first commit

backend/app/dataset/rules/java/null_pointer.py

@@ -0,0 +1,75 @@
+"""AST-level null-pointer injection for Java using javalang."""
+
+from typing import Optional
+
+import javalang
+
+from app.dataset.rules.base import Mutation, MutationRule
+
+
+class JavaNoneReferenceRule(MutationRule):
+    """Remove a null-check guard in Java source.
+
+    Uses the pure-Python `javalang` parser to locate an `if (x != null)` guard
+    and remove it, leaving the dereference unprotected. This keeps mutation
+    semantics precise without regex/text replacement.
+    """
+
+    name = "java_none_reference"
+    language = "java"
+    defect_type = "null_pointer"
+
+    def detect_and_mutate(self, source: str, filename: str = "") -> Optional[Mutation]:
+        try:
+            tree = javalang.parse.parse(source)
+        except Exception:
+            return None
+
+        for path, node in tree:
+            if not isinstance(node, javalang.tree.IfStatement):
+                continue
+            cond = node.condition
+            # Match: x != null
+            if (
+                isinstance(cond, javalang.tree.BinaryOperation)
+                and cond.operator == "!="
+                and isinstance(cond.operandr, javalang.tree.Literal)
+                and cond.operandr.value == "null"
+            ):
+                var_name = getattr(cond.operandl, "member", str(cond.operandl))
+                lines = source.splitlines(keepends=True)
+                start = node.position.line if node.position else 1
+                # Estimate end line by finding matching brace (simplistic)
+                end = start
+                depth = 0
+                for idx in range(start - 1, len(lines)):
+                    depth += lines[idx].count("{") - lines[idx].count("}")
+                    if depth > 0:
+                        end = idx + 1
+                    if depth <= 0 and idx > start - 1:
+                        end = idx + 1
+                        break
+                body_lines = lines[start - 1:end]
+                # keep body lines between header and closing brace, dedent one level
+                inner = body_lines[1:-1] if len(body_lines) > 2 else []
+                dedented = []
+                for line in inner:
+                    if line.startswith("            "):
+                        dedented.append("        " + line[12:])
+                    elif line.startswith("        "):
+                        dedented.append("    " + line[8:])
+                    elif line.startswith("    "):
+                        dedented.append(line[4:])
+                    else:
+                        dedented.append(line)
+                mutated = "".join(lines[: start - 1] + dedented + lines[end:])
+                return Mutation(
+                    defect_type=self.defect_type,
+                    language=self.language,
+                    line_start=start,
+                    line_end=end,
+                    mutated_source=mutated,
+                    reference_fix=f"Add `if ({var_name} != null)` guard before dereferencing.",
+                    description=f"Removed null-check guard for '{var_name}'.",
+                )
+        return None