Deltares/ddlpy/pull/63
+* catch accidentally switched start/end dates in https://github.com/Deltares/ddlpy/pull/65
 
 0.1.0 (2019-01-03)
 ------------------

ddlpy/ddlpy.py

@@ -66,6 +66,21 @@ def locations():
     return merged.set_index("Code")
 
 
+def _check_convert_dates(start_date, end_date, return_str=True):
+    start_date = pd.Timestamp(start_date)
+    end_date = pd.Timestamp(end_date)
+
+    if start_date > end_date:
+        raise ValueError("start_date is larger than end_date")
+
+    if return_str:
+        start_date_str = pytz.UTC.localize(start_date).isoformat(timespec='milliseconds')
+        end_date_str = pytz.UTC.localize(end_date).isoformat(timespec='milliseconds')
+        return start_date_str, end_date_str
+    else:
+        return start_date, end_date
+
+
 def _get_request_dicts(location):
     aquometadata_dict = {
         "Eenheid": {"Code": location["Eenheid.Code"]},
@@ -94,11 +109,7 @@ def measurements_available(location, start_date, end_date):
     """
     endpoint = ENDPOINTS['check_observations_available']
 
-    start_date = pd.Timestamp(start_date)
-    end_date = pd.Timestamp(end_date)
-
-    start_date_str = pytz.UTC.localize(start_date).isoformat(timespec='milliseconds')
-    end_date_str = pytz.UTC.localize(end_date).isoformat(timespec='milliseconds')
+    start_date_str, end_date_str = _check_convert_dates(start_date, end_date, return_str=True)
 
     request_dicts = _get_request_dicts(location)
 
@@ -140,12 +151,8 @@ def measurements_amount(location, start_date, end_date, period="Jaar"):
         raise ValueError(f"period should be one of {accepted_period}, not '{period}'")
 
     endpoint = ENDPOINTS['collect_number_of_observations']
-
-    start_date = pd.Timestamp(start_date)
-    end_date = pd.Timestamp(end_date)
 
-    start_date_str = pytz.UTC.localize(start_date).isoformat(timespec='milliseconds')
-    end_date_str = pytz.UTC.localize(end_date).isoformat(timespec='milliseconds')
+    start_date_str, end_date_str = _check_convert_dates(start_date, end_date, return_str=True)
 
     request_dicts = _get_request_dicts(location)
 
@@ -264,9 +271,8 @@ def _measurements_slice(location, start_date, end_date):
     """get measurements for location, for the period start_date, end_date, use measurements instead"""
     endpoint = ENDPOINTS["collect_observations"]
 
-    start_date_str = pytz.UTC.localize(start_date).isoformat(timespec="milliseconds")
-    end_date_str = pytz.UTC.localize(end_date).isoformat(timespec="milliseconds")
-
+    start_date_str, end_date_str = _check_convert_dates(start_date, end_date, return_str=True)
+
     request_dicts = _get_request_dicts(location)
 
     request = {
@@ -295,8 +301,7 @@ def _measurements_slice(location, start_date, end_date):
 
 def measurements(location, start_date, end_date, clean_df=True):
     """return measurements for the given location and time window (start_date, end_date)"""
-    start_date = pd.Timestamp(start_date)
-    end_date = pd.Timestamp(end_date)
+    start_date, end_date = _check_convert_dates(start_date, end_date, return_str=False)
 
     measurements = []
 

tests/test_ddlpy.py

@@ -68,14 +68,6 @@ def test_measurements_noindex(location):
     assert measurements.shape[0] > 1
 
 
-def test_measurements_string(location):
-    """measurements for a location """
-    start_date = "1953-01-01"
-    end_date = "1953-04-01"
-    measurements = ddlpy.measurements(location, start_date=start_date, end_date=end_date)
-    assert measurements.shape[0] > 1
-
-
 def test_measurements_latest(location):
     """measurements for a location """
     latest = ddlpy.measurements_latest(location)
@@ -138,7 +130,7 @@ def test_measurements_remove_duplicates_nottoomuch(location):
 
 
 def test_simplify_dataframe(location):
-    start_date  = dt.datetime(2019,11,24)
+    start_date = dt.datetime(2019,11,24)
     end_date = dt.datetime(2019,12,5)
     meas_wathte = ddlpy.measurements(location, start_date=start_date, end_date=end_date)
     assert len(meas_wathte.columns) == 53
@@ -148,6 +140,37 @@ def test_simplify_dataframe(location):
     assert len(meas_simple.columns) == 2
 
 
+datetype_list = ["string", "pd.Timestamp", "dt.datetime", "mixed"]
+@pytest.mark.parametrize("datetype", datetype_list)
+def test_check_convert_dates(datetype):
+    if datetype == "string":
+        start_date = "1953-01-01"
+        end_date = "1953-04-01"
+    elif datetype == "pd.Timestamp":
+        start_date = pd.Timestamp("1953-01-01")
+        end_date = pd.Timestamp("1953-04-01")
+    elif datetype == "dt.datetime":
+        start_date = dt.datetime(1953,1,1)
+        end_date = dt.datetime(1953,4,1)
+    elif datetype == "mixed":
+        start_date = "1953-01-01"
+        end_date = dt.datetime(1953,4,1)
+
+    # assert output
+    start_date_out, end_date_out = ddlpy.ddlpy._check_convert_dates(start_date, end_date)
+    assert start_date_out=='1953-01-01T00:00:00.000+00:00'
+    assert end_date_out=='1953-04-01T00:00:00.000+00:00'
+
+
+def test_check_convert_wrongorder():
+    start_date = "1953-01-01"
+    end_date = "1953-04-01"
+
+    # assert output
+    with pytest.raises(ValueError):
+        start_date_out, end_date_out = ddlpy.ddlpy._check_convert_dates(end_date, start_date)
+
+
 def test_command_line_interface():
     """Test the CLI."""
     runner = CliRunner()