obsidian-mcp 0.3.0

//! Configuration: env/CLI config for vault path, watch toggle, log level,
//! and optional search features (Tantivy BM25, embeddings).

use std::path::PathBuf;

const DEFAULT_EMBEDDINGS_MODEL: &str = "BAAI/bge-small-en-v1.5";

#[derive(Debug, Clone)]
pub struct Config {
    pub vault_path: PathBuf,
    pub watch: bool,
    pub log_level: String,
    /// Enable Tantivy BM25 full-text index (`OBSIDIAN_TANTIVY`, default `true`).
    pub tantivy: bool,
    /// Enable semantic embedding search (`OBSIDIAN_EMBEDDINGS`, default `false`).
    pub embeddings: bool,
    /// HuggingFace model name for embeddings (`OBSIDIAN_EMBEDDINGS_MODEL`).
    pub embeddings_model: String,
}

impl Config {
    /// Load configuration from CLI args and environment variables.
    ///
    /// Priority for vault path: CLI arg (first positional) > `OBSIDIAN_VAULT_PATH` env var.
    pub fn load() -> Result<Self, String> {
        let vault_path = std::env::args()
            .nth(1)
            .or_else(|| std::env::var("OBSIDIAN_VAULT_PATH").ok())
            .map(PathBuf::from)
            .ok_or_else(|| {
                "Vault path required: pass as first argument or set OBSIDIAN_VAULT_PATH".to_string()
            })?;

        let watch = std::env::var("OBSIDIAN_WATCH")
            .unwrap_or_else(|_| "true".into())
            .eq_ignore_ascii_case("true");

        let log_level = std::env::var("OBSIDIAN_LOG_LEVEL").unwrap_or_else(|_| "info".into());

        let tantivy = std::env::var("OBSIDIAN_TANTIVY")
            .unwrap_or_else(|_| "true".into())
            .eq_ignore_ascii_case("true");

        let embeddings = std::env::var("OBSIDIAN_EMBEDDINGS")
            .unwrap_or_else(|_| "false".into())
            .eq_ignore_ascii_case("true");

        let embeddings_model = std::env::var("OBSIDIAN_EMBEDDINGS_MODEL")
            .unwrap_or_else(|_| DEFAULT_EMBEDDINGS_MODEL.into());

        Ok(Self {
            vault_path,
            watch,
            log_level,
            tantivy,
            embeddings,
            embeddings_model,
        })
    }
}