Processing

M
Written By Melih SünbülLast updated 5 months ago

Once data is in the system, it needs to be transformed into business insight. Lex App provides building blocks for running calculations — from single-record transforms to batch generation across hundreds of combinations — with logging, parallel execution, and automatic fallback.

flowchart LR
    A["User clicks
    Calculate"] --> B["CalculationModel
    (single record)"]
    B --> C{"Triggers
    batch?"}
    C -- Yes --> D["CalculatedModelMixin
    (combinations)"]
    C -- No --> E["calculate()"]
    D --> F{"Celery
    active?"}
    F -- Yes --> G["Parallel
    workers"]
    F -- No --> H["Synchronous"]
    E --> I["Results saved
    + LexLogger output"]
    G --> I
    H --> I

Building Blocks

Calculations

The CalculationModel base class gives you a built-in state machine (NOT_CALCULATED → IN_PROGRESS → SUCCESS), automatic error capture, and a calculate button in the UI. You implement one method — calculate() — and the framework handles everything else.

Batch Calculations

CalculatedModelMixin is the framework's combination engine. Declare the dimensions (defining_fields), tell it what values each dimension can take (get_selected_key_list()), and it generates every combination — deduplicating, clustering for parallel execution, and dispatching to Celery or running synchronously.

Celery & Async Calculations

When a calculation triggers many children, the framework can dispatch them to Celery workers in parallel. If Celery isn't available, the framework falls back to synchronous processing automatically — your code doesn't change either way.

Logging

LexLogger produces rich, Markdown-formatted log entries during calculations. Tables, headings, DataFrames, code blocks — all stored in the database and displayed in the frontend's calculation log panel. Context-aware: it automatically links logs to the correct calculation and model instance.

Was this helpful?

Your feedback shapes what we write next.