pyldo 0.0.8

Linked Data Objects for Python - Making RDF as easy as working with plain objects

PyLDO - Linked Data Objects for Python

PyLDO makes working with RDF data as easy as working with plain Python objects. It's the Python equivalent of the JavaScript LDO library and designed for building Solid applications.

Features

• Seamless RDF to Python object mapping - Work with RDF as Pydantic models
• ShEx schema support - Generate typed Python classes from ShEx shapes
• Solid integration - Full support for Solid Pods with DPoP authentication
• Link traversal - Automatically fetch linked resources across documents
• Reactive updates - Subscribe to data changes with SubscribableDataset
• Type safe - Full type hints and Pydantic v2 integration

Installation

pip install pyldo

Quick Start

1. Generate Python types from a ShEx schema

# Generate types from a ShEx schema file
pyldo generate profile.shex --output ./ldo/

This creates:

• profile_types.py - Pydantic models
• profile_context.py - JSON-LD context
• profile_shapetypes.py - ShapeType definitions

2. Use the generated types

from pyldo import parse_rdf
from ldo.profile_types import ProfileShape

# Parse RDF data
dataset = parse_rdf('''
    @prefix foaf: <http://xmlns.com/foaf/0.1/> .
    <#me> a foaf:Person ;
        foaf:name "Alice" ;
        foaf:knows <https://bob.example/profile#me> .
''', base_iri="https://alice.example/profile")

# Get a typed Python object
profile = dataset.using(ProfileShape).from_subject("#me")

print(profile.name)  # "Alice"

# Modify and serialize back to RDF
profile.name = "Alice Smith"
profile.sync_to_graph()
print(dataset.to_turtle())

3. Work with Solid Pods

from pyldo import SolidClient, parse_rdf
from ldo.profile_types import ProfileShape

# Create a client (add auth for private data)
client = SolidClient()

# Fetch a profile from a Solid Pod
turtle = await client.get("https://alice.solidcommunity.net/profile/card")
dataset = parse_rdf(turtle, base_iri="https://alice.solidcommunity.net/profile/card")

profile = dataset.using(ProfileShape).from_subject("#me")
print(f"Hello, {profile.name}!")

Core Concepts

LdoDataset

The main entry point for working with RDF data:

from pyldo import LdoDataset
from ldo.profile_types import PersonShape

dataset = LdoDataset()
dataset.parse_turtle(turtle_string)

# Query for typed objects
person = dataset.using(PersonShape).from_subject("https://example.com/#me")

Transactions

Track changes and generate SPARQL updates:

dataset.start_transaction()

profile.name = "New Name"
profile.sync_to_graph()

# Generate SPARQL UPDATE for the changes
sparql = dataset.to_sparql_update()
print(sparql)
# DELETE DATA { <#me> <http://xmlns.com/foaf/0.1/name> "Old Name" }
# INSERT DATA { <#me> <http://xmlns.com/foaf/0.1/name> "New Name" }

dataset.commit()

SubscribableDataset

React to data changes:

from pyldo import SubscribableDataset

dataset = SubscribableDataset()

# Subscribe to changes
def on_change(event):
    print(f"Data changed: {event.type}")

unsubscribe = dataset.subscribe(on_change)

# Subscribe to specific subjects
dataset.subscribe_to_subject(subject_uri, on_change)

LinkQuery

Traverse links across multiple resources:

from pyldo import LinkQuery, LdoDataset
from ldo.profile_shapetypes import PersonShapeType

dataset = LdoDataset()

query = LinkQuery(
    dataset=dataset,
    shape_type=PersonShapeType,
    starting_resource="https://alice.example/profile",
    starting_subject="https://alice.example/profile#me",
)

# Fetch profile and friends
result = await query.run({
    "name": True,
    "knows": {
        "name": True,  # Fetches linked profiles
    }
})

CLI Commands

# Generate Python types from ShEx
pyldo generate schema.shex --output ./ldo/

# Show version
pyldo --version

Integration Examples

FastAPI

from fastapi import FastAPI
from pyldo import parse_rdf, SolidClient
from ldo.profile_types import ProfileShape

app = FastAPI()
client = SolidClient()

@app.get("/profile/{webid:path}")
async def get_profile(webid: str):
    # pyldo models ARE Pydantic models - FastAPI serializes them!
    turtle = await client.get(webid)
    dataset = parse_rdf(turtle, base_iri=webid)
    profile = dataset.using(ProfileShape).from_subject(webid)
    return profile  # Automatic JSON serialization

LangChain

from langchain.tools import tool
from pyldo import parse_rdf, SolidClient
from ldo.profile_types import ProfileShape

client = SolidClient()

@tool
def read_solid_profile(webid: str) -> str:
    """Read a person's profile from their Solid Pod."""
    turtle = client.get_sync(webid)
    dataset = parse_rdf(turtle, base_iri=webid)
    profile = dataset.using(ProfileShape).from_subject(webid)
    return f"Name: {profile.name}"

API Reference

Parsing & Serialization

• parse_rdf(data, format, base_iri) - Parse RDF to LdoDataset
• to_turtle(ldo) - Serialize to Turtle
• to_ntriples(ldo) - Serialize to N-Triples
• to_jsonld(ldo) - Serialize to JSON-LD
• to_sparql_update(ldo) - Generate SPARQL UPDATE

Transactions

• start_transaction(ldo) - Begin tracking changes
• commit_transaction(ldo) - Apply changes
• rollback_transaction(ldo) - Discard changes
• transaction_changes(ldo) - Get pending changes

Language Support

• languages_of(ldo, property) - Get available languages
• set_language_preferences(*langs) - Set preferred languages

Querying

• match_subject(dataset, predicate, object) - Find subjects
• match_object(dataset, subject, predicate) - Find objects

Requirements

• Python 3.10+
• pydantic >= 2.0.0
• rdflib >= 7.0.0
• httpx >= 0.25.0

License

MIT

Credits

Inspired by the JavaScript LDO library by Jackson Morgan.