Module stages 

Shared stage processors for the OCR pipeline.

This module provides reusable stage processors that eliminate code duplication across different pipeline flows (single-image, in-memory, dynamic batching, detection-from-memory). Each processor encapsulates the logic for a specific pipeline stage with consistent error handling, logging, and performance characteristics.

Extensible Pipeline Architecture

The pipeline supports extensible stages through the PipelineStage trait and StageRegistry system, allowing new stages to be added without modifying core pipeline code.

Stage Processor Pattern Abstraction

This module includes helper utilities that reduce code duplication across stage processors by abstracting common lifecycle patterns:

• Timing management - Automatic start/stop timing for operations
• Empty input handling - Consistent behavior for empty input collections
• Parallel processing - Automatic decisions between sequential and parallel processing
• Metrics collection - Standardized success/failure counting and metadata
• Result wrapping - Consistent StageResult creation with metrics

These abstractions reduce repetitive code by approximately 25-35% across the orientation, cropping, and recognition processors while maintaining the same functionality and API compatibility.

See [processor_helper] for the concrete helper implementations.

Re-exports

pub use config::ExtensiblePipelineConfig;

pub use config::GlobalPipelineSettings;

Modules

config

Configuration system for extensible pipeline stages.

Structs

BatchConfig

Configuration for batch processing behavior

BatchProcessor

Helper for processing collections with common lifecycle patterns

CroppingConfig

Configuration for cropping processing

CroppingResult

Result of cropping processing

CroppingStageProcessor

Processor for text box cropping stage.

ExtensibleCroppingStage

Extensible cropping stage that implements PipelineStage trait.

ExtensibleOrientationStage

Extensible orientation stage that implements PipelineStage trait.

ExtensiblePipeline

Extensible pipeline that uses the stage registry.

ExtensibleRecognitionStage

Extensible recognition stage that implements PipelineStage trait.

ExtensibleTextDetectionStage

Extensible text detection stage that implements PipelineStage trait.

ExtensibleTextLineOrientationStage

Text line orientation stage for the extensible pipeline.

GroupingStrategyFactory

Factory for creating grouping strategies

OrientationConfig

Configuration for orientation processing

OrientationCorrectionConfig

Configuration for orientation correction

OrientationCorrector

Component responsible for applying text line orientation corrections

OrientationResult

Result of orientation classification processing

OrientationStageProcessor

Processor for document orientation classification stage.

PipelineExecutor

Pipeline executor for running extensible pipelines.

RecognitionConfig

Configuration for recognition processing

RecognitionResult

Result of recognition processing

RecognitionStageProcessor

Processor for text recognition grouping and processing stage.

SingleItemProcessor

Helper for single item processing with timing and metrics

StageContext

Context information available to pipeline stages during execution.

StageData

Data that flows between pipeline stages.

StageId

Unique identifier for a pipeline stage.

StageMetrics

Metrics collected during stage processing.

StageProcessorHelper

Helper utility for common stage processing patterns.

StageResult

Result wrapper for stage processing operations.

TextDetectionConfig

Configuration for text detection processing.

TextDetectionResult

Result of text detection processing.

TextLineOrientationConfig

Configuration for text line orientation processing.

Enums

GroupingStrategyConfig

Configuration for different grouping strategies

StageDependency

Dependency specification for a pipeline stage.

Traits

GroupingStrategy

Trait for image grouping strategies

PipelineStage

Trait for extensible pipeline stages.

StageAlgorithm

Trait for stage algorithms that can be run with automatic timing and metrics.

StageProcessor

Trait for stage processors that follow the common lifecycle pattern.

Functions

process_items

Generic utility function for processing collections with common patterns.

run_with_metrics

Generic stage harness that runs an algorithm with automatic timing and metrics.

run_with_metrics_and_fallback

Generic stage harness with custom error handling.