feat(tui): filter usage by model

[jimyag]

Summary

Add an interactive TUI model filter and show each model's usage share for the selected period.

Key changes

• Press f to edit a case-insensitive, partial model-name filter; use Enter to apply, Esc to cancel, and Ctrl+U to clear.
• Recalculate aggregate tokens, cost, tool calls, messages, and conversation counts from matching model data.
• Filter the Sessions view to conversations that used a matching model.
• Display per-period and totals-row model shares, using input + output + cached tokens and falling back to message counts when token details are incomplete.

Validation

• cargo build --quiet
• cargo test --quiet (401 passed)
• cargo clippy --quiet -- -D warnings
• cargo doc --quiet
• cargo fmt --all --quiet
• git diff --check
• Ratatui TestBackend rendering coverage for filtered models, recalculated cost, and model shares

Notes

• The filter is session-local and does not change persistent configuration.
• Session rows are selected by model membership; their numeric values remain whole-session totals because session aggregates do not retain per-model token breakdowns.

Summary by CodeRabbit

• New Features

  • Added an interactive model filter accessible with the f key.
  • Filter model statistics using case-insensitive name matching.
  • Edit, clear, apply, or cancel filters while preserving them during updates.
  • Display active filter status, usage percentages, and estimated-model indicators.
  • Updated keyboard help and disabled quit/update shortcuts while editing.

• Bug Fixes

  • Filtering now recalculates statistics, costs, sessions, daily summaries, and usage shares accurately, including when no models match.
  • Aggregate totals now reflect only the currently visible time periods.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: bfd865c9-5929-47fd-99c6-720f8f48b281

📥 Commits

Reviewing files that changed from the base of the PR and between 582b2c8 and 6e58eeb.

📒 Files selected for processing (2)

• src/tui.rs
• src/tui/tests.rs

🚧 Files skipped from review as they are similar to previous changes (2)

• src/tui/tests.rs
• src/tui.rs

---

📝 Walkthrough

Walkthrough

The TUI adds case-insensitive model filtering with editable input, cancellation, persistent application, recalculated statistics, and filtered sessions. Aggregate rows display usage shares from visible-period statistics. Tests cover filtering, fallback data, percentages, and rendering.

Changes

Model filtering and usage display

Layer / File(s)	Summary
Filter analyzer data src/tui.rs	Model filtering retains matching models and recalculates daily statistics, session aggregates, conversations, and totals. The filter reapplies when statistics change.
Interactive filter workflow src/tui.rs	The f key opens filter input. Editing supports character entry, backspace, Ctrl+U, Enter, and Esc. Active editing guards quit and update shortcuts.
Usage share aggregation and validation src/tui.rs, src/tui/tests.rs	Aggregate rows calculate sorted token-based usage shares with message-count fallback and estimated-model markers. Tests cover filtered calculations, empty matches, percentages, and table rendering.

Estimated code review effort: 4 (Complex) | ~45 minutes

Sequence Diagram(s)

sequenceDiagram
  participant User
  participant run_app
  participant UiState
  participant filter_analyzer_view_by_model
  User->>run_app: press f
  run_app->>UiState: activate filter editing
  User->>run_app: edit or apply model text
  run_app->>filter_analyzer_view_by_model: apply model filter
  filter_analyzer_view_by_model-->>UiState: return filtered statistics
  UiState-->>User: redraw filtered views and usage shares

Loading

Possibly related PRs

• Piebald-AI/splitrail#189: Both changes modify TUI aggregate statistics and visible-period application handling.
• Piebald-AI/splitrail#215: Both changes modify aggregate statistics table rendering and model/stat columns.
• Piebald-AI/splitrail#219: Both changes modify TUI filtering and statistics aggregation.

Suggested reviewers: mike1858

Poem

A rabbit presses f beneath the moon,
Matching models gather in tune.
Totals show their shares with care,
Sessions follow matching data there.
Enter applies; Esc restores the view.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: adding model-based usage filtering to the TUI.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/tui.rs`:
- Around line 1514-1525: Update the non-empty model-filter branch in the
help_text construction to append the existing estimated-pricing note whenever
has_estimated_models is true, matching the behavior of the unfiltered help
branch while preserving the current filter controls and base_help_text.
- Around line 267-305: Update the DailyStats filtering logic around filtered_day
to preserve user_messages if it is consumed downstream, and handle days where
model_stats is incomplete relative to models without silently zeroing totals.
Inspect user_messages usages and the data-building paths, then apply the
established fallback behavior for matched models (at minimum preserving
message-count totals) while retaining accurate model_stats-derived metrics when
available. Ensure the reconstructed filtered_day and downstream totals remain
consistent for cached and current data.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 30232248-c9dc-42f2-b0d9-12edd4978ca4

📥 Commits

Reviewing files that changed from the base of the PR and between 9de0187 and 8e93d5c.

📒 Files selected for processing (2)

• src/tui.rs
• src/tui/tests.rs

[jimyag]

[mike1858]

Cool, this is amazing! Thank you for all of your PRs by the way - they're very useful, and the screenshots you attach help me understand the feature better and review faster 👍

[mike1858]

@jimyag Due to merging one of the other PRs, you have some new merge conflicts. Will you please fix them?

[jimyag]

Thank you! I’m glad the PRs have been useful, and I’ll keep including screenshots where they help make the behavior easier to review. I really appreciate your time and feedback! 🙏