DataExtractionAgentTools Class

Provides AI agent tools for extracting structured data from PDF documents and images. Supports extraction of text, forms, tables, and images as JSON output. Combines Smart Data Extractor and Smart Table Extractor capabilities.

Inheritance

System.Object

AgentToolBase

DataExtractionAgentTools

Inherited Members

AgentToolBase.GetTools()

Namespace: Syncfusion.AI.AgentTools.DataExtraction

Assembly: Syncfusion.DocumentSDK.AI.AgentTools.dll

Syntax

public class DataExtractionAgentTools : AgentToolBase

Constructors

DataExtractionAgentTools(DocumentStorageManager)

Initializes a new instance for DocumentStorage mode.

Declaration

public DataExtractionAgentTools(DocumentStorageManager storageManager)

Parameters

Type	Name	Description
DocumentStorageManager	storageManager	The document storage manager to read documents from.

DataExtractionAgentTools(String)

Initializes a new instance of the DataExtractionAgentTools class.

Declaration

public DataExtractionAgentTools(string outputDirectory = null)

Parameters

Type	Name	Description
System.String	outputDirectory	Optional output directory for saving JSON files.

Methods

ConvertPdfTableToMarkdown(String, Double, Int32, Int32, Boolean, String)

Converts tables from a PDF document into Markdown format.

Declaration

[Tool(Name = "ConvertPdfTableToMarkdown", Description = "Converts tables from PDF documents and scanned images into Markdown (MD) format. It analyzes visual table structures, including bordered and border-less tables, enabling developers to programmatically convert tabular content into clean and well-structured Markdown representations.")]
public AgentToolResult ConvertPdfTableToMarkdown([ToolParameter(Description = "Path to input PDF file")] string inputFilePath, [ToolParameter(Description = "Confidence threshold for table extraction (0.0-1.0)")] double confidenceThreshold = 0.6, [ToolParameter(Description = "Start page number for extraction (1-based). Use -1 for all pages")] int startPage = -1, [ToolParameter(Description = "End page number for extraction (1-based). Use -1 for all pages")] int endPage = -1, [ToolParameter(Description = "Enable detection of border-less tables")] bool detectBorderlessTables = true, [ToolParameter(Description = "Optional output Markdown (.md) file path")] string outputFilePath = null)

Parameters

Type	Name	Description
System.String	inputFilePath	Path to input PDF or image file.
System.Double	confidenceThreshold	Confidence threshold for table extraction (0.0-1.0).
System.Int32	startPage	Start page number for extraction (1-based). Use -1 for all pages.
System.Int32	endPage	End page number for extraction (1-based). Use -1 for all pages.
System.Boolean	detectBorderlessTables	Enable detection of border-less tables.
System.String	outputFilePath	Optional output Markdown (.md) file path.

Returns

Type	Description
AgentToolResult	Result containing the extracted table Markdown data or error message.

ConvertPdfToMarkdown(String, Boolean, Double, Int32, Int32, Boolean, String)

Converts a PDF document or image file into Markdown by extracting structured data.

Declaration

[Tool(Name = "ConvertPdfToMarkdown", Description = "Converts structured information from PDF documents and scanned images into Markdown (MD) format. It analyzes text blocks, tables and headers to preserve the original layout and formatting in the generated Markdown output.")]
public AgentToolResult ConvertPdfToMarkdown([ToolParameter(Description = "Path to input PDF file")] string inputFilePath, [ToolParameter(Description = "Enable table detection in the document")] bool enableTableDetection = true, [ToolParameter(Description = "Confidence threshold for extraction (0.0-1.0). Higher values return only high-confidence results")] double confidenceThreshold = 0.6, [ToolParameter(Description = "Start page number for extraction (1-based). Use -1 for all pages")] int startPage = -1, [ToolParameter(Description = "End page number for extraction (1-based). Use -1 for all pages")] int endPage = -1, [ToolParameter(Description = "Enable detection of border-less tables")] bool detectBorderlessTables = true, [ToolParameter(Description = "Optional output Markdown (.md) file path")] string outputFilePath = null)

Parameters

Type	Name	Description
System.String	inputFilePath	Path to input PDF or image file.
System.Boolean	enableTableDetection	Enable table detection in the document.
System.Double	confidenceThreshold	Confidence threshold for extraction (0.0-1.0). Higher values return only high-confidence results.
System.Int32	startPage	Start page number for extraction (1-based). Use -1 for all pages.
System.Int32	endPage	End page number for extraction (1-based). Use -1 for all pages.
System.Boolean	detectBorderlessTables	Enable detection of border-less tables.
System.String	outputFilePath	Optional output Markdown (.md) file path.

Returns

Type	Description
AgentToolResult	Result containing the extracted Markdown data or error message.

ExtractDataAsJson(String, Boolean, Boolean, Double, Int32, Int32, Boolean, Boolean, Boolean, Boolean, Boolean, String)

Extracts structured data including text, forms, tables, and images from a PDF document or image file.

Declaration

[Tool(Name = "ExtractDataAsJson", Description = "Extracts structured data including text, forms, tables, and images from a PDF document or image file and returns as JSON string. Supports form field detection with configurable options (signatures, textboxes, checkboxes, radio buttons) and table detection with border-less table support. Page range and confidence threshold can be specified for fine-tuned extraction.")]
public AgentToolResult ExtractDataAsJson([ToolParameter(Description = "Path to input PDF or image file (.pdf, .png, .jpg, .jpeg)")] string inputFilePath, [ToolParameter(Description = "Enable form field detection in the document")] bool enableFormDetection = true, [ToolParameter(Description = "Enable table detection in the document")] bool enableTableDetection = true, [ToolParameter(Description = "Confidence threshold for extraction (0.0-1.0). Higher values return only high-confidence results")] double confidenceThreshold = 0.6, [ToolParameter(Description = "Start page number for extraction (1-based). Use -1 for all pages")] int startPage = -1, [ToolParameter(Description = "End page number for extraction (1-based). Use -1 for all pages")] int endPage = -1, [ToolParameter(Description = "Detect signature fields in forms")] bool detectSignatures = true, [ToolParameter(Description = "Detect textbox fields in forms")] bool detectTextboxes = true, [ToolParameter(Description = "Detect checkbox fields in forms")] bool detectCheckboxes = true, [ToolParameter(Description = "Detect radio button fields in forms")] bool detectRadioButtons = true, [ToolParameter(Description = "Enable detection of border-less tables")] bool detectBorderlessTables = true, [ToolParameter(Description = "Optional output JSON file path to save the extracted data")] string outputFilePath = null)

Parameters

Type	Name	Description
System.String	inputFilePath	Path to input PDF or image file (.pdf, .png, .jpg, .jpeg).
System.Boolean	enableFormDetection	Enable form field detection in the document.
System.Boolean	enableTableDetection	Enable table detection in the document.
System.Double	confidenceThreshold	Confidence threshold for extraction (0.0-1.0). Higher values return only high-confidence results.
System.Int32	startPage	Start page number for extraction (1-based). Use -1 for all pages.
System.Int32	endPage	End page number for extraction (1-based). Use -1 for all pages.
System.Boolean	detectSignatures	Detect signature fields in forms.
System.Boolean	detectTextboxes	Detect textbox fields in forms.
System.Boolean	detectCheckboxes	Detect checkbox fields in forms.
System.Boolean	detectRadioButtons	Detect radio button fields in forms.
System.Boolean	detectBorderlessTables	Enable detection of border-less tables.
System.String	outputFilePath	Optional output JSON file path to save the extracted data.

Returns

Type	Description
AgentToolResult	Result containing the extracted JSON data or error message.

ExtractFormFieldsAsJson(String, Boolean, Boolean, Boolean, Boolean, Double, Int32, Int32, String)

Extracts only form field data from a PDF document and returns as JSON string.

Declaration

[Tool(Name = "ExtractFormFieldsAsJson", Description = "Extracts only form data from a PDF document and returns as JSON string. Optimized for form-focused extraction with support  page range specification, and confidence thresholding. Use this method when you only need form data without table or other content.")]
public AgentToolResult ExtractFormFieldsAsJson([ToolParameter(Description = "Path to input PDF file")] string inputFilePath, [ToolParameter(Description = "Detect signature fields in forms")] bool detectSignatures = true, [ToolParameter(Description = "Detect textbox fields in forms")] bool detectTextboxes = true, [ToolParameter(Description = "Detect checkbox fields in forms")] bool detectCheckboxes = true, [ToolParameter(Description = "Detect radio button fields in forms")] bool detectRadioButtons = true, [ToolParameter(Description = "Confidence threshold for form recognition (0.0-1.0). Higher values return only high-confidence form fields")] double confidenceThreshold = 0.6, [ToolParameter(Description = "Start page number for recognition (1-based). Use -1 for all pages")] int startPage = -1, [ToolParameter(Description = "End page number for recognition (1-based). Use -1 for all pages")] int endPage = -1, [ToolParameter(Description = "Optional output JSON file path to save the recognized form fields")] string outputFilePath = null)

Parameters

Type	Name	Description
System.String	inputFilePath	Path to input PDF file.
System.Boolean	detectSignatures	Detect signature fields in forms.
System.Boolean	detectTextboxes	Detect textbox fields in forms.
System.Boolean	detectCheckboxes	Detect checkbox fields in forms.
System.Boolean	detectRadioButtons	Detect radio button fields in forms.
System.Double	confidenceThreshold	Confidence threshold for form recognition (0.0-1.0). Higher values return only high-confidence tables.
System.Int32	startPage	Start page number for recognition (1-based). Use -1 for all pages.
System.Int32	endPage	End page number for recognition (1-based). Use -1 for all pages.
System.String	outputFilePath	Optional output JSON file path to save the recognized form fields data.

Returns

Type	Description
AgentToolResult	Result containing the recognized form fields JSON data or error message.

ExtractTableAsJson(String, Boolean, Double, Int32, Int32, String)

Extracts only table data from a PDF document and returns as JSON string.

Declaration

[Tool(Name = "ExtractTableAsJson", Description = "Extracts only table data from a PDF document and returns as JSON string. Optimized for table-focused extraction with support for border-less table detection, page range specification, and confidence thresholding. Use this method when you only need table data without form fields or other content.")]
public AgentToolResult ExtractTableAsJson([ToolParameter(Description = "Path to input PDF file")] string inputFilePath, [ToolParameter(Description = "Enable detection of border-less tables")] bool detectBorderlessTables = true, [ToolParameter(Description = "Confidence threshold for table extraction (0.0-1.0). Higher values return only high-confidence tables")] double confidenceThreshold = 0.6, [ToolParameter(Description = "Start page number for extraction (1-based). Use -1 for all pages")] int startPage = -1, [ToolParameter(Description = "End page number for extraction (1-based). Use -1 for all pages")] int endPage = -1, [ToolParameter(Description = "Optional output JSON file path to save the extracted table data")] string outputFilePath = null)

Parameters

Type	Name	Description
System.String	inputFilePath	Path to input PDF file.
System.Boolean	detectBorderlessTables	Enable detection of border-less tables.
System.Double	confidenceThreshold	Confidence threshold for table extraction (0.0-1.0). Higher values return only high-confidence tables.
System.Int32	startPage	Start page number for extraction (1-based). Use -1 for all pages.
System.Int32	endPage	End page number for extraction (1-based). Use -1 for all pages.
System.String	outputFilePath	Optional output JSON file path to save the extracted table data.

Returns

Type	Description
AgentToolResult	Result containing the extracted table JSON data or error message.