From 7ec22cd973acb850b7c84ade4f84424622b0c8a2 Mon Sep 17 00:00:00 2001
From: Jake LoRocco <jake.lorocco@ibm.com>
Date: Wed, 20 Aug 2025 11:47:23 -0400
Subject: [PATCH] copy reqs during validation; add test

---
 mellea/stdlib/requirement.py           | 15 ++++++++++-----
 test/stdlib_basics/test_requirement.py | 12 ++++++++++--
 2 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/mellea/stdlib/requirement.py b/mellea/stdlib/requirement.py
index ad4709fa..1aa849fa 100644
--- a/mellea/stdlib/requirement.py
+++ b/mellea/stdlib/requirement.py
@@ -3,6 +3,7 @@
 import inspect
 import re
 from collections.abc import Callable
+from copy import copy
 from typing import Any, overload
 
 from mellea.backends import (
@@ -97,6 +98,9 @@ def __init__(
         self.validation_fn = validation_fn
         self.check_only = check_only
 
+        # Used for validation. Do not manually populate.
+        self._output: str | None = None
+
     def validate(
         self,
         backend: Backend,
@@ -117,17 +121,18 @@ def validate(
             assert isinstance(last_output, ModelOutputThunk), (
                 " Context has no appropriate last output"
             )
-            self._output = last_output.value  # type: ignore
+
+            # Create a copy of the requirement that holds the output
+            # and its template gets populated with the output correctly.
+            req_copy = copy(self)
+            req_copy._output = last_output.value
             llm_as_a_judge_result = backend.generate_from_context(
-                self,
+                req_copy,
                 ctx,
                 format=format,
                 model_options=model_options,
                 generate_logs=generate_logs,
             )
-            # This is crucial, because requirements can get reused;
-            # this also means requirements are not thread-safe.
-            self._output = None
             return ValidationResult(
                 result=self.output_to_bool(llm_as_a_judge_result),
                 reason=llm_as_a_judge_result.value,
diff --git a/test/stdlib_basics/test_requirement.py b/test/stdlib_basics/test_requirement.py
index aa7845d4..12af105a 100644
--- a/test/stdlib_basics/test_requirement.py
+++ b/test/stdlib_basics/test_requirement.py
@@ -1,11 +1,19 @@
 import pytest
 from mellea.stdlib.base import ModelOutputThunk
-from mellea.stdlib.requirement import simple_validate
-from mellea.stdlib.session import SimpleContext
+from mellea.stdlib.requirement import Requirement, simple_validate
+from mellea.stdlib.session import SimpleContext, start_session
 
 ctx = SimpleContext()
 ctx.insert(ModelOutputThunk("test"))
 
+def test_llmaj_validation_req_output_field():
+    m = start_session(ctx=ctx)
+    req = Requirement("Must output test.")
+    assert req._output is None
+
+    _ = req.validate(m.backend,ctx=ctx)
+    assert req._output is None, "requirement's output shouldn't be updated during/after validation"
+
 def test_simple_validate_bool():
     validation_func = simple_validate(lambda x: False, reason="static reason")
     val_result = validation_func(ctx)