Changelog

All notable changes to this project will be documented in this file.

[1.4.1] - 2026-06-24

Session Pollution: Isolated active similarity thresholds (pg_trgm.word_similarity_threshold) using transaction-scoped SET LOCAL queries.
Robust Process Management: Replaced internal driver references with standardized SQL-based SELECT pg_backend_pid() calls.
Memory Security (OOM Guard): Extended MemoryCacheProvider with LRU eviction and maximum constraints (maxEntries: 10000).
Cache Eviction Cleanup: Added background sweep interval daemon in MemoryCacheProvider to free stale RAM assets.
Total Count Expressions: Suppressed heavy COUNT(*) OVER() window scans during deep keyset-linked page requests when skipTotalCount is active.
Cache Key Clashing: Implemented strict alphanumeric order sorting of query parameters before hashing keys.
Chronological Metrics: Patched sequence metrics array in MetricsCollector to output in strict chronological timeframe sequence.
Transactional Consistency: Integrated AbortSignal listeners internally within Active Transaction scopes.
Gemini Credentials Security: Moved client API keys from URL query arguments to x-goog-api-key headers to stop proxies logging plain text credentials.

DRY Refactoring: Extracted redundant filter syntax constructors and rows formatting loops into unified utility maps (result-mapper.ts).
Static Keyboards Mapping: Replaced dynamic hash generations inside keyboard translator (convertLayout) with pre-calculated key maps.
Instance Caching: Cached Strategy classes internally to avoid garbage collection stress on query-time allocations.
Latency Buffers: Upgraded collector buffer mechanics to use $O(1)$ ring buffer logic over flat arrays.
Cache Hit Logic: Refactored writing rules to save pages if total > 0 OR data.length > 0 (keeps caching active when skipTotalCount is active).

Database Scale Enhancements: Scaled query throughput to execute O(log N) lookup overhead over 10M+ datasets.
GiST/KNN Trigram Indices: Migrated fuzzy trigram scans to GiST arrays with KNN distance operator (<->) to enable database-backed fast sorting.
FTS Ranking Speedup: Replaced positional ts_rank_cd logic with ts_rank to cut FTS retrieval cold latencies by 20-30%.
Parallel Search Execution: Implemented direct Promise.race FTS and standard search query scheduling.
Active Zombie Suppression: Enabled on-the-fly execution abortion of the slower search branch via backend connection cancellation signal (pg_cancel_backend).
Dedicated cancelPool: Spawned isolated backend connection pool (size 2-3) reserved strictly for cancel interrupts.
Automatic Generated Vectors: Changed manual data indexing workflows to use GENERATED ALWAYS AS ... STORED vectors, speeding up seeding up to 10x.

skipTotalCount Parameter: Added config flag to omit heavy total records counts on target pages.

Reliability System: Built-in health() checks, OOM protections (MAX_ROWS).
Redlock-style Cache Deduplication: Prevents cache stampedes in distributed environments.
Intelligent Rate-Limiting Queues: Uses p-queue for AI APIs (OpenAI/Gemini) to handle 429 errors gracefully.
True Zero-Freeze Networking: Native AbortController propagation ensures timeouts actually kill underlying fetch requests.
ISO 9 Compliance: Standardized Cyrillic-to-Latin transliteration for cross-language search reliability.
Monitoring & Metrics: Real-time tracking of cacheHitRate, dbLatency, and providerErrors via engine.metrics.

Keyset Pagination: High-performance cursor-based pagination support for massive result sets.
CLI Migration Suite: Automated interactive database setup and index migration tool (npm run init).

Enterprise Security: Parameterized queries, strict SqlSanitizer whitelist for identifiers, and injection-safe filter keys.