Overview
Output parsers transform raw LLM text into structured data. They support both complete outputs and streaming partial outputs.
BaseOutputParser
Base class for parsing LLM outputs.
Source: langchain_core.output_parsers.base:136
Inherits: BaseLLMOutputParser, RunnableSerializable[LanguageModelOutput, T]
Type Parameters
The output type after parsing
Properties
Input type for the parser (str | BaseMessage)
Output type after parsing (inferred from generic)
Core Methods
parse
def parse(self, text: str) -> T
parse_result
def parse_result(
self,
result: list[Generation],
*,
partial: bool = False
) -> T
List of generation candidates from the LLM
Whether to parse partial results (useful for streaming)
aparse
async def aparse(self, text: str) -> T
parse.
aparse_result
async def aparse_result(
self,
result: list[Generation],
*,
partial: bool = False
) -> T
parse_result.
invoke
def invoke(
self,
input: str | BaseMessage,
config: RunnableConfig | None = None,
**kwargs: Any
) -> T
def get_format_instructions(self) -> str
Format instructions to include in the prompt
JsonOutputParser
Parse LLM output as JSON.
Source: langchain_core.output_parsers.json:31
Inherits: BaseCumulativeTransformOutputParser[Any]
Properties
pydantic_object
type[BaseModel] | None
default:"None"
Optional Pydantic model for validation
Methods
parse_result
def parse_result(
self,
result: list[Generation],
*,
partial: bool = False
) -> Any
If True, parse partial JSON (useful for streaming)
Parsed JSON object (dict or list)
parse
def parse(self, text: str) -> Any
Example
from langchain_core.output_parsers import JsonOutputParser
from langchain_core.prompts import PromptTemplate
from pydantic import BaseModel, Field
class Joke(BaseModel):
setup: str = Field(description="The setup of the joke")
punchline: str = Field(description="The punchline")
parser = JsonOutputParser(pydantic_object=Joke)
prompt = PromptTemplate(
template="Tell me a joke.\n{format_instructions}",
input_variables=[],
partial_variables={"format_instructions": parser.get_format_instructions()}
)
chain = prompt | model | parser
result = chain.invoke({}) # {"setup": "...", "punchline": "..."}
PydanticOutputParser
Parse LLM output into a Pydantic model.
Source: langchain_core.output_parsers.pydantic
Inherits: BaseOutputParser[TBaseModel]
Type Parameters
Pydantic model type to parse into
Properties
Pydantic model class to parse into
Constructor
def __init__(self, pydantic_object: type[TBaseModel])
Methods
parse
def parse(self, text: str) -> TBaseModel
Validated Pydantic model instance
Example
from langchain_core.output_parsers import PydanticOutputParser
from pydantic import BaseModel, Field
class Person(BaseModel):
name: str = Field(description="Person's name")
age: int = Field(description="Person's age")
parser = PydanticOutputParser(pydantic_object=Person)
chain = prompt | model | parser
result = chain.invoke({...}) # Person(name="...", age=...)
StrOutputParser
Simple parser that returns the text output as a string.
Source: langchain_core.output_parsers.string
Inherits: BaseOutputParser[str]
Methods
parse
def parse(self, text: str) -> str
Example
from langchain_core.output_parsers import StrOutputParser
chain = prompt | model | StrOutputParser()
result = chain.invoke({...}) # "string output"
CommaSeparatedListOutputParser
Parse output into a comma-separated list.
Source: langchain_core.output_parsers.list
Inherits: BaseOutputParser[list[str]]
Methods
parse
def parse(self, text: str) -> list[str]
Example
from langchain_core.output_parsers import CommaSeparatedListOutputParser
parser = CommaSeparatedListOutputParser()
parser.parse("apple, banana, orange") # ["apple", "banana", "orange"]
XMLOutputParser
Parse XML output into a dictionary.
Source: langchain_core.output_parsers.xml
Inherits: BaseOutputParser[dict]
Properties
tags
list[str] | None
default:"None"
Optional list of tags to parse. If None, parses all tags.
Methods
parse
def parse(self, text: str) -> dict
Example
from langchain_core.output_parsers import XMLOutputParser
parser = XMLOutputParser()
parser.parse(
"<person><name>John</name><age>30</age></person>"
)
# {"person": {"name": "John", "age": "30"}}
langchain_core.output_parsers.openai_tools
Inherits: BaseGenerationOutputParser
Methods
parse_result
def parse_result(
self,
result: list[Generation],
*,
partial: bool = False
) -> list[dict[str, Any]]
List of tool call dictionaries
langchain_core.output_parsers.openai_tools
Inherits: BaseGenerationOutputParser
Properties
Name of the tool to extract
Whether to return only the first matching tool call
Example
from langchain_core.output_parsers.openai_tools import JsonOutputKeyToolsParser
parser = JsonOutputKeyToolsParser(key_name="search", first_tool_only=True)
chain = model.bind_tools([search_tool]) | parser
result = chain.invoke("search for cats") # {"query": "cats"}
langchain_core.output_parsers.openai_tools
Inherits: BaseGenerationOutputParser
Properties
tools
list[type[BaseModel]]
required
List of Pydantic model classes corresponding to tools
Whether to return only the first tool call
Example
from langchain_core.output_parsers.openai_tools import PydanticToolsParser
from pydantic import BaseModel, Field
class SearchArgs(BaseModel):
query: str = Field(description="Search query")
parser = PydanticToolsParser(tools=[SearchArgs], first_tool_only=True)
chain = model.bind_tools([SearchArgs]) | parser
result = chain.invoke("search for cats") # SearchArgs(query="cats")
langchain_core.output_parsers.transform
Inherits: BaseOutputParser[T]
Methods
def transform(
self,
input: Iterator[str | BaseMessage],
config: RunnableConfig | None = None,
**kwargs: Any
) -> Iterator[T]
async def atransform(
self,
input: AsyncIterator[str | BaseMessage],
config: RunnableConfig | None = None,
**kwargs: Any
) -> AsyncIterator[T]
transform.
OutputParserException
Exception raised when output parsing fails.
Source: langchain_core.exceptions
Inherits: ValueError
Properties
The output from the LLM that failed to parse
Optional observation about why parsing failed
Whether to send the error back to the LLM for correction