Re-exports§
pub use jre_manager::JreManager;
Modules§
- analytics
- cache
- classifier
- config
- graphs
- jre_
manager - JRE Manager - Auto-download and manage Java Runtime for JNI backend
- preprocessors
- Document Preprocessors
- processor
- rules
- storage
- types
Structs§
- Bookmark
Data - Bookmark
Section - Bounding
Box - Cache
Defaults - Which cache points are enabled for writing.
- Classification
Result - Depth
Distribution - Document
Analysis - Composite output of the analytics pre-pass. Carries one finalized output per stat kind. Lives in pipeline memory; not serialized into the public graph output (a separate sidecar dump may serialize it for development purposes).
- Document
Graph - Document
Info - Aggregated document-level information computed during parsing. This is NOT a node in the tree — it is information about the document. Has proto-L1 character: one per document, invariant to tree structure. See 006-document-info-separation.md for design rationale.
- Document
Metadata - Document
Node - Document
Processor - Element
Group - Flat
Document - Font
Class - Graph
Analytics - Analytics computer that can analyze any subset of nodes in the graph
- Graph
Analytics Result - Result of analytics computation for any subset of nodes
- Histogram
Bin - List
Sequence - Node
Content - Node
IdGenerator - Generates deterministic node IDs scoped to a specific document parse.
- Node
Location - Node
Type Distribution - Parsed
PdfElement - Parsing
Config - PdfPreprocessor
- PDF Preprocessor with pluggable backend
- PdfText
Element - Physical
Location - Pipeline
Stages - Captured intermediate outputs from each pipeline stage Used for testing and diagnostics — lets you inspect/compare each boundary
- Placement
- Spatial and structural metadata about where a text element lives in its source.
Populated for PDF-sourced elements. Future HTML/Markdown preprocessors may produce
elements with
placement: None. - Preprocessor
Output - Complete output from document preprocessing
- Semantic
Location - Sequential
Document - Sequential
Segment - Sorted
Document Graph - The serialization-ready output format. Carries a schema version so consumers can detect and handle shape changes.
- Structural
Profile - Quantitative measurement of graph shape — deterministic, mechanically computed from structure. Travels with graph.json. Describes the L0 tree’s statistical properties. See AmplifyNotes/09-Profile-Types.md for design rationale.
- Style
Data - Style
Metadata - Tika
JniBackend - JNI-based Tika backend for PDF processing
- Token
Distribution - Histogram-based token distribution for comprehensive statistical analysis
- Token
Histogram - Histogram representation enabling statistical calculations (mean, median, mode, variance)
Enums§
- Cache
Point - A discrete cache point in the processing pipeline. Ordered by pipeline position: C0 < C1 < C2 < C3.
- Document
Type - Flow
Type - Signals whether physical location data is meaningful for this document
- Fresh
From - Controls which cache points to bypass during processing. Cascade: fresh-from C1 means skip C1, C2, C3 caches.
- Group
Type - Node
Type - Parsed
Element Type
Constants§
- SCHEMA_
VERSION - The schema version stamped on every graph output. Bump this when the output shape changes.
Traits§
- Preprocessor
- Preprocessor trait - converts documents to PreprocessorOutput
Functions§
- infer_
title - Infer a best-guess document title from parsed elements. Used as a fallback when Tika metadata doesn’t provide a title. Current strategy: first Section element’s text. Future candidates: largest font on page 1, first bold text, etc.