push-mkumqllmxxky #168

Merged
eric merged 50 commits from push-mkumqllmxxky into main 2026-10-28 18:32:04 +01:00
Owner
No description provided.
eric added 50 commits 2026-10-28 18:30:57 +01:00
This commit introduces the `pmouniverse::pump` module responsible for handling track embedding, metadata reading, and vector calculation. It includes significant changes to the data model, configuration handling for models and threads, and refactors the service startup logic to ensure models are present before operation.
This commit introduces significant structural changes to the service layer, including refactoring `UniverseService` and implementing asynchronous model fetching with resumable downloads and integrity checks. It also introduces deferred embedding logic for managing track availability.
Adds core mDNS functionality including daemon setup, unicast search logic, DNS wire parsing, and a new catalog structure for service resolution.
This commit introduces the proposal for a migration system to handle schema changes and ensure data survival. It defines the data model for `pmoident`, outlines the search cascade logic, specifies changes to the embedding model, and defines the API for MusicBrainz/AcoustID integration.
Introduces the core Store structure and a robust migration system featuring transactional steps, schema versioning, and a backup mechanism to ensure data integrity during schema changes.
Introduces a robust migration system featuring transactional steps, schema version validation, and specific error handling for migration failures and schema mismatches. This includes implementing checks for sequential migration steps, enforcing baseline rules, conditional backups, and expanding error types to handle schema versioning and space configuration issues safely.
This commit introduces a comprehensive system for schema evolution, including pre-migration backups and explicit space management. Changes include defining vector storage schemas, enforcing space mismatch constraints, updating the data model, and implementing persistence for space metadata.
This commit implements schema migrations, updates cache policies, and expands the public API. It introduces new configuration options for identification, refactors dependency management, and establishes new rules for handling high scores and data enrichment.
Refactor identification services to include the MusicBrainz client, introduce robust caching policies, enhance configuration validation, and improve error handling across the system.
Implement schema migration, refined caching, and service resilience patterns. This includes introducing a Gate mechanism for strict access control, enforcing rate limiting, and applying a Breaker pattern to manage service failures and cooldown periods. Configuration handling is also formalized to define explicit rules for cache behavior and data purging.
Refactor synchronized audio segment processing logic to ensure correct execution flow for fingerprinting.
Refactor the audio pipeline to introduce new decoding modules, implement robust file header reading, and refine the fingerprinting algorithm for accuracy.

This includes new API exposure for volume control, limited sample collection, and comprehensive testing for stereo/mono fingerprint consistency.
This commit updates the core identification pipeline, implementing a fingerprint lookup followed by a text fallback cascade. Changes include schema migration, refined identification constraints, new musicBrainz parsing features, and comprehensive unit tests for identification scenarios and error handling.
Introduces asynchronous identification workflows, refines album group selection rules, improves error handling for HTTP statuses, and sets up new test modules.
This update introduces a robust cascading identification strategy, refined matching criteria (including album and similarity checks), and improved resilience through service gating, error handling, and comprehensive test coverage across the identification and data models.
Introduces the full identification pipeline, including new data models, transactional identity replacement, and a command-line tool for cache processing.

This change refactors the project's core structure by introducing new modules for configuration, embedding, and identification. It implements the full indexing flow, allowing tracks to be fetched, metadata extracted, and identified. Key features include transactional identity replacement within database transactions, improved source handling for unreadable files, and a new CLI tool for running identification sweeps against the cache.
Implement new examples for calculating cosine distance matrices, nearest neighbors, random pairs, and recording groups.

Introduce auditing scripts for Shazam identification and text route verification, including robust error handling and statistical reporting.

Update project dependencies and configuration for core libraries.
Refactor the core identification pipeline to manage audio contradictions, introducing new `AudioVerdict` states and updating the API for candidate selection.

This change formalizes the decision rules, ensuring that audio data overrides labels when discrepancies occur, and enhances the reporting mechanism for contradictory results.
This commit introduces the core Shazam recognition service, including new configuration settings, data migration for the identification database, clip selection logic, and SHA-256 based caching mechanisms. It also expands the library's internal structure and public API for identification services.
Introduces the Pace service mechanism to manage call spacing, hourly limits, and failure resilience. This includes implementing the Gate structure for throttling, dynamic cooldown periods, and enforcing capacity constraints via configuration validation.
Restructures the identification process to incorporate Shazam and fingerprinting, introducing new `Standing`, `AudioVerdict`, and `Proof` states. Expands test coverage to validate all identification outcomes, including handling contradictory audio and verified results.
Introduces mandatory proof requirements for track identities, updates the schema for identity storage, and refines identification logic based on new outcome types.

This migration introduces a new `track_identity` schema, enforces proof requirements for identities entering the universe, and updates how track metadata is handled via new data models and identification outcomes.
This commit introduces enhancements to the identification process, including new methods for merging reports and updated logic for handling identification sweeps. It also refactors the audio cache API to allow for parameterized cache opening, and updates data migration constraints for identity tracking.
This change updates the Shazam identification system by implementing a new clip strategy that allows listening to continue unless a confirmation is received. It also introduces new API methods for Shazam identification, such as checking candidate status and retrieving ISRC information. Additionally, data model constraints are enforced, and the testing suite is expanded to validate complex recognition scenarios, error handling, and identification stopping conditions.
This commit introduces new utilities for text processing, including normalization, word splitting, and similarity calculation. It also adds features for identifying lead artists, generating album keys, and updating test assertions to ensure accurate song and artist matching.
Refactor identification logic to use Jaccard index and containment checks for title similarity. Introduce a persistent cache layer supporting schema migration, Gzip compression, and comprehensive testing for cache lookup states and data integrity.
Refactors the core Shazam recognition pipeline, introduces new data structures for time offsets, improves data integrity through model constraints, and adds new examples for signature comparison and neighbor analysis.
Refines the logic for mapping standing variants to proof names and introduces new test cases verifying identification scenarios involving fingerprint proof and service agreement.
Updates the Identification struct to include clip data and modifies the identification logic to format and display all recognized Shazam clips.
Introduces new judgment states, refined identification logic, and updates the data model to support affirmed results. Schema changes are implemented to track proof for identified titles and artists, alongside improvements to reporting output.
Introduce RwLock synchronization primitives for Soxr operations, ensuring thread safety during resampling. Update pipeline logic to correctly handle child process termination, and add tests verifying stable concurrent operations.
Introduces a multi-stage asynchronous pipeline for track processing, including indexing and identification workers. Adds support for track metadata fields (album, year, duration, ISRC) and implements new identification logic, including Qobuz ISRC fetching. Also extends the public API with new refresh functionality.
This commit finalizes the implementation of canonical music identities, requiring proof for titles and resolving concurrency issues related to track identification and data flow updates.
This change introduces Migration 6, adding the `term_external` table to the schema. It refactors the data model with new structures like `TermSpec` and `WikidataGenre`, enhances genre loading with cycle detection, expands error handling, and introduces the `UniverseService` for managing concurrent operations.
This commit introduces a multi-level genre annotation system, transactional management for annotations, and core logic for track identification. It includes mechanisms for handling identification backlogs, defining identification levels, and adding a command-line probe utility for querying the music universe.
This change introduces logic for identifying tracks based on vector proximity, classifying duplicates, and addressing issues related to data redundancy and identification threshold dependency.
Introduces a comprehensive deduplication system using Union-Find, allowing tracks to merge based on audio similarity and name criteria. Updates indexing logic to handle merges, ensures data persistence during merging, and expands the public API for new features.
Introduces mechanisms for tracking permanent ID validity, including state persistence, expiration deadlines, and explicit invalidation. Updates the database structure, cache registry, and core resolution logic to enforce fetchability checks and handle resolution failures appropriately.
Introduce new methods for permanent ID invalidation and fetchability checks. Update ingestion to support time-based URL validity, simplify playlist feeder architecture, and add tests for URL processing.
Introduces specific error types for source unavailability and header fetching failures. Implements robust cache invalidation for permanent IDs, supports transactional bulk deletion, and refines lazy resource resolution logic. Also updates client APIs for checking track existence and confirming resource status.
Introduces mechanisms for splitting media URIs into transformers and permanent IDs. Implements runtime-dependent playability checks and a `spawn` control point. Updates music renderer logic to handle non-playable tracks gracefully.
Adds a function to verify source types for specific audio paths, ensuring correct mapping for sources like Qobuz and URLs.
This commit introduces logic to determine if a video is gone by querying the YouTube oEmbed endpoint and handles cases where a definitive verdict cannot be provided. A new error variant is added to represent inconclusive checks, and lazy providers are updated to handle confirmed source disappearance errors.
This commit introduces error handling for ALSA output, specifically managing underrun events. It also updates dependency resolution in Cargo.lock and configures the build system to support cross-compilation for the Linux target.
Introduces comprehensive pipeline lifecycle management using CancellationToken, implements card state logic, and adds extensive testing support for failure scenarios and retry mechanisms.

This update expands the core audio library API, introduces new components for managing player pipelines and card operations, implements device opening error handling, and ensures that pipeline termination correctly signals systemd for restart.
Refine the music selection mechanism by replacing the Gaussian model with alternatives, introducing a temperature parameter for control, and proposing new metrics for testing coherence, diversity, and repetition.
This change enhances how the server determines the request base URL, prioritizing headers used in reverse proxy environments. It also introduces warnings for potential misconfigurations, such as requesting a host that differs from the configured base URL, which could expose internal IP addresses.
Introduces the stream API for handling incoming data, modifies HTTP responses, and implements robust cache retry logic using exponential backoff and listener state tracking. Also adds station re-discovery logic to handle provider errors gracefully.
Bump version to 0.5.23
All checks were successful
Release / create-release (push) Successful in 12s
Release / apk (push) Successful in 4m33s
bbaab4dc5c
Updates the package version in Cargo.toml and version.txt, and updates the dependency locking configuration in Cargo.lock.
eric merged commit a79c640982 into main 2026-10-28 18:32:04 +01:00
eric deleted branch push-mkumqllmxxky 2026-10-28 18:32:10 +01:00
Sign in to join this conversation.
No reviewers
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
pmo/pmomusic!168
No description provided.