What data for this category looks like
Daily documents for this category include territorial spatial overall plans, detailed plans, special plan texts, planning implementation evaluation reports, and public opinion summary materials. Data sources include the territorial spatial basic information platform, natural resource survey and monitoring data, population and economic statistical results, and remote sensing images. The update frequency is annual. Document structures mix structured statistical tables, spatial vector layer descriptions, unstructured text discussions, and public messages. Fields include planning scope, constraint indicators, implementation progress, and others. Units include square kilometers, person-times, and others.
What constraints these characteristics impose on the document parsing and chunking link
Multi-source mixed data structures require the parsing process to handle structured tables, spatial layer descriptions, and unstructured text simultaneously. This increases parsing complexity. Annual batch-updated documents have large volume. Reasonable single-file size and timeout thresholds must be adapted. Scattered public opinion summary messages require retaining contextual associations. Semantic integrity must not be broken during chunking. Scanned planning drawings and remote sensing image attachments require OCR recognition. Recognition errors are likely to occur if high-precision parsing is not enabled. Spatial constraint indicators in planning documents must accurately correspond to their respective chapters. Chunking logic must match the chapter hierarchy of the document.
Configuration settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the chapter paragraph length of planning documents, avoids splitting key indicator descriptions |
chunkSplitter | Split by title hierarchy | Matches the chapter structure of planning documents, retains contextual relevance |
customPdfParse | Enable high-precision OCR mode | Adapts to scanned planning drawings and remote sensing image attachments, reduces the probability of recognition errors |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Addresses the parsing time of single large planning reports, avoids timeout errors in private deployment scenarios |
chunkSettingMode | Manual configuration | Supports adjusting chunking rules for different document types, adapts to the differing needs of planning texts and public opinions |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapts to large annual batch-updated planning documents, meets single-file upload requirements |
Three common errors
- Background parsing function entry not displayed: After a PDF document is uploaded, no parsing fragments are generated. Background logs prompt that the corresponding parsing service cannot be found. The cause is that the
customPdfParseparameter is not enabled in the system configuration, or the parsing service connection address is not correctly configured during private deployment. - OCR Error error returned: After a planning document containing scanned drawings is uploaded, the parsing result has garbled characters or missing layer descriptions. The system returns the OCR Error status code. The cause is that high-precision OCR mode is not enabled, or the image recognition dependencies of the parsing service are not loaded normally.
- Large document parsing timeout: After a single planning report larger than 500MB is uploaded, the system returns a timeout error. The parsing service log shows that the parsing was successful. The cause is that the
PARSE_FILE_TIMEOUT_SECONDSconfiguration value is too low, and does not cover the parsing time of a single large document.
How to confirm the configuration is correct
- The system configuration page is accessed, and the
customPdfParseparameter is checked for enabled status. The high-precision OCR function is confirmed to be activated. - A small planning document containing scanned drawings is uploaded. The parsing result is checked for obvious garbled characters, verifying that the OCR configuration takes effect.
- A planning document of approximately 100MB is uploaded. After parsing completes, the system log is checked to confirm that the parsing time does not exceed the
PARSE_FILE_TIMEOUT_SECONDSconfiguration value. - The corresponding knowledge base management page is accessed. Chunked document fragments are checked to confirm that the fragment length falls within the
chunkSizeconfiguration range, and that the chapter hierarchy is not damaged.
The values provided are common starting points and should be evaluated against the samples of the deploying organization.
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.