Implement ensembl-py logging

pipelines/nextflow/modules/genome_metadata/compare_genome_stats.nf

@@ -27,6 +27,6 @@ process COMPARE_GENOME_STATS {
 
     script:
         """
-        genome_stats_compare --ncbi $ncbi_stats --core $core_stats > diff_stats.json
+        genome_stats_compare --ncbi $ncbi_stats --core $core_stats --debug > diff_stats.json
         """
 }

pyproject.toml

@@ -48,7 +48,7 @@ dependencies = [
     "bcbio-gff == 0.7.0",
     "biopython == 1.81",
     "ensembl-hive @ git+https://github.com/Ensembl/ensembl-hive.git",
-    "ensembl-py @ git+https://github.com/Ensembl/ensembl-py.git",  # minimum v1.2.0
+    "ensembl-py @ git+https://github.com/Ensembl/ensembl-py.git",  # minimum v1.2.1
     "jsonschema >= 4.6.0",
     "importlib_resources",  # not needed from Python 3.9+
     "intervaltree >= 3.1.0",

src/python/ensembl/io/genomio/genome_stats/compare.py

@@ -22,10 +22,11 @@
 import json
 import re
 from typing import Any, Dict
+import logging
 
 from ensembl.io.genomio.utils import get_json
 from ensembl.utils.argparse import ArgumentParser
-
+from ensembl.utils.logging import init_logging
 
 def _diff_dicts(ncbi: Dict[str, int], core: Dict[str, int]) -> Dict[str, Any]:
     """Compare two dicts with the same keys and compute the difference of their values.
@@ -189,17 +190,22 @@ def main() -> None:
     )
     parser.add_argument_src_path("--ncbi_stats", required=True, help="NCBI dataset JSON file")
     parser.add_argument_src_path("--core_stats", required=True, help="Core database JSON file")
+    parser.add_log_arguments(add_log_file=True)
     args = parser.parse_args()
 
+    # Configure and initialise logging
+    init_logging(args.log_level, args.log_file, args.log_file_level)
+
     try:
         ncbi_stats = get_json(args.ncbi_stats)["reports"][0]
-    except KeyError:
+        logging.info(f"{args.ncbi_stats} JSON .'reports' obtained")
+    except (json.decoder.JSONDecodeError, KeyError):
+        logging.warning(f"{args.ncbi_stats} JSON file is empty")
         ncbi_stats = {}
     core_stats = get_json(args.core_stats)
     all_stats = compare_stats(ncbi_stats, core_stats)
 
     print(json.dumps(all_stats, indent=2, sort_keys=True))
 
-
 if __name__ == "__main__":
     main()