fix: incorporate changes from nextstrain/mpox#146

ingest/bin/fetch-from-genbank

@@ -6,14 +6,15 @@
 set -euo pipefail
 
 bin="$(dirname "$0")"
-TAXID="${1:?NCBI taxon id is required.}"
+
 
 main() {
-    fetch "$TAXID" | "$bin"/csv-to-ndjson
+    local ncbi_taxon_id="${1:?NCBI taxon id is required.}"
+    fetch "$ncbi_taxon_id" | "$bin"/csv-to-ndjson
 }
 
 fetch() {
-    curl "$("$bin"/genbank-url --taxonid "$1")" \
+    curl "$("$bin"/genbank-url --ncbi-taxon-id "$1")" \
         --fail --silent --show-error --http1.1 \
         --header 'User-Agent: https://github.com/nextstrain/dengue (hello@nextstrain.org)'
 }

ingest/bin/genbank-url

@@ -17,22 +17,19 @@ import argparse
 
 
 def parse_args():
-    """
-    Define parse args
-    """
     parser = argparse.ArgumentParser(
         description="Given an NCBI taxon ID, generate URL to download "
         "all viral sequences and their curated metadata from GenBank via NCBI Virus."
     )
     parser.add_argument(
-        "--taxonid",
+        "--ncbi-taxon-id",
         help="NCBI Taxon ID.",
+        default="12637",
         required=True
     )
     return parser.parse_args()
 
-
-def build_query_url(ncbi_id: str):
+def build_query_url(ncbi_taxon_id: str):
     """
     Generate URL to download all viral sequences and their curated metadata
     from GenBank via NCBI Virus.
@@ -42,7 +39,7 @@ def build_query_url(ncbi_id: str):
         # Search criteria
         'fq': [
             '{!tag=SeqType_s}SeqType_s:("Nucleotide")', # Nucleotide sequences (as opposed to protein)
-            f'VirusLineageId_ss:({ncbi_id})', # NCBI Taxon id for virus
+            f'VirusLineageId_ss:({ncbi_taxon_id})', # NCBI Taxon id for virus
             'Division_s:("VRL")', # Restrict to viral sequences, avoid PAT
             '{!tag=SLen_i}SLen_i:([5000 TO 15000])', # Longer than 5K bp, shorter than 15k bp
             #'{!tag=UpdateDate_dt}UpdateDate_dt:([2022-01-01T00:00:00.00Z TO 2022-12-01T00:00:00.00Z ])', # recent
@@ -90,18 +87,13 @@ def build_query_url(ncbi_id: str):
         # to be nice.
         'email': 'hello@nextstrain.org',
     }
-    query = urlencode(params, doseq=True, encoding="utf-8")
+    query = urlencode(params, doseq = True, encoding = "utf-8")
 
     print(f"{endpoint}?{query}")
 
-
 def main():
-    """
-    Main method
-    """
     args = parse_args()
-    build_query_url(args.taxonid)
-
+    build_query_url(args.ncbi_taxon_id)
 
-if __name__ == "__main__":
+if __name__ == '__main__':
     main()