Getting Started

pip install parsec-llm

# OpenAI
from parsec.models.adapters import OpenAIAdapter
adapter = OpenAIAdapter(api_key="your-key", model="gpt-4o-mini")
 
# Anthropic
from parsec.models.adapters import AnthropicAdapter
adapter = AnthropicAdapter(api_key="your-key", model="claude-3-5-sonnet-20241022")
 
# Gemini
from parsec.models.adapters import GeminiAdapter
adapter = GeminiAdapter(api_key="your-key", model="gemini-pro")
 
# Ollama
from parsec.models.adapters import OllamaAdapter
adapter = OllamaAdapter(model="llama3", base_url="http://localhost:11434")

from parsec.validators import JSONValidator
from parsec import EnforcementEngine
 
# Set up enforcement
validator = JSONValidator()
engine = EnforcementEngine(adapter, validator, max_retries=3)
 
# Define what you want
schema = {
    "type": "object",
    "properties": {
        "name": {"type": "string"},
        "age": {"type": "integer"},
        "email": {"type": "string"}
    },
    "required": ["name", "email"]
}
 
# Get structured output
result = await engine.enforce(
    "Extract: John Doe is 30, contact at john@example.com",
    schema
)
 
print(result.parsed_output)
# {"name": "John Doe", "age": 30, "email": "john@example.com"}

from parsec.cache import InMemoryCache
 
# Create cache
cache = InMemoryCache(max_size=100, default_ttl=3600)
 
# Add to engine
engine = EnforcementEngine(adapter, validator, cache=cache)
 
# First call hits API
result1 = await engine.enforce(prompt, schema)
 
# Second identical call uses cache (no API call!)
result2 = await engine.enforce(prompt, schema)
 
# Check stats
print(cache.get_stats())
# {'hits': 1, 'misses': 1, 'hit_rate': '50.00%'}

from parsec.prompts import PromptTemplate, TemplateRegistry, TemplateManager
 
# Create template
template = PromptTemplate(
    name="extract_person",
    template="Extract person info from: {text}\n\nReturn as JSON.",
    variables={"text": str},
    required=["text"]
)
 
# Register with version
registry = TemplateRegistry()
registry.register(template, "1.0.0")
 
# Use with enforcement
manager = TemplateManager(registry, engine)
result = await manager.enforce_with_template(
    template_name="extract_person",
    variables={"text": "John Doe, age 30"},
    schema=schema
)
 
# Save templates to file
registry.save_to_disk("templates.yaml")