

Who this is for
- Form Extraction Developer
- Backend engineers who need to write model output to the database
What problem does this article solve?
The model output is mixed into interpreted text, field missing, or JSON parsing fails.
Key concepts
- structured output
- JSON Schema
- field type
- parse
- bug fix
Walkthrough
- Define the target schema
- Give field constraints in prompt
- Call the model to generate JSON
- Validate with Python and give a retry strategy
Python code example
The companion demo is generated in:
cd examples/llm-guides/009_json_schema_basics
python demo.py
The example follows two principles:
- From
OPENAI_API_KEY, load the key at runtime to avoid hard-coding it. - Without an API key, the example runs in mock mode so you can follow the workflow.
The essential call looks like this:
import os
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
print("未配置 OPENAI_API_KEY,进入 mock 模式。")
else:
# 真实项目中在这里初始化 SDK 客户端并发起请求。
print("已读取 OPENAI_API_KEY,可以调用模型。")
Common pitfalls
- Publishing after a single successful demo run leaves out error handling, logging and human review.
- Including API keys, real file paths, account details or business data in published articles or code.
- Readers can struggle to reproduce an example if it is unclear which parts are mocked and which run in production.
- Citing an API or framework feature without checking its documentation can mislead readers when versions change.
Summary
- LLM Guide 09 explains how to get consistent JSON output when a model adds explanatory prose, omits fields, or returns JSON that fails to parse.
- Make the workflow reproducible, reversible and easy to review. One good generation tells you little about how it will hold up in use.
- Before publishing, check current documentation for model capabilities, API parameters and third-party frameworks.
CSDN tags
JSON, large model, Python, data extraction, structured output
WeChat article summary
A field extraction example shows the basics of structured output and how to pass model results safely to a program.
Alternative title for Zhihu
- Why is the large model always outputting illegal JSON?
- How to make LLM output resolvable data?
- What large-model implementation problems can JSON Schema solve?
References and further reading
- Check the OpenAI documentation for your model, SDK and account permissions.
- Python documentation on environment variables, virtual environments and exception handling.
- Consult the documentation for the tools used in the example, such as LangChain, LlamaIndex, Haystack or the vector database.