Crate depyler_oracle 

ML-powered compile error classification and auto-fixing.

Uses aprender models (Decision Tree, N-gram, Random Forest) to:

• Classify transpilation errors into actionable categories
• Suggest fixes based on historical patterns
• Detect error drift requiring model retraining

Re-exports

pub use autofixer::AutoFixer;

pub use autofixer::FixContext;

pub use autofixer::FixResult;

pub use autofixer::TransformRule;

pub use automl_tuning::automl_full;

pub use automl_tuning::automl_optimize;

pub use automl_tuning::automl_quick;

pub use automl_tuning::AutoMLConfig;

pub use automl_tuning::AutoMLResult;

pub use citl_fixer::CITLFixer;

pub use citl_fixer::CITLFixerConfig;

pub use citl_fixer::IterativeFixResult;

pub use estimator::message_to_features;

pub use estimator::samples_to_features;

pub use estimator::OracleEstimator;

pub use graph_corpus::analyze_graph_corpus;

pub use graph_corpus::build_graph_corpus;

pub use graph_corpus::convert_to_training_samples;

pub use graph_corpus::load_vectorized_failures;

pub use graph_corpus::GraphCorpusStats;

pub use graph_corpus::VectorizedFailure;

pub use params_persistence::default_params_path;

pub use params_persistence::load_params;

pub use params_persistence::params_exist;

pub use params_persistence::save_params;

pub use params_persistence::OptimizedParams;

pub use synthetic::generate_synthetic_corpus;

pub use synthetic::generate_synthetic_corpus_sized;

pub use synthetic::SyntheticConfig;

pub use synthetic::SyntheticGenerator;

pub use tuning::find_best_config;

pub use tuning::quick_tune;

pub use tuning::TuningConfig;

pub use tuning::TuningResult;

pub use classifier::ErrorCategory;

pub use classifier::ErrorClassifier;

pub use features::ErrorFeatures;

pub use hansei::CategorySummary;

pub use hansei::HanseiConfig;

pub use hansei::HanseiReport;

pub use hansei::IssueSeverity;

pub use hansei::TranspileHanseiAnalyzer;

pub use hansei::TranspileIssue;

pub use hansei::TranspileOutcome;

pub use hansei::Trend;

pub use hybrid_retrieval::reciprocal_rank_fusion;

pub use hybrid_retrieval::Bm25Scorer;

pub use hybrid_retrieval::HybridRetriever;

pub use hybrid_retrieval::RrfResult;

pub use ngram::FixPattern;

pub use ngram::FixSuggestion;

pub use ngram::NgramFixPredictor;

pub use patterns::CodeTransform;

pub use patterns::FixTemplate;

pub use patterns::FixTemplateRegistry;

pub use tfidf::CombinedFeatureExtractor;

pub use tfidf::TfidfConfig;

pub use tfidf::TfidfFeatureExtractor;

pub use training::TrainingDataset;

pub use training::TrainingSample;

pub use depyler_training::classify_with_moe;

pub use depyler_training::load_real_corpus;

pub use depyler_training::train_moe_on_real_corpus;

pub use depyler_training::train_moe_oracle;

pub use moe_oracle::ExpertDomain;

pub use moe_oracle::MoeClassificationResult;

pub use moe_oracle::MoeOracle;

pub use moe_oracle::MoeOracleConfig;

pub use github_corpus::analyze_corpus;

pub use github_corpus::build_github_corpus;

pub use github_corpus::convert_oip_to_depyler;

pub use github_corpus::get_moe_samples_from_oip;

pub use github_corpus::load_oip_training_data;

pub use github_corpus::CorpusStats;

pub use github_corpus::OipDefectCategory;

pub use github_corpus::OipTrainingDataset;

pub use github_corpus::OipTrainingExample;

pub use unified_training::build_default_unified_corpus;

pub use unified_training::build_unified_corpus;

pub use unified_training::build_unified_corpus_with_oip;

pub use unified_training::print_merge_stats;

pub use unified_training::MergeStats;

pub use unified_training::UnifiedTrainingConfig;

pub use unified_training::UnifiedTrainingResult;

pub use curriculum::classify_error_difficulty;

pub use curriculum::classify_from_category;

pub use curriculum::CurriculumEntry;

pub use curriculum::CurriculumScheduler;

pub use curriculum::CurriculumStats;

pub use curriculum::DifficultyLevel;

pub use ast_embeddings::AstEmbedder;

pub use ast_embeddings::AstEmbedding;

pub use ast_embeddings::AstEmbeddingConfig;

pub use ast_embeddings::CombinedEmbeddingExtractor;

pub use ast_embeddings::CombinedFeatures;

pub use ast_embeddings::PathContext;

Modules

ast_embeddings

AST Embeddings for Code2Vec-style code representation (Issue #210).

autofixer

Automatic code transformation based on oracle classifications.

automl_tuning

AutoML-powered hyperparameter tuning using aprender.

citl_fixer

CITL (Compiler-in-the-Loop Learning) integration for Depyler.

classifier

Error classification types and logic.

corpus_extract

Corpus extraction for oracle training data

curriculum

Curriculum Learning for Error Processing (Strategy #3 - DEPYLER-0633)

depyler_training

Depyler-specific training data for error classification.

estimator

Aprender Estimator trait implementation for oracle predictor.

features

Feature extraction from error messages.

github_corpus

GitHub history corpus integration via organizational-intelligence-plugin.

graph_corpus

DEPYLER-1303: Graph-aware corpus integration for Oracle training.

hansei

Hansei (反省) Post-Transpilation Analysis

hybrid_retrieval

Hybrid Retrieval with BM25 + TF-IDF Reciprocal Rank Fusion

moe_oracle

MoE (Mixture of Experts) Oracle for Error Classification (DEPYLER-0580)

ngram

N-gram based fix pattern predictor.

params_persistence

Parameter persistence for optimized generation parameters.

patterns

Fix pattern templates with code transformations.

self_supervised

Self-Supervised Corpus Generation for Oracle Training.

synthetic

Synthetic data generation for scaling training corpus.

tfidf

TF-IDF based feature extraction for error classification.

training

Training data for the error classifier.

tuning

Hyperparameter tuning for oracle predictor.

unified_training

Unified training pipeline for merging all data sources deterministically.

utol

Unified Training Oracle Loop (UTOL)

verificar_integration

Integration with verificar for synthetic training data generation.

Structs

ClassificationResult

Classification result with confidence and suggested fix.

Oracle

OracleConfig

Oracle for compile error prediction and fixing.

RetrainConfig

Configuration for retrain trigger (mirrors aprender::online::orchestrator::RetrainConfig).

RetrainStats

Statistics from the retrain trigger (mirrors aprender::online::orchestrator::OrchestratorStats).

RetrainTrigger

Retrain trigger for Oracle (adapted from aprender::online::orchestrator::RetrainOrchestrator).

Enums

ObserveResult

Result of observing a prediction (mirrors aprender::online::orchestrator::ObserveResult).

OracleError

Error types for the oracle.

Functions

get_training_corpus_paths

Get training corpus file paths for hash computation.

print_drift_status

Print drift status to stdout with visual indicators.

print_retrain_status

Print retrain trigger status with Andon-style alerts.

Type Aliases

Result

Result type for oracle operations.