This document is written based on FastGPT v4.17.0.
What this category of data looks like
Daily documents for this category include performance test reports, device compatibility lists, optimization records, and crash analysis reports. Data sources include engine Profiler, crash platforms, real-device testing tools, and frame rate collection systems.
Documents are typically categorized by device, test time, and performance metrics. They include fields such as device identifiers, frame rate values, resource usage metrics, and crash timestamps. Update frequency is daily or per version iteration. Single document size may increase as the number of covered test devices rises.
What constraints do these characteristics impose on the document parsing and chunking link
Diverse document types require the parsing link to support multiple formats, to avoid structured test data being split into meaningless fragments. High-frequency update characteristics require parsing timeout settings to adapt to large, time-consuming documents.
Document structures categorized by device and test unit require chunking logic to retain field associations and grouping rules. Single test data groups must not be arbitrarily split. Some documents include embedded performance charts and tables, requiring the parsing link to retain original document layout and field structure, to avoid loss of key performance metrics.
Improper chunking may cause subsequent optimization links to miss critical data, leading to game client freezes, crashes, or store negative reviews.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapt to the single-file size limit of real-device test logs and Profiler export files commonly included in game client performance optimization documents, to avoid large file upload failures |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Game performance documents often contain large amounts of real-device test data, with long parsing times. Set a reasonable timeout to avoid mid-process interruptions |
chunkSize | 800–1200 characters | Adapt to the length of a single test data set plus analysis in performance test reports, retain context while avoiding overloaded single chunks |
chunkSplitter | Group by device + newline character | Game performance documents often organize test results by device. Group by device then split with newline characters to retain test unit integrity |
customPdfParse | Enabled | Most performance test reports are exported in PDF format. Enabling custom PDF parsing retains table and field structure from original documents |
chunkSettingMode | Manual configuration | Adjust chunking rules for different types of performance documents (such as crash analysis reports, frame rate test reports) to adapt to specific scenario requirements |
Three common misconfigurations
- Uploading a 3 MB performance test PDF document returns a parsing failure prompt. The cause is failure to adjust the
UPLOAD_FILE_MAX_SIZEconfiguration. The default value is insufficient to accommodate large performance documents. - Setting a custom delimiter results in unexpected chunk lengths, either merging multiple test paragraphs or splitting single test data groups. The cause is failure to adjust delimiter matching rules in conjunction with the
chunkSizeconfiguration, and failure to set custom delimiters according to the document's device grouping logic. - Deploying Marker and processing performance document PDFs results in errors. The symptom is timeout or format errors during parsing. The cause is incorrect configuration of
PARSE_FILE_TIMEOUT_SECONDS, or failure to adapt to parsing dependencies for embedded performance charts in PDFs. - Attempting to import an Excel-format device compatibility list results in empty parsing results or disorganized fields. The cause is failure to configure parsing rules adapted to structured table files. The default parsing logic only supports text and PDF formats.
How to confirm the configuration is properly set
- Upload a single performance test PDF document with a size not exceeding 1000 MB, check if upload and parsing status is normal.
- Upload a document containing test data for multiple devices, verify if chunked results are grouped by device, and single chunk lengths meet expectations.
- Adjust
chunkSizeand custom delimiters, manually verify that chunks retain complete context for single test data groups. - Check system logs to confirm no timeout errors occur during parsing, and the
PARSE_FILE_TIMEOUT_SECONDSconfiguration is effective.
The values provided are common starting points, and should be evaluated against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-10-19.