first commit

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

@@ -0,0 +1,57 @@
+"""AST-level boundary error injection for Java using javalang."""
+
+from typing import Optional
+
+import javalang
+
+from app.dataset.rules.base import Mutation, MutationRule
+
+
+class JavaBoundaryErrorRule(MutationRule):
+    """Mutate array length boundary from `<` to `<=`.
+
+    Uses `javalang` to locate a binary comparison against `.length` and flips
+    the operator to introduce an off-by-one access.
+    """
+
+    name = "java_boundary_error"
+    language = "java"
+    defect_type = "boundary_condition_error"
+
+    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.BinaryOperation):
+                continue
+            if node.operator != "<":
+                continue
+            right = node.operandr
+            if isinstance(right, javalang.tree.MemberReference) and right.member == "length":
+                # BinaryOperation itself may lack position; use enclosing statement
+                if_statement = next((n for n in path if isinstance(n, javalang.tree.IfStatement)), None)
+                pos = if_statement.position if if_statement else node.position
+                if not pos:
+                    continue
+                lines = source.splitlines(keepends=True)
+                line_no = pos.line
+                line = lines[line_no - 1]
+                mutated_line = line.replace("< ", "<= ", 1)
+                if mutated_line == line:
+                    mutated_line = line.replace("<", "<=", 1)
+                if mutated_line == line:
+                    continue
+                mutated = "".join(lines[:line_no - 1] + [mutated_line] + lines[line_no:])
+                return Mutation(
+                    defect_type=self.defect_type,
+                    language=self.language,
+                    line_start=line_no,
+                    line_end=line_no,
+                    mutated_source=mutated,
+                    reference_fix="Use strict `< length` to avoid ArrayIndexOutOfBoundsException.",
+                    description="Changed array boundary check to off-by-one (<= length).",
+                )
+        return None