From 230240d17e444b6c42b3396680e00f6265cd51de Mon Sep 17 00:00:00 2001
From: Jakub Dylag <jjd1c23@soton.ac.uk>
Date: Thu, 10 Apr 2025 08:44:38 +0100
Subject: [PATCH] (feat) flag to stop phen diff from check config.yml if it
doesn't exist
---
acmc/main.py | 8 +++++++-
acmc/phen.py | 57 ++++++++++++++++++++++++++++------------------------
2 files changed, 38 insertions(+), 27 deletions(-)
diff --git a/acmc/main.py b/acmc/main.py
index 4b7ac6c..5f796b0 100644
--- a/acmc/main.py
+++ b/acmc/main.py
@@ -78,7 +78,7 @@ def _phen_copy(args: argparse.Namespace):
def _phen_diff(args: argparse.Namespace):
"""Handle the `phen diff` command."""
- phen.diff(args.phen_dir, args.version, args.old_phen_dir, args.old_version)
+ phen.diff(args.phen_dir, args.version, args.old_phen_dir, args.old_version, args.not_check_config)
def main():
@@ -323,6 +323,12 @@ def main():
required=True,
help="Old phenotype version to compare with the changed version",
)
+ phen_diff_parser.add_argument(
+ "--not-check-config",
+ action='store_true',
+ default=False,
+ help="(Optional) Prevent loading and comparing config file, in the case where one does not exist",
+ )
phen_diff_parser.set_defaults(func=_phen_diff)
# Parse arguments
diff --git a/acmc/phen.py b/acmc/phen.py
index abea461..052e1a0 100644
--- a/acmc/phen.py
+++ b/acmc/phen.py
@@ -1375,37 +1375,42 @@ def diff_phen(
old_phen_path: Path,
old_version: str,
report_path: Path,
+ not_check_config:bool,
):
"""Compare the differences between two versions of a phenotype"""
- # validate phenotypes
- _logger.debug(f"Validating for diff old path: {str(old_phen_path.resolve())}")
- validate(str(old_phen_path.resolve()))
- _logger.debug(f"Validating for diff new path: {str(new_phen_path.resolve())}")
- validate(str(new_phen_path.resolve()))
-
- # get old and new config
- old_config_path = old_phen_path / CONFIG_FILE
- with old_config_path.open("r") as file:
- old_config = yaml.safe_load(file)
- new_config_path = new_phen_path / CONFIG_FILE
- with new_config_path.open("r") as file:
- new_config = yaml.safe_load(file)
-
# write report heading
report = f"# Phenotype Comparison Report\n"
- report += f"## Original phenotype\n"
- report += f" - {old_config['phenotype']['omop']['vocabulary_id']}\n"
- report += f" - {old_version}\n"
- report += f" - {str(old_phen_path.resolve())}\n"
- report += f"## Changed phenotype:\n"
- report += f" - {new_config['phenotype']['omop']['vocabulary_id']}\n"
- report += f" - {new_version}\n"
- report += f" - {str(new_phen_path.resolve())}\n"
# Step 1: check differences configuration files
- # Convert list of dicts into a dict: {name: file}
- report += diff_config(old_config, new_config)
+ if not not_check_config:
+ # validate phenotypes
+ _logger.debug(f"Validating for diff old path: {str(old_phen_path.resolve())}")
+ validate(str(old_phen_path.resolve()))
+ _logger.debug(f"Validating for diff new path: {str(new_phen_path.resolve())}")
+ validate(str(new_phen_path.resolve()))
+
+ # get old and new config
+ old_config_path = old_phen_path / CONFIG_FILE
+ with old_config_path.open("r") as file:
+ old_config = yaml.safe_load(file)
+ new_config_path = new_phen_path / CONFIG_FILE
+ with new_config_path.open("r") as file:
+ new_config = yaml.safe_load(file)
+
+ # write report
+ report += f"## Original phenotype\n"
+ report += f" - {old_config['phenotype']['omop']['vocabulary_id']}\n"
+ report += f" - {old_version}\n"
+ report += f" - {str(old_phen_path.resolve())}\n"
+ report += f"## Changed phenotype:\n"
+ report += f" - {new_config['phenotype']['omop']['vocabulary_id']}\n"
+ report += f" - {new_version}\n"
+ report += f" - {str(new_phen_path.resolve())}\n"
+
+
+ # Convert list of dicts into a dict: {name: file}
+ report += diff_config(old_config, new_config)
# Step 2: check differences between map files
# List files from output directories
@@ -1422,7 +1427,7 @@ def diff_phen(
_logger.info(f"Phenotypes diff'd successfully")
-def diff(phen_dir: str, version: str, old_phen_dir: str, old_version: str):
+def diff(phen_dir: str, version: str, old_phen_dir: str, old_version: str, not_check_config:bool):
# make tmp directory .acmc
timestamp = time.strftime("%Y%m%d_%H%M%S")
temp_dir = Path(f".acmc/diff_{timestamp}")
@@ -1483,7 +1488,7 @@ def diff(phen_dir: str, version: str, old_phen_dir: str, old_version: str):
report_filename = f"{version}_{old_version}_diff.md"
report_path = changed_phen_path / report_filename
# diff old with new
- diff_phen(changed_path, version, old_path, old_version, report_path)
+ diff_phen(changed_path, version, old_path, old_version, report_path, not_check_config)
finally:
# clean up tmp directory
--
GitLab