metadata-curation-client 0.7.0

API client for metadata curation platforms

Metadata Curation Client

API client for external partners to integrate with metadata curation platforms.

Installation

pip install metadata-curation-client

Basic Usage

from metadata_curation_client import CurationAPIClient, PropertyType

# Initialize client
client = CurationAPIClient("http://localhost:8000")

# Create source
source = client.create_source({
    "name": "My Archive",
    "description": "Digital editions from our collection"
})

# Create controlled vocabulary property
language_prop = client.create_property({
    "technical_name": "language",
    "name": "Language", 
    "type": PropertyType.MULTIPLE_CHOICE,
    "property_options": [{"name": "English"}, {"name": "German"}]
})

# Create free text property
description_prop = client.create_property({
    "technical_name": "description",
    "name": "Description", 
    "type": PropertyType.FREE_TEXT
})

# Create entity
entity = client.create_entity({
    "source_id": source["id"],
    "source_internal_id": "my_001"
})

# Create properties for each type
genre_prop = client.create_property({
    "technical_name": "genre",
    "name": "Genre", 
    "type": PropertyType.MULTIPLE_CHOICE,
    "property_options": [
        {"name": "Poetry"}, {"name": "Prose"}, {"name": "Drama"}
    ]
})

language_prop = client.create_property({
    "technical_name": "language",
    "name": "Language", 
    "type": PropertyType.SINGLE_CHOICE,
    "property_options": [
        {"name": "English"}, {"name": "German"}, {"name": "French"}
    ]
})

has_annotations_prop = client.create_property({
    "technical_name": "has_annotations",
    "name": "Has Annotations", 
    "type": PropertyType.BINARY
})

year_prop = client.create_property({
    "technical_name": "publication_year",
    "name": "Publication Year", 
    "type": PropertyType.NUMERICAL
})

description_prop = client.create_property({
    "technical_name": "description",
    "name": "Description", 
    "type": PropertyType.FREE_TEXT
})

# Example 1: MULTIPLE_CHOICE suggestion
properties = client.get_properties()
genre_prop = next(p for p in properties if p["technical_name"] == "genre")
poetry_option = next(opt for opt in genre_prop["property_options"] if opt["name"] == "Poetry")
client.create_suggestion({
    "source_id": source["id"],
    "entity_id": entity["id"],
    "property_id": genre_prop["id"],
    "property_option_id": poetry_option["id"]
})

# Example 2: SINGLE_CHOICE suggestion
language_prop = next(p for p in properties if p["technical_name"] == "language")
english_option = next(opt for opt in language_prop["property_options"] if opt["name"] == "English")
client.create_suggestion({
    "source_id": source["id"],
    "entity_id": entity["id"],
    "property_id": language_prop["id"],
    "property_option_id": english_option["id"]
})

# Example 3: BINARY suggestion (uses property_option_id)
# Binary properties always have options with name "1" (true/1) and "0" (false/0)
has_annotations_prop = next(p for p in properties if p["technical_name"] == "has_annotations")
true_option = next(opt for opt in has_annotations_prop["property_options"] if opt["name"] == "1")
client.create_suggestion({
    "source_id": source["id"],
    "entity_id": entity["id"],
    "property_id": has_annotations_prop["id"],
    "property_option_id": true_option["id"]  # For "yes"/"true" value
})

# Example 3: NUMERICAL suggestion (uses custom_value)
client.create_suggestion({
    "source_id": source["id"],
    "entity_id": entity["id"],
    "property_id": year_prop["id"],
    "custom_value": "2025"  # Note: numerical values are sent as strings
})

# Example 4: FREE_TEXT suggestion (uses custom_value)
client.create_suggestion({
    "source_id": source["id"],
    "entity_id": entity["id"],
    "property_id": description_prop["id"],
    "custom_value": "This is a detailed description of the entity."
})

# Mark ingestion complete
client.mark_ingestion_complete(source["id"])

Property Types

• PropertyType.SINGLE_CHOICE - Predefined options (only one allowed)
• PropertyType.MULTIPLE_CHOICE - Predefined options (multiple allowed)
• PropertyType.FREE_TEXT - Free text
• PropertyType.BINARY - True/false values
• PropertyType.NUMERICAL - Numeric values

API Reference

See the docstrings in curation_api_client.py for detailed method documentation.

Enhanced Integration with SourceManager

For more sophisticated integrations, we also provide a higher-level abstraction in source_manager.py that mirrors some of the conveniences of our internal extractors:

from metadata_curation_client import CurationAPIClient, PropertyType, SourceManager, PropertyBuilder

# Initialize client and create source
client = CurationAPIClient("http://localhost:8000")

# Define properties using helper builders
property_definitions = [
    PropertyBuilder.multiple_choice(
        "example_genre", "Genre", ["Poetry", "Prose", "Drama"]
    ),
    PropertyBuilder.binary(
        "example_has_annotations", "Has Annotations"
    ),
    PropertyBuilder.numerical(
        "example_year", "Publication Year"
    )
]

# Initialize the source manager - this will:
# - Fetch all existing data
# - Build lookup tables
# - Create any missing properties
manager = SourceManager(client, source_identifier, property_definitions)

# Efficiently get or create entity using lookup tables
entity = manager.get_or_create_entity("book_001")

# Create suggestions in a batch with validation and deduplication
manager.create_suggestions_batch(
    entity["id"],
    {
        "example_genre": "Poetry",
        "example_has_annotations": True,
        "example_year": 2022
    }
)

# Mark ingestion complete (updates timestamp)
manager.finish_ingestion()

Benefits of the SourceManager

The SourceManager provides several advantages for more complex integrations:

1. Reduced API Calls: Prefetches data to minimize API requests
2. Lookup Tables: Maintains efficient in-memory lookups for entities, properties, and suggestions
3. Automatic Property Creation: Creates properties from definitions as needed
4. Validation: Automatically validates values based on property types
5. Deduplication: Avoids creating duplicate suggestions
6. Builder Helpers: Provides convenient builder classes for creating properties and sources
7. Timestamp Management: Automatically updates the last ingestion timestamp

For a complete example, see example_with_source_manager.py.

Choosing the Right Approach

• Basic API Client: For simple integrations or when you need complete control over the process
• SourceManager: For more complex integrations where efficiency and convenience are priorities

Both approaches use the same underlying API endpoints and data models, so you can choose the one that best fits your needs or even mix them as required.