refactor(llc)!: route getUnreadCount through the generated client

[VelikovPetar]

Submit a pull request

Linear: FLU-935
Github Issue: -

CLA

• I have signed the Stream CLA (required).
• The code changes follow best practices
• Code changes are tested (add some information if not applicable)

Description of the pull request

Moves StreamChatClient.getUnreadCount onto the OpenAPI-generated client, as new plan group 18 (Unread Counts), split out of group 09 (Users).

• getUnreadCount returns Result<GetUnreadCountResponse> instead of throwing. On success it still updates the current user's totalUnreadCount, unreadChannels and unreadThreads (via result.onSuccess); a failure leaves them unchanged.
• GetUnreadCountResponse moves to models/response/ as a freezed class with a non-null duration.
• UnreadCountsChannel, UnreadCountsThread and UnreadCountsChannelType become freezed plain models, one per file, without fromJson/toJson; they gain value equality and copyWith.
• New UsersRepository, mirroring UserApi; the mappers go in the existing user_mapper.dart.
• StreamChatApi.user.getUnreadCount is removed.

Generated	Public
DefaultApi.unreadCounts (GET /api/v2/chat/unread)	StreamChatClient.getUnreadCount
WrappedUnreadCountsResponse	GetUnreadCountResponse (kept)
UnreadCountsChannel / UnreadCountsThread / UnreadCountsChannelType	same names (ours, freezed)
—	UsersRepository (internal)

Why

The migration moves every endpoint onto the generated client, which returns a Result for every call. This one is the smallest unblocked slice: no parameters, a response that matches ours field for field, and nothing persists it.

Notes for reviewers

• The v10 names stay. unreadCounts and WrappedUnreadCountsResponse say nothing a caller needs.
• Same backend handler. /unread and /api/v2/chat/unread both mount v1.UnreadCounts; it's gated by ClassicUnreadCountsEnabled identically on both, so the switch changes nothing about who may call it. Not in beta.
• v2 sends last_read as epoch nanoseconds rather than RFC 3339; the generated converter handles both.

Testing

• melos run analyze clean; stream_chat suite green (1,910 tests), flutter_core and persistence green. generate_plan.py --check reports 0 problems.
• New client_get_unread_count_test.dart: the full envelope against a fully populated generated response, the current-user update, the failure without throwing, and the counts left unchanged on failure.
• Live v1 vs v2 parity on the demo app for three users (no unread, 541 unread across 42 channels, one unread thread): identical counts, and the real v2 bodies decode through the generated model with the same instants as v1.

Screenshots / Videos

Not applicable: no visible UI change.

🤖 Generated with Claude Code

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

---

_{Comment @coderabbitai help to get the list of available commands.}