From 1d649b127eb10c426f9b9a67cbd1fe6ec8e6befa Mon Sep 17 00:00:00 2001
From: SvMax <Biancucci95@gmail.com>
Date: Sat, 8 Jul 2023 08:57:26 +0200
Subject: [PATCH] Added param to return only a structured json from the
 get_format_instructions method (#5848)

I just added a parameter to the method get_format_instructions, to
return directly the JSON instructions without the leading instruction
sentence. I'm planning to use it to define the structure of a JSON
object passed in input, the get_format_instructions().

---------

Co-authored-by: Bagatur <baskaryan@gmail.com>
---
 .../output_parsers/format_instructions.py     |  8 +++
 langchain/output_parsers/structured.py        | 52 +++++++++++++++++--
 2 files changed, 57 insertions(+), 3 deletions(-)

diff --git a/langchain/output_parsers/format_instructions.py b/langchain/output_parsers/format_instructions.py
index d9e5b6b69c..23c8328178 100644
--- a/langchain/output_parsers/format_instructions.py
+++ b/langchain/output_parsers/format_instructions.py
@@ -8,6 +8,14 @@ STRUCTURED_FORMAT_INSTRUCTIONS = """The output should be a markdown code snippet
 }}
 ```"""
 
+STRUCTURED_FORMAT_SIMPLE_INSTRUCTIONS = """
+```json
+{{
+{format}
+}}
+"""
+
+
 PYDANTIC_FORMAT_INSTRUCTIONS = """The output should be formatted as a JSON instance that conforms to the JSON schema below.
 
 As an example, for the schema {{"properties": {{"foo": {{"title": "Foo", "description": "a list of strings", "type": "array", "items": {{"type": "string"}}}}}}, "required": ["foo"]}}}}
diff --git a/langchain/output_parsers/structured.py b/langchain/output_parsers/structured.py
index a7b5c3910d..40ff75174a 100644
--- a/langchain/output_parsers/structured.py
+++ b/langchain/output_parsers/structured.py
@@ -4,7 +4,10 @@ from typing import Any, List
 
 from pydantic import BaseModel
 
-from langchain.output_parsers.format_instructions import STRUCTURED_FORMAT_INSTRUCTIONS
+from langchain.output_parsers.format_instructions import (
+    STRUCTURED_FORMAT_INSTRUCTIONS,
+    STRUCTURED_FORMAT_SIMPLE_INSTRUCTIONS,
+)
 from langchain.output_parsers.json import parse_and_check_json_markdown
 from langchain.schema import BaseOutputParser
 
@@ -32,11 +35,54 @@ class StructuredOutputParser(BaseOutputParser):
     ) -> StructuredOutputParser:
         return cls(response_schemas=response_schemas)
 
-    def get_format_instructions(self) -> str:
+    def get_format_instructions(self, only_json: bool = False) -> str:
+        """
+        Method to get the format instructions for the output parser.
+
+        example:
+        ```python
+        from langchain.output_parsers.structured import (
+            StructuredOutputParser, ResponseSchema
+        )
+
+        response_schemas = [
+            ResponseSchema(
+                name="foo",
+                description="a list of strings",
+                type="List[string]"
+                ),
+            ResponseSchema(
+                name="bar",
+                description="a string",
+                type="string"
+                ),
+        ]
+
+        parser = StructuredOutputParser.from_response_schemas(response_schemas)
+
+        print(parser.get_format_instructions())
+
+        output:
+        # The output should be a markdown code snippet formatted in the following
+        # schema, including the leading and trailing "```json" and "```":
+        #
+        # ```json
+        # {
+        #     "foo": List[string]  // a list of strings
+        #     "bar": string  // a string
+        # }
+
+        Args:
+            only_json (bool): If True, only the json in the markdown code snippet
+                will be returned, without the introducing text. Defaults to False.
+        """
         schema_str = "\n".join(
             [_get_sub_string(schema) for schema in self.response_schemas]
         )
-        return STRUCTURED_FORMAT_INSTRUCTIONS.format(format=schema_str)
+        if only_json:
+            return STRUCTURED_FORMAT_SIMPLE_INSTRUCTIONS.format(format=schema_str)
+        else:
+            return STRUCTURED_FORMAT_INSTRUCTIONS.format(format=schema_str)
 
     def parse(self, text: str) -> Any:
         expected_keys = [rs.name for rs in self.response_schemas]