benchloop-client 0.2.23__tar.gz → 0.2.25__tar.gz

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: benchloop_client
-Version: 0.2.23
+Version: 0.2.25
 Summary: Benchloop CLI — command line tool for benchmark analysis data (metrics, trajectory, reports, artifacts).
 Requires-Python: >=3.10
 Requires-Dist: httpx>=0.25.2
@@ -84,6 +84,9 @@ bl metadata download --task-id TASK_ID -o ./out
 bl report download -j J123 -o ./out
 bl report get -g G123
 bl report download -g G123 -o ./out
+bl rsi get -j J123
+bl rsi trigger -j J123
+bl rsi download -g G123 -o ./out
 bl jobs get -g G1 --all          # get supports --all (auto-paginate)
 bl jobs download -g G1           # download is always full
 bl artifacts download -g G1 -o ./out
@@ -99,6 +102,15 @@ bl brm status --task-id UID
 bl brm list -g G1
 ```
 
+Metrics and Job trajectory requests always explicitly use the Trace source. The
+CLI does not expose a `trace`/`artifacts` source selector.
+
+Group trajectory retrieval and downloads use the backend Trace export API. The
+CLI waits for the export URL, downloads the gzip JSON-lines payload, and keeps
+the existing `group_id`/`trajectories` JSON layout. The export wait timeout is
+30 minutes by default; override it in seconds with
+`--download-timeout SECONDS` on trajectory commands or `pull`.
+
 Group downloads always auto-paginate and include every Job in the Group. In
 particular, `trajectory`, detailed `metrics`, `report`, and `artifacts` enumerate
 the complete Job list instead of relying on the backend Group endpoints that

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/README.md

@@ -72,6 +72,9 @@ bl metadata download --task-id TASK_ID -o ./out
 bl report download -j J123 -o ./out
 bl report get -g G123
 bl report download -g G123 -o ./out
+bl rsi get -j J123
+bl rsi trigger -j J123
+bl rsi download -g G123 -o ./out
 bl jobs get -g G1 --all          # get supports --all (auto-paginate)
 bl jobs download -g G1           # download is always full
 bl artifacts download -g G1 -o ./out
@@ -87,6 +90,15 @@ bl brm status --task-id UID
 bl brm list -g G1
 ```
 
+Metrics and Job trajectory requests always explicitly use the Trace source. The
+CLI does not expose a `trace`/`artifacts` source selector.
+
+Group trajectory retrieval and downloads use the backend Trace export API. The
+CLI waits for the export URL, downloads the gzip JSON-lines payload, and keeps
+the existing `group_id`/`trajectories` JSON layout. The export wait timeout is
+30 minutes by default; override it in seconds with
+`--download-timeout SECONDS` on trajectory commands or `pull`.
+
 Group downloads always auto-paginate and include every Job in the Group. In
 particular, `trajectory`, detailed `metrics`, `report`, and `artifacts` enumerate
 the complete Job list instead of relying on the backend Group endpoints that

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "benchloop_client"
-version = "0.2.23"
+version = "0.2.25"
 description = "Benchloop CLI — command line tool for benchmark analysis data (metrics, trajectory, reports, artifacts)."
 readme = "README.md"
 requires-python = ">=3.10"

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/src/aliyun/bmcli/cli.py

@@ -17,6 +17,7 @@ from aliyun.bmcli.commands import (
     overview as overview_cmd,
     pull as pull_cmd,
     report as report_cmd,
+    rsi as rsi_cmd,
     tasks as tasks_cmd,
     trajectory as trajectory_cmd,
 )
@@ -139,6 +140,7 @@ app.add_typer(jobs_cmd.app, name="jobs")
 app.add_typer(artifacts_cmd.app, name="artifacts")
 app.add_typer(metadata_cmd.app, name="metadata")
 app.add_typer(brm_cmd.app, name="brm")
+app.add_typer(rsi_cmd.app, name="rsi")
 app.command(
     "ap-cmd",
     context_settings={"allow_extra_args": True, "ignore_unknown_options": True},

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/src/aliyun/bmcli/client.py

@@ -38,6 +38,9 @@ class ArtifactMissingError(ClientError):
     """
 
 
+_SSO_LOGIN_HOST = "login.alibaba-inc.com"
+
+
 class ApiClient:
     """Thin wrapper over ``httpx.Client`` for the benchmark openapi."""
 
@@ -148,11 +151,16 @@ class ApiClient:
         url: str,
         dest_path: str | Path,
         progress: Optional[Any] = None,
+        *,
+        detect_sso_redirect: bool = False,
     ) -> Path:
         """Stream ``url`` to ``dest_path`` on disk. Returns the path written.
 
         ``progress``（可选）需提供 ``begin(total)`` / ``advance(nbytes)`` 两个
         方法（如 ``output.TransferStream``）；每次重试会重新 ``begin`` 归零。
+
+        ``detect_sso_redirect`` 开启时，被网关 302 到 SSO 登录页抛 ``AuthError``；
+        默认关闭，保持"HTML 错误页 = 产物缺失"的既有分类，不影响其他调用方。
         """
         dest = Path(dest_path)
         dest.parent.mkdir(parents=True, exist_ok=True)
@@ -161,6 +169,13 @@ class ApiClient:
             try:
                 with httpx.Client(timeout=self.timeout, follow_redirects=True) as client:
                     with client.stream("GET", self._url(url), headers=self._headers()) as resp:
+                        if detect_sso_redirect and self._is_sso_login_redirect(resp):
+                            # 网关不认 X-API-Key 时会 302 到 SSO 登录页；这是
+                            # 认证问题，不能落入下面的“HTML 错误页=文件缺失”误判。
+                            raise AuthError(
+                                "endpoint requires SSO login (redirected to "
+                                f"{_SSO_LOGIN_HOST}); X-API-Key not accepted"
+                            )
                         if resp.status_code == 401:
                             raise AuthError("authentication failed (HTTP 401)")
                         if resp.status_code == 404:
@@ -199,6 +214,15 @@ class ApiClient:
             raise last_exc
         return dest  # pragma: no cover
 
+    @staticmethod
+    def _is_sso_login_redirect(resp: httpx.Response) -> bool:
+        targets = [r.headers.get("location", "") for r in resp.history]
+        try:
+            targets.append(str(resp.url))
+        except RuntimeError:
+            pass  # resp.request 未设置时（如测试里的裸 Response）没有 url
+        return any(_SSO_LOGIN_HOST in target for target in targets)
+
     @staticmethod
     def _safe_body(resp: httpx.Response) -> Any:
         try:

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/src/aliyun/bmcli/commands/_common.py

@@ -31,6 +31,12 @@ OutputDirOption = typer.Option(
 SkipExistingOption = typer.Option(
     False, "--skip-existing", help="Skip files that already exist on disk."
 )
+DownloadTimeoutOption = typer.Option(
+    1800,
+    "--download-timeout",
+    min=1,
+    help="Maximum seconds to wait for a Group Trace trajectory export.",
+)
 
 
 def expand(values: Optional[list[str]]) -> list[str]:

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/src/aliyun/bmcli/commands/metrics.py

@@ -25,6 +25,7 @@ from aliyun.bmcli.runtime import get_runtime, report_and_exit, safe_run_concurre
 app = typer.Typer(help="Run metrics (Group overall/detail, Job).", no_args_is_help=True)
 
 _SCOPES = ("overall", "detail", "all")
+_TRACE_SOURCE = "trace"
 
 _GROUP_RUN_METRIC_FIELDS = (
     "job_count",
@@ -112,7 +113,9 @@ def _validate_scope(scope: str) -> str:
 
 
 def _fetch_job(client, job_id: str):
-    data = client.get_json(f"/openapi/metrics/{job_id}")
+    data = client.get_json(
+        f"/openapi/metrics/{job_id}", params={"source": _TRACE_SOURCE}
+    )
     return _format_job_payload(data, job_id)
 
 
@@ -125,7 +128,12 @@ def fetch_group_detail(
         nonlocal tokenized_metrics
         data = client.get_json(
             "/openapi/metrics",
-            params={"group_id": group_id, "skip": skip, "limit": limit},
+            params={
+                "group_id": group_id,
+                "skip": skip,
+                "limit": limit,
+                "source": _TRACE_SOURCE,
+            },
         )
         if isinstance(data, dict) and tokenized_metrics is None:
             tokenized_metrics = data.get("tokenized_metrics")
@@ -154,7 +162,10 @@ def fetch_group_detail(
 
 
 def _fetch_group_overall(client, group_id: str):
-    data = client.get_json(f"/openapi/groups/{group_id}/metrics/overall")
+    data = client.get_json(
+        f"/openapi/groups/{group_id}/metrics/overall",
+        params={"source": _TRACE_SOURCE},
+    )
     return _format_group_overall(data, group_id)
 
 

{benchloop_client-0.2.23 → benchloop_client-0.2.25}/src/aliyun/bmcli/commands/pull.py

@@ -14,6 +14,7 @@ from aliyun.bmcli.commands._common import (
     JobOption,
     OutputDirOption,
     SkipExistingOption,
+    DownloadTimeoutOption,
     expand,
     resolve_output_dir,
     write_empty_tgz,
@@ -29,7 +30,10 @@ from aliyun.bmcli.commands.metrics import (
     fetch_group_detail,
 )
 from aliyun.bmcli.commands.tasks import _fetch_all as fetch_all_tasks
-from aliyun.bmcli.commands.trajectory import write_group_trajectories
+from aliyun.bmcli.commands.trajectory import (
+    fetch_job_trajectory,
+    write_group_trajectories,
+)
 from aliyun.bmcli.client import ArtifactMissingError, AuthError
 from aliyun.bmcli.ids import enforce_mutually_exclusive, require_any
 from aliyun.bmcli.runtime import get_runtime
@@ -61,6 +65,7 @@ def pull(
     ),
     output_dir: Optional[str] = OutputDirOption,
     skip_existing: bool = SkipExistingOption,
+    download_timeout: int = DownloadTimeoutOption,
 ) -> None:
     """Aggregate downloads for the given groups/jobs into the layout."""
     rt = get_runtime(ctx)
@@ -107,7 +112,7 @@ def pull(
                 f"{jid}:trajectory",
                 lambda jid=jid: _fetch_and_write_json(
                     root / f"{jid}_trajectory.json",
-                    lambda: client.get_json(f"/openapi/jobs/{jid}/trajectory"),
+                    lambda: fetch_job_trajectory(client, jid),
                 ),
             )
         if "report" in selected:
@@ -160,7 +165,7 @@ def pull(
                         rt.concurrency,
                         trajectory_dest,
                         pretty=rt.pretty,
-                        ids=job_ids(_group_jobs()),
+                        download_timeout=download_timeout,
                     ),
                 )
         if "report" in selected:

benchloop_client-0.2.25/src/aliyun/bmcli/commands/rsi.py

@@ -0,0 +1,235 @@
+"""`bmcli rsi` — RSI weakness analysis reports."""
+
+from __future__ import annotations
+
+from typing import Any, Optional
+
+import typer
+
+from aliyun.bmcli import output
+from aliyun.bmcli.client import AuthError, NetworkError
+from aliyun.bmcli.commands._common import (
+    GroupOption,
+    JobOption,
+    OutputDirOption,
+    SkipExistingOption,
+    expand,
+    resolve_output_dir,
+)
+from aliyun.bmcli.commands.group_data import (
+    fetch_all_jobs,
+    fetch_job_resources,
+    job_ids,
+    raise_for_failures,
+)
+from aliyun.bmcli.ids import enforce_mutually_exclusive, require_any
+from aliyun.bmcli.runtime import get_runtime, report_and_exit, safe_run_concurrent
+
+app = typer.Typer(help="RSI weakness analysis reports.", no_args_is_help=True)
+
+
+def _guard(groups: list[str], jobs: list[str]) -> None:
+    enforce_mutually_exclusive(groups, jobs)
+    require_any(groups, jobs)
+
+
+def _markdown(data: Any, key: str) -> str:
+    if not isinstance(data, dict):
+        return ""
+    report = data.get("report")
+    if not isinstance(report, dict):
+        return ""
+    content = report.get(key) or ""
+    return content if isinstance(content, str) else ""
+
+
+def _fetch_job(client, job_id: str):
+    return client.get_json(f"/openapi/jobs/{job_id}/rsi")
+
+
+def _fetch_group(client, group_id: str):
+    return client.get_json(f"/openapi/groups/{group_id}/rsi")
+
+
+def _trigger_job(client, job_id: str):
+    return client.post_json(f"/openapi/jobs/{job_id}/rsi/trigger")
+
+
+def _trigger_group(client, group_id: str, include_passed: bool):
+    return client.post_json(
+        f"/openapi/groups/{group_id}/rsi/trigger",
+        {"include_passed": include_passed},
+    )
+
+
+def _write_markdown(content: str, dest, skip_existing: bool):
+    if not content:
+        return None
+    if skip_existing and dest.exists():
+        output.log(f"skip existing {dest}")
+    else:
+        output.write_text_file(dest, content)
+        output.log(f"wrote {dest}")
+    return str(dest)
+
+
+@app.command("get")
+def get(
+    ctx: typer.Context,
+    group: Optional[list[str]] = GroupOption,
+    job: Optional[list[str]] = JobOption,
+) -> None:
+    """Query Job or Group RSI reports; prints a JSON array."""
+    rt = get_runtime(ctx)
+    groups = expand(group)
+    jobs = expand(job)
+    _guard(groups, jobs)
+
+    if jobs:
+        res = safe_run_concurrent(
+            jobs, lambda jid: _fetch_job(rt.client, jid), rt.concurrency
+        )
+    else:
+        res = safe_run_concurrent(
+            groups, lambda gid: _fetch_group(rt.client, gid), rt.concurrency
+        )
+    report_and_exit(res, emit=True, pretty=rt.pretty)
+
+
+@app.command("download")
+def download(
+    ctx: typer.Context,
+    group: Optional[list[str]] = GroupOption,
+    job: Optional[list[str]] = JobOption,
+    output_dir: Optional[str] = OutputDirOption,
+    skip_existing: bool = SkipExistingOption,
+) -> None:
+    """Download RSI report markdown returned inline by the get API."""
+    rt = get_runtime(ctx)
+    groups = expand(group)
+    jobs = expand(job)
+    _guard(groups, jobs)
+    root = resolve_output_dir(rt.output_dir, output_dir)
+
+    def job_worker(job_id: str):
+        try:
+            dest = root / f"{job_id}_rsi.md"
+            if skip_existing and dest.exists():
+                output.log(f"skip existing {dest}")
+                return str(dest)
+            return _write_markdown(
+                _markdown(_fetch_job(rt.client, job_id), "report_md"),
+                dest,
+                skip_existing,
+            )
+        finally:
+            job_progress.advance(job_id)
+
+    def download_group_reports(group_id: str):
+        written = []
+        failures = []
+        try:
+            summary_dest = root / f"{group_id}_rsi.md"
+            if skip_existing and summary_dest.exists():
+                output.log(f"skip existing {summary_dest}")
+                summary = str(summary_dest)
+            else:
+                summary = _write_markdown(
+                    _markdown(_fetch_group(rt.client, group_id), "summary_md"),
+                    summary_dest,
+                    skip_existing,
+                )
+            if summary:
+                written.append(summary)
+        except AuthError:
+            raise
+        except NetworkError as exc:
+            failures.append((f"{group_id}:summary", f"network: {exc}"))
+        except Exception as exc:  # noqa: BLE001 - continue with per-Job reports
+            failures.append((f"{group_id}:summary", str(exc)))
+
+        jobs_data = fetch_all_jobs(rt.client, group_id)
+        ids = job_ids(jobs_data)
+        report_progress = output.ProgressCounter(
+            f"group rsi jobs {group_id}",
+            len(ids),
+            "jobs",
+            action="processed",
+            show_eta=True,
+        )
+        report_progress.start()
+
+        def download_job_report(api, job_id: str):
+            try:
+                dest = root / group_id / f"{job_id}_rsi.md"
+                if skip_existing and dest.exists():
+                    output.log(f"skip existing {dest}")
+                    return str(dest)
+                return _write_markdown(
+                    _markdown(_fetch_job(api, job_id), "report_md"),
+                    dest,
+                    skip_existing,
+                )
+            finally:
+                report_progress.advance(job_id)
+
+        reports = fetch_job_resources(
+            rt.client,
+            ids,
+            download_job_report,
+            rt.concurrency,
+        )
+        written.extend(dest for dest in reports.items if dest)
+        failures.extend(reports.failures)
+        raise_for_failures("rsi reports", failures)
+        return written
+
+    def group_worker(group_id: str):
+        try:
+            return download_group_reports(group_id)
+        finally:
+            group_progress.advance(group_id)
+
+    if jobs:
+        job_progress = output.ProgressCounter(
+            "job rsi downloads", len(jobs), "jobs", action="processed"
+        )
+        job_progress.start()
+        res = safe_run_concurrent(jobs, job_worker, rt.concurrency)
+    else:
+        group_progress = output.ProgressCounter(
+            "group rsi downloads", len(groups), "groups", action="processed"
+        )
+        group_progress.start()
+        res = safe_run_concurrent(groups, group_worker, 1)
+    report_and_exit(res, emit=False)
+
+
+@app.command("trigger")
+def trigger(
+    ctx: typer.Context,
+    group: Optional[list[str]] = GroupOption,
+    job: Optional[list[str]] = JobOption,
+    include_passed: bool = typer.Option(
+        False,
+        "--include-passed",
+        help="Include passed jobs when triggering a Group RSI report.",
+    ),
+) -> None:
+    """Trigger Job or Group RSI report generation."""
+    rt = get_runtime(ctx)
+    groups = expand(group)
+    jobs = expand(job)
+    _guard(groups, jobs)
+
+    if jobs:
+        res = safe_run_concurrent(
+            jobs, lambda jid: _trigger_job(rt.client, jid), rt.concurrency
+        )
+    else:
+        res = safe_run_concurrent(
+            groups,
+            lambda gid: _trigger_group(rt.client, gid, include_passed),
+            rt.concurrency,
+        )
+    report_and_exit(res, emit=True, pretty=rt.pretty)