From 3c3ae71752d61930909a520f03d9c7bd93883c93 Mon Sep 17 00:00:00 2001
From: Jakub Dylag <jjd1c23@soton.ac.uk>
Date: Fri, 28 Mar 2025 15:11:11 +0000
Subject: [PATCH] Allow multiple files per concept set - validation passed
---
acmc/phen.py | 67 ++++++++++++++++++++++++++--------------------------
1 file changed, 34 insertions(+), 33 deletions(-)
diff --git a/acmc/phen.py b/acmc/phen.py
index ff2edc2..4893144 100644
--- a/acmc/phen.py
+++ b/acmc/phen.py
@@ -514,46 +514,47 @@ def validate(phen_dir: str):
# create a list of all the concept set names defined in the concept set configuration
concept_set_names = []
for item in phenotype["concept_sets"]:
- # if item["name"] in concept_set_names:
- # validation_errors.append(
- # f"Duplicate concept set defined in concept sets {item['name'] }"
- # )
- # else:
- concept_set_names.append(item["name"])
-
- # check codes definition
- for item in phenotype["concept_sets"]:
- # check concepte code file exists
- concept_code_file_path = concepts_path / item["file"]["path"]
- if not concept_code_file_path.exists():
- validation_errors.append(
- f"Coding file {str(concept_code_file_path.resolve())} does not exist"
- )
-
- # check concepte code file is not empty
- if concept_code_file_path.stat().st_size == 0:
+ if item["name"] in concept_set_names:
validation_errors.append(
- f"Coding file {str(concept_code_file_path.resolve())} is an empty file"
+ f"Duplicate concept set defined in concept sets {item['name'] }"
)
+ else:
+ concept_set_names.append(item["name"])
- # check code file type is supported
- if concept_code_file_path.suffix not in CODE_FILE_TYPES:
- raise ValueError(
- f"Unsupported filetype {concept_code_file_path.suffix}, only support csv, xlsx, xls code file types"
- )
+ # check codes definition
+ for files in phenotype["concept_sets"]:
+ for item in files["files"]:
+ # check concepte code file exists
+ concept_code_file_path = concepts_path / item["path"]
+ if not concept_code_file_path.exists():
+ validation_errors.append(
+ f"Coding file {str(concept_code_file_path.resolve())} does not exist"
+ )
- # check columns specified are a supported medical coding type
- for column in item["file"]["columns"]:
- if column not in code_types:
+ # check concepte code file is not empty
+ if concept_code_file_path.stat().st_size == 0:
validation_errors.append(
- f"Column type {column} for file {concept_code_file_path} is not supported"
+ f"Coding file {str(concept_code_file_path.resolve())} is an empty file"
)
- # check the actions are supported
- if "actions" in item["file"]:
- for action in item["file"]["actions"]:
- if action not in COL_ACTIONS:
- validation_errors.append(f"Action {action} is not supported")
+ # check code file type is supported
+ if concept_code_file_path.suffix not in CODE_FILE_TYPES:
+ raise ValueError(
+ f"Unsupported filetype {concept_code_file_path.suffix}, only support csv, xlsx, xls code file types"
+ )
+
+ # check columns specified are a supported medical coding type
+ for column in item["columns"]:
+ if column not in code_types:
+ validation_errors.append(
+ f"Column type {column} for file {concept_code_file_path} is not supported"
+ )
+
+ # check the actions are supported
+ if "actions" in item:
+ for action in item["actions"]:
+ if action not in COL_ACTIONS:
+ validation_errors.append(f"Action {action} is not supported")
if len(validation_errors) > 0:
_logger.error(validation_errors)
--
GitLab