Merge pull request #36 from nextstrain/noisify-genebank-transform-35

Noisify genebank transform

tests/transform-genbank-location/transform-genbank-location.t

@@ -0,0 +1,30 @@
+Verify behavior of `transform-genbank-location` around prescence/abscence of
+`database` and `location` fields.
+
+If `location` field is present, transform it.
+
+  $ echo '{"database":"GenBank", "location": "USA:Oregon, Salem" }' \
+  >  | $TESTDIR/../../transform-genbank-location
+  {"database":"GenBank","location":"Salem","country":"USA","division":"Oregon"}
+
+If `database` field is absent, complain.
+
+  $ echo '{"location": "USA:Oregon, Salem" }' \
+  >  | $TESTDIR/../../transform-genbank-location
+  Record must contain `database` field to use `transform-genbank-location.`
+  {"location":"USA:Oregon, Salem"}
+
+If `database` field has unsupported value, complain.
+
+  $ echo '{"database": "unsupported", "location": "USA:Oregon, Salem" }' \
+  >  | $TESTDIR/../../transform-genbank-location
+  Database value of unsupported not supported for `transform-genbank-location`; must be "GenBank" or "RefSeq".
+  {"database":"unsupported","location":"USA:Oregon, Salem"}
+
+
+If `location` field is absent, complain.
+
+  $ echo '{"database": "GenBank" }' \
+  >  | $TESTDIR/../../transform-genbank-location
+  `transform-genbank-location` requires a `location` field; this record does not have one.
+  {"database":"GenBank"}

transform-genbank-location

@@ -7,14 +7,23 @@ GenBank by verifying that the 'database' field has a value of "GenBank" or "RefS
 Outputs the modified record to stdout.
 """
 import json
-from sys import stdin, stdout
+from sys import stdin, stderr, stdout
 
 
 def parse_location(record: dict) -> dict:
     # Expected pattern for the location field is "<country_value>[:<region>][, <locality>]"
     # See GenBank docs for their "country" field:
     # https://www.ncbi.nlm.nih.gov/genbank/collab/country/
-    geographic_data = record['location'].split(':')
+    location_field = record.get("location", "")
+    if not location_field:
+        print(
+            "`transform-genbank-location` requires a `location` field; this record does not have one.",
+            file=stderr,
+        )
+        # bail early because we're not gonna make any changes
+        return record
+
+    geographic_data = location_field.split(':')
 
     country = geographic_data[0]
     division = ''
@@ -38,6 +47,13 @@ if __name__ == '__main__':
         database = record.get('database', '')
         if database in {'GenBank', 'RefSeq'}:
             parse_location(record)
+        else:
+            if database:
+                error_msg = f"""Database value of {database} not supported for `transform-genbank-location`; must be "GenBank" or "RefSeq"."""
+            else:
+                error_msg = "Record must contain `database` field to use `transform-genbank-location.`"
+
+            print(error_msg, file=stderr)
 
         json.dump(record, stdout, allow_nan=False, indent=None, separators=',:')
         print()