By the end of this chapter you will be able to:

• Merge directly from database collectors
• Popuate attributes with key values using the '_key[n]' notation

In this exercise you will repeat the comparison you did in the previous exercise, of the full channel list in your model against the list in the inventory database – but this time using a merge:

• Add a new merge stream to your model, by dragging in the icon .
• Call this stream Name Channel List Reconciliation 2
• Add an input pipe from Latest Package All Channels, with Name sales
• Drag the following attributes from Latest Package All Channels to Channel List Reconciliation 2:
  • CustomerRef
  • Channel
• Add an pipe from the database collector SOURCE_ALL_CHANNELS_LIST to Channel List Reconciliation 2

  • Call this pipe inv

  • Set up the grouping attributes on the input pipes:
    • Double click on the pipe sales
      • Go to the Sort/Group section
      • Press  and drag in CustomerRef and Channel
      • Press  to save your changes
    • Double-click on the pipe inv
      • Go to the Sort/Group section
      • Press 
      • In the new grouping attribute form that pops up:
        • Select the attribute CustomerRef from the drop down list
        • Tick the Group flag
        • Press 
      • Add the grouping attribute Channel in the same way
      • Back in the main pipe configuration form, press 

Now you will complete the configuration of the merge stream, using the _key internal variable to write key values into the stream:

• Double click on Channel List Reconciliation 2 to open the configuration form
• Double click the attribute CustomerRef
  • In the attribute details form that pops up, update Expression to: _key[1]
    • This writes the first key value into this attribute
  • Press 
  • Double click the attribute Channel
  • In the attribute details form that pops up, update Expression to: _key[2]
    • This writes the second key value into this attribute
  • Press 
  • Add an attribute SalesCheck, and set the Expression to

if (countElements(sales) == 1,
 "YES"
 ,
 "NO"
 )

• Add an attribute InvCheck, and set the Expression to

if (countElements(inv) == 1,
 "YES"
 ,
 "NO"
 )

• Press  in the main stream configuration form to save your changes
• Update the query in the database collector to:

select * from SOURCE_ALL_CHANNELS_LIST
 order by CustomerRef, Channel

(the ordering is needed to support the merge)

• Run Analysis on Channel List Reconciliation 2
• In your results you should see an additional 3 records that were not included in the reconciliation you did in the previous exercise – channels that are recorded in the inventory, but not in the channels list, are included in your output

This is because a merge will generate an output record for every key value it finds across all input pipes, whereas if you do a data enrichment using a lookup pipe – you will only get an output record for every record from the input pipe

Filter unmatched rows

To help you find unmatched rows in your output data, you will now add a filter

• Open the data set you have just generated in Channel List Reconciliation 2
• Create a filter:
  
  • Where ANY of the following are true
    
    • SalesCheck equals NO
    • InvCheck equals NO
• Press - Apply the filter without saving
• You will get a list of all non-matched records, that is, all records where the SalesCheck value is NO or the InvCheck value is NO