Frontmatter MCP

"""Semantic search context management.""" from dataclasses import dataclass from pathlib import Path from frontmatter_mcp.semantic.cache import EmbeddingCache from frontmatter_mcp.semantic.indexer import EmbeddingIndexer, IndexerState from frontmatter_mcp.semantic.model import EmbeddingModel from frontmatter_mcp.settings import Settings @dataclass class SemanticContext: """Context for semantic search operations.""" model: EmbeddingModel cache: EmbeddingCache indexer: EmbeddingIndexer @property def is_ready(self) -> bool: """Check if indexing is complete and ready for semantic search.""" return self.indexer.state == IndexerState.READY def get_semantic_context(settings: Settings) -> SemanticContext: """Create a semantic context instance. The model is lazy-loaded when first needed (e.g., when cache connects or when encoding text). This allows the MCP server to start quickly. Args: settings: Application settings. Returns: SemanticContext with model, cache, and indexer. """ model = EmbeddingModel(settings.embedding_model) cache = EmbeddingCache(cache_dir=settings.cache_dir, model=model) def get_files() -> list[Path]: return list(settings.base_dir.rglob("*.md")) indexer = EmbeddingIndexer(cache, model, get_files, settings.base_dir) return SemanticContext(model=model, cache=cache, indexer=indexer)