From 8d6cc90fc52195da99e633ddd1e9662c6c826d8b Mon Sep 17 00:00:00 2001
From: JongRok BAEK <54343137+L-cloud@users.noreply.github.com>
Date: Wed, 14 Feb 2024 06:30:53 +0900
Subject: [PATCH] langchain.core : Use shallow copy for schema manipulation in
 JsonOutputParser.get_format_instructions (#17162)

- **Description :**

Fix: Use shallow copy for schema manipulation in get_format_instructions

Prevents side effects on the original schema object by using a
dictionary comprehension for a safer and more controlled manipulation of
schema key-value pairs, enhancing code reliability.

  - **Issue:**  #17161
  - **Dependencies:** None
  -  **Twitter handle:** None
---
 libs/core/langchain_core/output_parsers/json.py  |  3 ++-
 .../tests/unit_tests/output_parsers/test_json.py | 16 ++++++++++++++++
 .../langchain/output_parsers/pydantic.py         |  3 ++-
 libs/langchain/langchain/output_parsers/yaml.py  |  3 ++-
 4 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/libs/core/langchain_core/output_parsers/json.py b/libs/core/langchain_core/output_parsers/json.py
index d16316596a7..c107cd0c765 100644
--- a/libs/core/langchain_core/output_parsers/json.py
+++ b/libs/core/langchain_core/output_parsers/json.py
@@ -221,7 +221,8 @@ class JsonOutputParser(BaseCumulativeTransformOutputParser[Any]):
         if self.pydantic_object is None:
             return "Return a JSON object."
         else:
-            schema = self.pydantic_object.schema()
+            # Copy schema to avoid altering original Pydantic schema.
+            schema = {k: v for k, v in self.pydantic_object.schema().items()}
 
             # Remove extraneous fields.
             reduced_schema = schema
diff --git a/libs/core/tests/unit_tests/output_parsers/test_json.py b/libs/core/tests/unit_tests/output_parsers/test_json.py
index 6d8cb4bee39..8cedc767699 100644
--- a/libs/core/tests/unit_tests/output_parsers/test_json.py
+++ b/libs/core/tests/unit_tests/output_parsers/test_json.py
@@ -8,6 +8,8 @@ from langchain_core.output_parsers.json import (
     parse_json_markdown,
     parse_partial_json,
 )
+from langchain_core.pydantic_v1 import BaseModel
+from langchain_core.utils.function_calling import convert_to_openai_function
 
 GOOD_JSON = """```json
 {
@@ -579,3 +581,17 @@ def test_partial_text_json_output_parser_with_json_code_block() -> None:
         {"country_name": "France", "population_size": 673915},
         {"country_name": "France", "population_size": 67391582},
     ]
+
+
+def test_base_model_schema_consistency() -> None:
+    class Joke(BaseModel):
+        setup: str
+        punchline: str
+
+    initial_joke_schema = {k: v for k, v in Joke.schema().items()}
+    SimpleJsonOutputParser(pydantic_object=Joke)
+    openai_func = convert_to_openai_function(Joke)
+    retrieved_joke_schema = {k: v for k, v in Joke.schema().items()}
+
+    assert initial_joke_schema == retrieved_joke_schema
+    assert openai_func.get("name", None) is not None
diff --git a/libs/langchain/langchain/output_parsers/pydantic.py b/libs/langchain/langchain/output_parsers/pydantic.py
index 1248560c0ca..9e415650425 100644
--- a/libs/langchain/langchain/output_parsers/pydantic.py
+++ b/libs/langchain/langchain/output_parsers/pydantic.py
@@ -29,7 +29,8 @@ class PydanticOutputParser(JsonOutputParser):
             raise OutputParserException(msg, llm_output=json_object)
 
     def get_format_instructions(self) -> str:
-        schema = self.pydantic_object.schema()
+        # Copy schema to avoid altering original Pydantic schema.
+        schema = {k: v for k, v in self.pydantic_object.schema().items()}
 
         # Remove extraneous fields.
         reduced_schema = schema
diff --git a/libs/langchain/langchain/output_parsers/yaml.py b/libs/langchain/langchain/output_parsers/yaml.py
index 528fc93f507..21bcf359a2c 100644
--- a/libs/langchain/langchain/output_parsers/yaml.py
+++ b/libs/langchain/langchain/output_parsers/yaml.py
@@ -43,7 +43,8 @@ class YamlOutputParser(BaseOutputParser[T]):
             raise OutputParserException(msg, llm_output=text) from e
 
     def get_format_instructions(self) -> str:
-        schema = self.pydantic_object.schema()
+        # Copy schema to avoid altering original Pydantic schema.
+        schema = {k: v for k, v in self.pydantic_object.schema().items()}
 
         # Remove extraneous fields.
         reduced_schema = schema