⚡ perf: avoid timeline bootstrap for lightweight PR views

src/gh_llm/commands/pr.py

@@ -10,7 +10,7 @@
 from gh_llm.github_api import GitHubClient
 from gh_llm.invocation import display_command_with
 from gh_llm.models import PullRequestDiffPage
-from gh_llm.pager import DEFAULT_PAGE_SIZE, TimelinePager
+from gh_llm.pager import DEFAULT_PAGE_SIZE, TimelinePager, build_context_from_meta
 from gh_llm.render import (
     render_checks_section,
     render_comment_node_detail,
@@ -325,16 +325,22 @@ def cmd_pr_view(args: Any) -> int:
     pager = TimelinePager(client)
 
     meta = client.resolve_pull_request(selector=args.pr, repo=args.repo)
-    context, first_page, last_page = pager.build_initial(
-        meta,
-        page_size=page_size,
-        show_resolved_details=expand.resolved,
-        show_outdated_details=True,
-        show_minimized_details=expand.minimized,
-        show_details_blocks=expand.details,
-        diff_hunk_lines=diff_hunk_lines,
-    )
-    shown_pages: set[int] = {1}
+    context = build_context_from_meta(meta=meta, page_size=page_size)
+    first_page: TimelinePage | None = None
+    last_page: TimelinePage | None = None
+    shown_pages: set[int] = set()
+
+    if show.timeline:
+        context, first_page, last_page = pager.build_initial(
+            meta,
+            page_size=page_size,
+            show_resolved_details=expand.resolved,
+            show_outdated_details=True,
+            show_minimized_details=expand.minimized,
+            show_details_blocks=expand.details,
+            diff_hunk_lines=diff_hunk_lines,
+        )
+        shown_pages.add(1)
 
     wrote_output = False
 
@@ -353,6 +359,7 @@ def print_block(lines: list[str]) -> None:
     if show.description:
         print_block(render_description(context))
     if show.timeline:
+        assert first_page is not None
         print_block(["## Timeline"])
         print_block(render_page(1, context, first_page))
 
@@ -565,9 +572,9 @@ def cmd_pr_thread_expand(args: Any) -> int:
 
 def cmd_pr_checks(args: Any) -> int:
     client = GitHubClient()
-    pager = TimelinePager(client)
-    context, meta = _resolve_context_and_meta(client=client, pager=pager, args=args)
-    checks = client.fetch_checks(meta.ref)
+    meta = _resolve_pr_meta(client=client, args=args)
+    context = build_context_from_meta(meta=meta, page_size=DEFAULT_PAGE_SIZE)
+    checks = client.fetch_checks(meta.ref) if meta.state == "OPEN" else []
     for line in render_checks_section(
         context=context,
         checks=checks,

src/gh_llm/models.py

@@ -155,6 +155,7 @@ class TimelineContext:
     is_draft: bool
     body: str
     updated_at: str
+    timeline_loaded: bool = True
     labels: tuple[str, ...] = ()
     kind: str = "pr"
     pr_reactions_summary: str | None = None
@@ -198,6 +199,7 @@ def to_dict(self) -> dict[str, object]:
             "is_draft": self.is_draft,
             "body": self.body,
             "updated_at": self.updated_at,
+            "timeline_loaded": self.timeline_loaded,
             "labels": list(self.labels),
             "kind": self.kind,
             "pr_reactions_summary": self.pr_reactions_summary,
@@ -243,6 +245,11 @@ def from_dict(cls, value: dict[str, object]) -> TimelineContext:
             is_draft=bool(value.get("is_draft")),
             body=_as_str(value.get("body"), ""),
             updated_at=_as_str(value.get("updated_at"), ""),
+            timeline_loaded=(
+                _as_int(value.get("total_pages"), 0) > 0
+                if value.get("timeline_loaded") is None
+                else bool(value.get("timeline_loaded"))
+            ),
             labels=tuple(_as_str(item, "") for item in _as_list(value.get("labels")) if item),
             kind=_as_str(value.get("kind"), "pr"),
             pr_reactions_summary=_as_str_optional(value.get("pr_reactions_summary")),

src/gh_llm/pager.py

@@ -12,6 +12,64 @@
 DEFAULT_PAGE_SIZE = 8
 
 
+def build_context_from_meta(
+    meta: PullRequestMeta,
+    page_size: int,
+    *,
+    total_count: int | None = None,
+    total_pages: int | None = None,
+) -> TimelineContext:
+    _validate_page_size(page_size)
+
+    timeline_loaded = total_count is not None and total_pages is not None
+    resolved_total_count = 0 if total_count is None else total_count
+    resolved_total_pages = 0 if total_pages is None else total_pages
+
+    return TimelineContext(
+        owner=meta.ref.owner,
+        name=meta.ref.name,
+        number=meta.ref.number,
+        page_size=page_size,
+        total_count=resolved_total_count,
+        total_pages=resolved_total_pages,
+        title=meta.title,
+        url=meta.url,
+        author=meta.author,
+        state=meta.state,
+        is_draft=meta.is_draft,
+        body=meta.body,
+        updated_at=meta.updated_at,
+        timeline_loaded=timeline_loaded,
+        labels=meta.labels,
+        kind=meta.kind,
+        pr_reactions_summary=meta.reactions_summary,
+        can_edit_pr_body=meta.can_edit_body,
+        is_merged=meta.is_merged,
+        head_ref_name=meta.head_ref_name,
+        head_ref_repo=meta.head_ref_repo,
+        head_ref_oid=meta.head_ref_oid,
+        head_ref_deleted=meta.head_ref_deleted,
+        pr_node_id=meta.node_id,
+        merge_state_status=meta.merge_state_status,
+        mergeable=meta.mergeable,
+        review_decision=meta.review_decision,
+        requires_approving_reviews=meta.requires_approving_reviews,
+        required_approving_review_count=meta.required_approving_review_count,
+        requires_code_owner_reviews=meta.requires_code_owner_reviews,
+        approved_review_count=meta.approved_review_count,
+        requires_status_checks=meta.requires_status_checks,
+        base_ref_name=meta.base_ref_name,
+        base_ref_oid=meta.base_ref_oid,
+        merge_commit_allowed=meta.merge_commit_allowed,
+        squash_merge_allowed=meta.squash_merge_allowed,
+        rebase_merge_allowed=meta.rebase_merge_allowed,
+        co_author_trailers=meta.co_author_trailers,
+        conflict_files=meta.conflict_files,
+        forward_after_by_page=({1: None} if timeline_loaded else {}),
+        backward_before_by_page=({resolved_total_pages: None} if timeline_loaded else {}),
+    )
+
+
 class TimelinePager:
     def __init__(self, client: GitHubClient) -> None:
         self._client = client
@@ -45,47 +103,11 @@ def build_initial(
         total_count = first_page.total_count
         total_pages = _page_count(total_count, page_size)
 
-        context = TimelineContext(
-            owner=meta.ref.owner,
-            name=meta.ref.name,
-            number=meta.ref.number,
+        context = build_context_from_meta(
+            meta=meta,
             page_size=page_size,
             total_count=total_count,
             total_pages=total_pages,
-            title=meta.title,
-            url=meta.url,
-            author=meta.author,
-            state=meta.state,
-            is_draft=meta.is_draft,
-            body=meta.body,
-            updated_at=meta.updated_at,
-            labels=meta.labels,
-            kind=meta.kind,
-            pr_reactions_summary=meta.reactions_summary,
-            can_edit_pr_body=meta.can_edit_body,
-            is_merged=meta.is_merged,
-            head_ref_name=meta.head_ref_name,
-            head_ref_repo=meta.head_ref_repo,
-            head_ref_oid=meta.head_ref_oid,
-            head_ref_deleted=meta.head_ref_deleted,
-            pr_node_id=meta.node_id,
-            merge_state_status=meta.merge_state_status,
-            mergeable=meta.mergeable,
-            review_decision=meta.review_decision,
-            requires_approving_reviews=meta.requires_approving_reviews,
-            required_approving_review_count=meta.required_approving_review_count,
-            requires_code_owner_reviews=meta.requires_code_owner_reviews,
-            approved_review_count=meta.approved_review_count,
-            requires_status_checks=meta.requires_status_checks,
-            base_ref_name=meta.base_ref_name,
-            base_ref_oid=meta.base_ref_oid,
-            merge_commit_allowed=meta.merge_commit_allowed,
-            squash_merge_allowed=meta.squash_merge_allowed,
-            rebase_merge_allowed=meta.rebase_merge_allowed,
-            co_author_trailers=meta.co_author_trailers,
-            conflict_files=meta.conflict_files,
-            forward_after_by_page={1: None},
-            backward_before_by_page={total_pages: None},
         )
         self._remember_forward(context, page=1, cursor_used=None, page_result=first_page)
 

src/gh_llm/render.py

@@ -35,10 +35,15 @@ def render_frontmatter(context: TimelineContext) -> list[str]:
         f"labels: {json.dumps(list(context.labels), ensure_ascii=False)}",
         f"draft: {str(context.is_draft).lower()}",
         f"updated_at: {context.updated_at}",
-        f"timeline_events: {context.total_count}",
-        f"page_size: {context.page_size}",
-        f"total_pages: {context.total_pages}",
     ]
+    if context.timeline_loaded:
+        lines.extend(
+            [
+                f"timeline_events: {context.total_count}",
+                f"page_size: {context.page_size}",
+                f"total_pages: {context.total_pages}",
+            ]
+        )
     if context.kind == "pr":
         lines.append(f"is_merged: {str(context.is_merged).lower()}")
         if context.head_ref_repo:

tests/test_cli.py

@@ -609,6 +609,91 @@ def test_view_and_expand_use_real_cursor_pagination(
     assert "END_MARKER" in out
 
 
+def test_pr_view_show_meta_skips_timeline_bootstrap(
+    monkeypatch: pytest.MonkeyPatch,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    responder = GhResponder()
+    monkeypatch.setattr(github_api.subprocess, "run", responder.run)
+
+    code = cli.run(["pr", "view", "77928", "--repo", "PaddlePaddle/Paddle", "--show", "meta"])
+    assert code == 0
+
+    out = capsys.readouterr().out
+    assert "pr: 77928" in out
+    assert "timeline_events:" not in out
+    assert "## Timeline" not in out
+    assert "## Checks" not in out
+
+    graphql_queries = [_extract_form(call, "query") for call in responder.calls if call[:3] == ["gh", "api", "graphql"]]
+    assert any("headRefName" in query and "timelineItems" not in query for query in graphql_queries)
+    assert not any("timelineItems(" in query for query in graphql_queries)
+    assert not any("reviewThreads(first:100" in query for query in graphql_queries)
+    assert not any("statusCheckRollup" in query for query in graphql_queries)
+
+
+def test_pr_view_show_checks_fetches_checks_without_timeline_bootstrap(
+    monkeypatch: pytest.MonkeyPatch,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    responder = GhResponder()
+    monkeypatch.setattr(github_api.subprocess, "run", responder.run)
+
+    code = cli.run(["pr", "view", "77928", "--repo", "PaddlePaddle/Paddle", "--show", "checks"])
+    assert code == 0
+
+    out = capsys.readouterr().out
+    assert "## Checks" in out
+    assert "[IN_PROGRESS/NONE] unit-tests (check-run)" in out
+    assert "## Timeline" not in out
+
+    graphql_queries = [_extract_form(call, "query") for call in responder.calls if call[:3] == ["gh", "api", "graphql"]]
+    assert any("statusCheckRollup" in query for query in graphql_queries)
+    assert not any("timelineItems(" in query for query in graphql_queries)
+    assert not any("reviewThreads(first:100" in query for query in graphql_queries)
+
+
+def test_pr_checks_command_skips_timeline_bootstrap(
+    monkeypatch: pytest.MonkeyPatch,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    responder = GhResponder()
+    monkeypatch.setattr(github_api.subprocess, "run", responder.run)
+
+    code = cli.run(["pr", "checks", "--pr", "77928", "--repo", "PaddlePaddle/Paddle", "--all"])
+    assert code == 0
+
+    out = capsys.readouterr().out
+    assert "## Checks" in out
+    assert "[COMPLETED/SUCCESS] lint (check-run)" in out
+
+    graphql_queries = [_extract_form(call, "query") for call in responder.calls if call[:3] == ["gh", "api", "graphql"]]
+    assert any("statusCheckRollup" in query for query in graphql_queries)
+    assert not any("timelineItems(" in query for query in graphql_queries)
+    assert not any("reviewThreads(first:100" in query for query in graphql_queries)
+
+
+def test_pr_view_show_mergeability_fetches_status_without_timeline_bootstrap(
+    monkeypatch: pytest.MonkeyPatch,
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    responder = GhResponder()
+    monkeypatch.setattr(github_api.subprocess, "run", responder.run)
+
+    code = cli.run(["pr", "view", "77971", "--repo", "PaddlePaddle/Paddle", "--show", "mergeability"])
+    assert code == 0
+
+    out = capsys.readouterr().out
+    assert "## Mergeability" in out
+    assert "Status: Merging is blocked" in out
+    assert "## Timeline" not in out
+
+    graphql_queries = [_extract_form(call, "query") for call in responder.calls if call[:3] == ["gh", "api", "graphql"]]
+    assert any("statusCheckRollup" in query for query in graphql_queries)
+    assert not any("timelineItems(" in query for query in graphql_queries)
+    assert not any("reviewThreads(first:100" in query for query in graphql_queries)
+
+
 def test_web_like_extra_timeline_events_are_rendered(
     monkeypatch: pytest.MonkeyPatch,
     tmp_path: Path,