

Who this is for
- Developers who need to deal with a large number of text materials
- Head of Corporate Documentation Automation
What problem does this article solve?
During batch extraction, some documents fail and fields are inconsistent, making it difficult to track and re-run.
Key concepts
- batch
- task queue
- structured extraction
- Failed to run again
- audit log
Walkthrough
- Prepare the input directory
- Generate JSON file by file
- Save intermediate results
- Summarize as CSV and record the list of failures
Python code example
The companion demo is generated in:
cd examples/llm-guides/012_json_batch_extraction
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 12 addresses inconsistent formats in batch document extraction. When some documents fail and others produce different fields, it becomes hard to track what happened and rerun the affected files.
- 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
Large model, JSON, batch processing, Python, document processing
WeChat article summary
The batch extraction example shows how to organize structured output files, log the outcome for each document, and rerun failed extractions.
Alternative title for Zhihu
- How to extract document information in batch call model?
- How to ensure that the JSON output of multiple documents is consistent?
- What can go wrong when processing large-model requests in batches?
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.