Extract Subdocuments

This procedure extracts subdocuments from a document or file.

Before using this procedure, ensure you have learned the Document Processing Book in your agent. After learning the Book, make sure to publish your Agent and create a new Playground for it to take effect.

Overview

This operation extracts one or more subdocuments from a document or file. A subdocument is a subsection of a larger document.

Syntax

The syntax for this operation begins with extract subdocuments from the document. Nested within this statement, you must specify the subdocuments to be extracted.

⚠️ If the subdocuments to be extracted are not specified, a question will be raised.

Ways to Specify Subdocuments

To specify which subdocuments should be extracted, include one or more of the following lines under extract subdocuments from the document:

Defining the Start Page Marker

extract subdocuments from the document
 the start page marker is "{start marker}"

Specifying the Subdocument Size

extract subdocuments from the document
 the subdocument size is {size}

Specifying the Subdocument Overlap Size

extract subdocuments from the document
 the subdocument size is {size}
 the subdocument overlap size is {overlap size}

OpenAI Model Specification

Optionally, the OpenAI model can be specified for the document extraction.

extract subdocuments from the document
 the openai model is "{model}"

Data

Components

This table lists the names of the data components in this operation and their properties.

Label	Renamable	Required
the document	Yes	Required
the start page marker	No	Optional
the subdocument size	No	Optional
the subdocument overlap size	No	Optional
the openai model	No	Optional

Values

The table below lists the parameters in the operation. Parameters are placeholders for data values. In the operation's syntax, replace the parameters with your own data values as needed.

Parameter	Description	Example Value	Required
start marker	Text indicating where a subdocument should begin.	Section 2	Optional
size	The maximum number of pages per subdocument.	10	Optional
overlap size	The number of pages that overlap between consecutive subdocuments.	1	Optional
model	The LLM model used for the extraction.	gpt-4o	Optional

Examples

1. Using Start Page Marker and Additional Fields

extract subdocuments from the file
    the start page marker is "Page containing the text 'Chapter'"
    The first field is "invoice number"
    The first field's format is "number"
    The second field is "invoice date"
    The second field's format is "string"

2. Using Subdocument Size & Overlap Size

extract subdocuments from the report
 the subdocument size is 5
 the subdocument overlap size is 1

3. Using Start Marker and OpenAI Model

extract subdocuments from the invoice
 the start page marker is "Page is the beginning of a new invoice"
 the openai model is gpt-"4o-mini"