Liquibase Enterprise and CSV Files
Liquibase Enterprise does have the ability to load and update data using CSV formatted files.
(Optional) Oracle Setup Data
The below samples are provided for an Oracle database. In order to run the below samples, the following Setup Data is required on your database.
Create a dbo schema
Create a dbo.contacts table, eg.
CREATE TABLE dbo.contacts ("id" NUMBER NOT NULL, "activeflag" NUMBER(1), "firstname" VARCHAR2(50), "lastname" VARCHAR2(50), "age" NUMBER, "lastcall" TIMESTAMP, CONSTRAINT "PK_CONTACTS" PRIMARY KEY ("id"));
Create a dbo.contacts_seq, eg.
CREATE SEQUENCE dbo.contacts_seq START WITH 1 INCREMENT BY 1 NOCACHE NOCYCLE;
Loading Data with a CSV file
Step 1:
Create the CSV file that contains the data you wish to have loaded. The first line needs to be the list of columns to be loaded. Liquibase Enterprise will create the INSERTs with the columns and data listed in the file.
CSV File and Format
id,activeflag,firstname,lastname,age,lastcall dbo.contacts_seq.nextval,0,Chris,Klackson,32,2025-07-04 12:33:27 dbo.contacts_seq.nextval,1,Samantha,Sallers,47,2025-07-15 01:01:16 dbo.contacts_seq.nextval,1,Pete,Prosser,50,2025-07-15 06:45:48 dbo.contacts_seq.nextval,0,Joeseph,Scala,70,2025-07-28 03:14:32 dbo.contacts_seq.nextval,0,Umbert,Klassen,31,2025-09-17 22:11:08 dbo.contacts_seq.nextval,1,Gary,Finer,35,2025-09-17 18:25:09 |
Step 2:
Place the file in the Resources directory of the Project Repo.
CSV File Path
{DaticalDBWorkspace}/{DaticalDBProjectName}/Resources/data/contacts.csv |
Path Considerations
If the CSV file is stored in the project resources directory - {DaticalDBWorkspace}/{DaticalDBProjectName}/Resources
, then you can use a relative path to reference the file.
Full Path:
/home/kevin/datical/ecomm/Resources/data/contacts.csv
Relative Path:
data/contacts.csv
It is a good practice to create a data
subdirectory in the Resources
directory and keep your CSV files there.
Step 3:
Create a changeset in the Changelog/changelog.xml file of the Project Repo.
Changeset
While specifying the columns in the loadData refactoring is not explicitly required, providing this information gives us the data types so the best possible SQL is generated.
Note you can include changeset attributes such as labels and contexts.
See Valid Data Types for supported values.
<changeSet author="Kevin" id="2024-06-25-data-load-1"> <loadData encoding="UTF-8" file="data/contacts.csv" tableName="contacts" schemaName="dbo"> <column header="id" name="id" type="NUMERIC"/> <column header="activeflag" name="activeflag" type="BOOLEAN"/> <column header="firstname" name="firstname" type="STRING"/> <column header="lastname" name="lastname" type="STRING"/> <column header="age" name="age" type="NUMERIC"/> <column header="lastcall" name="lastcall" type="DATETIME"/> </loadData> </changeSet> |
Step 4:
The changeset is now eligible to be deployed to the various databases on the pipeline. If you are using artifacts, be sure the package the changelog updates prior to running any deployments.
Generated SQL Code
The CSV file generates the following SQL code.
INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Chris', 'Klackson', 32, to_date('2025-07-04 12:33:27', 'YYYY-MM-DD HH24:MI:SS')); INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Samantha', 'Sallers', 47, to_date('2025-07-15 01:01:16', 'YYYY-MM-DD HH24:MI:SS')); INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Pete', 'Prosser', 50, to_date('2025-07-15 06:45:48', 'YYYY-MM-DD HH24:MI:SS')); INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Joeseph', 'Scala', 70, to_date('2025-07-28 03:14:32', 'YYYY-MM-DD HH24:MI:SS')); INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Umbert', 'Klassen', 31, to_date('2025-09-17 22:11:08', 'YYYY-MM-DD HH24:MI:SS')); INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (contacts_seq.nextval, 0, 'Gary', 'Finer', 35, to_date('2025-09-17 18:25:09', 'YYYY-MM-DD HH24:MI:SS')); |
Updating Data with a CSV file
The previous example works nicely when you're loading data for the first time. But, what do you do if you want to use a CSV file to update data? How would that work?
In this case, you'll want to use the loadUpdateData refactoring. It's similar to using loadData with one exception – you need to specify the primary key.
Once specified, Liquibase Enterprise will generate the appropriate "INSERT or UPDATE" SQL to be applied to your database.
Step 1:
Create the CSV file that contains the data you wish to have loaded. The first line needs to be the list of columns to be loaded. Liquibase Enterprise will create the INSERTs with the columns and data listed in the file.
CSV File and Format
id,activeflag,firstname,lastname,age,lastcall 1,1,Chris,Klackson,32,2025-07-04 12:33:27 2,0,Samantha,Sallers,47,2025-07-15 01:01:16 3,1,Pete,Prosser,50,2025-07-15 06:45:48 4,0,Joeseph,Scala,71,2025-07-28 03:14:32 5,0,Umbert,Klassenmass,31,2025-09-17 22:11:08 6,1,Gary,Finer,36,2025-09-17 18:25:09 |
Step 2:
Place the file in the Resources directory of the Project Repo.
CSV File Path
{DaticalDBWorkspace}/{DaticalDBProjectName}/Resources/data/updated_contacts_1.csv |
Step 3:
Create a changeset in the Changelog/changelog.xml file of the Project Repo.
Changeset
Make sure to specify the primary key column.
While specifying the columns in the loadUpdateData refactoring is not explicitly required, providing this information gives us the data types so the best possible SQL is generated.
Note you can include changeset attributes such as labels and contexts.
See Valid Data Types for supported values.
<changeSet id="2024-06-25-contacts-data-update-1" author="Kevin" > <loadUpdateData encoding="UTF-8" file="data/updated_contacts_1.csv" tableName="contacts" primaryKey="id"> <column name="id" header="id" type="NUMERIC"/> <column name="activeflag" header="activeflag" type="NUMERIC"/> <column name="firstname" header="firstname" type="STRING"/> <column name="lastname" header="lastname" type="STRING"/> <column name="age" header="age" type="NUMERIC"/> <column name="lastcall" header="lastcall" type="DATETIME"/> </loadUpdateData> </changeSet> |
Step 4:
The changeset is now eligible to be deployed to the various databases on the pipeline. If you are using artifacts, be sure the package the changelog updates prior to running any deployments.
Generated SQL Code
The CSV file generates the following SQL code.
DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 1; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (1, 1, 'Chris', 'Klackson', 32, to_date('2013-07-04 12:33:27', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 1, age = 32, firstname = 'Chris', lastcall = to_date('2025-07-04 12:33:27', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Klackson' WHERE id = 1; END IF; END; /; DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 2; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (2, 0, 'Samantha', 'Sallers', 47, to_date('2013-07-15 01:01:16', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 0, age = 47, firstname = 'Samantha', lastcall = to_date('2025-07-15 01:01:16', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Sallers' WHERE id = 2; END IF; END; /; DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 3; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (3, 1, 'Pete', 'Prosser', 50, to_date('2013-07-15 06:45:48', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 1, age = 50, firstname = 'Pete', lastcall = to_date('2025-07-15 06:45:48', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Prosser' WHERE id = 3; END IF; END; /; DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 4; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (4, 0, 'Joeseph', 'Scala', 71, to_date('2012-07-28 03:14:32', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 0, age = 71, firstname = 'Joeseph', lastcall = to_date('2025-07-28 03:14:32', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Scala' WHERE id = 4; END IF; END; /; DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 5; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (5, 0, 'Umbert', 'Klassenmass', 31, to_date('2012-09-17 22:11:08', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 0, age = 31, firstname = 'Umbert', lastcall = to_date('2025-09-17 22:11:08', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Klassenmass' WHERE id = 5; END IF; END; /; DECLARE v_reccount NUMBER := 0; BEGIN SELECT COUNT INTO v_reccount FROM INTUSER.contacts WHERE id = 6; IF v_reccount = 0 THEN INSERT INTO INTUSER.contacts (id, activeflag, firstname, lastname, age, lastcall) VALUES (6, 1, 'Gary', 'Finer', 36, to_date('2013-09-17 18:25:09', 'YYYY-MM-DD HH24:MI:SS')); ELSIF v_reccount = 1 THEN UPDATE INTUSER.contacts SET activeflag = 1, age = 36, firstname = 'Gary', lastcall = to_date('2025-09-17 18:25:09', 'YYYY-MM-DD HH24:MI:SS'), lastname = 'Finer' WHERE id = 6; END IF; END; /; |
General Considerations for Loading Data With a CSV File
Date and Datetime Considerations
Note, that date and datetime must be in one of the following ISO formats.
yyyy-MM-dd'T'HH:mm:ss
yyyy-MM-dd HH:mm:ss
yyyy-MM-dd'T'HH:mm:ss.S
Valid Data Types
When specifying columns in the loadData refactoring, you may choose from the following generic data types.
BOOLEAN
NUMERIC
DATE
DATETIME
STRING
COMPUTED
SKIP