This document is written for FastGPT v4.17.0.
What the data for this category looks like
Core data for this category comes from raw material quality certificates, furnace heating system data, press collection data, MES systems, non-destructive testing instruments, and customer forging part drawings. Daily documents include forging process specifications, cutting material cards, furnace heating records, forging process cards, mold life records, heat treatment work orders, and ultrasonic testing reports. Data updates follow the rhythm of each forged part, each batch of materials, or each furnace run. Most documents contain structured fields such as furnace batch number, forging part batch number, heating temperature, pressure value, and testing defect type. Some documents include standardized tables and process step descriptions.
Constraints imposed on document parsing and chunking
The data characteristics of this category impose multiple constraints on document parsing and chunking. First, data is associated by batch and updated frequently. Chunking must retain key associated fields such as furnace batch number and forging part batch number to avoid loss of batch information due to chunk breaks. Second, documents contain a large number of structured tables and mixed text. Parsing must support complete table extraction to avoid failure of process parameter association caused by table splitting. Third, document formats from multiple sources are inconsistent. Some testing reports contain non-standard encoded content, so custom parsing rules must be adapted to accurately extract core fields. Fourth, the collection of documents associated with a single batch has a large volume. Reasonable file upload and parsing timeout thresholds must be adapted.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Forging documents contain process parameters, testing data, and text descriptions. This length supports complete extraction of mixed content and prevents key fields from being split |
chunkSplitter | Split by paragraph + table | Forging documents contain a large number of process tables and step descriptions. This rule preserves table integrity and paragraph logic |
customPdfParse | Enabled | Most forging documents are standardized process forms and testing reports. Custom parsing accurately extracts associated fields such as furnace batch number and forging part batch number |
PARSE_FILE_TIMEOUT_SECONDS | 300–600 seconds | Parsing large ultrasonic testing reports and multi-page process specifications takes a long time. This range covers parsing needs for most documents |
UPLOAD_FILE_MAX_SIZE | 500 MB | The collection of documents associated with a single batch of forgings may include multiple reports and specifications. This threshold meets batch upload requirements |
Three common misconfigurations
- Parsing timeout errors occur when uploading multi-page ultrasonic testing PDF files. The cause is failure to adjust the
PARSE_FILE_TIMEOUT_SECONDSparameter. The default timeout duration is insufficient to complete full parsing of large testing reports. - Encoding error prompts appear when uploading PPT-format forging process documents. The cause is that the document contains non-standard encoded annotations or notes, and the custom parsing encoding adaptation logic is not enabled.
- Forging part furnace batch number association information is lost after chunking. The cause is that
customPdfParseis not enabled. The default parsing logic does not retain key associated fields in structured forms, leading to a disconnect between chunked content and batch information.
How to confirm configurations are correctly set
- Upload a single forging process specification PDF, and check if the parsed text completely retains key information such as process parameters, step descriptions, and furnace batch number.
- Upload a heating furnace record XLSX containing multiple rows of tables, and check if the chunking result retains complete table rows and columns, with no parameter breaks caused by table splitting.
- View parsing logs to confirm that the
customPdfParseconfiguration is active, and that there are no encoding error or field missing prompts. - Upload a collection of documents associated with a single batch, and confirm that the upload proceeds normally without triggering file size limit prompts.
The values provided here are common starting points and should be evaluated against the reader's own samples.