What this category’s data looks like
Precast concrete component daily documents include component detail drawings, steel bar cutting lists, mold inspection records, pouring and vibration records, curing records, and demolding factory inspection reports. Data sources include BIM or detailed design files, steel bar and concrete batch ledgers, mold table and curing kiln operation data, laboratory test records, and QR code traceability systems. Data update frequency is per single component, per batch, or per kiln. Each document contains fields such as component number, dimensional parameters, steel bar model, embedded part position, strength grade, and test values. Common units include millimeters, cubic meters, degrees Celsius, and megapascals.
Constraints on document parsing and chunking
The multi-type documents, high-frequency updates, and precise field requirements for precast concrete components impose multiple constraints on document parsing and chunking. First, document formats vary widely across sources. These include CAD-to-PDF detail drawings exported from BIM, cutting lists in Excel format, and scanned paper test records. Multiple parsing logics must be supported. Second, the per-component or per-batch update feature requires chunking to precisely match the minimum unit of business queries. This avoids splitting associated information such as steel bar models and embedded part positions. Third, errors directly affect on-site installation and component scrapping. Parsed data fields must be complete and free of misalignment. Chunking must avoid mixing content from different components.
Configuration settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Precast concrete component documents contain associated parameters such as precise dimensions and embedded part positions. Chunking must cover a complete single business group to avoid splitting critical data |
chunkSplitter | By paragraph + semantic boundary | Adapts to the logical structure of detail drawings and cutting lists, avoids splitting associated content such as steel bar models and pouring parameters |
customPdfParse | Enabled | Adapts to CAD-to-PDF documents exported from BIM, retains layer annotations and component numbers and other key information, avoids loss of core fields during parsing |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Supports uploading large component detail drawing PDFs with multiple pages, adapts to the batch drawing needs of precast component factories |
PARSE_FILE_TIMEOUT_SECONDS | 600–900 seconds | Handles parsing of hundreds-page component detail drawings, avoids timeout errors caused by oversized files |
chunkSettingMode | Manual configuration | Adjust chunking parameters for different document types (such as record-type and drawing-type) to adapt to multi-source document scenarios |
Three common mistakes
- Symptom: Timeout errors occur when calling
customPdfParseto parse large PDFs. Cause: ThePARSE_FILE_TIMEOUT_SECONDSparameter was not adjusted. The default timeout duration is insufficient to complete parsing of hundreds-page component detail drawings. - Symptom: Content from adjacent components appears mixed in chunking results. Cause:
chunkSplitterwas configured to split by fixed character count, not by semantic boundary. This breaks the association between embedded part positions and corresponding component numbers. - Symptom: The interface displays the
Cannot read properties of undefinederror after uploading a file. Cause: The locally deployedcustomPdfParseversion does not match FastGPT v4.17.0, and dependency package adaptation configuration was not completed.
How to verify correct configuration
- Upload a PDF of a precast concrete component detail drawing with more than 100 pages. Check whether parsing progress completes within the duration set by
PARSE_FILE_TIMEOUT_SECONDS. - Randomly select one chunked content entry. Verify that it contains complete associated information such as steel bar models and embedded part positions, and is not split by character boundaries.
- After enabling
customPdfParse, parse a CAD-to-PDF document exported from BIM. Confirm that layer annotations, component numbers and other fields are not lost. - Adjust
chunkSizeto different ranges. Compare the semantic completeness of chunking results, and select a value that meets business requirements.
The provided values are common starting points, and should be evaluated against local samples.