Expand description
Shared API for the MinerU VLM client.
Re-exports§
pub use command::RunContext;pub use command::RunError;pub use command::RunOptions;pub use command::RunReport;pub use command::run;pub use command::run_with_context;
Modules§
- command
- Shared high-level command execution.
- models
- official
- Additive SDK for the pinned official Python MinerU parser.
- official_
api - Strict wire DTOs for the additive official-v1 HTTP API.
Structs§
- Asset
- Block
Kind - Content
Block - Document
- Error
Context - Limits
- MinerU
VlmClient - MinerU
VlmConfig - MinerU
VlmPreprocessor - Model
Block - Model
Info - Normalized
Bbox - Official
Output Manifest - Official
PdfOptions - Output
Manifest - Page
Range - Page
Result - Parse
Options - Sampling
Params - VlmCompletion
- VlmEncoded
Image - VlmExtract
Result - VlmHeader
- VlmHttp
Client - VlmHttp
Config - VlmLayout
Block - VlmPrepared
Extraction - VlmPrepared
Layout - VlmRequest
- VlmSse
Stream
Enums§
- Asset
Kind - Concurrency
Model - Request-level concurrency model for the official route.
Classicis the long-standing single-encoder pipeline and the default;TwoPhasesplits each page’s semantic work into an encode-all stage and a request-all stage, removing the CPU-encode serialization in front of request dispatch.MINERU_OFFICIAL_CONCURRENCY_MODEL=two-phaseopts in. - Error
- PdfInput
- Rotation
- VlmBatch
Priority - VlmError
- VlmImage
Input
Functions§
- canonical_
stem - Canonical portable output directory name shared by route preflight and publication.
- markdown_
output_ path - Locates the markdown output published under
directoryforstem. - write_
outputs