API Overview

The Open Reviewer API is organized into four main areas:

Core Components

Module	Description
Evaluators	Core evaluation classes for single and multi-model review
Data Models	Pydantic models for test cases, results, and configuration
Semantic Analysis	AST parsing, repository mapping, and code search
Embeddings	Vector storage and embedding generation

Quick Import

from review_eval import (
    # Core evaluators
    ReviewEvaluator,
    MultiModelEvaluator,
    DocsAwareEvaluator,
    SemanticEvaluator,
    create_semantic_evaluator,

    # Data models
    GoldenTestCase,
    ModelConfig,
    ModelReviewResult,
    MultiModelResult,
    ReviewResult,

    # Model configurations
    DEFAULT_MODELS,
    BENCHMARK_MODELS,

    # Utilities
    print_multi_model_report,
)

Package Structure

review_eval/
├── __init__.py              # Public API exports
├── evaluator.py             # ReviewEvaluator
├── multi_model_evaluator.py # MultiModelEvaluator
├── docs_aware_evaluator.py  # DocsAwareEvaluator
├── semantic_evaluator.py    # SemanticEvaluator
├── models.py                # Pydantic data models
├── docs_loader.py           # Documentation discovery
├── index_repo.py            # CLI: index repository
├── update_embeddings.py     # CLI: update embeddings
└── semantic/
    ├── ast_parser.py        # Python AST parsing
    ├── repo_map.py          # Repository mapping
    ├── search.py            # Semantic search
    ├── models.py            # Semantic data models
    └── embeddings/
        ├── chunker.py       # Code chunking
        ├── client.py        # Embedding client
        └── vector_store.py  # Qdrant integration