> ## Documentation Index
> Fetch the complete documentation index at: https://docs.jaantonio.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Source data

> Understand the XML evidence, field catalog, ownership, and processing history.

`source_data` is the low-level evidence layer between the CargoWise XML archive and the current
business model.

The private archive keeps the original payload. `source_data` keeps the parsed structure needed to
trace reported values back to that payload.

## Evidence model

| Table               | Purpose                                                                                 |
| ------------------- | --------------------------------------------------------------------------------------- |
| `payloads`          | Payload ID, content hash, size, parser version, and node counts                         |
| `nodes`             | XML elements, attributes, text, paths, raw values, inferred types, and parent positions |
| `field_definitions` | Observed semantic paths and inferred value types                                        |
| `node_associations` | Proven ownership by a supported root, component, or collection                          |

The node tree preserves scalar fields even when no reporting column exists for them.

<Note>
  If Analytics cannot relate a scalar node, the node remains available and its
  `source_fields.owners` value is an empty list.
</Note>

## Processing history

| Table or pair                        | Purpose                                               |
| ------------------------------------ | ----------------------------------------------------- |
| `projection_jobs`, `projection_runs` | Live processing attempts and outcomes                 |
| `replay_jobs`, `replay_results`      | Durable progress and results for a historical rebuild |

Replay changes the current business model without changing the preserved source evidence.

## When to use it

Start with `reporting.source_fields` for scalar XML evidence in a simpler shape. Use `source_data`
directly when a consultation needs deeper field discovery, lineage, parser evidence, or replay
history.

Next, read [CargoWise mirror](/analytics/cargowise-mirror) to see how Analytics builds current
business records from this evidence.