Need configurable schemas for different vendor invoice formats

Yes — the right way to handle multiple vendor invoice formats is to make your schema configurable per vendor instead of hardcoding one invoice shape.

Recommended approach

1) Define a common core invoice model

Keep a shared base structure for fields that most invoices have:

• invoice_number
• invoice_date
• vendor_name
• vendor_address
• subtotal
• tax
• total
• currency
• line_items

2) Add vendor-specific schema extensions

Each vendor can override or extend the base schema with its own fields and parsing rules, for example:

• PO number
• customer account ID
• service period
• payment terms
• tax breakdown
• custom line-item attributes

3) Store schema config separately

Use a schema registry or configuration file per vendor, such as:

• JSON Schema
• YAML config
• database-backed schema definitions

Example:

{
  "vendor_id": "acme_corp",
  "base_schema": "invoice_v1",
  "fields": {
    "invoice_number": {"required": true, "path": "$.header.inv_no"},
    "invoice_date": {"required": true, "path": "$.header.date"},
    "total": {"required": true, "path": "$.summary.total_amount"},
    "po_number": {"required": false, "path": "$.header.po"}
  }
}

4) Support mapping rules

If invoices come from OCR or EDI/PDF extraction, add mapping rules like:

• field aliases
• regex patterns
• conditional logic
• default values
• nested paths

5) Version your schemas

Vendors often change formats. Use versioning like:

• vendor_x_invoice_v1
• vendor_x_invoice_v2

This makes changes safe and traceable.

---

Example JSON Schema pattern

Base schema

{
  "$id": "invoice_base",
  "type": "object",
  "required": ["invoice_number", "invoice_date", "vendor_name", "total"],
  "properties": {
    "invoice_number": { "type": "string" },
    "invoice_date": { "type": "string", "format": "date" },
    "vendor_name": { "type": "string" },
    "total": { "type": "number" },
    "currency": { "type": "string" }
  }
}

Vendor-specific schema

{
  "$id": "vendor_acme_invoice",
  "allOf": [
    { "$ref": "invoice_base" },
    {
      "type": "object",
      "properties": {
        "po_number": { "type": "string" },
        "service_period": { "type": "string" }
      }
    }
  ]
}

---

Best practices

• Use a canonical invoice model internally
• Keep vendor-specific parsing isolated
• Make fields optional by default, required per vendor
• Add validation and error reporting
• Support schema versioning
• Keep sample documents for each vendor to test extraction

---

If you want implementation options

I can help you build this in any of these ways:

1. JSON Schema-based design
2. Database schema registry
3. Python/Pydantic models
4. Node.js/TypeScript interfaces
5. OCR/invoice extraction pipeline

If you want, I can generate a complete configurable invoice schema design with examples for 2–3 different vendors.