HPD-TextExtract 0.5.5

HPD-TextExtract

.NET-native text and document extraction library with rich PDF structure, diagnostics, OCR planning, and layout-aware outputs.

Install

dotnet add package HPD-TextExtract

Use When

Use this package when an app or library needs to turn files, byte payloads, or URLs into text plus structured extraction metadata.

Supported inputs include:

• PDF
• Plain text, Markdown, JSON, and XML
• HTML and web URLs
• Microsoft Word, Excel, and PowerPoint Open XML documents
• Images through an injected OCR engine

For HPD agent middleware, use HPD-Agent.TextExtraction, which builds on this package.

Quick Start

using HPD.TextExtract;

using var extractor = new TextExtractionUtility();
var result = await extractor.ExtractTextAsync("document.pdf");

if (!result.IsSuccess)
{
    throw new InvalidOperationException(result.ErrorMessage);
}

Console.WriteLine(result.ExtractedText);

Binary Payloads

using HPD.TextExtract;
using HPD.TextExtract.Models;

var bytes = await File.ReadAllBytesAsync("document.pdf");

using var extractor = new TextExtractionUtility();
var result = await extractor.ExtractTextAsync(
    bytes,
    mimeType: MimeTypes.Pdf,
    fileName: "document.pdf");

Dependency Injection

using HPD.TextExtract;

builder.Services.AddTextExtraction();

Register custom decoders or OCR engines when the built-in behavior is not enough:

using HPD.TextExtract;
using HPD.TextExtract.Decoders;

builder.Services.AddTextExtractionWithOcr<MyOcrEngine>();

PDF Notes

PDF extraction is powered by PDFium through PDFiumCore.

The PDF pipeline extracts native text, glyph geometry, font metadata, colors, embedded image regions, optional screenshots, and diagnostics. It can also plan OCR for scanned or low-quality pages when an OCR executor is configured.

PDFium is a native dependency. The NuGet dependency brings platform-specific native assets, including macOS arm64 and x64, Windows x64/x86, and Linux x64 through the upstream PDFium packages. Native libraries are deployed as sidecar runtime assets; they are not embedded inside HPD.TextExtract.dll.

Output Shape

TextExtractionResult gives the simple text view:

• IsSuccess
• ExtractedText
• FileName
• MimeType
• ProcessingTime
• ErrorMessage

For richer callers, TextExtractionResult.Extraction exposes:

• Content.Sections
• Pages
• TextItems
• Assets
• Diagnostics
• Metadata

Target Frameworks

This package targets the repo-standard modern frameworks:

• net8.0
• net9.0
• net10.0

It is configured for trimming, single-file analysis, and Native AOT analysis.