v2.2.0

Bug Fixes

• Fixed database record loading when libbook.keywords is NULL.
• Fixed database record loading when libbook.md5 is NULL.
• Both batch book lookup and single book lookup now scan nullable string columns through sql.NullString.
• NULL keywords and NULL md5 now become empty strings in merged/database manifest records instead of failing with converting NULL to string is unsupported.
• Existing MD5 NUL-byte trimming still applies after NULL conversion.

SQL Import And Fetch Behavior: default generated config avoids downloading unused Librusec tables

• Librusec SQL dump selection moved fully into default fetch config regex, matching existing Flibusta model.
• Default Librusec profiles now fetch only SQL dumps actually needed by current importer/query paths.
• Removed code-level Librusec SQL import allow-list from db/importer.go.
• Import now validates dump format, then imports discovered .sql files without table-name filtering.
• Fetch path remains config-driven: whatever sql_pattern matches is downloaded.

Behavior Change

• If user manually points import/cache/merge at a SQL directory containing extra Librusec .sql files, those files are no longer silently skipped by importer code.
• To avoid importing extra Librusec dumps, filter at fetch/config level via fetch.libraries[].sql_pattern.
• Existing user config files are not auto-updated; regenerate config or copy new Librusec sql_pattern if needed.

Documentation

• Added INPX Record Construction section to README.

Maintenance

• Dependency Updates

Migration Notes

• For default config users: regenerate config or use bundled defaults to get narrower Librusec SQL fetch behavior.
• For custom Librusec fetch profiles: update sql_pattern if you want to avoid unused tables.
• For manual SQL dump directories: remove unwanted .sql files before import if you do not want them loaded.

Full Changelog: v2.1.0...v2.2.0