curate-andersen-lab-data: Ensure dates are valid

New data now include collection dates that are the string
"missing" or date ranges (e.g. "2022-04-22/2022-04-24").
Ensure that the returned date is a valid date that can be used to
construct the strain name.

Resolves <#129>

ingest/build-configs/ncbi/bin/curate-andersen-lab-data

@@ -68,15 +68,35 @@ def create_new_record(anderson_record: dict) -> dict:
 
 def use_date_when_available(andersen_record: dict) -> str:
     """
-    Give the old date field `Date` precedence since they are more specific
+    Give the old date field `Date` precedence since they are more specific.
+
+    If using the new date field `Collection_Date`, then verify that it's a
+    valid date.
     """
     old_date_field = andersen_record.get("Date", "")
     old_date_uncertain = "NA" in old_date_field or "?" in old_date_field
 
     if old_date_field and not old_date_uncertain:
         return old_date_field
 
-    return andersen_record["Collection_Date"]
+    new_date = andersen_record["Collection_Date"]
+    if new_date.lower() == "missing":
+        new_date = "XXXX-XX-XX"
+
+    # Certain dates are date ranges, e.g. "2022-04-22/2022-04-24"
+    # Only keep the first date for our metadata
+    date_range_pattern = r"([\d]{4}-[\d]{2}-[\d]{2})\/[\d]{4}-[\d]{2}-[\d]{2}"
+    matches = re.match(date_range_pattern, new_date)
+    if matches:
+        new_date = matches.group(1)
+
+    try:
+        parse_date(new_date)
+    except ValueError as err:
+        print(f"WARNING: {err}", file=stderr)
+        new_date = "XXXX-XX-XX"
+
+    return new_date
 
 
 def parse_center_name(center_name: str) -> str:
@@ -101,24 +121,24 @@ def construct_strain_name(record: dict, sample_id: str) -> str:
     """
     host = record['host']
     country = record['country']
-    year = parse_year(record['date'])
+    year = str(parse_date(record['date']).year)
     strain = f"A/{host}/{country}/{sample_id}/{year}".replace(" ", "")
     return re.sub(r'[^\w\_\-\.\|\/]', '_', strain)
 
 
-def parse_year(date_string: str) -> str:
+def parse_date(date_string: str) -> datetime:
     """
-    Parse the year from the provided `date_string`
+    Parse the provided `date_string` as a datetime object.
     """
-    date_formats = ['%Y-%m-%d', '%Y']
+    date_formats = ['%Y-%m-%d', '%Y', 'XXXX-XX-XX']
     for date_format in date_formats:
         try:
             parsed_date = datetime.strptime(date_string, date_format)
-            return str(parsed_date.year)
+            return parsed_date
         except ValueError:
             continue
 
-    raise ValueError(f"Could not parse year from date string {date_string!r}")
+    raise ValueError(f"Could not parse date from date string {date_string!r}")
 
 
 if __name__ == '__main__':