Optimize mail fetching with batch DB queries and in-memory caching

[bkaankose]

Summary

Refactored the FetchMailsAsync method in MailService to significantly reduce database round-trips by implementing batch queries and pre-loading data into in-memory caches. This improves performance when fetching large numbers of mails, especially when threading is enabled.

Key Changes

• Batch folder loading: Instead of loading folders one-per-mail, folders are now loaded in batches using Task.WhenAll, with initial seeding from options to avoid redundant DB calls
• Batch account loading: All accounts are fetched once via GetAccountsAsync() instead of per-mail lookups
• Batch contact loading: Implemented new GetContactsByAddressesAsync() method in ContactService to fetch all sender contacts in a single SQL IN(...) query instead of individual queries per mail
• Refactored property assignment: Extracted property assignment logic into new AssignPropertiesFromCaches() method that works entirely from in-memory dictionaries with no DB calls
• Optimized thread expansion: Thread mail fetching now uses batch queries and reuses the same caching strategy as initial mail loading
• Early exit optimization: Added early returns when mail lists are empty to avoid unnecessary processing

Implementation Details

• Removed ConcurrentDictionary usage in favor of regular Dictionary since all operations are now synchronous after pre-loading
• Changed from per-mail async property loading to a two-phase approach: batch pre-load all data, then synchronously assign from caches
• Self-sent mails (e.g., in Sent folder) now construct contacts directly from account metadata to avoid unnecessary DB lookups
• Updated GetMailItemsAsync() to use the same optimized caching pattern
• Improved cancellation token handling with additional ThrowIfCancellationRequested() calls

Performance Impact

Reduces DB round-trips from O(n) per mail to O(1) for folders, accounts, and contacts, resulting in significantly faster mail list loading especially for large mailboxes and merged views.

https://claude.ai/code/session_018bqahGc6zi95JJhc2aARKS