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

# Extraction Options

Both synchronous and asynchronous extraction endpoints accept the same request body and support the extraction options described below.

These options allow you to control:

* How documents are selected and classified
* Whether documents are split
* What additional data is returned
* How large or complex documents are handled

<Accordion title="File (Required)" icon="sparkles" iconType="regular">
  The file parameter contains the document to be extracted.

  * **Required**
  * Only **one file** may be sent per request
  * Files must be Base64-encoded

  **Supported formats**

  * pdf
  * docx
  * txt
  * tiff / tif
  * jpg / jpeg
  * png

  | **Key** | **Type** | **Value**              |
  | :------ | :------- | :--------------------- |
  | file    | File     | \[Base64-encoded file] |
</Accordion>

<Accordion title="Template Name" icon="sparkles">
  Specifies the name of the template created in the UI that should be used for extraction.

  * Required \*\*if \*\*`filterName`**is not provided**
  * Must match the template name **exactly**, including case and spacing

  Templates define the expected structure and fields extracted from a document.

  | **Key**      | **Type** | **Value**        |
  | :----------- | :------- | :--------------- |
  | templateName | Text     | \[template name] |
</Accordion>

<Accordion title="Filter Name" icon="sparkles">
  Specifies the name of a filter created in the UI.

  * Required \*\*if \*\*`templateName`**is not provided**
  * Must match the filter name **exactly**, including case and spacing

  Filters automatically determine which template to use based on the rules defined during filter creation.

  | **Key**    | **Type** | **Value**      |
  | :--------- | :------- | :------------- |
  | filterName | Text     | \[filter name] |
</Accordion>

<Accordion title="Document Splitting" icon="sparkles">
  Enables splitting a single file containing multiple documents into individual documents, which are then extracted separately within the same request.

  * Accepts `true` or `false`
  * Defaults to `false` if not specified

  This option is useful for multi-document PDFs such as scanned batches.

  | **Key**           | **Type** | **Value**  |
  | :---------------- | :------- | :--------- |
  | documentSplitting | Text     | true/false |
</Accordion>

<Accordion title="Returning Documents" icon="sparkles">
  Controls whether the extracted documents are returned as Base64-encoded PDF files.

  * Accepts `true` or `false`
  * Defaults to `false` if not specified

  This option is commonly used in combination with document splitting when storing the separated PDFs is required.

  | **Key**         | **Type** | **Value**  |
  | :-------------- | :------- | :--------- |
  | returnDocuments | Text     | true/false |
</Accordion>

<Accordion title="Returning Text" icon="sparkles">
  Controls whether all extracted text is returned as a single text block.

  * Accepts `true` or `false`
  * Defaults to `false` if not specified

  This is useful for:

  * Full-text search
  * Indexing
  * Archival or audit purposes

  | Key        | Type | Value      |
  | :--------- | :--- | :--------- |
  | returnText | Text | true/false |
</Accordion>

<Accordion title="Schema Chunking (Template Chunking)" icon="sparkles">
  Schema chunking controls how line items are extracted from documents containing repeating or tabular data.

  * Defaults to `auto` if not specified

  **Modes**

  * **Auto**\
    Extracts line items on a per-page basis, allowing documents to contain a large number of line items.\
    Each individual line item must exist entirely within a single page.
  * **Combined**\
    Combines pages during extraction, allowing line items to span multiple pages.\
    This mode limits the total number of line items that can be extracted.

  > **Recommendation:** Keep schema chunking set to `auto` unless line items span multiple pages.

  | **Key**        | **Type** | **Value**     |
  | :------------- | :------- | :------------ |
  | schemaChunking | Text     | auto/combined |
</Accordion>

<Accordion title="Password" icon="sparkles">
  Allows extraction of password-protected documents.

  If the document is password protected, include the password in the request and the system will attempt extraction using the provided password.

  | Key      | Type | Value       |
  | :------- | :--- | :---------- |
  | password | Text | \[password] |
</Accordion>