Structure

A MettleCI Unit Test Specification (often abbreviated ‘Spec’') is a YAML-formatted file which uses a grammar modelled loosely on the Gherkin syntax of a testing tool called Cucumber. The overall structure follows the Gherkin pattern…

given:
  # This source of input data
when:
  # The specified job is executed with these parameter values
then:
  # Expect the Job to produce data that looks like this

Given

The given section defined a list of stage nodes (or sparseLookup nodes, see below) defining input links whose values you wish to replace with test data.

given:
  - stage: My_First_Sequential_File          # Stage 'My_First_Sequential_File' has...
    link: Input_Link                         # an output link 'Input_Link' which... 
    path: Sequential_File_1.csv              # will be injected with data from the test data file 'Sequential_File_1.csv'
  - stage: My_Sequential_Other_File
    link: Other_Input_Link
    path: Sequential_File_2.csv  

Each link is specified using a combination of stage and link properties (to uniquely identify an incoming link which supplies data to your Job) and a path node to identify the test data CSV file containing the test data that is to be injected on that incoming link.

Sparse Lookup sources

When an input source is used with a Sparse Lookup stage then rather than using the stage node to specify the input you will use the sparseLookup node.

given:
  - stage: Some_Non_Lookup_Stage
    link: Some_Input_Link
    path: Some_File.csv
  - sparseLookup: SparseLookup
    path: Database-Reference.csv
    key:
      - KEY_COLUMN_1
      - KEY_COLUMN_2

The sparseLookup node specifies …

1. the value defining the name of the sparse lookup reference stage,

2. a path to the relevant CSV unit test data file, and

3. a list of key columns to be used for the sparse lookup.

See this page for more information.

When

The when node specifies which job will be executed during testing as well as any parameters (including job macros) that affect the data produced by the job.

when:
  job: KeyGeneratorExample                   # The DataStage Job with which this test spec is associated 
  parameters:
    DSJobStartDate: 2012-01-15               # Run the test using this value for the DSJobStartDate macro
    DSJobStartTime: 11:05:01                 # Run the test using this value for the DSJobStartTime macro 
    paramStartKey: 100                       # Run the test using this value for the paramStartKey Job parameter
    

Then

The then section associates unit test data files with each of your Job’s input links.

then:
  # Compare output
  - stage: ODBC_customer
    link: customer_out
    path: ODBC_customers.csv
  - stage: ODBC_order
    link: order_out
    path: ODBC_orders.csv
    cluster:
      - ACCOUNT_ID
      - TYPE_CODE

Similar to the Given section, each link in the Then section is specified using a combination of stage and link nodes (to uniquely identify an outgoing link which produces data from your Job) and a path node to identify the test data CSV file containing the test data that is to be injected on that incoming link.

Cluster keys

The cluster node is used to assist MettleCI’s resource management when using high volumes of test data. Setting a Cluster Key will prompt MettleCI to split the actual output and expected output using multiple, smaller subsets (based on the supplied keys) before the data is compared. Data is split such that each subset will only contain records that have the same values for all columns that make up the Cluster Key. In general, cluster keys should only be used when necessary, and not specified by default. Read more about the using the cluster node in High Volume Unit Tests.

Row count comparisons

You can configure as test to only compare output row counts, rather than the content of those rows, by setting the checkRowCountOnly node to true.

then:
  - stage: Output
    link: Write1
    path: Output-Write1.csv
    checkRowCountOnly: true

See this page for more information.

Excluding columns from unit tests

You can omit selected columns from the output comparison by listing them under an ignore node for the relevant output.

then:
  - stage: Transform
    link: Output
    path: Transform-Output.csv
    ignore: 
      - CREATION_DATE
      - LAST_UPDATED

See this page for more information.

Test Specification Types

Test specification patterns