Enriching Data¶
Introduction¶
Enriching data involves integrating the data received in the stream with data from c8db or another data stream, or an external service to derive an expected result. To understand the different ways in which this is done, follow the sections below.
Enrich data with a Collection¶
This section explains how to enrich the data in a specific stream by joining it with c8db collection. For this purpose, consider a scenario where you receive sales records generated from multiple locations as events via a system.
- Open the GUI. Click on
Stream Apps
tab. - Click on New to start defining a new stream application.
- Enter a Name as
EnrichingTransactionsApp
or feel free to chose any other name for the stream application. - Enter a Description.
-
Add the following sample stream application.
-
Define the input stream and the c8db collection that need to be joined as follows.
-
Define the stream as follows.
define stream TrasanctionStream (userId long, transactionAmount double, location string);
-
Define the table as follows.
define table UserTable (userId long, firstName string, lastName string);
-
-
Then define the stream query to join the stream and the table, and handle the result as required.
-
Add the
from
clause as follows with thejoin
keyword to join the table and the stream.from TransactionStream as t join UserTable as u on t.userId == u.userId
Info
Note the following about the
from
clause:- In this example, the input data is taken from both a stream and a table. You need to assign a unique reference for each of them to allow the query to differentiate between the common attributes. In this example,
TransactionStream
stream is referred to ast
, and theUserTable
table is referred to asu
. - The
join
keyword joins the stream and the table together while specifying the unique references. - The condition for the stream and the table to be joined is
t.userId == u.userId
, which means that for an event to be taken from theTransactionStream
for the join, one or more events that have the same value for theuserId
must exist in theUserTable
table and vice versa.
- In this example, the input data is taken from both a stream and a table. You need to assign a unique reference for each of them to allow the query to differentiate between the common attributes. In this example,
-
To specify how the value for each attribute in the output stream is derived, add a
select
clause as follows.select t.userId, str:concat( u.firstName, " ", u.lastName) as userName, transactionAmount, location
Info
Note the following in the
select
statement:- The
userId
attribute name is common to both the stream and the table. Therefore, you need to specify from where this attribute needs gto be taken. Here, you can also specifyu.userId
instead oft.userId
. - You are specifying the output generated to include an attribute named
userName
. The value for that is derived by concatenating the values of two attributes in theUserTable
table (i.e.,firstName
andlastName
attributes) by applying thestr:concat()
function. - Similarly, you can apply any of the range of streams functions available to further enrich the joined output.
- The
-
To infer an output stream into which the enriched data must be directed, add the
insert into
clause as follows.insert into EnrichedTrasanctionStream;
-
The completed stream application is as follows.
@App:name("EnrichingTransactionsApp") define stream TrasanctionStream (userId long, transactionAmount double, location string); define table UserTable (userId long, firstName string, lastName string); @sink(type= 'c8streams', stream='EnrichedTrasanctionStream', @map(type='json')) define stream EnrichedTrasanctionStream(userId long, userName string, transactionAmount double, location string); select t.userId, str:concat( u.firstName, " ", u.lastName) as userName, transactionAmount, location from TrasanctionStream as t join UserTable as u on t.userId == u.userId insert into EnrichedTrasanctionStream;
-
To check whether the above stream application works as expected follow below steps
-
Load
UserTable
Collection with User Data{"userId":1200001,"firstName":"Raleigh","lastName":"McGilvra"} {"userId":1200002,"firstName":"Marty","lastName":"Mueller"} {"userId":1200003,"firstName":"Kelby","lastName":"Mattholie"}
-
Publish events on the
TrasanctionStream
{"userId":1200002,"transactionAmount":803,"location":"Chicago"} {"userId":1200001,"transactionAmount":1023,"location":"New York"}
-
You can observe the following events on
EnrichedTrasanctionStream
{"event":{"userId":1200002,"userName":"Marty Mueller","transactionAmount":803.0,"location":"Chicago"}} {"event":{"userId":1200001,"userName":"Raleigh McGilvra","transactionAmount":1023.0,"location":"New York"}}
-
-
Enrich data with another Stream¶
This section explains how to enrich the data in a specific stream by joining it with another stream.
To understand how this is done, consider a scenario where you receive information about cash withdrawals and cash deposits at different bank branches from two separate applications. Therefore, this two types of information are captured via two separate streams. To compare the withdrawals with deposits and observe whether enough deposits are being made to manage the withdrawals, you need to join both these streams. To do this, follow the procedure below.
-
Open the GUI. Click on
Stream Apps
tab. -
Click on New to start defining a new stream application.
-
Enter a Name as
BankTransactionsApp
or feel free to chose any other name for the stream application. -
Enter a Description.
-
First, define the two input streams via which you are receiving informations about withdrawals and deposits.
-
Define a stream named
CashWithdrawalStream
to capture information about withdrawals as follows.define stream CashWithdrawalStream(branchID int, amount long);
-
Define a stream named
CashDepositsStream
to capture information about deposits as follows.define stream CashDepositsStream(branchID string, amount long);
-
-
Now let's define an output stream to which the combined information from both the input streams need to be directed after the join.
@sink(type='c8stream', stream.list='CashFlowStream') define stream CashFlowStream(branchID string, withdrawalAmount long, depositAmount long);
Info
A sink annotation is connected to the output stream to publish the output events. For more information about adding sinks to publish events, see the Publishing Data guide.
-
To specify how the join is performed, and how to use the combined information, write a stream query as follows.
-
To perform the join, add the
from
clause as follows.from CashWithdrawalStream as w join CashDepositStream as d on w.branchID == d.branchID
Info
Observe the following about the above
from
clause:- Both the input streams have attributes of the same name. To identify each name, you must specify a reference for each stream. In this example, the reference for the
CashWithdrawalStream
isw
, and the reference for theCashDepositsStream
stream isd
. - You need to use
join
as the keyword to join two streams. The join condition isw.branchID == d.branchID
where branch IDs are matched. An event in theCashWithdrawalStream
stream is directed to theCashFlowStream
if there are events with the same branch ID in theCashDepositStream
and vice versa.
- Both the input streams have attributes of the same name. To identify each name, you must specify a reference for each stream. In this example, the reference for the
-
To specify how the value for each attribute is derived, add a
select
statement as follows.select w.branchID as branchID, w.amount as withdrawals, d.amount as deposits
Info
The
branchID
attribute name is common to both input streams. Therefore, you can also specifyd.branchID as branchID
instead ofw.branchId as branchId
. -
To filter only events where total cash withdrawals are greater than 95% of the cash deposits, add a
having
clause as follows.having w.amount > d.amount * 0.95
-
To insert the results into the
CashFlowStream
output stream, add theinsert into
clause as follows.insert into CashFlowStream;
-
The completed stream application is as follows:
@App:name("BankTransactionsApp") define stream CashWithdrawalStream(branchID string, amount long); define stream CashDepositsStream(branchID string, amount long); @sink(type='c8streams', stream='CashFlowStream') define stream CashFlowStream(branchID string, withdrawalAmount long, depositAmount long); select w.branchID as branchID, w.amount as withdrawalAmount, d.amount as depositAmount from CashWithdrawalStream as w join CashDepositsStream as d on w.branchID == d.branchID having w.amount > d.amount * 0.95 insert into CashFlowStream;
-
For the different types of joins you can perform via streams, see Stream Query Guide - Join
Enrich data with External Services¶
This section explains how to enrich the data in a specific stream by connecting with an external service and adding information received from that service to the existing data.
To understand how this is done, consider an example where you have some credit card numbers, but need to connect with an external service to identify the credit card companies that issued them, and then save that information in a database.
-
Start creating a new stream application. You can name it
CCTypeIdentificationApp
For instructions, see Creating a Stream Application. -
Define the input stream from which the input data (i.e., the credit card no in this example) must be taken.
define stream CreditCardStream (creditCardNo string);
-
To publish the input data to the external application, connect a sink to the stream you created as shown below. For more information about publishing information, see the Publishing Data guide.
@sink(type='http-request',publisher.url='https://secure.ftipgw.com/ArgoFire/validate.asmx/GetCardType',method='POST', headers="'Content-Type:application/x-www-form-urlencoded'", sink.id="cardTypeSink", @map(type='keyvalue', @payload(CardNumber='{{creditCardNo}}'))) define stream CreditCardStream (creditCardNo string);
Info
Note the following about the above sink definition: - It is assumed that the external application receives requests in HTTP. Therefore, the sink type is
http-request
. - Thepublisher.url
parameter specifies the URL to which the outgoing events need to be published via HTTP. - For more information about the HTTP transport, see Plugins - HTTP. -
To capture the response of the external application once it returns the credit card type, define a stream as follows. For more information about consuming data, see the Consuming Data guide.
define stream EnrichedCreditCardStream (creditCardNo string, creditCardType string);
-
Assuming that the external application sends its output via the HTTP transport, connect a source of the
http
type to theEnrichedCreditCardStream
stream as follows. For more information about consuming events, see the Consuming Data guide.@source(type='http-response' ,sink.id='cardTypeSink', @map(type='xml', namespaces = "xmlns=http://localhost/SmartPayments/", @attributes(creditCardNo = 'trp:creditCardNo',creditCardType = "."))) define stream EnrichedCreditCardInfoStream (creditCardNo string,creditCardType string);
Info
It is assumed that the external application sends requests in HTTP. Therefore, the source type is
http-request
. For more information about the HTTP transport, see Plugins - HTTP. -
To save the response of the external application, define a table named
CCInfoTable
.define table CCInfoTable (cardNo long, cardType string);
-
To save the data enriched by integrating the information received from the external service, add a stream query as follows.
select * from EnrichedCreditCardInfoStream update or insert into CCInfoTable on CCInfoTable.creditCardNo == creditCardNo;
The above query selects all the attributes in the
EnrichedCreditCardInfoStream
and inserts them into theCCInfoTable
table. If a specific record already exists,the query updates it by replacing the attribute values with the latest values taken from theEnrichedCreditCardInfoStream
. -
The completed stream application is as follows:
@App:name("CCTypeIdentificationApp") define stream CreditCardStream (creditCardNo string); @sink(type='http-call',publisher.url='http://secure.ftipgw.com/ArgoFire/validate.asmx/GetCardType',method='POST', headers="'Content-Type:application/x-www-form-urlencoded'", sink.id="cardTypeSink", @map(type='keyvalue', @payload(CardNumber='{{creditCardNo}}'))) define stream GetCreditCardInfoStream (creditCardNo string); @source(type='http-call-response' ,sink.id='cardTypeSink', @map(type='json', namespaces = "xmlns=http://localhost/SmartPayments/", @attributes(creditCardNo = 'trp:creditCardNo',creditCardType = "."))) define stream EnrichedCreditCardInfoStream (creditCardNo string,creditCardType string); -- Define `SampleFilteringByValueDestInvalidDataTable` define Table CCInfoTable(creditCardNo string,creditCardType string); select creditCardNo from CreditCardStream insert into GetCreditCardInfoStream; select * from EnrichedCreditCardInfoStream update or insert into CCInfoTable on CCInfoTable.creditCardNo == creditCardNo;
Enrich data using built-in Plugins¶
The following is a list of stream plugins with which you can enrich data.