Danbooru Tag Graph

A NetworkX-based graph system for efficiently managing Danbooru tag relationships including implications, aliases, and deprecated status.

Features

• High-performance graph operations using NetworkX
• Tag relationship management (implications and aliases)
• Deprecated tag tracking
• Efficient serialization with pickle
• Transitive relationship queries
• Batch operations for optimal performance

Installation

pip install danbooru-tag-graph

Quick Start

from danbooru_tag_graph import DanbooruTagGraph

# Create a new graph
graph = DanbooruTagGraph()

# Add tags
graph.add_tag("cat", is_deprecated=False, fetched=True)
graph.add_tag("animal", is_deprecated=False, fetched=True)
graph.add_tag("feline", is_deprecated=False, fetched=True)

# Add relationships
graph.add_implication("cat", "animal")  # cat implies animal
graph.add_alias("cat", "feline")        # cat and feline are aliases

# Query relationships
implications = graph.get_implications("cat")
aliases = graph.get_aliases("cat")
transitive = graph.get_transitive_implications("cat")

# Expand tags with frequencies
expanded_tags, frequencies = graph.expand_tags(["cat"])
print(f"Expanded: {expanded_tags}")
print(f"Frequencies: {frequencies}")

# Save/load graph
graph.save_graph("my_tags.pickle")
new_graph = DanbooruTagGraph("my_tags.pickle")

Graph Structure

The tag graph uses NetworkX's MultiDiGraph with:

• Nodes: Individual tags with metadata (deprecated status, fetched status, etc.)
• Edges: Two types:
  1. 'implication': Directed edges from antecedent to consequent tags
  2. 'alias': Bidirectional edges between alias tags

Performance Benefits

• Single file load vs thousands of individual files
• In-memory graph operations instead of repeated file I/O
• Batch relationship queries and transitive closure pre-computation
• Efficient serialization with pickle

API Reference

DanbooruTagGraph

Core Methods

• add_tag(tag, is_deprecated=False, fetched=False, **metadata) - Add a tag node
• add_implication(antecedent, consequent) - Add implication relationship
• add_alias(tag1, tag2) - Add alias relationship
• is_tag_deprecated(tag) - Check if tag is deprecated
• mark_tag_fetched(tag) - Mark tag as having relationships fetched
• is_tag_fetched(tag) - Check if tag relationships are fetched

Query Methods

• get_implications(tag, include_deprecated=False) - Get direct implications
• get_aliases(tag, include_deprecated=False) - Get direct aliases
• get_transitive_implications(tag, include_deprecated=False) - Get all transitive implications
• get_alias_group(tag, include_deprecated=False) - Get complete alias group
• expand_tags(tags, include_deprecated=False) - Expand tags with frequencies

Persistence Methods

• load_graph(cache_file) - Load graph from pickle file
• save_graph(cache_file=None) - Save graph to pickle file
• auto_save() - Save if graph has been modified

Utility Methods

• stats() - Get graph statistics
• get_unfetched_tags(tags) - Get list of tags needing data fetch
• import_from_json_cache(cache_dir) - Import from old JSON cache format

License

MIT License - see LICENSE file for details.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.