Auto-fixing malformed output in LangChain - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ component_behavior

intermediate

What is the output of this LangChain output parser when given malformed JSON?

Consider a LangChain output parser designed to auto-fix malformed JSON outputs from a language model. Given the input string:

{"name": "Alice", "age": 30,}

What will the parser output after auto-fixing?

LangChain

from langchain.output_parsers import JsonOutputKeyToolsParser

parser = JsonOutputKeyToolsParser()
input_str = '{"name": "Alice", "age": 30,}'
result = parser.parse(input_str)
print(result)

A {'name': 'Alice', 'age': 30}

B None

C {'name': 'Alice', 'age': '30,'}

D SyntaxError: Expecting property name enclosed in double quotes

💡 Hint

Think about how auto-fixing parsers handle trailing commas in JSON.

The parser auto-fixes the trailing comma in the JSON string, resulting in a valid dictionary with keys 'name' and 'age'.

📝 Syntax

intermediate

Which option correctly fixes this malformed LangChain output with missing quotes?

Given this malformed output from a language model:

{name: "Bob", age: 25}

Which option shows the correctly auto-fixed output parser result?

LangChain

from langchain.output_parsers import JsonOutputKeyToolsParser

parser = JsonOutputKeyToolsParser()
input_str = '{name: "Bob", age: 25}'
result = parser.parse(input_str)
print(result)

A {'name': Bob, 'age': 25}

B SyntaxError: Expecting property name enclosed in double quotes

C {'name': 'Bob', 'age': 25}

D {name: 'Bob', age: 25}

💡 Hint

JSON keys must be strings with double quotes; auto-fix adds missing quotes.

The parser adds missing double quotes around keys to produce valid JSON, then returns the dictionary.

🔧 Debug

advanced

What error does this LangChain output parser raise with unfixable malformed output?

Given this input string with unbalanced brackets:

{"city": "Paris", "country": "France"

What error will the LangChain output parser raise when trying to auto-fix?

LangChain

from langchain.output_parsers import JsonOutputKeyToolsParser

parser = JsonOutputKeyToolsParser()
input_str = '{"city": "Paris", "country": "France"'
result = parser.parse(input_str)

A No error, returns {'city': 'Paris', 'country': 'France'}

B ValueError: Could not auto-fix malformed output

C TypeError: 'NoneType' object is not subscriptable

D json.decoder.JSONDecodeError: Expecting ',' delimiter

💡 Hint

Think about what happens when JSON is incomplete and cannot be fixed automatically.

The parser tries to parse but fails due to missing closing bracket, raising JSONDecodeError.

❓ state_output

advanced

What is the value of the 'age' key after auto-fixing this LangChain output?

Given this malformed output string:

{"name": "Eve", "age": "twenty-five"}

and a parser that expects 'age' as an integer and auto-fixes types, what will be the value of 'age' in the parsed output?

LangChain

from langchain.output_parsers import JsonOutputKeyToolsParser

class CustomParser(JsonOutputKeyToolsParser):
    def parse(self, text):
        result = super().parse(text)
        if isinstance(result.get('age'), str):
            try:
                result['age'] = int(result['age'])
            except ValueError:
                result['age'] = None
        return result

parser = CustomParser()
input_str = '{"name": "Eve", "age": "twenty-five"}'
result = parser.parse(input_str)
print(result['age'])

A "twenty-five"

B None

C 25

D TypeError

💡 Hint

The string 'twenty-five' cannot convert to int, so the parser sets age to None.

The parser tries to convert 'age' to int, fails, and assigns None to 'age'.

🧠 Conceptual

expert

Which option best describes the main limitation of auto-fixing malformed LangChain outputs?

Auto-fixing malformed outputs helps handle minor errors in language model responses. What is the main limitation of this approach?

A It cannot fix semantic errors or incorrect data types beyond simple syntax issues.

B It always produces perfect outputs regardless of input quality.

C It slows down the language model response time significantly.

D It requires manual intervention for every output to confirm fixes.

💡 Hint

Think about what auto-fixing can and cannot do automatically.

Auto-fixing mainly addresses syntax errors but cannot correct wrong meanings or complex data type mismatches.