Inserting Parent and Child Records
Use Case
When inserting records, often there is a need to fill in details about child records that have a dependency on a parent.
For instance, when dealing with a CRM system, Invoices often may not be entered without at least one line item.
Invoices may have many line items, with each line item made up of several fields. This presents a unique challenge when offering the data as relational tables.
When reading the data, it is easy enough to model an Invoice and an InvoiceLineItem table with a foreign key connecting the two.
But during inserts, the CRM system will require both the Invoice and the InvoiceLineItems to be created in a single submission.
To solve this sort of problem, our tools offer child collection columns on the parent.
These columns may be used to submit insert statements that include details of both the parent and the child records.
From our example, the Invoice table may have a column called InvoiceLineItems.
During the insert, we can pass in the details of the records that would need to be inserted to the InvoiceLineItems table into the InvoiceLineItems column of the Invoice record.
This can be done using the following methods:
Methods for Inserting Parent/Child Records
The driver facilitates two methods for inserting parent/child records: temporary table insertion and XML/JSON aggregate insertion.
Temporary (#TEMP) tables
The simplest way to enter data would be to use a #TEMP table, or temporary table, which the driver will store in memory.
Reference the #TEMP table with the following syntax:
TableName#TEMP
#TEMP tables are stored in memory for the duration of a connection.
Therefore, in order to use them, you cannot close the connection between submitting inserts to them, and they cannot be used in environments where a different connection may be used for each query.
Within that single connection, the table remains in memory until the bulk insert is successful, at which point the temporary table will be wiped from memory.
For example:
INSERT INTO InvoiceLineItems#TEMP (ReferenceNumber, Item, Quantity, Amount) VALUES ('INV001', 'Basketball', 10, 9.99) INSERT INTO InvoiceLineItems#TEMP (ReferenceNumber, Item, Quantity, Amount) VALUES ('INV001', 'Football', 5, 12.99)
Once the InvoiceLineItems table is populated, the #TEMP table may be referenced during an insert into the Invoice table:
INSERT INTO Invoices (ReferenceNumber, Customer, InvoiceLines) VALUES ('INV001', 'John Doe', 'InvoiceLineItems#TEMP')
Under the hood, the driver will read in values from the #TEMP table.
Notice that the ReferenceNumber was used to identify what Invoice the lines are tied to.
This is because the #TEMP table may be populated and used with a bulk insert, where you will have different lines for different Invoices.
This allows the #TEMP tables to be used with a bulk insert. For example:
INSERT INTO Invoices#TEMP (ReferenceNumber, Customer, InvoiceLines) VALUES ('INV001', 'John Doe', 'InvoiceLineItems#TEMP') INSERT INTO Invoices#TEMP (ReferenceNumber, Customer, InvoiceLines) VALUES ('INV002', 'Jane Doe', 'InvoiceLineItems#TEMP') INSERT INTO Invoices SELECT ReferenceNumber, Customer, InvoiceLines FROM Invoices#TEMP
In this case, we are inserting two different Invoices. The ReferenceNumber is how we determine which Lines go with which Invoice.
Note: The tables and columns presented here are an example of how the driver works in general. The specific table and column names may be different in the driver.
XML/JSON Aggregates
As an alternative to #TEMP tables, direct XML/JSON may be used. Since #TEMP tables are not used to construct them, it does not matter if you use the same connection or close the connection after insert.
For example:
[ { "Item", "Basketball", "Quantity": 10 "Amount": 9.99 }, { "Item", "Football", "Quantity": 5 "Amount": 12.99 } ]
OR
<Row> <Item>Basketball</Item> <Quantity>10</Quantity> <Amount>9.99</Amount> </Row> <Row> <Item>Football</Item> <Quantity>5</Quantity> <Amount>12.99</Amount> </Row>
Note that the ReferenceNumber is not present in these examples.
That is because the XML/JSON by its nature is not being passed by reference, but passed in full per insert against the parent record.
There is no need to provide something to tie the child back to the parent since the complete XML/JSON must be constructed and submitted for each row.
Then, insert the values:
INSERT INTO Invoices (ReferenceNumber, Customer, InvoiceLines) VALUES ('INV001', 'John Doe', '{...}')
OR
INSERT INTO Invoices (ReferenceNumber, Customer, InvoiceLines) VALUES ('INV001', 'John Doe', '<Row>...</Row>')
Example for GoogleContacts
Note: the key references such as Id may be different in your environment:
// Execute bulk insert // Perform a bulk insert into the parent table [My Contacts]. INSERT INTO ContactOrganizations#TEMP (Current,Department,Domain,JobDescrption,Location,Name,StartDate) VALUES (true,'Engineering','QA','Big Data Engineer','Bangalore','Tech Mahindra','03-01-2021') INSERT INTO ContactOrganizations#TEMP (Current,Department,Domain,JobDescrption,Location,Name,StartDate,EndDate) VALUES (false,'Engineering','QA','Software Consultant','Bangalore','CData','05-23-2019','02-17-2020') INSERT INTO [My Contacts] (GivenName, Organizations) VALUES ('John','ContactOrganizations#TEMP') // Perform a bulk insert into the table ContactGroups. INSERT INTO ContactGroups#TEMP (Name) VALUES ('Test Bulk') INSERT INTO ContactGroups#TEMP (Name) VALUES ('Test Bulk 2') INSERT INTO ContactGroups (Name) SELECT (Name) from ContactGroups#TEMP // Peform a bulk insert into the table ContactGroups. INSERT INTO ClientData#TEMP (ClientDataKey, ClientDataValue) VALUES ('Client Data 1','sfsdfsds45rf') INSERT INTO ClientData#TEMP (ClientDataKey, ClientDataValue) VALUES ('Client Data 2','2jfk24g78w2jbdk') INSERT INTO ContactGroups (Name,ClientData) VALUES ('Insert ClientData to a Contact Group','ClientData#TEMP') // Peform a bulk insert into the table [My Contacts]. INSERT INTO MyContacts#TEMP(GivenName, HomeAddresses) VALUES ('Elexandar', 'Bangalore India') INSERT INTO MyContacts#TEMP(GivenName, HomeAddresses) VALUES ('Altin', 'Bangalore India') INSERT INTO [My Contacts] (GivenName, HomeAddresses) SELECT GivenName, HomeAddresses FROM MyContacts#TEMP // Execute bulk update INSERT INTO MyContacts#TEMP(Id, HomeAddresses) VALUES ('c970377457588673806', 'Bangalore India') INSERT INTO MyContacts#TEMP(Id, HomeAddresses) VALUES ('c928048274343381622', 'Bangalore India') UPDATE [My Contacts] (Id, HomeAddresses) SELECT Id, HomeAddresses FROM MyContacts#TEMP // Execute bulk delete INSERT INTO MyContacts#TEMP(Id) VALUES ('c970377457588673806') INSERT INTO MyContacts#TEMP(Id) VALUES ('c928048274343381622') DELETE FROM [My Contacts] WHERE EXISTS SELECT Id FROM MyContacts#TEMP