# Output Parsers and Structured Output — LangChain / LlamaIndex

Source: https://www.geekswithgeeks.com/en/langchain-llamaindex/c-parsers

> Turn model text into validated Python objects.

## Parse, validate, fail loudly

Models return text; programs need objects. **Output parsers** convert the reply: `StrOutputParser` for plain text, `JsonOutputParser` for JSON, and **`PydanticOutputParser`** to validate against a **Pydantic** schema with types and constraints (for example urgency between 1 and 5). A parser can also produce **format instructions** to put in the prompt. If the reply breaks the schema the parser raises an `OutputParserException`, which you can catch and retry. Many chat models also support **native structured output** through `model.with_structured_output(Schema)`, which uses the provider's tool-calling or JSON-schema feature and is usually more reliable than parsing free text.

## Validating with Pydantic, run

I ran this offline in a Python virtual environment with langchain-core 1.6.6, langchain-text-splitters 1.1.2 and llama-index-core 0.14.25. No API key or network call is needed because a fake model or a toy embedding stands in for the real one. The valid reply becomes a `Ticket` object. A reply with urgency 9 violates the `le=5` constraint, so the parser raises `OutputParserException`.

```python
from pydantic import BaseModel, Field
from langchain_core.output_parsers import PydanticOutputParser
from langchain_core.exceptions import OutputParserException

class Ticket(BaseModel):
    category: str = Field(description="billing, technical or other")
    urgency: int = Field(ge=1, le=5)

parser = PydanticOutputParser(pydantic_object=Ticket)
print(parser.parse('{"category": "billing", "urgency": 4}'))
try:
    parser.parse('{"category": "billing", "urgency": 9}')
except OutputParserException as e:
    print("rejected:", type(e).__name__)
print("format hint mentions:", "urgency" in parser.get_format_instructions())

```

Output:

```
category='billing' urgency=4
rejected: OutputParserException
format hint mentions: True
```

**Quiz:** What happens when the reply breaks the Pydantic schema?

- [ ] The model is retrained
- [ ] The bad value is silently accepted
- [x] The parser raises an exception you can catch and retry
- [ ] The program ends without error

*Answer:* The parser raises an exception you can catch and retry. Failing loudly lets you handle bad output explicitly.