Replace hardcoded percentages with values from a file

CMakeLists.txt

@@ -21,6 +21,7 @@
 set(EXTENSION_SOURCES src/sqlsmith_extension.cpp
         src/statement_generator.cpp
         src/statement_simplifier.cpp
+        src/random_nums_config.cpp
         src/fuzzyduck.cpp ${EXTENSION_OBJECT_FILES})
 
 build_static_extension(${TARGET_NAME} ${EXTENSION_SOURCES})

config.json

@@ -0,0 +1,8 @@
+{
+    "select_percentage": "70",
+    "attach_percentage": "20",
+    "attach_use_percentage": "80",
+    "detach_percentage": "15",
+    "set_percentage": "5",
+    "delete_percentage": "25"
+}

config_nested.json

@@ -0,0 +1,100 @@
+{
+    "attach": {
+        "attach_percentage": "20",
+        "attach_use_percentage": "80",
+        "attach_read_only": "30"
+    },
+    "detach": {
+        "detach_percentage": "15",
+        "detach_random_name": "20"
+    },
+    "set": {
+        "set_percentage": "5",
+        "set_attached_db": "90"
+    },
+    "delete_percentage": "25",
+    "select": {
+        "select_percentage": "70",
+        "select_node": {
+            "select_node_perc": "5",
+            "select_node_is_distinct_perc": "30",
+            "select_node_from_table_perc": "95",
+            "select_node_where_perc": "60",
+            "select_node_having_perc": "25",
+            "select_node_groups": {
+                "select_node_groups_perc": "30",
+                "select_node_group_by_perc": "70"
+            },
+            "select_node_qualify_perc": "10",
+            "select_node_aggregate_perc": "10",
+            "select_node_sample": {
+                "select_node_sample_perc": "10",
+                "select_node_sample_is_perc": "50",
+                "select_node_sample_size": "100"
+            }
+        },
+        "setop": "40",
+        "result_modifiers": "5",
+        "limit_percent_modifier": "50",
+        "limit_percent_modifier_limit": "30",
+        "limit_percent_modifier_offset": "30",
+        "limit_modifier_limit": "30", 
+        "limit_modifier_offset": "30"
+    },
+    "create": {
+        "create_percentage": "0",
+        "create_generate_select": "50",
+        "create_num_cols": "1000"
+    },
+    "table_ref": {
+        "table_ref_base_table_ref_perc": "60",
+        "table_ref_expression_list_ref": "20",
+        "table_ref_join_ref": "40" 
+    },
+    "join_ref": {
+        "join_ref_cross": "10",
+        "join_ref_asof": "10",
+        "join_ref_natural": "10",
+        "join_ref_positional": "10",
+        "join_ref_general_expression": "70"
+    },
+    "expression": {
+        "expression_column_ref": "50",
+        "expression_constant": "30",
+        "expression_subquery": "3"
+    },
+    "constant_value": {
+        "constant_value_bigint": "50",
+        "constant_value_to_string": "30"
+    },
+    "function": {
+        "function_aggregate": {
+            "function_aggregate_window_function": "10",
+            "function_aggregate_order_by": "10",
+            "function_aggregate_random_expression": "10",
+            "function_aggregate_distinct": "10"
+        }
+    },
+    "window_function": {
+        "window_function_partitions": "50",
+        "window_function_orders": "30",
+        "window_function_ignore_nulls": "30",
+        "window_function_result_offset": "30",
+        "window_function_result_default": "30"
+    },
+    "star": {
+        "star_relation_name": "10",
+        "star_column_name_exclude_list": "20",
+        "star_column_name": "20",
+        "star_columns": {
+            "star_columns_true": "50",
+            "star_columns_true_lambda": "50"
+        }
+    },
+    "relational_name": {
+        "relational_name_choose_current": "80"
+    },
+    "column_names": {
+        "column_names_choose_current": "80"
+    }
+}

scripts/run_fuzzer.py

@@ -34,6 +34,8 @@
         perform_checks = False
     elif param.startswith('--enable_verification'):
         verification = param.replace('--enable_verification=', '').lower() == 'true'
+    elif param.startswith('--randoms_config_filepath'):
+        randoms_config_filepath = param.replace('--randoms_config_filepath=', '')
     elif param.startswith('--shell='):
         shell = param.replace('--shell=', '')
     elif param.startswith('--seed='):
@@ -76,7 +78,8 @@ def run_fuzzer_script(fuzzer):
     if fuzzer == 'sqlsmith':
         return "call sqlsmith(max_queries=${MAX_QUERIES}, seed=${SEED}, verbose_output=1, log='${LAST_LOG_FILE}', complete_log='${COMPLETE_LOG_FILE}');"
     elif fuzzer == 'duckfuzz':
-        return "call fuzzyduck(max_queries=${MAX_QUERIES}, seed=${SEED}, verbose_output=1, log='${LAST_LOG_FILE}', complete_log='${COMPLETE_LOG_FILE}', enable_verification='${ENABLE_VERIFICATION}');"
+        return "call fuzzyduck(max_queries=${MAX_QUERIES}, seed=${SEED}, verbose_output=1, log='${LAST_LOG_FILE}', complete_log='${COMPLETE_LOG_FILE}', \
+        enable_verification='${ENABLE_VERIFICATION}', randoms_config_filepath='${RANDOMS_CONFIG_FILEPATH}');"
     elif fuzzer == 'duckfuzz_functions':
         return "call fuzz_all_functions(seed=${SEED}, verbose_output=1, log='${LAST_LOG_FILE}', complete_log='${COMPLETE_LOG_FILE}');"
     else:
@@ -132,6 +135,7 @@ def run_shell_command(cmd):
     .replace('${COMPLETE_LOG_FILE}', complete_log_file)
     .replace('${SEED}', str(seed))
     .replace('${ENABLE_VERIFICATION}', str(verification))
+    .replace('${RANDOMS_CONFIG_FILEPATH}', randoms_config_filepath)
 )
 
 print(load_script)

src/CMakeLists.txt

@@ -7,7 +7,7 @@ include_directories(third_party/sqlsmith/include)
 add_subdirectory(third_party)
 
 set(SQLSMITH_SOURCES
-    sqlsmith_extension.cpp statement_generator.cpp statement_simplifier.cpp
+    sqlsmith_extension.cpp statement_generator.cpp statement_simplifier.cpp random_nums_config.cpp
     fuzzyduck.cpp ${SQLSMITH_OBJECT_FILES})
 
 build_static_extension(sqlsmith ${SQLSMITH_SOURCES})

src/fuzzyduck.cpp

@@ -22,15 +22,20 @@ void FuzzyDuck::BeginFuzzing() {
 	if (max_queries == 0) {
 		throw BinderException("Provide a max_queries argument greater than 0");
 	}
+	auto &fs = FileSystem::GetFileSystem(context);
 	if (!complete_log.empty()) {
-		auto &fs = FileSystem::GetFileSystem(context);
 		TryRemoveFile(complete_log);
 		complete_log_handle =
 			fs.OpenFile(complete_log, FileFlags::FILE_FLAGS_WRITE | FileFlags::FILE_FLAGS_FILE_CREATE_NEW);
 	}
 	if (enable_verification) {
 		RunQuery("PRAGMA enable_verification");
 	}
+	if (!randoms_config_filepath.empty()) {
+		config = RandomNumsConfig().GetConfigFromFile(randoms_config_filepath.c_str());
+	} else {
+		config = RandomNumsConfig().GetDefaultConfig();
+	}
 }
 
 void FuzzyDuck::EndFuzzing() {
@@ -70,6 +75,7 @@ string FuzzyDuck::GenerateQuery() {
 	// generate statement
 	StatementGenerator generator(context);
 	generator.verification_enabled = enable_verification;
+	generator.config = config;
 	// accumulate statement(s)
 	auto statement = string("");
 	if (generator.RandomPercentage(10)) {

src/include/fuzzyduck.hpp

@@ -10,8 +10,10 @@
 
 #include "duckdb.hpp"
 #include "duckdb/parser/query_node.hpp"
+#include "random_nums_config.hpp"
 
 namespace duckdb {
+
 struct FileHandle;
 
 class FuzzyDuck {
@@ -27,6 +29,9 @@ class FuzzyDuck {
 	bool verbose_output = false;
 	bool enable_verification = false;
 	idx_t timeout = 30;
+	string randoms_config_filepath;
+	// RandomNumsConfig config;
+	unordered_map<RandomPercentagesEnum, idx_t> config;
 
 public:
 	void Fuzz();
@@ -50,6 +55,7 @@ class FuzzyDuck {
 
 private:
 	unique_ptr<FileHandle> complete_log_handle;
+	unique_ptr<FileHandle> randoms_config_handle;
 };
 
 } // namespace duckdb

src/include/random_nums_config.hpp

@@ -0,0 +1,145 @@
+// ===----------------------------------------------------------------------===//
+//                         DuckDB
+
+// random_nums_config.hpp
+
+
+// ===----------------------------------------------------------------------===//
+
+#pragma once
+
+#include "duckdb.hpp"
+#include "yyjson.hpp"
+
+namespace duckdb {
+
+enum class RandomPercentagesEnum  : idx_t {
+    // ----------------------------------
+	// Generate Attach Percentages
+	// ----------------------------------
+    ATTACH = 0,
+    ATTACH_USE = 1,
+    ATTACH_READ_ONLY = 2,
+    // ----------------------------------
+	// Generate Detach Percentages
+	// ----------------------------------
+    DETACH = 3,
+    DETACH_RANDOM_NAME = 4,
+    // ----------------------------------
+	// Generate Set Percentages
+	// ----------------------------------
+    SET = 5,
+    SET_ATTACHED_DB = 6,
+    // ----------------------------------
+	// Generate Select Percentages
+	// ----------------------------------
+    DELETE = 7,
+
+    // ----------------------------------
+	// Generate Select Percentages
+	// ----------------------------------
+	SELECT = 8,
+    SELECT_NODE = 9,
+    SELECT_NODE_IS_DISTINCT = 10,
+    SELECT_NODE_FROM_TABLE = 11,
+    SELECT_NODE_WHERE = 12,
+    SELECT_NODE_HAVING = 13,
+    SELECT_NODE_GROUPS = 14,
+    SELECT_NODE_GROUP_BY = 15,
+    SELECT_NODE_QUALIFY = 16,
+    SELECT_NODE_AGGREGATE = 17,
+    SELECT_NODE_SAMPLE = 18,
+    SELECT_NODE_SAMPLE_IS_PERC = 19,
+    SELECT_NODE_SAMPLE_SIZE = 20,
+    SETOP = 21,
+    RESULT_MODIFIERS = 22,
+    LIMIT_PERCENT_MODIFIER = 23,
+    LIMIT_PERCENT_MODIFIER_LIMIT = 24,
+    LIMIT_PERCENT_MODIFIER_OFFSET = 25,
+    LIMIT_MODIFIER_LIMIT = 26,
+    LIMIT_MODIFIER_OFFSET = 27,
+
+    // ----------------------------------
+	// Generate Create Percentages
+	// ----------------------------------
+    CREATE = 28,
+    CREATE_GENERATE_SELECT = 29,
+    CREATE_NUM_COLS = 30,
+    // -----------------------------------
+	// Generate Table Ref Percentages
+	// -----------------------------------
+    TABLE_REF_BASE_TABLE_REF_PERC = 31,
+    TABLE_REF_EXPRESSION_LIST_REF = 32,
+    TABLE_REF_JOIN_REF = 33,
+
+    // -----------------------------------
+	// Generate Join Ref Percentages
+	// -----------------------------------
+    JOIN_REF_CROSS = 34,
+    JOIN_REF_ASOF = 35,
+    JOIN_REF_NATURAL = 36,
+    JOIN_REF_POSITIONAL = 37,
+    JOIN_REF_GENERAL_EXPRESSION = 38,
+
+    // -----------------------------------
+	// Generate Expression Percentages
+	// -----------------------------------
+    EXPRESSION_COLUMN_REF = 39,
+    EXPRESSION_CONSTANT = 40,
+    EXPRESSION_SUBQUERY = 41,
+
+    // -----------------------------------
+	// Generate Constant Value Percentages
+	// -----------------------------------
+    CONSTANT_VALUE_BIGINT = 42,
+    CONSTANT_VALUE_TO_STRING = 43,
+
+    // -----------------------------------
+	// Generate Function Percentages
+	// -----------------------------------    
+    FUNCTION_AGGREGATE_WINDOW_FUNCTION = 44,
+    FUNCTION_AGGREGATE_ORDER_BY = 45,
+    FUNCTION_AGGREGATE_DISTINCT = 46,
+    FUNCTION_AGGREGATE_RANDOM_EXPRESSION = 47,
+
+    // -----------------------------------
+	// Generate Window Function Percentages
+	// -----------------------------------    
+    WINDOW_FUNCTION_PARTITIONS = 48,
+    WINDOW_FUNCTION_ORDERS = 49,
+    WINDOW_FUNCTION_IGNORE_NULLS = 50,
+    WINDOW_FUNCTION_RESULT_OFFSET = 51,
+    WINDOW_FUNCTION_RESULT_DEFAULT = 52,
+
+    // -----------------------------------
+	// Generate Star Percentages
+	// -----------------------------------
+    STAR_RELATION_NAME = 53,
+    STAR_COLUMN_NAME = 54,
+    STAR_COLUMN_NAME_EXCLUDE_LIST = 55,
+    STAR_COLUMNS_TRUE = 56,
+    STAR_COLUMNS_TRUE_LAMBDA = 57,
+
+    RELATIONAL_NAME_CHOOSE_CURRENT = 58,
+    COLUMN_NAMES_CHOOSE_CURRENT = 59,
+
+    COUNT
+
+};
+
+class RandomNumsConfig {
+public: 
+    RandomNumsConfig();
+    RandomNumsConfig(const char *config_file_path);
+    ~RandomNumsConfig();
+
+    RandomPercentagesEnum percentages_selector;
+    // unordered_map<RandomPercentagesEnum, idx_t> config;
+
+    unordered_map<RandomPercentagesEnum, idx_t> GetDefaultConfig();
+    unordered_map<RandomPercentagesEnum, idx_t> GetConfigFromFile(const char *json_string);
+    string RandomPercentagesEnumToString(RandomPercentagesEnum type);
+
+};
+
+}// namespace duckdb

src/include/statement_generator.hpp

@@ -11,6 +11,7 @@
 #include "duckdb.hpp"
 #include "duckdb/parser/parsed_data/detach_info.hpp"
 #include "duckdb/parser/query_node.hpp"
+#include "random_nums_config.hpp"
 
 #define TESTING_DIRECTORY_NAME "duckdb_unittest_tempdir"
 
@@ -40,6 +41,7 @@ class StatementGenerator {
 	friend class ExpressionDepthChecker;
 	friend class AggregateChecker;
 	friend class WindowChecker;
+	unordered_map<RandomPercentagesEnum, idx_t> config;
 
 public:
 	StatementGenerator(ClientContext &context);
@@ -55,6 +57,8 @@ class StatementGenerator {
 	bool RandomPercentage(idx_t percentage);
 	bool verification_enabled = false;
 	idx_t RandomValue(idx_t max);
+
+
 	string GetRandomAttachedDataBase();
 	unique_ptr<SQLStatement> GenerateStatement(StatementType type); // came from private