structured-output

README

Structured Output Example

This example demonstrates the structured output parsing and validation capabilities of go-llms.

Features Demonstrated

1. JSON Parsing with Recovery - Parse LLM outputs that may contain markdown formatting or common JSON errors
2. Schema Validation - Validate parsed data against defined schemas with detailed error reporting
3. Format Conversion - Convert between JSON, YAML, and XML formats
4. Bridge Integration - Integration with go-llmspell scripting engine

Running the Example

go run main.go

Example Output

The example shows four different use cases:

1. JSON Parsing with Recovery

Demonstrates parsing JSON from LLM output that includes markdown code blocks and trailing commas:

• Extracts JSON from markdown code blocks
• Fixes common JSON issues (trailing commas, etc.)
• Provides multiple recovery strategies

2. Schema Validation

Shows how to validate data against a schema:

• Define schemas with type constraints
• Validate required fields
• Check format constraints (email, etc.)
• Get detailed error messages and fix suggestions

3. Format Conversion

Converts data between different formats:

• JSON to YAML conversion
• JSON to XML conversion with custom root element
• Pretty printing options

4. Bridge Integration

Demonstrates go-llmspell integration:

• Convert bridge schemas to OutputSchema format
• Parse and validate LLM outputs
• Get parser information and capabilities

Code Structure

The example is organized into separate functions for each feature:

• parseJSONWithRecovery() - JSON parsing with error recovery
• validateAgainstSchema() - Schema-based validation
• convertFormats() - Format conversion examples
• bridgeIntegration() - Bridge adapter usage

Related Documentation

• Structured Output Package Documentation
• Output Parser API
• Validator API