Azure Content Understanding C# Client Sample

A focused .NET 8 sample application demonstrating Azure Content Understanding capabilities with quick-start guidance and examples. Draws heavily from other samples available, like https://github.com/Azure-Samples/data-extraction-using-azure-content-understanding/ and https://github.com/Azure-Samples/azure-ai-content-understanding-dotnet.

Features

• Terraform based iac deploy/provision all necessary Azure resources
• Azure Content Understanding API integration with authentication and configurable endpoints
• Health checks for key Azure resources (Content Understanding, Key Vault)
• Analyzer and Classifier samples exploring several use cases
• End-to-end document analysis pipeline
  • Create/Upload Analayzers and Classifiers
  • Run analysis of single documents
  • Run classification and analysis on single/multiple documents
• Simple CLI operations with local export of results

📋 Prerequisites

• .NET 8 SDK
• (Optional) Azure CLI for authentication (az login)
• (Optional) Terraform if you want to deploy the sample infra in iac/

🚀 Quick Start

1. (Optional) Deploy infrastructure in iac/ using the provided scripts.

2. Build and run the client

# From the repo root
cd src/ContentUnderstanding.Client

# Build
dotnet build

# Run (interactive by default)
dotnet run

# Preferred: subcommands (System.CommandLine)
dotnet run -- --use-cli health
dotnet run -- --use-cli analyzers
dotnet run -- --use-cli analyze --analyzer receipt --document receipt1.pdf
dotnet run -- --use-cli classifiers
dotnet run -- --use-cli classify --classifier <name> --document <file>
dotnet run -- --use-cli classify-dir --classifier <name> --directory <subfolder>
### Classify a whole directory

Classify all supported files in a subfolder under Data/SampleDocuments using a classifier:

dotnet run --project .\src\ContentUnderstanding.Client -- --use-cli classify-dir --classifier <name> --directory <subfolder>

• Non-recursive: only files directly in <subfolder> are processed.
• Supported types: .pdf, .png, .jpg, .jpeg, .tif, .tiff, .bmp.
• Sequential processing with per-file error logging; the run continues on errors.
• Outputs: per-file JSON and formatted text results in Output/ plus a mandatory batch summary:
  • batch_<directory>_<classifier>_<timestamp>_summary.json

📖 Usage Guide

Command-Line Interface

Preferred: subcommands via System.CommandLine

# Show comprehensive help
dotnet run -- --use-cli

# Health check core Azure resources (Content Understanding, Key Vault, Managed Identity)
dotnet run -- --use-cli health

# List all available analyzers
dotnet run -- --use-cli analyzers

Creating Analyzers

# Create default analyzer (receipt)
dotnet run -- --use-cli create-analyzer

# Create specific analyzer by file name
dotnet run -- --use-cli create-analyzer --analyzer-file receipt.json

Document Analysis

# Use all defaults (receipt1.pdf + receipt analyzer)
dotnet run -- --use-cli analyze

# Document-specific analysis
dotnet run -- --use-cli analyze --document receipt.png

# Analyzer-specific analysis
dotnet run -- --use-cli analyze --analyzer enginemanual

# Full control
dotnet run -- --use-cli analyze --analyzer receipt --document receipt.png

# Use absolute paths for documents outside the project
dotnet run -- --use-cli analyze --document "C:\\path\\to\\my\\document.pdf"

Supported File Formats

The application automatically detects content types for:

• PDF: .pdf (application/pdf)
• Images: .png, .jpg, .jpeg, .tiff, .bmp

📋 CLI Reference

Preferred subcommands:

dotnet run -- --use-cli <command> [options]

Common commands:

• health
• analyzers
• analyze --analyzer --document
• check-operation --operation-id
• classifiers
• create-classifier --classifier --classifier-file
• create-analyzer --analyzer --analyzer-file
• classify --classifier --document
• classify-dir --classifier --directory

Project layout

azure-ai-content-understanding-basic/
├── src/
│   └── ContentUnderstanding.Client/       # Main console application (namespace ContentUnderstanding.Client)
│       ├── Program.cs                     # Main entry point with parameterized CLI
│       ├── Services/                      # HTTP service layer (ContentUnderstanding.Client.Services)
│       ├── Data/                          # Analyzer schemas and sample documents
│       ├── Models/                        # DTOs (ContentUnderstanding.Models namespace)
│       └── Output/                        # Analysis results export (git-ignored)
├── iac/                                   # Infrastructure as Code (Terraform)
├── docs/                                  # Documentation
└── README.md

Configuration

The application uses appsettings.json for non-sensitive configuration and Azure Key Vault or environment variables for secrets. See docs/CONFIGURATION.md for details.

Authentication

The application uses DefaultAzureCredential for authentication, supporting:

• Azure CLI (az login)
• Visual Studio credentials
• Environment variables
• Managed Identity (in Azure)

Secret Management

• API keys are typically stored in Azure Key Vault
• The application looks for the ai-services-key secret by default
• No hardcoded credentials in source code

Note: Key Vault access may be restricted by network rules. For local development you can set the API key via environment variables or appsettings.Development.json if preferred.

Documentation

• docs/CONFIGURATION.md - Configuration guide
• docs/initial_plan.md - Project plan and status
• Azure Content Understanding docs: https://learn.microsoft.com/en-us/azure/ai-services/content-understanding/

License

This project is licensed under the MIT License.