Is a small library that helps developers generate test data for the purpose of writing test/specs against their code which read and write data to the database.
This library currently makes some technical assumptions listed below. Some of these may be addressed in a future version.
- This library only works with Postgres version 9.1+. It uses the Haskell
postgresql-simple
library. - It only works with tables that uses the
Integer
type for the primary key, and by virtue foreign keys which are based on the same type. - It also relies on schemas which follow certain conventions;
- Primary keys in your database need to be of type
Integer
, and also need to be namedId
. - Foreign keys need to be named using the convention
tableName_id
where tableName is the singular form of the table name. So for example if you have ausers
table and anaccounts
table, and theusers
table has a foreign key to the accounts table, then the foreign key in theaccounts
table needs to be nameduser_id
.
- Primary keys in your database need to be of type
To get started you will need to run the script ./bin/db-refresh
, this script
requires an up and running Postgres server which can be installed by following the
steps below;
## install postgresql
brew install postgresql
## to initialize (once only)
initdb /usr/local/var/postgres -E utf8
## to start
pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start
## to stop
pg_ctl -D /usr/local/var/postgres stop -s -m fast
Setting up DB for tests.
./bin/db-refresh
## install postgresql
apt-get install postgresql postgresql-client libpq-dev
# For the bin/db* scripts to work you have to setup your user to have superuser permissions
# Something like:
# Login as the default database user postgres
sudo -u postgres bash
# Start the client
psql
# Create a user with your username and set permissions
create user <your-username>;
alter role <your-username> with superuser;
alter user <your-username> with encrypted password '<password>';
If you are the type of person that feels reading documentation is a waste of time, then I highly recommend you look in the test folder of this project to learn how to use this particular library, otherwise please read on.
To get started with this library you need to define your ParentTable
and ChildTable
tables. The ParentTable
type represents a table in your relational database that
has no foreign keys, and has a primary key column named Id
of type Integer
.
The ChildTable
type is table in your relational database that has 1 or more
foreign keys, and has a primary key column named Id
of type Integer
. As mentioned
in the Limitations section foreign keys currently need to be named using
a particular convention.
Here are few examples of how to define your tables;
userTable :: ParentTable
userTable = ParentTable "users" ["name", "age", "active"]
accountTableAsParent :: ParentTable
accountTableAsParent = ParentTable "accounts" ["name", "description"]
accountTableAsChild :: ChildTable
accountTableAsChild = ChildTable "accounts" ["name", "description"]
profileTable :: ChildTable
profileTable = ChildTable "profiles" ["active"]
As you can see from the example above the accounts
table has been defined
as both a parent table and child table. This is an example of how to represent
as associative table (Table which represents a many to many relationship)
using this library.
Once your tables are defined then you can define your functions which can
take input data as a list of tuples which represent your data. Let's start
with the users
table. The following example demonstrates how to define a
function to insert users, and then how to use this function;
insertUsers :: (ToRow q) => Connection -> [q] -> IO [Id]
insertUsers conn = build conn userTable
insertUsers conn [("John" :: Text, 40 :: Int, False :: Bool)
,("Jane" :: Text, 32 :: Int, True :: Bool)]
The recommended pattern for setting this all up is to place such code in a factory to create the data.
Populating child tables requires the resolution of foreign keys. Quokka
derives relationships through the ParentTable
type. This type captures the
related table through the type thereby faciliating a mechanism through which
child rows can be inserted into the table. The following example demonstrates
how to define a function to insert child rows, and then how to use the
function. This example builds on the last example;
insertAccounts :: (ToRow q) => Connection -> [q] -> IO [Id]
insertAccounts conn = buildWith1Rel conn userTable accountTableAsChild
userIds <- insertUsers conn [("John" :: Text, 40 :: Int, False :: Bool)]
accountIds <- insertAccounts conn [("Johns Account" :: Text, "Description" :: Text, id' userIds)]
The function id'
is a helper function available in the library to extract the
foreign key value of the parent entity when generating the insert
statement
for the database.
When we use functions like buildWith1Rel
Quokka
relies on a simple convention when
generating insert statements to populate foreign key columns. This convention will
not suit all scenarios. For schemas which have foreign keys which do not follow convention
Quokka
comes with alternate functions such as buildWith1CustomRel
. Let's look at
a concrete example to better understand what we are saying. Let's take two tables users
and accounts
. If the accounts
table has a foreign key named user_id
to the users
table then to populate both users
and accounts
we can use the function buildWith1Rel
.
But if the accounts
table has a foreign key column named usersid
then the function
buildWith1Rel
will fail as it will try to insert the foreign key into the column user_id
.
So in this case we use the function buildWith1CustomRel
, this function takes a Relation
as
an argument. The Relation
type takes the ParentTable
and the FK
type in its constructor,
and we set the name of the foreign key column in the FK
type.
Associate tables are a special case of child table with 2 foreign key associations.
Populating associate tables requires the resolution of 2 foreign keys. Similar
to child tables an associate table can be modelled using two ParentTable
types
and a single ChildTable
type. This is illustrated in the example below which
builds on the example from the Populating Parent Tables
section;
insertProfiles :: (ToRow q) => Connection -> [q] -> IO [Id]
insertProfiles conn = buildWithManyRels conn [userTable, accountTableAsParent] profileTable
userIds <- insertUsers conn [("John" :: Text, 40 :: Int, False :: Bool)]
accountIds <- insertAccounts conn [("Johns Account" :: Text, "Description" :: Text, id' userIds)]
profileIds <- insertProfiles conn [(True :: Bool, id' userIds, id' accountIds)]
buildWithManyRels
in this case returns a Query
type which is aware of the
associative nature of profiles
with users
and accounts
. So when we invoke
the insertProfiles
function the foreign key values are resolved during the
insert. Quokka
uses concrete types in a very restrictive way. So the definition
of the Account
child table cannot be used in the function call buildWithManyRels
so we had to define the accounts
table twice, once as ParentTable
type, and
once as a ChildTable
type.
In the examples presented we see that even though we insert a single row the API
treats everything as a collection. This can be a bit tiresome when we only want
to insert one tuple. For this we have a whole series of functions that deal with
a single record. For example to insert 1 single parent tuple with no relations we
use the function build1
instead of build
. Similarly to build a parent and
child relationship with a single tuple you can use the build1With1Rel
instead
off the buildWith1Rel
function so on and so forth.
MIT License
Copyright (c) 2019 Shirren Premaratne
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.