What data for this category looks like
Tender-related documents for furniture and wood product manufacturing include cutting drawings, cutting lists, edge banding and drilling programs, coating process documents, hardware inventory lists, assembly and packaging records, and more. Data sources include order design systems, sheet and timber batch records, CNC machining equipment, ERP/MES systems, and moisture content testing equipment. Data update frequency varies by order, daily, or per material batch. A single cutting list may include size specifications, processing procedures, and corresponding sheet batch information for dozens to hundreds of board components.
How these characteristics impose constraints on document parsing and chunking
Documents in this category have diverse types. They include both plain text lists and process descriptions, as well as drawing documents embedded with CAD text. This requires parsing tools to support multi-format content. Frequently updated order data requires parsing processes to have high efficiency, to avoid delaying tender preparation cycles. The large number of board components and dense associated fields require that chunking must retain complete associated content such as size, batch, and hardware information for the same board component, to avoid splitting that breaks the ability to restore business logic.
How to set configurations
These configurations are formulated based on the rules of FastGPT v4.17.0.
| Configuration Item | Recommended Value Range | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the length of associated information for a single board component in a cutting list, avoiding excessively long single chunks that reduce retrieval accuracy, or excessively short chunks that break business associations |
chunkSplitter | Custom delimiters (line breaks, tabs, "board part number" keyword) | Documents in this category mostly follow fixed field separation rules. Splitting by business keywords avoids splitting that separates board components from their corresponding batch and size information |
customPdfParse | Enabled | For cutting drawing PDFs embedded with CAD text, enabling this preserves structured metadata from the original document, avoiding loss of field information during plain text parsing |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Covers parsing duration for large cutting drawing PDFs under 10MB, avoiding timeout errors |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Supports batch uploading multiple process documents and cutting drawings, adapting to enterprise needs for batch preparation of tender materials |
chunkSettingMode | Manual mode | Allows separate adjustment of chunking rules for different document types such as hardware inventory lists and cutting drawings, adapting to differentiated parsing needs for multi-format documents in this category |
Three common mistakes
- When uploading a cutting list PDF larger than 10MB, the error
timeout of 360000ms exceededappears. The cause is failing to adjust thePARSE_FILE_TIMEOUT_SECONDSparameter, using the default shorter duration that cannot cover the parsing cycle of large documents. - Board component sizes are separated from their corresponding sheet batches in chunking results. The cause is failing to configure
chunkSplitterto split by business keywords, only using the default line break splitting, which breaks the field association logic within the document. - Custom parsing scripts do not take effect. The cause is failing to enable
customPdfParse, causing the written JavaScript parsing logic to not be called by the system, making it unable to process PDF documents embedded with CAD text.
How to confirm configurations are set correctly
- Upload a cutting list PDF of around 10MB, check the completion status of the parsing task, confirm that the parsing duration does not exceed the duration set by
PARSE_FILE_TIMEOUT_SECONDS. - Export the chunking results, check whether a single chunk contains complete board part number, size, and corresponding sheet batch information, confirm that the
chunkSizeandchunkSplitterconfigurations are effective. - After enabling
customPdfParse, upload a cutting drawing PDF embedded with CAD text, check whether the parsing result retains field information such as hardware model and installation position from the original document. - Batch upload multiple different types of tender documents, confirm that the upload progress proceeds normally, and that uploads are not blocked due to total file size exceeding the
UPLOAD_FILE_MAX_SIZElimit.
The values provided here are common starting points, and should be evaluated against the reader's own samples.