From 6366819f120f5fd108acd60079921ff6aa595116 Mon Sep 17 00:00:00 2001
From: avogar <avogar@clickhouse.com>
Date: Fri, 17 Nov 2023 16:52:20 +0000
Subject: [PATCH] Fix generating deep nested columns in CapnProto/Protobuf
 schemas

---
 src/Formats/StructureToFormatSchemaUtils.cpp  |  4 ++
 ...apnp_protobuf_auto_schema_nested.reference | 52 +++++++++++++++++++
 ...02920_capnp_protobuf_auto_schema_nested.sh | 21 ++++++++
 3 files changed, 77 insertions(+)
 create mode 100644 tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.reference
 create mode 100755 tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.sh

diff --git a/src/Formats/StructureToFormatSchemaUtils.cpp b/src/Formats/StructureToFormatSchemaUtils.cpp
index 47701fa4f81..c56ff821a4a 100644
--- a/src/Formats/StructureToFormatSchemaUtils.cpp
+++ b/src/Formats/StructureToFormatSchemaUtils.cpp
@@ -96,6 +96,10 @@ NamesAndTypesList collectNested(const NamesAndTypesList & names_and_types, bool
             nested[field_name].emplace_back(nested_name, type);
     }
 
+    /// Collect nested recursively.
+    for (auto & [field_name, elements] : nested)
+        elements = collectNested(elements, allow_split_by_underscore, format_name);
+
     for (const auto & [field_name, elements]: nested)
         result.emplace_back(field_name, std::make_shared<DataTypeTuple>(elements.getTypes(), elements.getNames()));
 
diff --git a/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.reference b/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.reference
new file mode 100644
index 00000000000..9874bc57142
--- /dev/null
+++ b/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.reference
@@ -0,0 +1,52 @@
+
+message Message
+{
+    message H
+    {
+        uint32 k = 1;
+    }
+    H h = 1;
+    message A
+    {
+        uint32 g = 1;
+        message B
+        {
+            uint32 c = 1;
+            uint32 f = 2;
+            message D
+            {
+                uint32 e = 1;
+            }
+            D d = 3;
+        }
+        B b = 2;
+    }
+    A a = 2;
+}
+46	(45,(42,44,43))
+
+struct Message
+{
+    struct H
+    {
+        k @0 : UInt8;
+    }
+    h @0 : H;
+    struct A
+    {
+        g @0 : UInt8;
+        struct B
+        {
+            c @0 : UInt8;
+            f @1 : UInt8;
+            struct D
+            {
+                e @0 : UInt8;
+            }
+            d @2 : D;
+        }
+        b @1 : B;
+    }
+    a @1 : A;
+}
+(46)	(45,(42,44,(43)))
diff --git a/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.sh b/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.sh
new file mode 100755
index 00000000000..aee6b866719
--- /dev/null
+++ b/tests/queries/0_stateless/02920_capnp_protobuf_auto_schema_nested.sh
@@ -0,0 +1,21 @@
+#!/usr/bin/env bash
+# Tags: no-fasttest
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CURDIR"/../shell_config.sh
+
+SCHEMA_FILE=$CLICKHOUSE_TEST_UNIQUE_NAME-schema
+FILE=$CLICKHOUSE_TEST_UNIQUE_NAME
+
+$CLICKHOUSE_LOCAL -q "select 42 as \`a.b.c\`, 43 as \`a.b.d.e\`, 44 as \`a.b.f\`, 45 as \`a.g\`, 46 as \`h.k\` format Protobuf settings output_format_schema='$SCHEMA_FILE.proto'" > $FILE.pb
+tail -n +2 $SCHEMA_FILE.proto
+$CLICKHOUSE_LOCAL -q "select * from file('$FILE.pb') settings format_schema='$SCHEMA_FILE:Message'"
+
+$CLICKHOUSE_LOCAL -q "select 42 as a_b_c, 43 as a_b_d_e, 44 as a_b_f, 45 as a_g, 46 as h_k format CapnProto settings output_format_schema='$SCHEMA_FILE.capnp'" > $FILE.capnp
+tail -n +2 $SCHEMA_FILE.capnp
+$CLICKHOUSE_LOCAL -q "select * from file('$FILE.capnp') settings format_schema='$SCHEMA_FILE:Message'"
+
+rm $SCHEMA_FILE*
+rm $FILE.*
+