fix: render_live_sessions shows active fields for resumed sessions (#139)

[microsasa]

Fixes #139

Problem

render_live_sessions used s.user_messages (all-time total) and _estimated_output_tokens(s) (model_metrics sum from shutdown data only) for all active sessions. For resumed sessions, this showed stale historical data instead of the current post-resume activity.

Fix

When a session has active_user_messages or active_output_tokens set (indicating it was resumed), use those fields instead. Pure-active sessions (never shut down, where both are 0) continue using the existing totals. This matches the pattern already used by _render_active_section.

Tests added

• test_resumed_session_shows_active_fields — resumed session with historical model_metrics and non-zero active fields shows only the active-period values
• test_pure_active_session_uses_totals — pure-active session (no prior shutdown) still uses user_messages and model_metrics totals

All 409 tests pass, 98% coverage, ruff/pyright clean.

Generated by Issue Implementer · ◷

Warning

⚠️ Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• astral.sh

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "astral.sh"

See Network Configuration for more information.

[Copilot]

Pull request overview

Fixes render_live_sessions so that resumed active sessions display post-resume activity (active_user_messages, active_output_tokens) instead of stale historical totals, aligning it with the existing _render_active_section behavior.

Changes:

• Update render_live_sessions to choose between active-period stats vs. historical totals based on presence of active_* fields.
• Add unit tests covering resumed-session behavior and ensuring pure-active sessions keep using totals.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
src/copilot_usage/report.py	Switch live-session Messages/Output Tokens to active_* fields for resumed sessions, otherwise keep existing totals logic.
tests/copilot_usage/test_report.py	Add coverage for resumed sessions vs. pure-active sessions in render_live_sessions.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

Commit pushed: 69f5961

Generated by Review Responder

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

Commit pushed: f0246c4

Generated by Review Responder

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

test_resumed_session_zero_activity_shows_zeros currently only asserts that historical totals are absent, but it doesn’t assert that the active-period zeros are actually rendered. If the renderer regresses to showing a dash/blank for zeros, this test would still pass. Consider asserting for whole-word 0 in the output (and ideally in both Messages and Output Tokens columns) to match the test’s docstring/intent.

Suggested change

	# And should explicitly render zeros for the active period
	session_line = next(
	(line for line in output.splitlines() if "Just Resumed" in line),
	"",
	)
	# Expect at least two whole-word zeros on the session row (Messages and Output Tokens)
	zeros_on_row = re.findall(r"\b0\b", session_line)
	assert len(zeros_on_row) >= 2, (
	"resumed session row should show 0 for both messages and output tokens"
	)

[microsasa]

Good catch — added the suggested assertions. The test now finds the "Just Resumed" row and verifies at least two whole-word 0 values appear (Messages and Output Tokens), so a regression to dash/blank would be caught.

Generated by Review Responder

[github-actions]

Quality Gate: APPROVED ✅

Code Quality: Good — clean conditional using last_resume_time is not None to distinguish resumed vs. pure-active sessions, consistent with the existing pattern in _render_active_section.

Impact: LOW — display-only bug fix in render_live_sessions table rendering (+32/−3 lines across report.py and test_report.py). No changes to data models, APIs, or dependencies.

Tests: Three meaningful test cases covering the resumed session case, the pure-active fallback, and the zero-activity edge case.

Auto-approving for merge.

[github-actions]

Quality Gate: APPROVED ✅

Low-impact bug fix with good test coverage. Auto-approving for merge.

Evaluated:

• Code quality: Clean, focused fix. Uses last_resume_time is not None to detect resumed sessions — correctly handles the zero-activity edge case. Follows existing patterns (_render_active_section).
• Tests: 3 new tests covering resumed session (active values shown), pure-active regression (totals preserved), and zero-activity edge case.
• Impact: LOW — display-only change in render_live_sessions(), no data mutation or API changes. 2 files, ~12 lines of source.

[microsasa]

❌ Pipeline orchestrator: review-response loop reached 3 rounds. Marking as stuck for human review.

[github-actions]

Commit pushed: 53b58c3

Generated by Review Responder