From 0847889db6e56f50d70e167915a69d4118b778b8 Mon Sep 17 00:00:00 2001
From: Yakov Olkhovskiy <yakov@clickhouse.com>
Date: Mon, 11 Sep 2023 19:06:00 +0000
Subject: [PATCH] libFuzzer infrastructure

---
 CMakeLists.txt                   |  1 +
 docker/packager/binary/build.sh  |  1 +
 tests/ci/libfuzzer_test_check.py | 18 ++++--------------
 tests/fuzz/build.sh              | 28 ++++++++++++++++++++++++++++
 4 files changed, 34 insertions(+), 14 deletions(-)
 create mode 100755 tests/fuzz/build.sh

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 781a9efe64a..b4e13e8ab5b 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -588,6 +588,7 @@ if (FUZZER)
             endif()
         endif()
     endforeach()
+    add_custom_command(TARGET fuzzers POST_BUILD COMMAND SRC=${CMAKE_SOURCE_DIR} BIN=${CMAKE_BINARY_DIR} OUT=${CMAKE_BINARY_DIR}/programs ${CMAKE_SOURCE_DIR}/tests/fuzz/build.sh VERBATIM)
 endif()
 
 include (cmake/sanitize_targets.cmake)
diff --git a/docker/packager/binary/build.sh b/docker/packager/binary/build.sh
index 75a18528e65..11efffd592c 100755
--- a/docker/packager/binary/build.sh
+++ b/docker/packager/binary/build.sh
@@ -100,6 +100,7 @@ fi
 mv ./programs/clickhouse* /output || mv ./programs/*_fuzzer /output
 [ -x ./programs/self-extracting/clickhouse ] && mv ./programs/self-extracting/clickhouse /output
 mv ./src/unit_tests_dbms /output ||: # may not exist for some binary builds
+mv ./programs/*.dict ./programs/*.options ./programs/*_seed_corpus.zip /output ||: # libFuzzer oss-fuzz compatible infrastructure
 
 prepare_combined_output () {
     local OUTPUT
diff --git a/tests/ci/libfuzzer_test_check.py b/tests/ci/libfuzzer_test_check.py
index 148b6e6d1e4..75af6ddf5d9 100644
--- a/tests/ci/libfuzzer_test_check.py
+++ b/tests/ci/libfuzzer_test_check.py
@@ -330,25 +330,15 @@ def main():
         image_name, Path("../../docker/test/libfuzzer/")
     )  # get_image_with_version(reports_path, image_name)
 
-    fuzzers_tmp_path = os.path.join(temp_path, "fuzzers_tmp")
-    if not os.path.exists(fuzzers_tmp_path):
-        os.makedirs(fuzzers_tmp_path)
+    fuzzers_path = os.path.join(temp_path, "fuzzers")
+    if not os.path.exists(fuzzers_path):
+        os.makedirs(fuzzers_path)
 
     # if validate_bugfix_check:
     #     download_last_release(packages_path)
     # else:
     #     download_all_deb_packages(check_name, reports_path, packages_path)
-    download_fuzzers(check_name, reports_path, fuzzers_tmp_path)
-
-    fuzzers_path = os.path.join(temp_path, "fuzzers")
-    for fuzzer in os.listdir(fuzzers_tmp_path):
-        fuzzer_path = os.path.join(fuzzers_path, fuzzer)
-        os.makedirs(fuzzer_path)
-        os.rename(
-            os.path.join(fuzzers_tmp_path, fuzzer), os.path.join(fuzzer_path, fuzzer)
-        )
-
-    os.rmdir(fuzzers_tmp_path)
+    download_fuzzers(check_name, reports_path, fuzzers_path)
 
     # server_log_path = os.path.join(temp_path, "server_log")
     # if not os.path.exists(server_log_path):
diff --git a/tests/fuzz/build.sh b/tests/fuzz/build.sh
new file mode 100755
index 00000000000..12f41f6e079
--- /dev/null
+++ b/tests/fuzz/build.sh
@@ -0,0 +1,28 @@
+#!/bin/bash -eu
+
+# copy fuzzer options and dictionaries
+cp $SRC/tests/fuzz/*.dict $OUT/
+cp $SRC/tests/fuzz/*.options $OUT/
+
+# prepare corpus dirs
+mkdir -p $BIN/tests/fuzz/lexer_fuzzer.in/
+mkdir -p $BIN/tests/fuzz/select_parser_fuzzer.in/
+mkdir -p $BIN/tests/fuzz/create_parser_fuzzer.in/
+mkdir -p $BIN/tests/fuzz/execute_query_fuzzer.in/
+
+# prepare corpus
+cp $SRC/tests/queries/0_stateless/*.sql $BIN/tests/fuzz/lexer_fuzzer.in/
+cp $SRC/tests/queries/0_stateless/*.sql $BIN/tests/fuzz/select_parser_fuzzer.in/
+cp $SRC/tests/queries/0_stateless/*.sql $BIN/tests/fuzz/create_parser_fuzzer.in/
+cp $SRC/tests/queries/0_stateless/*.sql $BIN/tests/fuzz/execute_query_fuzzer.in/
+cp $SRC/tests/queries/1_stateful/*.sql $BIN/tests/fuzz/lexer_fuzzer.in/
+cp $SRC/tests/queries/1_stateful/*.sql $BIN/tests/fuzz/select_parser_fuzzer.in/
+cp $SRC/tests/queries/1_stateful/*.sql $BIN/tests/fuzz/create_parser_fuzzer.in/
+cp $SRC/tests/queries/1_stateful/*.sql $BIN/tests/fuzz/execute_query_fuzzer.in/
+
+# build corpus archives
+cd $BIN/tests/fuzz
+for dir in *_fuzzer.in; do
+    fuzzer=$(basename $dir .in)
+    zip -rj "$OUT/${fuzzer}_seed_corpus.zip" "${dir}/"
+done