first commit

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

@@ -0,0 +1,63 @@
+"""AST-level boundary error injection for JavaScript using esprima."""
+
+from typing import Optional
+
+import esprima
+
+from app.dataset.rules.base import Mutation, MutationRule
+
+
+class JSBoundaryErrorRule(MutationRule):
+    """Mutate array length boundary from `<` to `<=`.
+
+    Uses `esprima` to locate a binary expression comparing against `.length`
+    and flips the operator.
+    """
+
+    name = "js_boundary_error"
+    language = "javascript"
+    defect_type = "boundary_condition_error"
+
+    def detect_and_mutate(self, source: str, filename: str = "") -> Optional[Mutation]:
+        # Modern JS is usually ESM: try module grammar first, then script.
+        try:
+            tree = esprima.parseModule(source, loc=True)
+        except Exception:
+            try:
+                tree = esprima.parseScript(source, loc=True)
+            except Exception:
+                return None
+
+        for node in self._walk(tree):
+            if node.type != "BinaryExpression" or node.operator != "<":
+                continue
+            right = node.right
+            if right.type == "MemberExpression" and getattr(right.property, "name", None) == "length":
+                line_no = node.loc.start.line
+                lines = source.splitlines(keepends=True)
+                line = lines[line_no - 1]
+                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 out-of-bounds access.",
+                    description="Changed array boundary check to off-by-one (<= length).",
+                )
+        return None
+
+    def _walk(self, node):
+        yield node
+        for key in getattr(node, "__dict__", {}):
+            child = getattr(node, key)
+            if isinstance(child, list):
+                for item in child:
+                    if hasattr(item, "type"):
+                        yield from self._walk(item)
+            elif hasattr(child, "type"):
+                yield from self._walk(child)