Merge pull request #84 from sanger-tol/check_ncbidatasets

Check that the JSON coming from `datasets` contains an assembly

bin/validate_datasets_json.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python3
+
+import argparse
+import json
+import sys
+
+
+def parse_args(args=None):
+    Description = "Verify the integrity of a JSON file coming from NCBI datasets"
+
+    parser = argparse.ArgumentParser(description=Description)
+    parser.add_argument("NCBI_SUMMARY_JSON", help="NCBI entry for this assembly (in JSON).")
+    parser.add_argument("--version", action="version", version="%(prog)s 1.0")
+    return parser.parse_args(args)
+
+
+def check_json(ncbi_summary):
+    with open(ncbi_summary) as file_in:
+        data = json.load(file_in)
+
+    assert "reports" in data
+
+
+def main(args=None):
+    args = parse_args(args)
+    check_json(args.NCBI_SUMMARY_JSON)
+
+
+if __name__ == "__main__":
+    sys.exit(main())

modules/local/ncbidatasets/summarygenome.nf

@@ -27,6 +27,8 @@ process NCBIDATASETS_SUMMARYGENOME {
         ${args} \\
         > ${prefix}.json
 
+    validate_datasets_json.py ${prefix}.json
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         ncbi-datasets-cli: \$(datasets --version | sed 's/^.*datasets version: //')