Merge remote-tracking branch 'upstream/main' into make-fuzzer-ci-use-…

…reusable-wf

scripts/reduce_sql.py

@@ -18,6 +18,7 @@
 SELECT * FROM reduce_sql_statement('${QUERY}');
 '''
 
+verbose = False
 
 class MultiStatementManager:
     delimiter = ';'
@@ -33,7 +34,8 @@ def __init__(self, multi_statement):
                 self.statements.append(stmt.strip() + ";")
 
     def is_multi_statement(sql_statement):
-        if len(sql_statement.split(';')) > 1:
+        splits = [x for x in sql_statement.split(';') if len(x.strip()) > 0]
+        if len(splits) > 1:
             return True
         return False
 
@@ -45,10 +47,12 @@ def sanitize_error(err):
     err = re.sub(r'Error: near line \d+: ', '', err)
     err = err.replace(os.getcwd() + '/', '')
     err = err.replace(os.getcwd(), '')
+    err = re.sub(r'LINE \d+:.*\n', '', err)
+    err = re.sub(r' *\^ *', '', err)
     if 'AddressSanitizer' in err:
         match = re.search(r'[ \t]+[#]0 ([A-Za-z0-9]+) ([^\n]+)', err).groups()[1]
         err = 'AddressSanitizer error ' + match
-    return err
+    return err.strip()
 
 
 def run_shell_command(shell, cmd):
@@ -69,7 +73,13 @@ def get_reduced_sql(shell, sql_query):
         raise Exception("Failed to reduce query")
     reduce_candidates = []
     for line in stdout.split('\n'):
-        reduce_candidates.append(line.strip('"').replace('""', '"'))
+        if len(line) <= 2:
+            continue
+        if line[0] == '"':
+            line = line[1:]
+        if line[len(line) - 1] == '"':
+            line = line[:len(line) - 1]
+        reduce_candidates.append(line.replace('""', '"'))
     return reduce_candidates[1:]
 
 
@@ -95,6 +105,15 @@ def reduce(sql_query, data_load, shell, error_msg, max_time_seconds=300):
                 print(sql_query)
                 print("=======================")
                 break
+            elif verbose:
+                print("Failed to reduce query")
+                print("=======================")
+                print(reduce_candidate)
+                print("=====Target error======")
+                print(error_msg)
+                print("=====Actual error======")
+                print(new_error)
+                print("=======================")
         if not found_new_candidate:
             break
     return sql_query
@@ -271,13 +290,16 @@ def reduce_query_log(queries, shell, data_load=[], max_time_seconds=300):
     parser.add_argument(
         '--max-time', dest='max_time', action='store', help='Maximum time in seconds to run the reducer', default=300
     )
+    parser.add_argument(
+        '--verbose', dest='verbose', action='store_true', help='Verbose output')
 
     args = parser.parse_args()
     print("Starting reduce process")
 
     shell = args.shell
     data_load = open(args.load).read()
     sql_query = open(args.exec).read()
+    verbose = args.verbose
     (stdout, stderr, returncode) = run_shell_command(shell, data_load + sql_query)
     expected_error = sanitize_error(stderr).strip()
     if len(expected_error) == 0:

src/include/statement_simplifier.hpp

@@ -16,6 +16,7 @@ class SQLStatement;
 class SelectStatement;
 class InsertStatement;
 class UpdateStatement;
+class PrepareStatement;
 class DeleteStatement;
 class TableRef;
 class SelectNode;
@@ -41,6 +42,7 @@ class StatementSimplifier {
 	void Simplify(SelectStatement &stmt);
 	void Simplify(InsertStatement &stmt);
 	void Simplify(UpdateStatement &stmt);
+	void Simplify(PrepareStatement &stmt);
 	void Simplify(DeleteStatement &stmt);
 
 	void Simplification();

src/statement_simplifier.cpp

@@ -8,6 +8,7 @@
 #include "duckdb/parser/expression/list.hpp"
 #include "duckdb/parser/statement/delete_statement.hpp"
 #include "duckdb/parser/statement/insert_statement.hpp"
+#include "duckdb/parser/statement/prepare_statement.hpp"
 #include "duckdb/parser/statement/update_statement.hpp"
 #include "duckdb/parser/statement/select_statement.hpp"
 #endif
@@ -308,6 +309,7 @@ void StatementSimplifier::SimplifyExpression(duckdb::unique_ptr<ParsedExpression
 			SimplifyChildExpression(expr, case_check.then_expr);
 			SimplifyChildExpression(expr, case_check.when_expr);
 		}
+		SimplifyList(op.case_checks, false);
 		break;
 	}
 	case ExpressionClass::CAST: {
@@ -423,6 +425,10 @@ void StatementSimplifier::Simplify(UpdateSetInfo &info) {
 	}
 }
 
+void StatementSimplifier::Simplify(PrepareStatement &stmt) {
+	Simplify(*stmt.statement);
+}
+
 void StatementSimplifier::Simplify(UpdateStatement &stmt) {
 	Simplify(stmt.cte_map);
 	SimplifyOptional(stmt.from_table);
@@ -445,6 +451,9 @@ void StatementSimplifier::Simplify(SQLStatement &stmt) {
 	case StatementType::DELETE_STATEMENT:
 		Simplify(stmt.Cast<DeleteStatement>());
 		break;
+	case StatementType::PREPARE_STATEMENT:
+		Simplify(stmt.Cast<PrepareStatement>());
+		break;
 	default:
 		throw InvalidInputException("Expected a single SELECT, INSERT or UPDATE statement");
 	}

test/sql/sql_reduce.test

@@ -1,4 +1,4 @@
-# name: test/sql_reduce.test
+# name: test/sql/sql_reduce.test
 # description: Test reduce SQL statement
 # group: [sqlsmith]
 
@@ -77,3 +77,16 @@ DELETE FROM a WHERE (i < 5000)
 DELETE FROM a WHERE (i >= 2000)
 DELETE FROM a WHERE (i AND (i < 5000))
 DELETE FROM a WHERE NULL
+
+query I
+SELECT * FROM reduce_sql_statement('PREPARE v1 AS SELECT a, b FROM tbl') ORDER BY 1
+----
+PREPARE v1 AS SELECT NULL, b FROM tbl
+PREPARE v1 AS SELECT NULL, b FROM tbl
+PREPARE v1 AS SELECT a FROM tbl
+PREPARE v1 AS SELECT a, NULL FROM tbl
+PREPARE v1 AS SELECT a, NULL FROM tbl
+PREPARE v1 AS SELECT a, b
+PREPARE v1 AS SELECT a, b FROM tbl
+PREPARE v1 AS SELECT a, b FROM tbl
+PREPARE v1 AS SELECT b FROM tbl