OCR

The ocr_call function extracts text from an image or PDF document using an OCR Model provider.

• Function name: extensions.com.cloud-apim.llm-extension.ocr_call

Parameters

Parameter	Type	Required	Description
provider	string	yes	The OCR model entity id
file_in	string	no	A file path on disk to read the document from
payload	object	yes	The payload object (the document and options)
payload.model	string	no	The model name (defaults to the model configured on the entity)
payload.document_url	string	no	A remote URL of the document (image or pdf)
payload.image / payload.document	string | array	no	The document content, either as a base64 string (or data-uri) or as a raw byte array
payload.bytes	array	no	The document content as a raw byte array
payload.image_base64 / payload.document_base64	string	no	The document content, base64 encoded
payload.content_type	string	no	The document content type (e.g. image/png, application/pdf)
payload.filename	string	no	The document file name
payload.pages	array	no	Optional list of page indices to process (provider dependent)
payload.pdf_password	string	no	Optional password for protected PDFs (AlphaEdge)

How the document is resolved

The document can be passed in several ways. They are resolved in the following priority order:

1. file_in — a file path read from disk
2. a raw byte array in payload.bytes, payload.image, or payload.document
3. a base64 string in payload.image_base64 or payload.document_base64
4. a base64 string or data-uri in payload.image, payload.document, or payload.content
5. a remote URL (or Mistral-style document object) resolved from the payload

A field named image or document is auto-detected: a JSON array is treated as raw bytes, a string is treated as base64 (or a data: uri).

Output

Returns the OCR result object:

{
  "model": "alpha-digit-max",
  "text": "The full extracted text...",
  "pages": [
    { "index": 0, "markdown": "The full extracted text..." }
  ],
  "usage_info": {
    "pages_processed": 1
  }
}

Example with a remote URL

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.ocr_call",
  "args": {
    "provider": "ocr-model_xxxxx",
    "payload": {
      "model": "alpha-digit-max",
      "document_url": "https://example.com/scan.pdf"
    }
  },
  "result": "ocr_result"
}

Example with a file input

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.ocr_call",
  "args": {
    "provider": "ocr-model_xxxxx",
    "file_in": "/path/to/scan.png",
    "payload": {
      "model": "alpha-digit-max",
      "content_type": "image/png"
    }
  },
  "result": "ocr_result"
}

Example with a base64 input

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.ocr_call",
  "args": {
    "provider": "ocr-model_xxxxx",
    "payload": {
      "image_base64": "JVBERi0xLjcKJ...",
      "content_type": "application/pdf",
      "model": "alpha-digit-max"
    }
  },
  "result": "ocr_result"
}

Example with a raw byte array

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.ocr_call",
  "args": {
    "provider": "ocr-model_xxxxx",
    "payload": {
      "bytes": [37, 80, 68, 70, 45, 49, 46, 55],
      "content_type": "application/pdf"
    }
  },
  "result": "ocr_result"
}