Make time span parsing a util

And add a test.
Fixes #76

network_wrangler/TransitNetwork.py

@@ -16,7 +16,7 @@
 from partridge.gtfs import Feed
 
 from .Logger import WranglerLogger
-
+from .Utils import parse_time_spans
 
 class TransitNetwork(object):
     """
@@ -158,14 +158,7 @@ def select_transit_features(self, selection: dict) -> pd.Series:
 
         # If a time key exists, filter trips using frequency table
         if selection.get('time') is not None:
-            # If time is given without seconds, add 00
-            if len(selection['time'][0]) <= 5:
-                selection['time'] = [i + ':00' for i in selection['time']]
-
-            # Convert times to seconds from midnight (Partride's time storage)
-            for i, val in enumerate(selection['time']):
-                h, m, s = val.split(":")
-                selection['time'][i] = int(h) * 3600 + int(m) * 60 + int(s)
+            selection['time'] = parse_time_spans(selection['time'])
 
             # Filter freq to trips in selection
             freq = freq[freq.trip_id.isin(trips['trip_id'])]

network_wrangler/Utils.py

@@ -1,5 +1,5 @@
 import pandas as pd
-
+from .Logger import WranglerLogger
 
 def point_df_to_geojson(df: pd.DataFrame, properties: list):
     """
@@ -66,3 +66,42 @@ def make_slug(text, delimiter: str = "_"):
 
     text = re.sub("[,.;@#?!&$']+", "", text.lower())
     return re.sub("[\ ]+", delimiter, text)
+
+def parse_time_spans(times):
+    """
+    parse time spans into tuples of seconds from midnight
+    can also be used as an apply function for a pandas series
+    Parameters
+    -----------
+    times: tuple(string)
+
+    returns
+    --------
+    tuple(integer)
+      time span as seconds from midnight
+    """
+    try:
+        start_time, end_time = times
+    except:
+        WranglerLogger.error(
+            "ERROR: times should be a tuple or list of two strings"
+        )
+        raise ValueError()
+
+    start_time=start_time.strip()
+    end_time=end_time.strip()
+
+    # If time is given without seconds, add 00
+    if len(start_time) <= 5:
+        start_time += ':00'
+    if len(end_time) <= 5:
+        end_time += ':00'
+
+    # Convert times to seconds from midnight (Partride's time storage)
+    h0, m0, s0 = start_time.split(":")
+    start_time_sec = int(h0) * 3600 + int(m0) * 60 + int(s0)
+
+    h1, m1, s1 = end_time.split(":")
+    end_time_sec = int(h1) * 3600 + int(m1) * 60 + int(s1)
+
+    return (start_time_sec, end_time_sec)

tests/test_utils.py

@@ -1,8 +1,6 @@
 import os
 import pytest
 
-
-
 slug_test_list = [
     {'text':"I am a roadway", 'delim':"_", "answer": "i_am_a_roadway"},
     {'text':"I'm a roadway", 'delim':"_", "answer": "im_a_roadway"},
@@ -18,3 +16,26 @@ def test_get_slug(request, slug_test):
     print('From: {} \nTo: {}'.format(slug_test['text'],slug))
     print('Expected: {}'.format(slug_test['answer']))
     assert(slug == slug_test['answer'])
+
+@pytest.mark.travis
+@pytest.mark.menow
+def test_time_convert(request):
+    print("\n--Starting:", request.node.name)
+
+    time_tests =  [
+        ( ("00:00:00","00:00:10"), (0,10) ),
+        ( ("0:00","0:10:00") , (0,600) ),
+        ( ("01:02:03","01:02:23"), (3723,3743) ),
+        ( ("1:02","1:02:13") ,  (3720, 3733) ),
+        ( ("25:24:23","25:24:33"), (91463, 91473) ),
+        ( ("250:24:23","250:24:25"), (901463,901465) ) ]
+
+    from pandas import DataFrame
+    df = DataFrame(time_tests, columns=['time','time_results'])
+    print("Original Time Series",df)
+
+    from network_wrangler.Utils import parse_time_spans
+    df["time"]= df["time"].apply(parse_time_spans)
+    print("Result Time Series",df)
+    from pandas.testing import assert_series_equal
+    assert_series_equal(df["time"], df["time_results"], check_names=False)