first commit

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

@@ -0,0 +1,69 @@
+"""AST-level concurrency issue injection for Java using javalang."""
+
+from typing import Optional
+
+import javalang
+
+from app.dataset.rules.base import Mutation, MutationRule
+
+
+class JavaConcurrencyRule(MutationRule):
+    """Remove a synchronized block to expose a race condition.
+
+    Uses `javalang` to locate `synchronized (lock) { ... }` and replaces it
+    with the bare block body.
+    """
+
+    name = "java_concurrency"
+    language = "java"
+    defect_type = "concurrency_issue"
+
+    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.SynchronizedStatement):
+                continue
+            pos = node.position
+            if not pos:
+                continue
+            lines = source.splitlines(keepends=True)
+            start = pos.line
+            # Estimate end by braces of the synchronized block
+            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]
+            # drop header line and closing brace line, keep body; body is at same
+            # indentation as the synchronized header minus 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="Restore `synchronized (lock)` to protect the critical section.",
+                description="Removed synchronized block, exposing a race condition.",
+            )
+        return None