[PYTHON/LANGCHAIN] 커스텀 출력 파서 만들기 2

■ 커스텀 출력 파서를 만드는 방법을 보여준다.

※ OPENAI_API_KEY 환경 변수 값은 .env 파일에 정의한다.

▶ main.py


import re
import json

from dotenv                  import load_dotenv
from pydantic                import BaseModel
from pydantic                import Field
from typing                  import List
from langchain_core.prompts  import ChatPromptTemplate
from langchain_openai        import ChatOpenAI
from langchain_core.messages import AIMessage

load_dotenv()

class Person(BaseModel):
    """Information about a person."""

    name             : str   = Field(..., description = "The name of the person"                       )
    height_in_meters : float = Field(..., description = "The height of the person expressed in meters.")

class People(BaseModel):
    """Identifying information about all people in a text."""

    people : List[Person]

chatPromptTemplate1 = ChatPromptTemplate.from_messages(
    [
        ("system", "Answer the user query. Output your answer as JSON that matches the given schema: ```json\n{schema}\n```. Make sure to wrap the answer in ```json and ``` tags"),
        ("human" , "{query}"),
    ]
)

chatPromptTemplate2 = chatPromptTemplate1.partial(schema = People.model_json_schema())

chatOpenAI = ChatOpenAI(model = "gpt-4o-mini")

def extractJSON(aiMessage : AIMessage) -> List[dict]:
    """Extracts JSON content from a string where JSON is embedded between ```json and ``` tags.

    Parameters :
        text (str) : The text containing the JSON content.

    Returns :
        list : A list of extracted JSON strings.
    """
    text = aiMessage.content
    # JSON 블록과 일치하도록 정규 표현식 패턴을 정의한다.
    pattern = r"```json(.*?)```"
    # 문자열에서 패턴의 겹치지 않는 모든 일치 항목을 찾는다.
    matcheStringList = re.findall(pattern, text, re.DOTALL)
    # 앞이나 뒤의 공백을 제거하고 일치하는 JSON 문자열 목록을 반환한다.
    try:
        return [json.loads(matchString.strip()) for matchString in matcheStringList]
    except Exception:
        raise ValueError(f"Failed to parse: {aiMessage}")

runnableSequence = chatPromptTemplate2 | chatOpenAI | extractJSON

responsePeopleList = runnableSequence.invoke({"query" : "Anna is 23 years old and she is 6 feet tall"})

print(responsePeopleList)

"""
[{'people': [{'name': 'Anna', 'height_in_meters': 1.8288}]}]
"""

import re

import json

from dotenv import load_dotenv

from pydantic import BaseModel

from pydantic import Field

from typing import List

from langchain_core.prompts import ChatPromptTemplate

from langchain_openai import ChatOpenAI

from langchain_core.messages import AIMessage

load_dotenv()

class Person(BaseModel):

"""Information about a person."""

name : str = Field(..., description = "The name of the person" )

height_in_meters : float = Field(..., description = "The height of the person expressed in meters.")

class People(BaseModel):

"""Identifying information about all people in a text."""

people : List[Person]

chatPromptTemplate1 = ChatPromptTemplate.from_messages(

[

("system", "Answer the user query. Output your answer as JSON that matches the given schema: ```json\n{schema}\n```. Make sure to wrap the answer in ```json and ``` tags"),

("human" , "{query}"),

]

)

chatPromptTemplate2 = chatPromptTemplate1.partial(schema = People.model_json_schema())

chatOpenAI = ChatOpenAI(model = "gpt-4o-mini")

def extractJSON(aiMessage : AIMessage) -> List[dict]:

"""Extracts JSON content from a string where JSON is embedded between ```json and ``` tags.

Parameters :

text (str) : The text containing the JSON content.

Returns :

list : A list of extracted JSON strings.

"""

text = aiMessage.content

# JSON 블록과 일치하도록 정규 표현식 패턴을 정의한다.

pattern = r"```json(.*?)```"

# 문자열에서 패턴의 겹치지 않는 모든 일치 항목을 찾는다.

matcheStringList = re.findall(pattern, text, re.DOTALL)

# 앞이나 뒤의 공백을 제거하고 일치하는 JSON 문자열 목록을 반환한다.

try:

return [json.loads(matchString.strip()) for matchString in matcheStringList]

except Exception:

raise ValueError(f"Failed to parse: {aiMessage}")

runnableSequence = chatPromptTemplate2 | chatOpenAI | extractJSON

responsePeopleList = runnableSequence.invoke({"query" : "Anna is 23 years old and she is 6 feet tall"})

print(responsePeopleList)

"""

[{'people': [{'name': 'Anna', 'height_in_meters': 1.8288}]}]

"""

▶ requirements.txt


annotated-types==0.7.0
anyio==4.7.0
certifi==2024.8.30
charset-normalizer==3.4.0
colorama==0.4.6
distro==1.9.0
h11==0.14.0
httpcore==1.0.7
httpx==0.28.1
idna==3.10
jiter==0.8.0
jsonpatch==1.33
jsonpointer==3.0.0
langchain-core==0.3.22
langchain-openai==0.2.11
langsmith==0.1.147
openai==1.57.0
orjson==3.10.12
packaging==24.2
pydantic==2.10.3
pydantic_core==2.27.1
python-dotenv==1.0.1
PyYAML==6.0.2
regex==2024.11.6
requests==2.32.3
requests-toolbelt==1.0.0
sniffio==1.3.1
tenacity==9.0.0
tiktoken==0.8.0
tqdm==4.67.1
typing_extensions==4.12.2
urllib3==2.2.3

annotated-types==0.7.0

anyio==4.7.0

certifi==2024.8.30

charset-normalizer==3.4.0

colorama==0.4.6

distro==1.9.0

h11==0.14.0

httpcore==1.0.7

httpx==0.28.1

idna==3.10

jiter==0.8.0

jsonpatch==1.33

jsonpointer==3.0.0

langchain-core==0.3.22

langchain-openai==0.2.11

langsmith==0.1.147

openai==1.57.0

orjson==3.10.12

packaging==24.2

pydantic==2.10.3

pydantic_core==2.27.1

python-dotenv==1.0.1

PyYAML==6.0.2

regex==2024.11.6

requests==2.32.3

requests-toolbelt==1.0.0

sniffio==1.3.1

tenacity==9.0.0

tiktoken==0.8.0

tqdm==4.67.1

typing_extensions==4.12.2

urllib3==2.2.3

※ pip install python-dotenv langchain-openai 명령을 실행했다.