tp: _interval_create! propagates ids

src/trace_processor/perfetto_sql/generator/structured_query_generator.cc

@@ -1202,11 +1202,13 @@ base::StatusOr<std::string> GeneratorImpl::CreateSlices(
 
   // Use _interval_create! macro which delegates to
   // __intrinsic_interval_create, an O(n+m) two-pointer C++ implementation.
-  // The macro expects inputs with a `ts` column, so we rename if needed.
+  // The macro expects inputs with `id` and `ts` columns. We pass a dummy id
+  // since structured queries don't need start_id/end_id, and exclude those
+  // columns from the output.
   return base::StackString<1024>(
-             "(SELECT * FROM _interval_create!("
-             "(SELECT %s AS ts FROM %s), "
-             "(SELECT %s AS ts FROM %s)))",
+             "(SELECT ts, dur FROM _interval_create!("
+             "(SELECT 1 AS id, %s AS ts FROM %s), "
+             "(SELECT 1 AS id, %s AS ts FROM %s)))",
              starts_ts_col.c_str(), starts_table.c_str(), ends_ts_col.c_str(),
              ends_table.c_str())
       .ToStdString();

src/trace_processor/perfetto_sql/generator/structured_query_generator_unittest.cc

@@ -4150,7 +4150,7 @@ TEST(StructuredQueryGeneratorTest, ExperimentalCreateSlicesBasic) {
     sq_2 AS (SELECT * FROM end_events),
     sq_1 AS (SELECT * FROM start_events),
     sq_0 AS (
-      SELECT * FROM (SELECT * FROM _interval_create!((SELECT ts AS ts FROM sq_1), (SELECT ts AS ts FROM sq_2)))
+      SELECT * FROM (SELECT ts, dur FROM _interval_create!((SELECT 1 AS id, ts AS ts FROM sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2)))
     )
     SELECT * FROM sq_0
   )"));
@@ -4200,7 +4200,7 @@ TEST(StructuredQueryGeneratorTest,
     sq_2 AS (SELECT * FROM slice WHERE name GLOB '*_end'),
     sq_1 AS (SELECT * FROM slice WHERE name GLOB '*_begin'),
     sq_0 AS (
-      SELECT * FROM (SELECT * FROM _interval_create!((SELECT ts AS ts FROM sq_1), (SELECT ts AS ts FROM sq_2)))
+      SELECT * FROM (SELECT ts, dur FROM _interval_create!((SELECT 1 AS id, ts AS ts FROM sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2)))
     )
     SELECT * FROM sq_0
   )"));
@@ -4229,9 +4229,10 @@ TEST(StructuredQueryGeneratorTest,
   )");
   auto ret = gen.Generate(proto.data(), proto.size());
   ASSERT_OK_AND_ASSIGN(std::string res, ret);
-  EXPECT_THAT(
-      res, testing::HasSubstr("_interval_create!((SELECT acquire_ts AS ts FROM "
-                              "sq_1), (SELECT release_ts AS ts FROM sq_2))"));
+  EXPECT_THAT(res,
+              testing::HasSubstr(
+                  "_interval_create!((SELECT 1 AS id, acquire_ts AS ts FROM "
+                  "sq_1), (SELECT 1 AS id, release_ts AS ts FROM sq_2))"));
 }
 
 TEST(StructuredQueryGeneratorTest, ExperimentalCreateSlicesWithFilters) {
@@ -4496,8 +4497,9 @@ TEST(StructuredQueryGeneratorTest,
   auto ret = gen.Generate(proto.data(), proto.size());
   ASSERT_OK_AND_ASSIGN(std::string res, ret);
   // Should default starts_ts_column to "ts"
-  EXPECT_THAT(res, testing::HasSubstr("_interval_create!((SELECT ts AS ts FROM "
-                                      "sq_1), (SELECT ts AS ts FROM sq_2))"));
+  EXPECT_THAT(res, testing::HasSubstr(
+                       "_interval_create!((SELECT 1 AS id, ts AS ts FROM "
+                       "sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2))"));
 }
 
 TEST(StructuredQueryGeneratorTest,
@@ -4523,8 +4525,9 @@ TEST(StructuredQueryGeneratorTest,
   auto ret = gen.Generate(proto.data(), proto.size());
   ASSERT_OK_AND_ASSIGN(std::string res, ret);
   // Should default ends_ts_column to "ts"
-  EXPECT_THAT(res, testing::HasSubstr("_interval_create!((SELECT ts AS ts FROM "
-                                      "sq_1), (SELECT ts AS ts FROM sq_2))"));
+  EXPECT_THAT(res, testing::HasSubstr(
+                       "_interval_create!((SELECT 1 AS id, ts AS ts FROM "
+                       "sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2))"));
 }
 
 TEST(StructuredQueryGeneratorTest,
@@ -4554,7 +4557,7 @@ TEST(StructuredQueryGeneratorTest,
     sq_2 AS (SELECT * FROM end_events),
     sq_1 AS (SELECT * FROM start_events),
     sq_0 AS (
-      SELECT * FROM (SELECT * FROM _interval_create!((SELECT ts AS ts FROM sq_1), (SELECT ts AS ts FROM sq_2)))
+      SELECT * FROM (SELECT ts, dur FROM _interval_create!((SELECT 1 AS id, ts AS ts FROM sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2)))
     )
     SELECT * FROM sq_0
   )"));
@@ -4642,8 +4645,9 @@ TEST(StructuredQueryGeneratorTest,
   auto ret = gen.Generate(proto.data(), proto.size());
   ASSERT_OK_AND_ASSIGN(std::string res, ret);
   // Empty string should default to "ts"
-  EXPECT_THAT(res, testing::HasSubstr("_interval_create!((SELECT ts AS ts FROM "
-                                      "sq_1), (SELECT ts AS ts FROM sq_2))"));
+  EXPECT_THAT(res, testing::HasSubstr(
+                       "_interval_create!((SELECT 1 AS id, ts AS ts FROM "
+                       "sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2))"));
 }
 
 TEST(StructuredQueryGeneratorTest, ExperimentalCreateSlicesEmptyEndsTsColumn) {
@@ -4669,8 +4673,9 @@ TEST(StructuredQueryGeneratorTest, ExperimentalCreateSlicesEmptyEndsTsColumn) {
   auto ret = gen.Generate(proto.data(), proto.size());
   ASSERT_OK_AND_ASSIGN(std::string res, ret);
   // Empty string should default to "ts"
-  EXPECT_THAT(res, testing::HasSubstr("_interval_create!((SELECT ts AS ts FROM "
-                                      "sq_1), (SELECT ts AS ts FROM sq_2))"));
+  EXPECT_THAT(res, testing::HasSubstr(
+                       "_interval_create!((SELECT 1 AS id, ts AS ts FROM "
+                       "sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2))"));
 }
 
 TEST(StructuredQueryGeneratorTest,
@@ -4714,7 +4719,7 @@ TEST(StructuredQueryGeneratorTest,
     sq_2 AS (SELECT * FROM end_events WHERE ts < 0),
     sq_1 AS (SELECT * FROM start_events WHERE ts < 0),
     sq_0 AS (
-      SELECT * FROM (SELECT * FROM _interval_create!((SELECT ts AS ts FROM sq_1), (SELECT ts AS ts FROM sq_2)))
+      SELECT * FROM (SELECT ts, dur FROM _interval_create!((SELECT 1 AS id, ts AS ts FROM sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2)))
     )
     SELECT * FROM sq_0
   )"));
@@ -4759,7 +4764,7 @@ TEST(StructuredQueryGeneratorTest, ExperimentalCreateSlicesNoMatchingEnds) {
     sq_2 AS (SELECT * FROM events WHERE ts > 10000),
     sq_1 AS (SELECT * FROM events WHERE ts < 1000),
     sq_0 AS (
-      SELECT * FROM (SELECT * FROM _interval_create!((SELECT ts AS ts FROM sq_1), (SELECT ts AS ts FROM sq_2)))
+      SELECT * FROM (SELECT ts, dur FROM _interval_create!((SELECT 1 AS id, ts AS ts FROM sq_1), (SELECT 1 AS id, ts AS ts FROM sq_2)))
     )
     SELECT * FROM sq_0
   )"));

src/trace_processor/perfetto_sql/intrinsics/functions/create_intervals.cc

@@ -63,8 +63,9 @@ struct IntervalCreate : public sqlite::Function<IntervalCreate> {
     auto* ends = sqlite::value::Pointer<SortedTimestamps>(
         argv[1], SortedTimestamps::kName);
 
-    std::vector<std::string> col_names{"ts", "dur"};
-    std::vector<ColType> col_types{ColType::kInt64, ColType::kInt64};
+    std::vector<std::string> col_names{"ts", "dur", "start_id", "end_id"};
+    std::vector<ColType> col_types{ColType::kInt64, ColType::kInt64,
+                                   ColType::kInt64, ColType::kInt64};
 
     dataframe::AdhocDataframeBuilder builder(
         col_names, GetUserData(ctx)->pool,
@@ -80,7 +81,9 @@ struct IntervalCreate : public sqlite::Function<IntervalCreate> {
     }
 
     const auto& start_ts = starts->timestamps;
+    const auto& start_ids = starts->ids;
     const auto& end_ts = ends->timestamps;
+    const auto& end_ids = ends->ids;
 
     // Two-pointer matching: O(n + m).
     // Both arrays are already sorted (guaranteed by ORDER BY in the SQL macro).
@@ -96,6 +99,8 @@ struct IntervalCreate : public sqlite::Function<IntervalCreate> {
       }
       builder.PushNonNullUnchecked(0, start_ts[i]);
       builder.PushNonNullUnchecked(1, end_ts[end_idx] - start_ts[i]);
+      builder.PushNonNullUnchecked(2, start_ids[i]);
+      builder.PushNonNullUnchecked(3, end_ids[end_idx]);
     }
 
     SQLITE_ASSIGN_OR_RETURN(ctx, auto ret_table, std::move(builder).Build());

src/trace_processor/perfetto_sql/intrinsics/functions/type_builders.cc

@@ -436,21 +436,22 @@ struct IntervalTreeIntervalsAgg
   }
 };
 
-// An SQL aggregate function which collects timestamps into a vector.
+// An SQL aggregate function which collects (id, ts) pairs into vectors.
 // Used as input to __intrinsic_interval_create. The caller is responsible
-// for ensuring timestamps are passed in sorted order (e.g. via ORDER BY).
+// for ensuring rows are passed in sorted order by ts (e.g. via ORDER BY).
 struct TimestampSetAgg
     : public sqlite::AggregateFunction<perfetto_sql::SortedTimestamps> {
   static constexpr char kName[] = "__intrinsic_timestamp_set_agg";
-  static constexpr int kArgCount = 1;
+  static constexpr int kArgCount = 2;
   struct AggCtx : sqlite::AggregateContext<AggCtx> {
     perfetto_sql::SortedTimestamps data;
   };
 
   static void Step(sqlite3_context* ctx, int argc, sqlite3_value** argv) {
     PERFETTO_DCHECK(argc == kArgCount);
     auto& data = AggCtx::GetOrCreateContextForStep(ctx).data;
-    data.timestamps.push_back(sqlite::value::Int64(argv[0]));
+    data.ids.push_back(sqlite::value::Int64(argv[0]));
+    data.timestamps.push_back(sqlite::value::Int64(argv[1]));
   }
 
   static void Final(sqlite3_context* ctx) {

src/trace_processor/perfetto_sql/intrinsics/types/sorted_timestamps.h

@@ -22,13 +22,16 @@
 
 namespace perfetto::trace_processor::perfetto_sql {
 
-// A sorted collection of timestamps used as an intermediate type for the
-// interval_create intrinsic function. Timestamps are collected via an
-// aggregate function and must be passed in ascending order (via ORDER BY)
-// to the scalar function.
+// A sorted collection of timestamps (with associated IDs) used as an
+// intermediate type for the interval_create intrinsic function. Timestamps
+// are collected via an aggregate function and must be passed in ascending
+// order (via ORDER BY) to the scalar function.
 struct SortedTimestamps {
   static constexpr char kName[] = "SORTED_TIMESTAMPS";
   std::vector<int64_t> timestamps;
+  // Row IDs corresponding 1:1 with timestamps. These are the original row
+  // IDs from the input table, used to produce start_id/end_id in output.
+  std::vector<int64_t> ids;
 };
 
 }  // namespace perfetto::trace_processor::perfetto_sql

src/trace_processor/perfetto_sql/stdlib/intervals/create_intervals.sql

@@ -17,39 +17,48 @@
 -- intervals by matching each start with the next end timestamp strictly greater
 -- than it.
 --
--- Both input tables must have a column named `ts`. Uses an efficient O(n+m)
--- two-pointer algorithm implemented in C++.
+-- Both input tables must have columns named `id` and `ts`. Uses an efficient
+-- O(n+m) two-pointer algorithm implemented in C++.
 --
 -- Example:
 -- ```
 -- SELECT * FROM _interval_create!(
---   (SELECT ts FROM starts_table),
---   (SELECT ts FROM ends_table)
+--   (SELECT id, ts FROM starts_table),
+--   (SELECT id, ts FROM ends_table)
 -- )
 -- ```
 CREATE PERFETTO MACRO _interval_create(
-    -- Table or subquery containing start timestamps (must have a `ts` column).
+    -- Table or subquery containing start timestamps (must have `id` and `ts`
+    -- columns).
     starts_table TableOrSubquery,
-    -- Table or subquery containing end timestamps (must have a `ts` column).
+    -- Table or subquery containing end timestamps (must have `id` and `ts`
+    -- columns).
     ends_table TableOrSubquery
 )
 -- Table with the schema:
 -- ts TIMESTAMP,
 --     The start timestamp.
 -- dur DURATION,
 --     The duration from start to the matched end.
+-- start_id LONG,
+--     The id of the matched start row.
+-- end_id LONG,
+--     The id of the matched end row.
 RETURNS TableOrSubquery AS
 (
   SELECT
     c0 AS ts,
-    c1 AS dur
+    c1 AS dur,
+    c2 AS start_id,
+    c3 AS end_id
   FROM __intrinsic_table_ptr(
     __intrinsic_interval_create(
       (
         SELECT
-          __intrinsic_timestamp_set_agg(ordered_s.ts)
+          __intrinsic_timestamp_set_agg(ordered_s.id, ordered_s.ts)
         FROM (
           SELECT
+            id,
             ts
           FROM $starts_table
           ORDER BY
@@ -58,9 +67,10 @@ RETURNS TableOrSubquery AS
       ),
       (
         SELECT
-          __intrinsic_timestamp_set_agg(ordered_e.ts)
+          __intrinsic_timestamp_set_agg(ordered_e.id, ordered_e.ts)
         FROM (
           SELECT
+            id,
             ts
           FROM $ends_table
           ORDER BY
@@ -70,5 +80,8 @@ RETURNS TableOrSubquery AS
     )
   )
   WHERE
-    __intrinsic_table_ptr_bind(c0, 'ts') AND __intrinsic_table_ptr_bind(c1, 'dur')
+    __intrinsic_table_ptr_bind(c0, 'ts')
+    AND __intrinsic_table_ptr_bind(c1, 'dur')
+    AND __intrinsic_table_ptr_bind(c2, 'start_id')
+    AND __intrinsic_table_ptr_bind(c3, 'end_id')
 );