From 7e03621a409c5bce283b71269e50c41ce193a347 Mon Sep 17 00:00:00 2001
From: Julia Kartseva <yulia.kartseva@gmail.com>
Date: Wed, 4 Sep 2024 05:43:56 +0000
Subject: [PATCH] Fix INFILE file format detection for async inserts

---
 src/Interpreters/AsynchronousInsertQueue.cpp  | 13 +++++++++-
 ...03233_async_insert_infile_format.reference |  2 ++
 .../03233_async_insert_infile_format.sh       | 25 +++++++++++++++++++
 3 files changed, 39 insertions(+), 1 deletion(-)
 create mode 100644 tests/queries/0_stateless/03233_async_insert_infile_format.reference
 create mode 100755 tests/queries/0_stateless/03233_async_insert_infile_format.sh

diff --git a/src/Interpreters/AsynchronousInsertQueue.cpp b/src/Interpreters/AsynchronousInsertQueue.cpp
index 461700dcb75..25cd1d0bfa2 100644
--- a/src/Interpreters/AsynchronousInsertQueue.cpp
+++ b/src/Interpreters/AsynchronousInsertQueue.cpp
@@ -315,7 +315,18 @@ void AsynchronousInsertQueue::preprocessInsertQuery(const ASTPtr & query, const
     auto sample_block = InterpreterInsertQuery::getSampleBlock(insert_query, table, table->getInMemoryMetadataPtr(), query_context);
 
     if (!FormatFactory::instance().isInputFormat(insert_query.format))
-        throw Exception(ErrorCodes::UNKNOWN_FORMAT, "Unknown input format {}", insert_query.format);
+    {
+        if (insert_query.format.empty() && insert_query.infile)
+        {
+            const auto & in_file_node = insert_query.infile->as<ASTLiteral &>();
+            const auto in_file = in_file_node.value.safeGet<std::string>();
+            const auto in_file_format = FormatFactory::instance().getFormatFromFileName(in_file);
+            if (!FormatFactory::instance().isInputFormat(in_file_format))
+                throw Exception(ErrorCodes::UNKNOWN_FORMAT, "Unknown input INFILE format {}", in_file_format);
+        }
+        else
+            throw Exception(ErrorCodes::UNKNOWN_FORMAT, "Unknown input format {}", insert_query.format);
+    }
 
     /// For table functions we check access while executing
     /// InterpreterInsertQuery::getTable() -> ITableFunction::execute().
diff --git a/tests/queries/0_stateless/03233_async_insert_infile_format.reference b/tests/queries/0_stateless/03233_async_insert_infile_format.reference
new file mode 100644
index 00000000000..b236c1b8330
--- /dev/null
+++ b/tests/queries/0_stateless/03233_async_insert_infile_format.reference
@@ -0,0 +1,2 @@
+1	ClickHouse
+2	HelloWorld
diff --git a/tests/queries/0_stateless/03233_async_insert_infile_format.sh b/tests/queries/0_stateless/03233_async_insert_infile_format.sh
new file mode 100755
index 00000000000..29ec87799bb
--- /dev/null
+++ b/tests/queries/0_stateless/03233_async_insert_infile_format.sh
@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CURDIR"/../shell_config.sh
+
+function cleanup()
+{
+    [ -e "${CLICKHOUSE_TMP}"/test_infile.csv ] && rm "${CLICKHOUSE_TMP}"/test_infile.csv
+}
+
+trap cleanup EXIT
+
+cleanup
+
+echo -e "id,\"word\"\n1,\"ClickHouse\"\n2,\"HelloWorld\"" > "${CLICKHOUSE_TMP}"/test_infile.csv
+
+${CLICKHOUSE_CLIENT} --query "DROP TABLE IF EXISTS async_insert_infile_data;"
+${CLICKHOUSE_CLIENT} --query "CREATE TABLE async_insert_infile_data (id UInt32, word String) ENGINE=Memory();"
+${CLICKHOUSE_CLIENT} --query "INSERT INTO async_insert_infile_data FROM INFILE '${CLICKHOUSE_TMP}/test_infile.csv' SETTINGS async_insert=1;"
+${CLICKHOUSE_CLIENT} --query "SELECT * FROM async_insert_infile_data ORDER BY id;"
+
+${CLICKHOUSE_CLIENT} --query "INSERT INTO async_insert_infile_data FROM INFILE '${CLICKHOUSE_TMP}/test_infile.csv' FORMAT NotExists SETTINGS async_insert=1;" 2>&1 | grep -q "UNKOWN_FORMAT"
+
+${CLICKHOUSE_CLIENT} --query "DROP TABLE async_insert_infile_data SYNC;"