What the data for this category looks like
Data sources for gas due diligence include internal operation systems of local gas operating enterprises, public utility supervision databases of housing and urban-rural development departments, and gas supply settlement vouchers from upstream gas suppliers. Updates follow a fixed schedule: core operation data is updated monthly, safety inspection data is archived quarterly, and annual revenue and user scale data is aggregated. Document structures include four types of attachments: detailed gas supply ledgers, user tiered files, safety inspection records, and gas source settlement vouchers. Fields include gas supply pressure (kPa), current total gas supply (10,000 cubic meters), number of registered users (households), number of households with completed inspections (households), settled revenue amount (10,000 yuan), and others. Units follow measurement standards for the corresponding business scenarios.
Constraints imposed on the tool calling and plugins workflow
Multi-source data docking requires plugins to support cross-system API calls. This includes internal operation interfaces of gas enterprises, housing and urban-rural development supervision platform interfaces, and gas settlement voucher parsing interfaces. The monthly update schedule requires tool calling scheduled triggers to be set to monthly intervals. It also requires adding data timeliness verification logic to filter historical data older than 90 days. Special measurement fields require plugins to include built-in unit verification rules. This prevents extraction errors caused by mixed unit usage. The fixed document structure requires plugins to preset field mapping rules. Structured content in attachments is mapped directly to corresponding fields in due diligence reports, without needing general text extraction.
Configuration settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
PARSE_FILE_TIMEOUT_SECONDS | 900 seconds | Gas due diligence report attachments include multi-page gas supply ledgers and inspection records. Single files have large size, requiring extended parsing timeout |
UPLOAD_FILE_MAX_SIZE | 2000 MB | Some gas supply ledgers contain multiple years of historical data. Single files may exceed general thresholds, requiring increased maximum upload size |
tool_call_trigger_mode | auto_select | Gas due diligence requires calling multiple tool types. Configure model automatic tool selection to complete tasks |
field_extract_template | Preset gas due diligence field mapping | For special measurement fields and document structures of gas data, preset dedicated extraction templates |
webhook_request_timeout | 60 seconds | When connecting to gas enterprise APIs, adapt to their interface response speed and set a reasonable timeout |
rag_top_k | Top 8 entries | Due diligence reports need to cover multi-dimensional data such as gas supply, users, and safety. This ensures sufficient associated document recall |
The parameter values provided on this page are common recommended starting points for configuration. Actual values are affected by material form, data volume, and business rules. Specific issues require case-by-case analysis, and it is recommended to test on your own samples before finalizing settings.
Three common errors
- After parsing uploaded gas due diligence Word drafts, revision marks and custom layout formatting are lost. Cause: A dedicated document parsing plugin that supports revision content extraction is not enabled, and only general text extraction tools are called.
- When using DingTalk webhook to push due diligence reports, the target platform cannot open attachments. Cause: The attachment transmission format is not adjusted to the compatible PDF format, and the original Word file is pushed directly.
- Tool calling tasks time out, returning
504 Gateway Timeout. Cause: Thewebhook_request_timeoutparameter is not adjusted, and the default timeout period is too short to adapt to the response speed of gas enterprise interfaces.
How to confirm configuration is complete
- Upload a gas due diligence Word draft containing revision marks. Verify whether the parsing result retains revision content and original layout, to confirm the document parsing plugin configuration.
- Call the test interface of the gas enterprise API. Check whether the fields of the returned data conform to the preset gas-specific mapping rules, to confirm the parameter configuration.
- Trigger the tool calling process. Check whether the model automatically selects the corresponding tool, and whether due diligence content is directly generated, to confirm the trigger mode configuration.
- Test the DingTalk webhook push function. Confirm that the pushed attachments can be opened normally on the target platform, and adjust the transmission format parameters.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.