ap-client 0.6.1.dev9__tar.gz → 0.6.1.dev11__tar.gz

{ap_client-0.6.1.dev9 → ap_client-0.6.1.dev11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: ap-client
-Version: 0.6.1.dev9
+Version: 0.6.1.dev11
 Summary: Agent Platform API Client & CLI
 Requires-Python: >=3.10
 Requires-Dist: irepo-artifact==0.2.2

{ap_client-0.6.1.dev9 → ap_client-0.6.1.dev11}/ap_client/fs_commands.py

@@ -3,8 +3,7 @@
 命令与后端映射:
 
 * ``fs ls`` → ``repo.list_user_data(uid, subpath=...)``
-* ``fs get`` → ``repo.download_user_data(uid, remote_subpath, local_path)``
-* ``fs put`` → ``repo.upload_user_data`` / ``repo.upload_user_data_dir``
+* ``fs get/put`` → CLI 并发传输层,沿用 SDK 精确对象 STS 与网络路由
 * ``fs access ls`` → ``repo.list_user_data_acl(uid, subpath=...)``
 * ``fs access grant`` → ``repo.grant_user_data(..., subpath=...)``
 * ``fs access revoke`` → ``repo.revoke_user_data(..., subpath=...)``
@@ -310,9 +309,18 @@ def fs_get(
     target: Optional[str] = typer.Option(
         None, "--target", help="Local target path (default: the remote file name in CWD)"
     ),
+    concurrency: int = typer.Option(
+        8, "--concurrency", "-j", min=1, max=64, help="Maximum concurrent transfer requests"
+    ),
+    progress: Optional[bool] = typer.Option(
+        None,
+        "--progress/--no-progress",
+        help="Show transfer progress on stderr (default: terminal only)",
+    ),
     output_format: str = typer.Option(None, "--format", help=_FORMAT_HELP),
 ):
     """Download one file from a user storage space."""
+    from .fs_transfer import UserDataTransfer
     from .irepo_sdk import sdk_error_exit
 
     output_format = _fmt(output_format)
@@ -325,7 +333,9 @@ def fs_get(
     try:
         if local.parent and str(local.parent) not in ("", "."):
             local.parent.mkdir(parents=True, exist_ok=True)
-        written = repo.download_user_data(uid, remote, str(local))
+        written = UserDataTransfer(repo, concurrency=concurrency, progress=progress).get(
+            uid, remote, local
+        )
     except Exception as exc:  # noqa: BLE001
         sdk_error_exit(exc, what=f"user storage {uid} {remote}", stage="download")
 
@@ -344,9 +354,18 @@ def fs_put(
         None, "--source", help="Local source path (default: <filepath> relative to CWD)"
     ),
     overwrite: bool = typer.Option(False, "--overwrite", help="Allow overwriting remote objects"),
+    concurrency: int = typer.Option(
+        8, "--concurrency", "-j", min=1, max=64, help="Maximum concurrent transfer requests"
+    ),
+    progress: Optional[bool] = typer.Option(
+        None,
+        "--progress/--no-progress",
+        help="Show transfer progress on stderr (default: terminal only)",
+    ),
     output_format: str = typer.Option(None, "--format", help=_FORMAT_HELP),
 ):
     """Upload a local file (or directory) into a user storage space."""
+    from .fs_transfer import UserDataTransfer
     from .irepo_sdk import sdk_error_exit
 
     output_format = _fmt(output_format)
@@ -363,10 +382,9 @@ def fs_put(
     remote = _remote_subpath(filepath, directory)
     is_dir = local.is_dir()
     try:
-        if is_dir:
-            outcome: Any = repo.upload_user_data_dir(uid, str(local), remote, overwrite=overwrite)
-        else:
-            outcome = repo.upload_user_data(uid, str(local), remote, overwrite=overwrite)
+        outcome: Any = UserDataTransfer(repo, concurrency=concurrency, progress=progress).put(
+            uid, local, remote, overwrite=overwrite
+        )
     except Exception as exc:  # noqa: BLE001
         sdk_error_exit(exc, what=f"user storage {uid} {remote}", stage="upload")
 

ap_client-0.6.1.dev11/ap_client/fs_transfer.py

@@ -0,0 +1,245 @@
+"""Bounded user-storage transfers through the SDK's authorization and routing."""
+
+from __future__ import annotations
+
+import os
+import tempfile
+import threading
+from concurrent.futures import FIRST_COMPLETED, ThreadPoolExecutor, wait
+from pathlib import Path
+
+from rich.console import Console
+from rich.progress import (
+    BarColumn,
+    DownloadColumn,
+    Progress,
+    TaskProgressColumn,
+    TextColumn,
+    TimeRemainingColumn,
+    TransferSpeedColumn,
+)
+
+from .ossutil_io import OssOperationError, _WriteBucket
+
+PART_SIZE = 8 * 1024 * 1024
+
+
+class TransferProgress:
+    def __init__(self, description: str, enabled: bool | None):
+        console = Console(stderr=True, markup=False, highlight=False)
+        self.display = Progress(
+            TextColumn("{task.description}"),
+            BarColumn(),
+            TaskProgressColumn(),
+            DownloadColumn(),
+            TransferSpeedColumn(),
+            TimeRemainingColumn(),
+            TextColumn("files {task.fields[files_done]}/{task.fields[files_total]}"),
+            console=console,
+            disable=not (console.is_terminal if enabled is None else enabled),
+        )
+        self.description = description
+        self.lock = threading.Lock()
+        self.sizes: dict[str, int] = {}
+        self.done: dict[str, int] = {}
+        self.files_done = 0
+        self.total_bytes = 0
+        self.completed_bytes = 0
+
+    def start(self, sizes: dict[str, int], *, unknown: bool = False):
+        self.sizes = dict(sizes)
+        self.total_bytes = sum(sizes.values())
+        self.known_total = not unknown
+        self.task = self.display.add_task(
+            self.description,
+            total=None if unknown else self.total_bytes,
+            files_done=0,
+            files_total=len(sizes),
+        )
+        return self
+
+    def __enter__(self):
+        self.display.start()
+        return self
+
+    def __exit__(self, kind, value, traceback):
+        if kind is not None:
+            self.display.update(self.task, description=self.description + " failed")
+        self.display.stop()
+
+    def callback(self, key: str):
+        def update(consumed, total):
+            with self.lock:
+                size = max(0, int(total or 0))
+                self.known_total = True
+                self.total_bytes += size - self.sizes[key]
+                self.sizes[key] = size
+                # Multipart callbacks and retries can arrive out of order.
+                old = self.done.get(key, 0)
+                self.done[key] = min(size, max(old, consumed))
+                self.completed_bytes += self.done[key] - old
+                self._render()
+
+        return update
+
+    def finish(self, key: str, *, skipped: bool = False):
+        with self.lock:
+            self.known_total = True
+            if skipped:
+                # Skipped bytes were not transferred; do not inflate the speed.
+                self.total_bytes -= self.sizes[key]
+                self.sizes[key] = 0
+            old = self.done.get(key, 0)
+            self.done[key] = self.sizes[key]
+            self.completed_bytes += self.done[key] - old
+            self.files_done += 1
+            self._render()
+
+    def _render(self):
+        self.display.update(
+            self.task,
+            completed=self.completed_bytes,
+            total=self.total_bytes if self.known_total else None,
+            files_done=self.files_done,
+        )
+
+
+class UserDataTransfer:
+    def __init__(self, repo, *, concurrency: int = 8, progress: bool | None = None):
+        if not 1 <= concurrency <= 64:
+            raise ValueError("concurrency must be between 1 and 64")
+        self.repo, self.concurrency, self.progress = repo, concurrency, progress
+        self.content = getattr(repo, "_content", None)
+        if not callable(getattr(self.content, "_execute_oss", None)):
+            raise OssOperationError("User storage transfer requires a compatible instance-repo SDK")
+        from instance_repo.store.oss import _import_oss2
+
+        self.oss = _import_oss2()
+
+    def _key(self, uid: str, path: str) -> str:
+        if not str(uid).strip():
+            raise ValueError("uid must not be empty")
+        return self.repo._user_data_prefix(uid) + self.repo._clean_user_data_subpath(path)
+
+    def put(self, uid: str, source: Path, remote: str, *, overwrite: bool = False):
+        from instance_repo._routing import _OperationIntent
+
+        source = Path(source)
+        base = self._key(uid, remote)
+        directory = source.is_dir()
+        files = []
+        if directory:
+            for root, dirs, names in os.walk(source):
+                dirs.sort()
+                for name in sorted(names):
+                    local = Path(root) / name
+                    key = self._key(
+                        uid, remote.rstrip("/") + "/" + local.relative_to(source).as_posix()
+                    )
+                    files.append((local, key))
+        else:
+            files.append((source, base))
+        if any(not local.is_file() for local, _ in files):
+            raise ValueError("Upload sources must be regular files")
+        sizes = {key: local.stat().st_size for local, key in files}
+        workers = min(self.concurrency, max(1, len(files)))
+        part_threads = max(1, self.concurrency // workers)
+        report = TransferProgress("Uploading", self.progress).start(sizes)
+
+        def transfer(item):
+            local, key = item
+            write_started = False
+            # Keep the SDK's resume records across failures and later invocations.
+            store = self.oss.ResumableStore()
+
+            def upload(bucket):
+                nonlocal write_started
+                if not overwrite and not write_started and bucket.object_exists(key):
+                    return {"key": key, "skipped": True}
+                write_started = True
+                self.oss.resumable_upload(
+                    _WriteBucket(bucket, overwrite),
+                    key,
+                    str(local),
+                    store=store,
+                    num_threads=part_threads,
+                    part_size=PART_SIZE,
+                    multipart_threshold=PART_SIZE,
+                    headers={} if overwrite else {"x-oss-forbid-overwrite": "true"},
+                    progress_callback=report.callback(key),
+                )
+                return {"key": key, "skipped": False}
+
+            # Exact object scope retains subdirectory/file ACL behavior.
+            result = self.content._execute_oss(_OperationIntent.NON_DOWNLOAD, key, None, upload)
+            report.finish(key, skipped=result["skipped"])
+            return result
+
+        with report:
+            if not files:
+                return []
+            if workers == 1:
+                outcomes = [transfer(item) for item in files]
+            else:
+                outcomes = self._parallel(files, transfer, workers)
+        return outcomes if directory else outcomes[0]
+
+    @staticmethod
+    def _parallel(files, transfer, workers):
+        outcomes = [None] * len(files)
+        items = iter(enumerate(files))
+        with ThreadPoolExecutor(max_workers=workers) as pool:
+            pending = {
+                pool.submit(transfer, item): index
+                for index, item in [next(items) for _ in range(workers)]
+            }
+            try:
+                while pending:
+                    done, _ = wait(pending, return_when=FIRST_COMPLETED)
+                    for future in done:
+                        outcomes[pending.pop(future)] = future.result()
+                    for _ in done:
+                        entry = next(items, None)
+                        if entry is not None:
+                            index, item = entry
+                            pending[pool.submit(transfer, item)] = index
+            except BaseException:
+                for future in pending:
+                    future.cancel()
+                raise
+        return outcomes
+
+    def get(self, uid: str, remote: str, target: Path) -> str:
+        from instance_repo._routing import _OperationIntent
+
+        key = self._key(uid, remote)
+        target = Path(target)
+        if target.exists() and not target.is_file():
+            raise ValueError("Download target must be a regular file")
+        target.parent.mkdir(parents=True, exist_ok=True)
+        report = TransferProgress("Downloading", self.progress).start({key: 0}, unknown=True)
+        # Keep an existing destination intact until every range has completed.
+        with (
+            report,
+            tempfile.TemporaryDirectory(prefix=".ap-fs-get-", dir=target.parent) as scratch,
+        ):
+            partial = Path(scratch) / "data"
+            store = self.oss.ResumableDownloadStore(root=scratch)
+
+            def download(bucket):
+                self.oss.resumable_download(
+                    bucket,
+                    key,
+                    str(partial),
+                    store=store,
+                    num_threads=self.concurrency,
+                    part_size=PART_SIZE,
+                    multiget_threshold=PART_SIZE,
+                    progress_callback=report.callback(key),
+                )
+
+            self.content._execute_oss(_OperationIntent.DOWNLOAD, key, None, download)
+            partial.chmod(0o600)
+            os.replace(partial, target)
+            report.finish(key)
+        return str(target)

{ap_client-0.6.1.dev9 → ap_client-0.6.1.dev11}/ap_client/irepo_sdk.py

@@ -955,7 +955,9 @@ def load_repo(
             _Transport = None
         if _Transport is not None:
             kwargs["transport"] = _Transport(
-                cfg.base_url, cfg.token_key or "", env=api_env,
+                cfg.base_url,
+                cfg.token_key or "",
+                env=api_env,
                 timeout=request_timeout,
             )
 
@@ -968,9 +970,45 @@ def load_repo(
     repo = _Repo(**kwargs)
     _bind_dataset_profile(repo)
     _install_oss_probe_cache(repo, verbose=cfg.verbose)
+    _install_user_data_directory_listing(repo)
     return repo
 
 
+def _install_user_data_directory_listing(repo: Any) -> bool:
+    """Keep the directory boundary lost by instance-repo 1.4.0 list_user_data.
+
+    Use the SDK's path validation and content store so STS authorization,
+    endpoint selection and pagination retain their normal behavior. Normalize
+    here until the SDK dependency includes its directory-prefix fix.
+    """
+    content = getattr(repo, "_content", None)
+    prefix_for = getattr(repo, "_user_data_prefix", None)
+    clean = getattr(repo, "_clean_user_data_subpath", None)
+    if not all(
+        callable(fn)
+        for fn in (
+            getattr(repo, "list_user_data", None),
+            prefix_for,
+            clean,
+            getattr(content, "list_objects", None),
+        )
+    ):
+        return False
+
+    def list_user_data(uid: str, subpath: str = "") -> list[str]:
+        from instance_repo.errors import SchemaError
+
+        if not str(uid).strip():
+            raise SchemaError("uid must not be empty")
+        prefix = prefix_for(uid)
+        if str(subpath).strip():
+            prefix += clean(subpath) + "/"
+        return content.list_objects(prefix)
+
+    repo.list_user_data = list_user_data
+    return True
+
+
 def _bind_dataset_profile(repo: Any) -> None:
     """Supply the configured context missing from SDK 1.3.5 dataset discovery.
 

{ap_client-0.6.1.dev9 → ap_client-0.6.1.dev11}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "ap-client"
-version = "0.6.1.dev9"
+version = "0.6.1.dev11"
 description = "Agent Platform API Client & CLI"
 readme = { text = "A lightweight Python SDK and command line interface for Agent Platform. It provides helpers for configuring API access and managing templates, datasets, jobs, and groups.", content-type = "text/markdown" }
 requires-python = ">=3.10"