AI search engine and MCP server for bots and agents. One Rust binary indexes code, docs, logs, JSONL/CSV and PDFs for BM25 full-text, vector and hybrid search, code search without grep, agent memory and RAG. Elasticsearch-compatible API; Corpus Hub serves hundreds of reference memory corpora.
Rust agent-memoryai-agentsbm25claude-codeclicode-searchcursordatabaseelasticsearchfull-text-searchhybrid-searchjsonllocal-firstmcpmcp-servermodel-context-protocolragrustsearch-enginevector-search
perf(query): admit keyword match_phrase into the scored-columnar plan (2.7x -> WIN, bool-wrapped mp scores ES-exact)
xerj-org886d804151ff
diff --git a/engine/crates/xerj-engine/src/index.rs b/engine/crates/xerj-engine/src/index.rsindex 8f68ecd24..97611b538 100644--- a/engine/crates/xerj-engine/src/index.rs+++ b/engine/crates/xerj-engine/src/index.rs@@ -21225,6 +21225,27 @@ fn scored_fast_plan(kind: ScoredLeafKind::NumericEq(query.parse::<f64>().ok()?),boost: mult * scoring_boost(boost)?,}),+ // match_phrase on a KEYWORD field: the keyword analyzer emits one+ // whole-value token, so the "phrase" degenerates to exact term+ // equality — identical semantics AND scoring to `match` on the+ // same keyword field (the arm above; ES builds the same+ // TermQuery). Mirrors the FTS lowering in `query_node_to_fts`+ // (which routes keyword match_phrase to `FtsTerm`), but lands on+ // the columnar TopKCand path instead of the O(matches) postings+ // enumeration. `slop` is irrelevant for a single-position term+ // (Lucene ignores slop on 1-term phrases); a custom `analyzer`+ // bails to the brute path.+ QueryNode::MatchPhrase {+ field,+ query,+ analyzer: None,+ boost,+ ..+ } if fs.kw.contains(field) => Some(ScoredLeaf {+ field: field.clone(),+ kind: ScoredLeafKind::Keyword(query.clone()),+ boost: mult * scoring_boost(boost)?,+ }),_ => None,}}@@ -21366,7 +21387,7 @@ fn scored_fast_plan(tie_breaker: *tie_breaker,})}- QueryNode::Term { .. } | QueryNode::Match { .. } => {+ QueryNode::Term { .. } | QueryNode::Match { .. } | QueryNode::MatchPhrase { .. } => {Some(ScoredClause::Leaf(scoring_leaf(node, mult, fs)?))}// multi_match(best_fields) over keyword/numeric fields lowers to@@ -21610,7 +21631,8 @@ fn scored_fast_plan(| QueryNode::Boosted { .. }| QueryNode::Term { .. }| QueryNode::MultiMatch { .. }- | QueryNode::Match { .. } => {+ | QueryNode::Match { .. }+ | QueryNode::MatchPhrase { .. } => {let root = match scoring_clause(query, 1.0, &fs)? {leaf @ ScoredClause::Leaf(_) => ScoredClause::Bool {must: vec![leaf],