ValueError(f"Missing some input keys: {missing_keys}")

Question

I am facing an error:

ValueError(f"Missing some input keys: {missing_keys}")

from this code. I am trying to parse resumes that are not supposed to have the keywords Name, phone, about_me in them.

However, it parses the keyword it finds similar to the Pydantic class, if prompt is used where I mention setting the field to null if the fields have no information.

Traceback:

Traceback (most recent call last):
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\main.py", line 188, in <module>
    info = ExtractInformationFromResume(extracted_text)
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\main.py", line 120, in ExtractInformationFromResume
    response = chain.run({"resume": resume})
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain_core\_api\deprecation.py", line 179, in warning_emitting_wrapper
    return wrapped(*args, **kwargs)
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain\chains\base.py", line 606, in run
    return self(args[0], callbacks=callbacks, tags=tags, metadata=metadata)[
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain_core\_api\deprecation.py", line 179, in warning_emitting_wrapper
    return wrapped(*args, **kwargs)
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain\chains\base.py", line 389, in __call__
    return self.invoke(
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain\chains\base.py", line 170, in invoke
    raise e
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain\chains\base.py", line 158, in invoke
    self._validate_inputs(inputs)
  File "C:\Users\Sarthak\PycharmProjects\API Resume parser\venv\lib\site-packages\langchain\chains\base.py", line 290, in _validate_inputs
    raise ValueError(f"Missing some input keys: {missing_keys}")
ValueError: Missing some input keys: {'\n    "ContactInformation"'}

class ContactInformation(BaseModel):
    Name: Optional[str] = None
    Email: Optional[str] = None
    Contact: Optional[str] = None
    Links: Optional[List[str]] = None

class Experience(BaseModel):
    title: Optional[str] = None
    company: Optional[str] = None
    duration: Optional[str] = None

class Education(BaseModel):
    course: Optional[str] = None
    branch: Optional[str] = None
    institute: Optional[str] = None

class Projects(BaseModel):
    name: Optional[str] = None
    description: Optional[str] = None
    link: Optional[str] = None

class OutputFormat(BaseModel):
    ContactInformation: Optional[ContactInformation] = None
    About_Me: Optional[str] = None
    Experience: Optional[Union[None, List[Experience]]] = None
    Education: Optional[Union[None, List[Education]]] = None
    Skills: Optional[List[str]] = None
    Certificates: Optional[List[str]] = None
    Projects: Optional[Union[None, List[Projects]]] = None
    Achievements: Optional[List[str]] = None
    Volunteer: Optional[List[str]] = None

Prompt:

PROMPT_2 = """You are given resume : ```{resume}```

Based on the given resume, extract information about the person.

The output should strictly follow the JSON schema below. If a field does not have information, set it to null. Ensure that the email contains an "@" symbol and the contact number is a string of digits. The name is usually found at the beginning of the resume without any labels. Social links should be valid URLs.

{
    "ContactInformation": {
        "Name": string, // Name of the person, usually at the beginning
        "Email": string, // Email address containing "@"
        "Contact": string, // Contact number as a string of digits
        "Links": [string] // Array of social profile URLs
    },
    "About_Me": string, // A brief summary about the person
    "Experience": [
        {
            "title": string, // Title of the position
            "company": string, // Company name
            "duration": string // Duration of employment
        }
    ],
    "Education": [
        {
            "course": string, // Name of the course or subject studied
            "branch": string, // Branch or faculty of study
            "institute": string // Name of the educational institute
        }
    ],
    "Skills": [string], // Array of skills
    "Certificates": [string], // Array of certificates
    "Projects": [
        {
            "name": string, // Name of the project
            "description": string, // Description of the project
            "link": string // URL to the project
        }
    ],
    "Achievements": [string], // Array of achievements
    "Volunteer": [string] // Array of volunteer work
}
"""

ExtractInformationFromResume function:

def ExtractInformationFromResume(resume: str) -> OutputFormat:
    llm = ChatOpenAI(
        openai_api_key='*********************************************',
        temperature=0.5,
        model_name="gpt-3.5-turbo"
    )

    # Define the output parser
    parser = PydanticOutputParser(pydantic_object=OutputFormat)

    # Create the prompt template with format instructions
    prompt_template = PromptTemplate(
        input_variables=["resume"],
        template=PROMPT_2,
        partial_variables={"format_instructions": parser.get_format_instructions()},
    )

    # Chain the LLM with the prompt
    chain = LLMChain(llm=llm, prompt=prompt_template)

    # Run the chain with the resume as input
    print(f"Resume Length: {len(resume)} characters")
    response = chain.run({"resume": resume})
    try:
        print("Raw LLM Response:", response)
    except Exception as e:
        print(f"Error: {str(e)}")

    # Parse the LLM response into the expected output format
    try:
        return OutputFormat(**json.loads(response))
    except json.JSONDecodeError:
        raise ValueError("Failed to parse response as valid JSON: " + response)