Fetching Phase
Overview
It is the responsibility of this phase to load into memory everything the Process Phase needs to process the document. We’ve already loaded the config file into memory. But is that the last word? No. Here we fetch any assets that may be in Orchestrator to override the config settings. We also load in the taxonomy file. If we’re using a queue, we fetch the document, which is a transaction item, from the queue.
Fetch Phase In A Nutshell
- Call Initialize Process
- Load Taxonomy
- Fetch Assets from Orchestrator
- Call Get Transaction Item (if using Queue)
Setting Up Assets
The DU Template looks to shared settings in Orchestrator to finish initializing. Set up a new folder in your tenant called DU. Within that folder, add the following Assets:
| Asset Name | Value |
|---|---|
| DocumentUnderstandingApiKey | Get from Admin > Licenses > Consumables > AI Center |
| AlwaysValidateClassification | False |
| AlwaysValidateExtraction | False |
| SkipClassifierTraining | False |
| SkipExtractorTraining | False |
Workflow 1: Initialize Process
In Arguments
| Argument | Type | Value |
|---|---|---|
| in_AssetsSheet | String | “Assets” |
| in_ConfigFile | String | “Data\config.xlsx” |
| in_MaxAttempts | Int32 | maxAttempts |
| in_RetryInterval | TimeSpan | retryInterval |
| io_Config | Dictionary | config |
What Happens
| Step | What Happens | Config Key Used | Section | Value |
|---|---|---|---|---|
| 1 | Log Message | LogMessage_InitStarted | Constants | “Project initialization started” |
| 2 | Retry Scope | MaxExecutionAttempts | Constants | 3 |
| RetryInterval | Constants | 30 | ||
| 3 | Load Taxonomy | taxonomy.json file | ||
| 4 | Read Range | Reads just Asset sheet from config into configDT | ||
| 5 | For Each Row in configDT | ApiKey | Assets | Required – Store in Asset called DocumentUnderstandingApiKey |
| Any other asset found will be added | ||||