scryer-engine 0.2.0

//! Generation-checked LRU cache of per-project symbol search indexes.
//!
//! Every successful index write bumps the owning project's generation; a cached index built
//! at an older generation is rebuilt wholesale on its next query. Symbol IDs change on every
//! re-ingest, so nothing is ever patched in place.

use std::collections::HashMap;
use std::num::NonZeroUsize;
use std::sync::{Arc, Mutex};

use lru::LruCache;
use rayon::iter::{IntoParallelIterator, ParallelIterator};
use scryer_db::{ScryerDb, SourceFile, Symbol};

use super::bm25::{Bm25Builder, Bm25Index};
use super::tokenize::{tokenize_identifier, tokenize_text};

/// `(project_id, dependency_package_id)`. Project 0 with `None` covers every dependency.
pub type IndexKey = (u64, Option<u64>);

/// Field weights: name, qualified-path segments, signature, docstring, file path.
const SYMBOL_FIELD_WEIGHTS: [f32; 5] = [3.0, 2.0, 1.0, 1.0, 1.0];

/// Docstring prefix retained per symbol (used for `like_symbol` queries).
const DOC_SNIPPET_CHARS: usize = 200;

/// Compact per-symbol payload stored in the index; results are served straight from it.
#[derive(Debug, Clone)]
pub struct SymbolDoc {
    pub id: u64,
    pub file_id: u64,
    pub kind: String,
    pub name: String,
    pub qualified_name: String,
    pub signature: String,
    pub doc_snippet: String,
    pub start_line: u32,
    pub end_line: u32,
    pub dependency_package_id: Option<u64>,
    pub file_path: String,
}

impl SymbolDoc {
    /// The qualified name without the trailing symbol name (`a::b::c` → `a::b`).
    pub fn parent_path(&self) -> &str {
        self.qualified_name
            .rsplit_once("::")
            .or_else(|| self.qualified_name.rsplit_once('.'))
            .map(|(parent, _)| parent)
            .unwrap_or("")
    }
}

/// A built index tagged with the generation it reflects.
#[derive(Debug)]
pub struct SymbolIndex {
    pub generation: u64,
    pub index: Bm25Index<SymbolDoc>,
}

/// LRU (16 entries) of symbol indexes with per-project generations and per-key build locks.
pub struct SearchIndexCache {
    generations: Mutex<HashMap<u64, u64>>,
    entries: tokio::sync::Mutex<LruCache<IndexKey, Arc<SymbolIndex>>>,
    build_locks: Mutex<HashMap<IndexKey, Arc<tokio::sync::Mutex<()>>>>,
}

impl Default for SearchIndexCache {
    fn default() -> Self {
        Self::new(16)
    }
}

impl SearchIndexCache {
    pub fn new(capacity: usize) -> Self {
        Self {
            generations: Mutex::default(),
            entries: tokio::sync::Mutex::new(LruCache::new(
                NonZeroUsize::new(capacity.max(1)).expect("capacity is non-zero"),
            )),
            build_locks: Mutex::default(),
        }
    }

    /// Current generation of `project_id`.
    pub fn generation(&self, project_id: u64) -> u64 {
        let gens = self.generations.lock().unwrap_or_else(|e| e.into_inner());
        gens.get(&project_id).copied().unwrap_or(0)
    }

    /// Mark every cached index of `project_id` stale.
    pub fn bump(&self, project_id: u64) {
        let mut gens = self.generations.lock().unwrap_or_else(|e| e.into_inner());
        *gens.entry(project_id).or_insert(0) += 1;
    }

    /// Bump the generation and drop every cached index of `project_id`.
    pub async fn invalidate(&self, project_id: u64) {
        self.bump(project_id);
        let mut entries = self.entries.lock().await;
        let stale: Vec<IndexKey> = entries
            .iter()
            .map(|(k, _)| *k)
            .filter(|k| k.0 == project_id)
            .collect();
        for key in stale {
            entries.pop(&key);
        }
    }

    async fn lookup(&self, key: IndexKey, generation: u64) -> Option<Arc<SymbolIndex>> {
        let mut entries = self.entries.lock().await;
        entries
            .get(&key)
            .filter(|e| e.generation == generation)
            .cloned()
    }

    /// Return the index for `key`, rebuilding it from the database if missing or stale.
    ///
    /// The DB lock is held only while loading rows; tokenization runs on `rayon_pool` under
    /// `spawn_blocking`. Concurrent callers for the same key wait for a single build.
    pub async fn get_or_build(
        &self,
        db: &ScryerDb,
        rayon_pool: &Arc<rayon::ThreadPool>,
        key: IndexKey,
    ) -> anyhow::Result<Arc<SymbolIndex>> {
        if let Some(hit) = self.lookup(key, self.generation(key.0)).await {
            return Ok(hit);
        }

        let build_lock = {
            let mut locks = self.build_locks.lock().unwrap_or_else(|e| e.into_inner());
            Arc::clone(locks.entry(key).or_default())
        };
        let _build_guard = build_lock.lock().await;

        let generation = self.generation(key.0);
        if let Some(hit) = self.lookup(key, generation).await {
            return Ok(hit);
        }

        let timer = std::time::Instant::now();
        let (symbols, paths) = load_rows(db, key).await?;
        let load_ms = timer.elapsed().as_millis();
        let pool = Arc::clone(rayon_pool);
        let index =
            tokio::task::spawn_blocking(move || pool.install(|| build_index(symbols, paths)))
                .await?;
        tracing::debug!(
            "search index built project={} package={:?} symbols={} load={}ms total={}ms",
            key.0,
            key.1,
            index.len(),
            load_ms,
            timer.elapsed().as_millis()
        );

        let built = Arc::new(SymbolIndex { generation, index });
        self.entries.lock().await.put(key, Arc::clone(&built));
        Ok(built)
    }
}

async fn load_rows(
    db: &ScryerDb,
    (project_id, package_id): IndexKey,
) -> anyhow::Result<(Vec<Symbol>, HashMap<u64, String>)> {
    let mut guard = db.lock().await;
    let (symbols, files) = match package_id {
        Some(pkg) => {
            let symbols = Symbol::filter(
                Symbol::fields()
                    .project_id()
                    .eq(project_id)
                    .and(Symbol::fields().dependency_package_id().eq(Some(pkg))),
            )
            .exec(&mut *guard)
            .await?;
            let files = SourceFile::filter(
                SourceFile::fields()
                    .project_id()
                    .eq(project_id)
                    .and(SourceFile::fields().dependency_package_id().eq(Some(pkg))),
            )
            .exec(&mut *guard)
            .await?;
            (symbols, files)
        }
        None => {
            let symbols = Symbol::filter(Symbol::fields().project_id().eq(project_id))
                .exec(&mut *guard)
                .await?;
            let files = SourceFile::filter(SourceFile::fields().project_id().eq(project_id))
                .exec(&mut *guard)
                .await?;
            (symbols, files)
        }
    };
    let paths = files.into_iter().map(|f| (f.id, f.path)).collect();
    Ok((symbols, paths))
}

fn build_index(symbols: Vec<Symbol>, paths: HashMap<u64, String>) -> Bm25Index<SymbolDoc> {
    let docs: Vec<(SymbolDoc, Vec<Vec<String>>)> = symbols
        .into_par_iter()
        .filter(|s| s.kind != "reexport")
        .map(|s| {
            let doc = SymbolDoc {
                id: s.id,
                file_id: s.file_id,
                file_path: paths.get(&s.file_id).cloned().unwrap_or_default(),
                doc_snippet: s
                    .docstring
                    .as_deref()
                    .map(|d| d.chars().take(DOC_SNIPPET_CHARS).collect())
                    .unwrap_or_default(),
                kind: s.kind,
                name: s.name,
                qualified_name: s.qualified_name,
                signature: s.signature,
                start_line: s.start_line,
                end_line: s.end_line,
                dependency_package_id: s.dependency_package_id,
            };
            let fields = vec![
                tokenize_identifier(&doc.name),
                tokenize_text(doc.parent_path()),
                tokenize_text(&doc.signature),
                tokenize_text(s.docstring.as_deref().unwrap_or("")),
                tokenize_text(&doc.file_path),
            ];
            (doc, fields)
        })
        .collect();

    let mut builder = Bm25Builder::new(&SYMBOL_FIELD_WEIGHTS);
    for (doc, fields) in docs {
        builder.add(doc, &fields);
    }
    builder.build()
}