aboutsummaryrefslogtreecommitdiff
path: root/gn3/db
diff options
context:
space:
mode:
Diffstat (limited to 'gn3/db')
-rw-r--r--gn3/db/traits.py47
1 files changed, 46 insertions, 1 deletions
diff --git a/gn3/db/traits.py b/gn3/db/traits.py
index 4baac67..a77e6a1 100644
--- a/gn3/db/traits.py
+++ b/gn3/db/traits.py
@@ -1,5 +1,5 @@
"""This class contains functions relating to trait data manipulation"""
-from typing import Any
+from typing import Any, Union
def get_trait_csv_sample_data(conn: Any,
@@ -30,3 +30,48 @@ def get_trait_csv_sample_data(conn: Any,
for val in (strain_id, strain_name,
value, error, count)]))
return f"# Publish Data Id: {publishdata_id}\n\n" + "\n".join(csv_data)
+
+
+def update_sample_data(conn: Any,
+ strain_name: str,
+ strain_id: int,
+ publish_data_id: int,
+ value: Union[int, float, str],
+ error: Union[int, float, str],
+ count: Union[int, str]):
+ """Given the right parameters, update sample-data from the relevant
+ table."""
+ STRAIN_ID_SQL: str = "UPDATE Strain SET Name = %s WHERE Id = %s"
+ PUBLISH_DATA_SQL: str = ("UPDATE PublishData SET value = %s "
+ "WHERE StrainId = %s AND Id = %s")
+ PUBLISH_SE_SQL: str = ("UPDATE PublishSE SET error = %s "
+ "WHERE StrainId = %s AND DataId = %s")
+ N_STRAIN_SQL: str = ("UPDATE NStrain SET count = %s "
+ "WHERE StrainId = %s AND DataId = %s")
+
+ updated_strains: int = 0
+ updated_published_data: int = 0
+ updated_se_data: int = 0
+ updated_n_strains: int = 0
+
+ with conn.cursor() as cursor:
+ # Update the Strains table
+ cursor.execute(STRAIN_ID_SQL, (strain_name, strain_id))
+ updated_strains: int = cursor.rowcount
+ # Update the PublishData table
+ cursor.execute(PUBLISH_DATA_SQL,
+ (None if value == "x" else value,
+ strain_id, publish_data_id))
+ updated_published_data: int = cursor.rowcount
+ # Update the PublishSE table
+ cursor.execute(PUBLISH_SE_SQL,
+ (None if error == "x" else error,
+ strain_id, publish_data_id))
+ updated_se_data: int = cursor.rowcount
+ # Update the NStrain table
+ cursor.execute(N_STRAIN_SQL,
+ (None if count == "x" else count,
+ strain_id, publish_data_id))
+ updated_n_strains: int = cursor.rowcount
+ return (updated_strains, updated_published_data,
+ updated_se_data, updated_n_strains)