What this category of data looks like
Urban gas utility application materials and case review data originate from submitted documents and archived case files for services including new installation, ownership transfer, safety inspection, and similar operations. Sources include PDF and Word files exported from business systems, offline scanned paper materials, and Feishu collaborative project documents. Data is updated per individual business batch. A single document typically contains multi-level approval nodes, user qualification information, site inspection records, and other content. Fields cover gas user identification number, gas usage address, installation capacity, approval duration, and more. Units such as cubic meters, square meters, and working days are used.
What constraints these characteristics impose on the document parsing and chunking step
Urban gas utility application materials and case review data have a multi-level directory structure, including nested content such as approval nodes and business forms. Full retention of hierarchical links is required during document parsing, otherwise process information will be lost during case review. A single document covers multiple types of business fields, so chunking must retain key identifier associations to avoid inability to match core information such as user ID after splitting. Incomplete parsing of Feishu multi-level documents with embedded docs and image formats will lead to missing approval records. Large-volume scanned PDFs and multi-page case documents may exceed the performance limits of parsing and chunking.
Configuration Settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | A single chunk of urban gas application materials must accommodate a complete business form or approval node to avoid splitting critical fields |
chunkSplitter | Split by headings and paragraphs | Adapts to the multi-level directory document structure and retains hierarchical links of approval workflows |
customPdfParse | Enable OCR mode | Adapts to offline scanned paper case documents and improves parsing accuracy for scanned files |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Covers the upload requirements of large-volume case PDF files and avoids 413 errors |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Adapts to the parsing duration of multi-page case documents and avoids timeout interruptions |
chunkSettingMode | Manual configuration | Adjust chunking parameters for different types of application materials such as new installation, safety inspection |
Three Common Mistakes
- A 413 error is triggered when uploading a file. The cause is that the
UPLOAD_FILE_MAX_SIZEconfiguration value is smaller than the actual size of the uploaded file, failing to meet the upload requirements of large-volume case PDF files. - Feishu multi-level directory documents cannot be fully parsed and retrieved. The cause is that chunking rules adapted to multi-level structures are not enabled, and fixed-length splitting alone results in loss of hierarchical links.
- Manual upload can be parsed normally but the API call returns no results. The cause is that the
PARSE_FILE_TIMEOUT_SECONDSparameter is not configured correctly, and the API request times out without waiting for parsing to complete.
How to Confirm Proper Configuration
- Upload the largest-volume single case PDF file and check if a 413 error is triggered, confirming that the
UPLOAD_FILE_MAX_SIZEconfiguration meets the actual file size requirements. - Upload a Feishu document with a multi-level directory and check if the parsed result includes all levels of approval node content, confirming that the chunking rules adapt to the document structure.
- Call the API to upload and parse a large-volume document and check if the returned result includes complete parsed content, confirming that the
PARSE_FILE_TIMEOUT_SECONDSconfiguration sufficiently covers the parsing duration. - Upload a Feishu document with embedded documents and images and check if the parsed result includes image text and doc content, confirming that the
customPdfParseconfiguration is enabled correctly.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.