Add parsing for non 1-minute data to UO SRML parser

docs/sphinx/source/whatsnew/v0.6.2.rst

@@ -51,6 +51,8 @@ Bug fixes
   :py:func:`~pvlib.irradiance.klucher` and
   :py:func:`~pvlib.pvsystem.calcparams_desoto`. (:issue:`698`)
 * Fix :py:class:`~pvlib.forecast.NDFD` model by updating variables.
+* Fix :py:func:`~pvlib.iotools.srml.format_index` to parse non
+  one-minute data correctly. (:issue:`709`)
 
 
 Testing
@@ -67,3 +69,4 @@ Contributors
 * Kevin Anderson (:ghuser:`kevinsa5`)
 * :ghuser:`bentomlinson`
 * Jonathan Gaffiot (:ghuser:`jgaffiot`)
+* Leland Boeman (:ghuser: `lboeman`)

pvlib/iotools/srml.py

@@ -42,11 +42,11 @@ def read_srml(filename):
 
     Notes
     -----
-    The time index is shifted back one minute to account for 2400 hours,
-    and to avoid time parsing errors on leap years. The returned data
-    values should be understood to occur during the interval from the
-    time of the row until the time of the next row. This is consistent
-    with pandas' default labeling behavior.
+    The time index is shifted back by one interval to account for the
+    daily endtime of 2400, and to avoid time parsing errors on leap
+    years. The returned data values should be understood to occur
+    during the interval from the time of the row until the time of the
+    next row. This is consistent with pandas' default labeling behavior.
 
     See SRML's `Archival Files`_ page for more information.
 
@@ -134,11 +134,17 @@ def format_index(df):
     year = int(df.columns[1])
     df_doy = df[df.columns[0]]
     # Times are expressed as integers from 1-2400, we convert to 0-2359 by
-    # subracting one and then correcting the minutes at each former hour.
-    df_time = df[df.columns[1]] - 1
-    fifty_nines = df_time % 100 == 99
-    times = df_time.where(~fifty_nines, df_time - 40)
-
+    # subracting the length of one interval and then correcting the times
+    # at each former hour. interval_length is determined by taking the
+    # difference of the first two rows of the time column.
+    interval_length = int(df[df.columns[1]][:2].diff()[1])
+    df_time = df[df.columns[1]] - interval_length
+    if interval_length == 100:
+        # Hourly files do not require fixing the former hour timestamps.
+        times = df_time
+    else:
+        old_hours = df_time % 100 == (100 - interval_length)
+        times = df_time.where(~old_hours, df_time - 40)
     times = times.apply(lambda x: '{:04.0f}'.format(x))
     doy = df_doy.apply(lambda x: '{:03.0f}'.format(x))
     dts = pd.to_datetime(str(year) + '-' + doy + '-' + times,
@@ -161,14 +167,30 @@ def read_srml_month_from_solardat(station, year, month, filetype='PO'):
     month: int
         Month to request data for.
     filetype: string
-        SRML file type to gather. 'RO' and 'PO' are the
-        only minute resolution files.
+        SRML file type to gather. See notes for explanation.
 
     Returns
     -------
     data: pd.DataFrame
         One month of data from SRML.
 
+    Notes
+    -----
+    File types designate the time interval of a file and if it contains
+    raw or processed data. For instance, `RO` designates raw, one minute
+    data and `PO` designates processed one minute data. The availability
+    of file types varies between sites. Below is a table of file types
+    and their time intervals. See [1] for site information.
+
+    ============= ============ ==================
+    time interval raw filetype processed filetype
+    ============= ============ ==================
+    1 minute      RO           PO
+    5 minute      RF           PF
+    15 minute     RQ           PQ
+    hourly        RH           PH
+    ============= ============ ==================
+
     References
     ----------
     [1] University of Oregon Solar Radiation Measurement Laboratory

pvlib/test/test_srml.py

@@ -73,3 +73,35 @@ def test_read_srml_month_from_solardat():
     file_data = srml.read_srml(url)
     requested = srml.read_srml_month_from_solardat('EU', 2018, 1)
     assert file_data.equals(requested)
+
+
+@network
+@pytest.mark.parametrize('station, year, month, filetype', [
+    ('TW', 2019, 4, 'RQ'),
+])
+def test_15_minute_dt_index(
+        station, year, month, filetype):
+    data = srml.read_srml_month_from_solardat(station, year, month, filetype)
+    start = pd.Timestamp('{:04d}{:02d}01 00:00'.format(year, month))
+    start = start.tz_localize('Etc/GMT+8')
+    end = pd.Timestamp('{:04d}{:02d}30 23:45'.format(year, month))
+    end = end.tz_localize('Etc/GMT+8')
+    assert data.index[0] == start
+    assert data.index[-1] == end
+    assert (data.index[3::4].minute == 45).all()
+
+
+@network
+@pytest.mark.parametrize('station, year, month, filetype', [
+    ('CD', 1986, 4, 'PH'),
+])
+def test_hourly_dt_index(
+        station, year, month, filetype):
+    data = srml.read_srml_month_from_solardat(station, year, month, filetype)
+    start = pd.Timestamp('{:04d}{:02d}01 00:00'.format(year, month))
+    start = start.tz_localize('Etc/GMT+8')
+    end = pd.Timestamp('{:04d}{:02d}30 23:00'.format(year, month))
+    end = end.tz_localize('Etc/GMT+8')
+    assert data.index[0] == start
+    assert data.index[-1] == end
+    assert (data.index.minute == 0).all()