Best practices for do-while nodes

Procedure

Inside the loop node, use the following DAG variables to access parameter values: ${dag.inputs} to get the complete dataset, ${dag.inputs[${dag.offset}]} to get the data for the current row, ${dag.offset} to get the offset, ${dag.loopTimes} to get the loop count, ${dag.inputs.length} to get the dataset length, and ${dag.inputs[0][1]} to retrieve data from a specific row and column as you would in a two-dimensional array.

1. Configure node dependencies

   The do-while node depends on the assignment node.

2. Pass the assigned result set

   The built-in Parameters output parameter outputs of the assignment node must be used as the Parameters input parameter for the do-while loop node.

3. Retrieve parameters within the do-while node's internal nodes

   Customize the internal workflow of the do-while node based on your business requirements, and use variables in the internal nodes to retrieve the required parameter values.

Create and configure an assignment node

Key points:

• Assignment code and context parameters: Select a language for the assignment node and use a statement to define the final result set. The assignment node assigns the output of its last statement to the built-in outputs output parameter.

  Note

  The do-while node uses the assignment node's output as its input. The following code is for testing only — you cannot directly access the table used here. After you understand the loop node, replace the table with one in your workspace.

  Example assignment code is select * from xc_dpe_e2.xc_rpt_user_info_d limit 2;. In the Output Parameters of the node context, the system generates a default parameter named outputs, with a type of variable and a value of ${outputs}. The value is determined at runtime.

• Upstream and downstream dependencies: Create an assignment node in your workflow and connect it to the do-while node to set it as the upstream node.

For detailed steps, see Assignment node.

Configure the do-while node input

On the editor page for the do-while node, click Scheduling Settings on the right. In the Parameters section, click Add. Set Parameter name to input and Value Source to the output of the upstream assignment node.

Define the loop body

Double-click the do-while node to open its editor page and define the loop body.

By default, a do-while node contains a start node, a shell node, and an end node. You can edit the code in the shell node to print the loop parameters. The key steps are as follows:

• Upstream and downstream dependencies: Connect the internal nodes to establish their dependencies.

• Loop task code: When writing the code for the internal shell node, you can use built-in variables to print various loop parameters. For available built-in variables for a do-while node, see Built-in variables. The following is sample code for the shell node:

  echo '${dag.input}';
  echo 'Get the row data for the current loop:'${dag.input[${dag.offset}]};
  echo 'Get the offset:'${dag.offset};
  echo 'Get the number of loops:'${dag.loopTimes};
  echo 'Get the length of the dataset passed from the upstream assignment node:'${dag.input.length};
  echo 'To get data from a specific row and column in the dataset from the assignment node, access it as a two-dimensional array:'${dag.input[0][1]};

Define the loop exit condition

Use the do-while node's built-in variables for loop control. For example, compare the dag.loopTimes variable (number of loops) with the dag.input.length variable (input length). If dag.loopTimes is less than dag.input.length, the node outputs True and the loop continues. Otherwise, it outputs False and the loop exits. Sample code:

if ${dag.loopTimes}<${dag.input.length}:
    print True;
else:
    print False;

Run the node and view the results

Go to Operation Center, right-click the node, and select Retroactive Instance > Current and Descendant Nodes Retroactively. Select the assignment node and the loop node. After the run completes, check the run log for the results.

• The assignment node's run log shows the following output. Output Result is the value passed to the downstream node:

  odps output: WARNING:[0,0]  Table(xc_dpe_e2,xc_rpt_user_info_d) is full scan with all partitions, please specify partit...
  odps output:
  odps output: resource cost: cpu 0.00 Core * Min, memory 0.00 GB * Min
  odps output: inputs:
  odps output: outputs:
  odps output:
  2021-08-12 16:58:39.784 INFO - ===Output Result: [["00163359810821","Hubei Province","windows_pc","1","Female","30-40 years","Cancer"]]
  2021-08-12 16:58:40.105 INFO - cost time: 5
  2021-08-12 16:58:40 INFO ===================================================================
  2021-08-12 16:58:40 INFO Exit code of the Shell command 0
  2021-08-12 16:58:40 INFO --- Invocation of Shell command completed ---
  2021-08-12 16:58:40 INFO Shell run successfully!
  2021-08-12 16:58:40 INFO Current task status: FINISH

• In the first iteration, the end node's run log shows that the Python code if 1 < 2: print True was executed, which produced an output of True and an exit code of 0. The loop has run twice with a duration of 5s.

• In the second iteration, the end node's run log shows that the conditional statement in codeContent is if 2 < 2: print True else: print False. The Python output is False, and the log shows ===>Output Result: False. Because the loop condition is not met, the do-while loop terminates. The loop ran a total of 2 times.