[WIP] Semantic aware trans

reducer/code/executor.py

@@ -15,9 +15,9 @@ def execute_query(sql_query, test_script, query_path="query.sql"):
     try:
        # result = subprocess.run([test_script], env=env, capture_output=True, shell=True)
         result = subprocess.run(f"./{test_script}", env=env, capture_output=True, shell=True)
-        # print("STDOUT:", result.stdout.decode())
-        # print("STDERR:", result.stderr.decode())
-        # print("Return code:", result.returncode)
+        print("STDOUT:", result.stdout.decode())
+        print("STDERR:", result.stderr.decode())
+        print("Return code:", result.returncode)
 
         #print("result returncode: ", result)
         return result.returncode

reducer/code/reduce_query.py

@@ -1,14 +1,19 @@
 from code.parser import SQLParser
 from code.executor import execute_query
 from code.delta_debugging import delta_debugging
+from code.semantic import reduce_where_clause
 
 def reduce_query(query_path, test_script, output_path):
     with open(f"{query_path}/original_test.sql", "r") as original_query:
-        query_string = original_query.readlines()
+        query_string = original_query.read()
+
+
+    reduced_query = reduce_where_clause(query_string, test_script, output_path)
+    print("reduced_query ", reduced_query)
 
     # Parse the query to an AST
     parser = SQLParser()
-    token_tree = parser.parse(query_string)
+    token_tree = parser.parse([reduced_query])
     token_tree_size = sum(len(parser.flatten_tokens(tree)) for tree in token_tree)
 
     if not token_tree:

reducer/code/semantic.py

@@ -0,0 +1,67 @@
+# import sqlparse
+# from code.executor import execute_query
+
+
+# def extract_where_expressions(where_clause):
+#     return [expr.strip() for expr in where_clause.split("AND")]
+
+# def rebuild_query(base_query, where_exprs):
+#     if not where_exprs:
+#         return base_query.split("WHERE")[0].strip() + ";"
+#     new_where = " AND ".join(where_exprs)
+#     return base_query.split("WHERE")[0].strip() + f" WHERE {new_where};"
+
+# def reduce_where_clause(query_str, test_script, query_output_path="query.sql"):
+#     if "WHERE" not in query_str.upper():
+#         return query_str
+
+#     where_start = query_str.upper().find("WHERE")
+#     base_query = query_str[:where_start]
+#     where_clause = query_str[where_start + len("WHERE"):].strip().rstrip(";")
+
+#     expressions = extract_where_expressions(where_clause)
+#     i = 0
+
+#     while i < len(expressions):
+#         trial_exprs = expressions[:i] + expressions[i+1:]
+#         trial_query = rebuild_query(base_query, trial_exprs)
+
+#         result_code = execute_query(trial_query, test_script, query_output_path)
+
+#         if result_code == 0:
+#             print(f"[REDUCED] Removed WHERE clause: {expressions[i]}")
+#             expressions = trial_exprs
+#         else:
+#             print(f"[RETAINED] Keeping WHERE clause: {expressions[i]}")
+#             i += 1
+
+#     return rebuild_query(base_query, expressions)
+import re
+from code.executor import execute_query
+
+def remove_where_clause(query_string: str) -> str:
+    """
+    Removes the WHERE clause from the SQL query using regex (simplified).
+    Assumes only one WHERE clause exists and is not nested in subqueries.
+    """
+    pattern = re.compile(r"\bWHERE\b.+?(?=(GROUP BY|ORDER BY|LIMIT|;|$))", re.IGNORECASE | re.DOTALL)
+    return pattern.sub('', query_string)
+
+
+def reduce_where_clause(query_string: str, test_script: str, query_path: str) -> str:
+    """
+    Try to semantically reduce the WHERE clause. Only keep the reduction
+    if the test script returns exit code 0 (bug still happens).
+    """
+    reduced_query = remove_where_clause(query_string)
+
+    # Test the reduced query
+    result = execute_query(reduced_query, test_script, query_path)
+
+    if result == 0:
+        print("[✔] Removed WHERE clause — bug still happens.")
+        return reduced_query
+    else:
+        print("[✘] Removing WHERE clause changed behavior — keeping original.")
+        return query_string
+

reducer/code/utils.py

@@ -1,6 +1,5 @@
 import shutil
 import os
-import sqlparse
 
 def prepare_workspace(query_path):
     src = os.path.abspath(query_path)

reducer/reducer.py

@@ -6,7 +6,6 @@
 from code.reduce_query import reduce_query
 from code.utils import prepare_workspace
 import time
-from code.utils import count_tokens
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Reduce bug-triggering SQL queries")

reducer/test.sql

@@ -0,0 +1,2 @@
+CREATE TABLE F (p BOOLEAN NOT NULL NULL NOT NULL, i BOOLEAN);
+INSERT INTO F SELECT * FROM (VALUES ((NOT false), false), (NULL, (NOT (NOT true)))) AS L;