Business Logic Engine

TODO:

• Derive calculated fields
• Action definition (transitions without exit state?)
• Automatic audit history log
• References between resources
• Decide if expressions need to be async (for loading references). Or use continuations?
• DFS / Topo sort for calculated fields and references
• Input objects to transitions & actions
• Input validation on actions
• Refactor input validation to specify field that error should be reported on
• E-mails and other side effects in actions
• Automatic role assignment (ball-in-court if a condition is met, etc)
• Demo frontend (run engine client-side)
• Expression standard library
• Custom functions in system definition -> expression context
• User objects in expressions
• updateResource
• i18n strings (for error messages or otherwise)
• Visibility control (allow / deny / hide ?) for properties and objects
• Per-resource read permissions
• Deletion rules (cascade, conditions)
• Consistent exception handling
• Cache & transaction control for data loader API
• SQL data loader
• Cleaner internal object validation
• Expression language docs
• Service layer generator (REST & GraphQL)
• Expression test runner

Usage

• First, make sure you have yarn installed.
• Install dependencies with

  yarn install
  

• Run tests with

  yarn test
  

• Run tests on every change with

  yarn test:live
  

• Run only one set of tests by specifying its filename i.e.:

  yarn test:live expressions.test.ts
  

• Run the script in index.ts with

  yarn start
  

• Run on every file change with

  yarn build:live
  

Order of operations

When performing an action, the following steps are evaluated in order

1. Retrieve resource properties
2. Calculate resource calculated properties
3. TODO: Evaluate resource-relationship role expressions (i.e. "ball-in-court" or "creator")
4. TODO: Evaluate resource-level write permissions
5. Evaluate action-level conditions
6. Evaluate action-level permissions
7. Update resource state (if applicable)
8. Apply defined effects (in order)
9. Write action and effects list to history log (if applicable)

Input validation

Something like this?

{
  "editDocument": {
    "from": [
      "authoring",
      "revising"
    ],
    "input": {
      "title": {
        "type": "string",
        "constraints": [
          { "minLength": 1 },
          { "regex": "*." }
        ]
      },
      "text": {
        "type": "string"
      }
    },
    "validation": [
      {
        "invalidIf": {
          ">": [
            {
              "stringLength": {
                "getInput": "title"
              }
            },
            {
              "stringLength": {
                "getInput": "text"
              }
            }
          ]
        },
        "invalidMessage": "Document text must be longer than the title.",
        "reportOnFields": ["text", "title"]
      }
    ]
  }