What the data for this category looks like
Daily documents for this category include packaging process cards, bonding and molding parameters, wafer incoming material drawings, test program versions, yield reports, and failure analysis reports. Data is sourced from wafer test maps, MES systems, bonding and molding equipment, ATE test machines, X-ray/SAM testing equipment, and customer specification requirements. Document update frequency adjusts based on production batches, single-chip test results, and daily operations. Some fields contain exclusive identifiers such as equipment numbers, test bin values, yield proportions, and failure types. The structure falls into three categories based on document type: parameter tables, process steps, and test results.
What constraints these characteristics impose on the document parsing and chunking link
Structural differences between document types require chunking logic to adapt to varying formats. Wafer incoming material drawings include bitmaps and parameter bindings, and cannot be split using general paragraph rules directly. Frequently updated single-batch documents require controlled chunk granularity to match real-time production data alignment needs. Documents with mixed multi-source data must retain field association relationships, to avoid losing binding information between equipment numbers and test results after splitting. Large-volume failure analysis reports require limiting single chunk length to prevent parsing timeouts. Daily updated operation and maintenance documents need fast parsing configurations to support real-time calls in production links.
How to configure parameters
This configuration is adapted for FastGPT v4.17.0. The specific parameter settings are listed below:
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the paragraph length of documents such as packaging process cards and yield reports, and preserves field association integrity |
chunkSplitter | Split by paragraph + heading | Matches the hierarchical structure of documents in this category, preventing loss of binding relationships between test program versions and corresponding parameters after splitting |
customPdfParse | Enabled | Adapts to documents with bitmap bindings such as wafer incoming material drawings, and retains original typesetting and field positions |
UPLOAD_FILE_MAX_SIZE | 500 MB | Covers requirements for single-batch large-volume failure analysis reports and batch uploads of multiple documents |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Accommodates parsing time for large-volume documents, avoiding timeout errors triggered by oversized files |
chunkSettingMode | Manual configuration | Adjusts chunking parameters for different document types to meet parsing needs for multi-source data |
Three common misconfigurations
- A 504 status code appears for file parsing timeouts. The cause is failure to adjust the
PARSE_FILE_TIMEOUT_SECONDSparameter, with parsing time for large-volume failure analysis reports exceeding the default threshold. - A JSON error is returned after online search parsing, with empty fields. The cause is failure to enable the
customPdfParseconfiguration, which cannot correctly parse wafer incoming material drawings with bitmap bindings, leading to failed field extraction. - Parsing of large-volume PDF documents continues waiting for results until timeout. The cause is failure to set reasonable
UPLOAD_FILE_MAX_SIZEandPARSE_FILE_TIMEOUT_SECONDSparameters, and lack of a configured queue mechanism, resulting in conflicts between multiple file parsing tasks.
How to confirm configurations are properly set
- Upload a single packaging process card document, check the paragraph integrity of the parsed chunks, and confirm the chunk granularity meets business requirements.
- Upload a wafer incoming material drawing with a bitmap, check if the parsed results retain original typesetting and field binding relationships, and verify the
customPdfParseconfiguration is active. - Upload a 500 MB-level failure analysis report, confirm parsing time does not exceed the preset threshold, and verify the rationality of the
PARSE_FILE_TIMEOUT_SECONDSconfiguration. - Upload multiple different types of documents simultaneously, check if parsing tasks execute in sequence, and verify the queue mechanism operates normally.
The values provided are common starting points and should be measured against the reader's own samples.