Merge pull request #62 from deardurham/61-midnight

Fix parsing midnight date time

ciprs_reader/parser/section/case_information.py

@@ -31,12 +31,16 @@ def clean(self, matches):
 
 class OffenseDateTime(CaseInformationParser):
 
-    pattern = [r"\s*Offense", r"Date/Time:\s*(?P<value>[\w/ :]+[AaPp][Mm])"]
+    pattern = [r"\s*Offense", r"Date\/Time:\s*(?P<value>[\w/ :]+([AaPp][Mm])?)"]
     section = ("Case Information", "Offense Date")
 
     def clean(self, matches):
         """Parse and convert to the date and time in ISO 8601 format"""
-        date = dt.datetime.strptime(matches["value"], "%m/%d/%Y %I:%M %p")
+        try:
+            date = dt.datetime.strptime(matches["value"], "%m/%d/%Y %I:%M %p")
+        except ValueError:
+            # fallback to parsing time like midnight, e.g. 05/17/2015 00:00
+            date = dt.datetime.strptime(matches["value"], "%m/%d/%Y %H:%M")
         matches["value"] = date.isoformat()
         return matches
 

requirements.txt

@@ -1,4 +1,4 @@
-pytest==6.2.2
+pytest==7.1.2
 pytest-cov==2.11.1
 pytest-pylint==0.18.0
 pre-commit==2.11.1

tests/parsers/test_case_information.py

@@ -17,6 +17,13 @@ def test_offense_date_time(report, state):
     assert matches["value"] == "2015-05-17T21:59:00"
 
 
+def test_offense_date_time_midnight(report, state):
+    string = "    Offense Date/Time: 05/17/2015 00:00   "
+    matches = case_information.OffenseDateTime(report, state).match(string)
+    assert matches is not None, "Regex match failed"
+    assert matches["value"] == "2015-05-17T00:00:00"
+
+
 @pytest.mark.parametrize(
     "expected,val",
     (

tests/test_records/expected_output/test_redacted_2.json

@@ -1430,6 +1430,7 @@
         },
         "Case Information": {
             "Case Status": "DISPOSED",
+            "Offense Date": "2005-03-04T00:00:00",
             "Arrest Date": "2005-12-03"
         },
         "Defendant": {
@@ -1463,4 +1464,4 @@
         "Superior Court Offense Information": [],
         "_meta": {}
     }
-]
+]