themis docs src query query_engine.cpp

query_engine.cpp

Path: src/query/query_engine.cpp

Purpose: Executes query plans, integrates BM25 scoring, vector search fusion, and result ranking.

Public functions / symbols:

• if (!st.ok) {
• for (const auto& res : results) {
• if (!structStatus.ok) {
• for (const auto& pk : keys) {
• catch (...) { THEMIS_WARN("executeAndEntities: Deserialisierung fehlgeschlagen für PK={}", pk); }
• ``
• for (auto& batch : batches) {
• catch (...) { THEMIS_WARN("executeOrEntitiesWithFallback: Deserialisierung fehlgeschlagen für PK={}", pk); }
• for (size_t i = start; i < end; ++i) {
• catch (...) { THEMIS_WARN("executeOrEntities: Deserialisierung fehlgeschlagen für PK={}", pk); }
• if (!st0.ok) { child.setStatus(false, st0.message); return {Status::Error("sequential: " + st0.message), {}}; }
• if (!st.ok) { child2.setStatus(false, st.message); return {Status::Error("sequential: " + st.message), {}}; }
• catch (...) { THEMIS_WARN("executeAndEntitiesSequential: Deserialisierung fehlgeschlagen für PK={}", pk); }
• catch (...) { /* Silent failure in parallel context */ }
• for (const auto& p : q.predicates) {
• if (!v || *v != p.value) { match = false; break; }
• if (match) {
• for (const auto& r : q.rangePredicates) {
• if (!v) { match = false; break; }
• if (eff < bestEst) { bestEst = eff; bestIdx = i; bestCapped = capped; }
• if (!st.ok) { span.setStatus(false, st.message); return {st, {}}; }
• if (optimize) {
• catch (...) { THEMIS_WARN("executeAndEntitiesWithFallback: Deserialisierung fehlgeschlagen für PK={}", pk); }
• static inline size_t bigLimit() { return static_cast<size_t>(1000000000ULL); }
• if (r.column == ob.column) { lb = r.lower; ub = r.upper; il = r.includeLower; iu = r.includeUpper; break; }
• for (const auto& k : scan) {
• catch (...) { THEMIS_WARN("executeAndEntitiesRangeAware_: Deserialisierung fehlgeschlagen für PK={}", pk); }
• if constexpr (std::is_same_v<T, std::nullptr_t>) {
• for (const auto& [key, valExpr] : objConst->fields) {
• for (const auto& elemExpr : arrLit->elements) {
• if (name == "bm25") {
• if (name == "fulltext_score") {
• static void collectVariables(
• for (const auto& arg : fn->arguments) {
• for (const auto& elem : arr->elements) {
• for (const auto& [key, val] : obj->fields) {
• if (equiJoin.found) {
• for (const auto& filter : build_filters->second) {
• for (const auto& filter : probe_filters->second) {
• for (const auto& build_doc : it->second) {
• for (const auto& let : let_nodes) {
• for (const auto& filter : multi_var_filters) {
• if (bin->op == query::BinaryOperator::Eq) {
• if (return_node) {
• for (const auto& filter : push_filters->second) {
• for (const auto& filter : filters) {
• for (const auto& agg : collect->aggregations) {
• for (const auto& doc : docs) {
• for (const auto& node : reachableNodes) {
• if (node != q.start_node) {

Notes / TODOs:

• Document scoring fusion, BM25 parameterization, and how vector indices are invoked.