feat: emit parent_query_id to link nested SPI queries

.github/workflows/ci-tap.yml

@@ -42,6 +42,13 @@ jobs:
 
       - uses: ./.github/actions/setup-vcpkg
 
+      - name: Install uv
+        # uv is used by Arrow-IPC TAP tests (t/026, t/028) to run inline
+        # pyarrow scripts that decode the debug_arrow_dump_dir output.
+        uses: astral-sh/setup-uv@v6
+        with:
+          version: "latest"
+
       - name: Cache PostgreSQL build
         id: cache-pg
         uses: actions/cache@v4

docker/init/00-schema.sql

@@ -53,6 +53,8 @@ CREATE TABLE pg_stat_ch.events_raw
 
     query_id Int64 COMMENT '64-bit hash identifying normalized queries. Queries differing only in constants share the same query_id. Use for aggregating statistics across similar queries.',
 
+    parent_query_id Int64 COMMENT 'query_id of the calling query (e.g. the plpgsql function that issued this SPI statement). 0 for top-level queries. Use WHERE parent_query_id = 0 to restrict aggregations to top-level queries and avoid double-counting CPU and duration. Signedness matches query_id so the two columns compare/join without explicit casts.',
+
     cmd_type LowCardinality(String) COMMENT 'Command type: SELECT, INSERT, UPDATE, DELETE, MERGE, UTILITY, or UNKNOWN. Use for workload characterization (read-heavy vs write-heavy).',
 
     rows UInt64 COMMENT 'Rows returned (SELECT) or affected (INSERT/UPDATE/DELETE). HIGH: large result sets or bulk operations. LOW: point queries. Watch for unexpected HIGH values indicating missing WHERE clauses.',

migrations/001_add_parent_query_id.sql

@@ -0,0 +1,12 @@
+-- Migration: add parent_query_id column
+--
+-- Introduced in pg_stat_ch 0.4.x. Each event now carries the query_id of its
+-- calling query (e.g. the plpgsql function that issued an SPI statement).
+-- Top-level queries emit 0. Use WHERE parent_query_id = 0 in aggregations to
+-- avoid double-counting CPU and duration across nested calls.
+--
+-- Run against your ClickHouse instance before upgrading the extension:
+--   clickhouse-client < migrations/001_add_parent_query_id.sql
+
+ALTER TABLE pg_stat_ch.events_raw
+    ADD COLUMN IF NOT EXISTS parent_query_id Int64 DEFAULT 0;

src/export/arrow_batch.cc

@@ -149,6 +149,7 @@ struct ArrowBatchBuilder::Impl {
   arrow::StringBuilder trace_id_builder;
   arrow::StringBuilder span_id_builder;
   DictBuilder query_id_builder;
+  DictBuilder parent_query_id_builder;
   DictBuilder db_name_builder;
   DictBuilder db_user_builder;
   DictBuilder db_operation_builder;
@@ -216,6 +217,7 @@ struct ArrowBatchBuilder::Impl {
         arrow::field("trace_id", arrow::utf8()),
         arrow::field("span_id", arrow::utf8()),
         arrow::field("query_id", DictionaryUtf8Type()),
+        arrow::field("parent_query_id", DictionaryUtf8Type()),
         arrow::field("db_name", DictionaryUtf8Type()),
         arrow::field("db_user", DictionaryUtf8Type()),
         arrow::field("db_operation", DictionaryUtf8Type()),
@@ -309,10 +311,15 @@ struct ArrowBatchBuilder::Impl {
 
     char queryid_buf[24];
     snprintf(queryid_buf, sizeof(queryid_buf), "%" PRIu64, static_cast<uint64_t>(event.queryid));
+    char parent_query_id_buf[24];
+    snprintf(parent_query_id_buf, sizeof(parent_query_id_buf), "%" PRIu64,
+             static_cast<uint64_t>(event.parent_query_id));
     char pid_buf[12];
     snprintf(pid_buf, sizeof(pid_buf), "%d", event.pid);
 
     if (!AppendString(&query_id_builder, queryid_buf, "Arrow query_id append") ||
+        !AppendString(&parent_query_id_builder, parent_query_id_buf,
+                      "Arrow parent_query_id append") ||
         !AppendString(&db_name_builder, db_name, "Arrow db_name append") ||
         !AppendString(&db_user_builder, db_user, "Arrow db_user append") ||
         !AppendString(&db_operation_builder, PschCmdTypeToString(event.cmd_type),
@@ -475,6 +482,7 @@ struct ArrowBatchBuilder::Impl {
         !add_array(&trace_id_builder, "Arrow trace_id finish") ||
         !add_array(&span_id_builder, "Arrow span_id finish") ||
         !add_dict_array(&query_id_builder, "Arrow query_id finish") ||
+        !add_dict_array(&parent_query_id_builder, "Arrow parent_query_id finish") ||
         !add_dict_array(&db_name_builder, "Arrow db_name finish") ||
         !add_dict_array(&db_user_builder, "Arrow db_user finish") ||
         !add_dict_array(&db_operation_builder, "Arrow db_operation finish") ||
@@ -581,6 +589,7 @@ struct ArrowBatchBuilder::Impl {
     trace_id_builder.Reset();
     span_id_builder.Reset();
     query_id_builder.ResetFull();
+    parent_query_id_builder.ResetFull();
     db_name_builder.ResetFull();
     db_user_builder.ResetFull();
     db_operation_builder.ResetFull();

src/export/stats_exporter.cc

@@ -241,6 +241,7 @@ void ExportEventStatsInternal(const std::vector<PschEvent>& events, StatsExporte
   auto col_username = exporter->DbUserColumn();
   auto col_pid = exporter->RecordInt32("pid");
   auto col_query_id = exporter->RecordInt64("query_id");
+  auto col_parent_query_id = exporter->RecordInt64("parent_query_id");
   auto col_cmd_type = exporter->DbOperationColumn();
   auto col_rows = exporter->MetricUInt64("rows");
   auto col_query = exporter->DbQueryTextColumn();
@@ -296,6 +297,7 @@ void ExportEventStatsInternal(const std::vector<PschEvent>& events, StatsExporte
     col_username->Append(std::string(ev.username, ev.username_len));
     col_pid->Append(ev.pid);
     col_query_id->Append(static_cast<int64_t>(ev.queryid));
+    col_parent_query_id->Append(static_cast<int64_t>(ev.parent_query_id));
     col_cmd_type->Append(PschCmdTypeToString(ev.cmd_type));
     col_rows->Append(ev.rows);