dsh-scnet 0.6.2 → 0.6.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -4,11 +4,18 @@
4
4
 
5
5
  DSH-SCNet (`dsh-scnet` on npm) is a community-maintained DSH bundle for operating Supercomputing Network (SCNet). It packages the source [`scnet-hpc`](https://github.com/lql341/scnet-hpc) Skill, profile-aware utilities, and bounded tools for SSH, Slurm, OpenAPI files, regional resources, and Notebook discovery.
6
6
 
7
- Current release: **0.6.2**
7
+ Current release: **0.6.3**
8
8
 
9
9
  > This is an independent community project. It is compatible with DeepSeek Harness but is not an official DeepSeek product and does not imply endorsement, partnership, or authorization by DeepSeek.
10
10
 
11
- ## What's new in 0.6.2
11
+ ## What's new in 0.6.3
12
+
13
+ - Added read-only account and resource summaries plus bounded `scnet_job_wait`.
14
+ - Submission queue preflight now fails closed and never silently selects a queue.
15
+ - Download paths can default to the remote filename; ordinary and chunked uploads return the same
16
+ directory/filename result shape.
17
+
18
+ ## 0.6.2 highlights
12
19
 
13
20
  - Completed jobs now resolve through the fast filtered history-list endpoint rather than timing
14
21
  out on the empty history-detail response.
@@ -67,11 +74,14 @@ See [Installation](./docs/installation.md) for npm, GitHub, and local-checkout w
67
74
  | `scnet_run_compute_probe` | Submit a minimal compute-node capability probe |
68
75
  | `scnet_status` | Read the saved backend, SSH profile, and OpenAPI configuration |
69
76
  | `scnet_openapi_regions` | List authorized OpenAPI regions without exposing tokens |
77
+ | `scnet_account_summary` | Read account status and balance |
78
+ | `scnet_resource_summary` | Summarize regional queues and resource limits |
70
79
  | `scnet_job_queues` | Query regional Slurm queues and live resource availability |
71
80
  | `scnet_job_list` | List active or historical jobs with compact normalized records |
72
81
  | `scnet_limits` | Read user and scheduler resource limits |
73
82
  | `scnet_submit_job` | Submit a job through OpenAPI or SSH; `dry_run=true` only previews the request |
74
83
  | `scnet_job_show` | Read one job's state, resources, and log paths |
84
+ | `scnet_job_wait` | Wait for a job to reach a terminal state with a bounded timeout |
75
85
  | `scnet_job_logs` | Read job logs by explicit path, or by `job_id` + `work_dir` |
76
86
  | `scnet_job_cancel` | Cancel a job; `dry_run=true` only previews the request |
77
87
  | `scnet_file_list` | List shared-storage files through the common file API |
@@ -120,7 +130,7 @@ synchronizes the skill and version into `dsh-scnet` and `codex-scnet-hpc`. Publi
120
130
  `dsh-scnet` npm package only after that downstream change is reviewed.
121
131
 
122
132
  For a synchronized release, run the package validation on `main`, then trigger the repository's
123
- `Publish DSH-SCNet` workflow with the exact value from `package.json` (currently `0.6.2`). The
133
+ `Publish DSH-SCNet` workflow with the exact value from `package.json` (currently `0.6.3`). The
124
134
  workflow verifies the package/Skill version match, runs validation, creates the tarball, publishes
125
135
  stable releases with the `latest` tag, and creates the matching Git tag. Do not bump only
126
136
  `package.json`: `skills/scnet-hpc/VERSION`, the package version, and both README release markers
package/README_CN.md CHANGED
@@ -4,10 +4,16 @@
4
4
 
5
5
  <p align="center">在 DeepSeek Harness 里操作超算互联网(SCNet):配置 SSH、管理 Slurm 作业、查询 OpenAPI 区域和文件、检查 Notebook 资源与实例,并保留海光 DCU/DTK 诊断能力。</p>
6
6
 
7
- 当前版本:**0.6.2**
7
+ 当前版本:**0.6.3**
8
8
 
9
9
  > 本项目是独立维护的社区项目,与 DeepSeek Harness 兼容,但不是 DeepSeek 官方产品,也不表示 DeepSeek 对本项目提供背书、合作或授权。
10
10
 
11
+ ## 0.6.3 更新
12
+
13
+ - 新增只读账户摘要、资源摘要和有界等待作业工具 `scnet_job_wait`。
14
+ - 提交前队列预检失败会停止提交,不再静默选择队列。
15
+ - 下载时可省略本地路径自动使用远端文件名;普通上传和分片上传返回统一结果结构。
16
+
11
17
  ## 0.6.2 更新
12
18
 
13
19
  - 已结束作业改用快速的历史列表过滤查询,不再卡在返回空数据的历史详情接口。
@@ -71,11 +77,14 @@ git 源、本地目录安装以及验证步骤见[安装说明](./docs/installat
71
77
  | scnet_run_compute_probe | 在计算节点运行最小能力探针 |
72
78
  | scnet_status | 只读查看 backend、SSH profile 和 OpenAPI 配置 |
73
79
  | scnet_openapi_regions | 列出授权 OpenAPI 区域,不暴露 token |
80
+ | scnet_account_summary | 查询账户状态和余额 |
81
+ | scnet_resource_summary | 汇总区域队列和资源限制 |
74
82
  | scnet_job_queues | 查询区域 Slurm 队列和实时空闲资源 |
75
83
  | scnet_job_list | 紧凑列出当前或历史作业 |
76
84
  | scnet_limits | 查询用户和调度器资源限制 |
77
85
  | scnet_submit_job | 通过 OpenAPI 或 SSH 提交作业;dry_run=true 只预览请求 |
78
86
  | scnet_job_show | 查询单个作业的状态、资源与日志路径 |
87
+ | scnet_job_wait | 在限定时间内等待作业进入终态 |
79
88
  | scnet_job_logs | 按显式 path,或按 job_id + work_dir 读取作业日志 |
80
89
  | scnet_job_cancel | 取消作业;dry_run=true 只预览请求 |
81
90
  | scnet_file_list | 通过通用文件 API 列出共享存储 |
@@ -132,7 +141,7 @@ git 源、本地目录安装以及验证步骤见[安装说明](./docs/installat
132
141
  才能发布对应版本的 `dsh-scnet` npm 包。
133
142
 
134
143
  下游同步合并到 `main` 后,先运行本地校验,再手动触发仓库的 `Publish DSH-SCNet` workflow,
135
- 输入 `package.json` 中的精确版本(当前为 `0.6.2`)。workflow 会校验 package 与 Skill
144
+ 输入 `package.json` 中的精确版本(当前为 `0.6.3`)。workflow 会校验 package 与 Skill
136
145
  版本一致,执行验证、生成 tarball、以稳定版 `latest` 标签发布,并创建对应 Git tag。
137
146
  不要只修改 `package.json`;`skills/scnet-hpc/VERSION`、package 版本和两份 README 的版本
138
147
  标记必须保持一致。
package/docs/testing.md CHANGED
@@ -39,11 +39,14 @@ Confirm that the composed configuration includes:
39
39
  | `scnet_generate_job` | Writes a profile-aware Slurm script in the current directory |
40
40
  | `scnet_status` | Returns redacted local SCNet configuration |
41
41
  | `scnet_openapi_regions` | Returns authorized regions without tokens |
42
+ | `scnet_account_summary` | Returns account status and balance |
43
+ | `scnet_resource_summary` | Returns queues and resource limits for a region |
42
44
  | `scnet_job_queues` | Returns live regional Slurm queues |
43
45
  | `scnet_job_list` | Lists active or historical jobs |
44
46
  | `scnet_limits` | Returns user and scheduler resource limits |
45
47
  | `scnet_submit_job` | Submits a job through OpenAPI or SSH; `dry_run=true` returns the pending request only |
46
48
  | `scnet_job_show` | Returns one job's state, resources, and log paths |
49
+ | `scnet_job_wait` | Waits for a job to reach a terminal state with a bounded timeout |
47
50
  | `scnet_job_logs` | Reads job logs by explicit path, or by `job_id` + `work_dir` |
48
51
  | `scnet_job_cancel` | Cancels one job; `dry_run=true` returns the pending request only |
49
52
  | `scnet_file_list` | Lists shared-storage files without modification |
@@ -60,12 +60,15 @@ DEEPSEEK_API_KEY: sk-你的key
60
60
  | 7 | 用 scnet_run_compute_probe 跑计算节点探针 | 返回 PROBE_ 结果 | 高(提交作业并等待) |
61
61
  | 8 | 用 scnet_status 查看当前配置 | 返回脱敏配置 | 只读 |
62
62
  | 9 | 用 scnet_openapi_regions 查看授权区域 | 返回区域且不含 token | 只读 |
63
+ | 9a | 用 scnet_account_summary 查看账户摘要 | 返回账户状态和余额 | 只读 |
64
+ | 9b | 用 scnet_resource_summary 查看资源摘要 | 返回队列和资源限制 | 只读 |
63
65
  | 10 | 用 scnet_job_queues 查看昆山队列 | 返回实时队列和空闲资源 | 只读 |
64
66
  | 11 | 用 scnet_job_list 列出当前或历史作业 | 返回紧凑作业记录 | 只读 |
65
67
  | 12 | 用 scnet_limits 查询资源限制 | 返回用户和调度器限制 | 只读 |
66
68
  | 13 | 用 scnet_submit_job(dry_run=true)预览一次提交 | 返回将要发送的请求,不真正提交 | 只读(dry_run) |
67
69
  | 14 | 用 scnet_submit_job 提交一个 5 秒小作业 | 返回 job_id、作业目录与日志路径 | 高(会消耗配额) |
68
70
  | 15 | 用 scnet_job_show 查该 job_id | 返回状态与 stdout/stderr 路径 | 只读 |
71
+ | 15a | 用 scnet_job_wait 等待作业终态 | 最多等待限定时间并返回状态 | 只读 |
69
72
  | 16 | 用 scnet_job_logs 读该作业日志 | 返回日志内容(同时验证 path 与 job_id+work_dir 两种模式结果一致) | 只读 |
70
73
  | 17 | 用 scnet_job_cancel 取消一个作业 | 建议先用 dry_run=true 预览 | 高(变更) |
71
74
  | 18 | 用 scnet_file_list 查看默认目录 | 返回共享存储文件 | 只读 |
package/index.mjs CHANGED
@@ -182,6 +182,16 @@ function errorTextFrom(res) {
182
182
  return matched ? matched[1].trim() : raw
183
183
  }
184
184
 
185
+ function errorCodeFrom(res) {
186
+ const raw = String(res.stdout || '').trim() || String(res.stderr || '').trim()
187
+ try {
188
+ const parsed = JSON.parse(raw)
189
+ return parsed && typeof parsed.error_code === 'string' ? parsed.error_code : ''
190
+ } catch {
191
+ return ''
192
+ }
193
+ }
194
+
185
195
  // 一次只给一条可执行的提示(不展示整套参数清单)
186
196
  function translateScnetError(text) {
187
197
  const value = String(text || '')
@@ -203,14 +213,23 @@ function translateScnetError(text) {
203
213
  return value
204
214
  }
205
215
 
216
+ function translateScnetFailure(res) {
217
+ const code = errorCodeFrom(res)
218
+ if (code === 'AUTHENTICATION_FAILED') return 'SCNet 认证失败:请检查 OpenAPI 配置或重新运行 setup。'
219
+ if (code === 'QUEUE_QUERY_FAILED') return '队列查询失败,已停止提交;请先检查区域、调度器和凭据。'
220
+ if (code === 'NO_QUEUE_AVAILABLE') return '当前资源条件下没有可用队列,请调整资源或选择其他区域。'
221
+ if (code === 'NETWORK_TIMEOUT') return 'SCNet 请求超时:请稍后重试;提交/取消类操作不要盲目重试。'
222
+ return translateScnetError(errorTextFrom(res))
223
+ }
224
+
206
225
  function scnetJobResult(res, action) {
207
226
  if (res.ok) return res.stdout.trim() || '{}'
208
- return `${action}失败(exit ${res.code}):\n${translateScnetError(errorTextFrom(res))}`
227
+ return `${action}失败(exit ${res.code}):\n${translateScnetFailure(res)}`
209
228
  }
210
229
 
211
230
  async function runScnetEnvelope(argv) {
212
231
  const res = await runScnet(argv)
213
- if (!res.ok) return { ok: false, error: translateScnetError(errorTextFrom(res)) }
232
+ if (!res.ok) return { ok: false, error: translateScnetFailure(res) }
214
233
  try {
215
234
  const parsed = JSON.parse(res.stdout)
216
235
  return { ok: true, envelope: parsed, data: parsed?.data }
@@ -236,22 +255,18 @@ function queueCandidates(queues, wantsAccelerator) {
236
255
  .sort((a, b) => Number(b?.free_nodes ?? 0) - Number(a?.free_nodes ?? 0))
237
256
  }
238
257
 
239
- // best-effort 队列预检:查得到就校验,查不到不阻塞提交
258
+ // 提交前队列预检:查询失败或空结果都必须阻止提交,不能把失败当成可提交。
240
259
  async function precheckQueue(target, args) {
241
260
  const check = await runScnetEnvelope([...target.argv, 'queues'])
242
- if (!check.ok) return { note: `队列预检已跳过(${check.error})` }
261
+ if (!check.ok) return { error: `队列查询失败,已停止提交:${check.error}` }
243
262
  const queues = Array.isArray(check.data) ? check.data : []
244
- if (queues.length === 0) return { note: '队列预检已跳过(该区域没有返回队列信息)' }
263
+ if (queues.length === 0) return { error: '队列查询成功但没有可用队列,已停止提交。' }
245
264
 
246
265
  const queue = String(args.queue ?? '').trim()
247
266
  const dcus = Number(String(args.dcus ?? '').trim() || 0)
248
267
 
249
268
  if (!queue) {
250
269
  const candidates = queueCandidates(queues, dcus > 0)
251
- if (candidates.length === 1) {
252
- const only = candidates[0]
253
- return { queue: String(only.partition), note: `已自动选择队列 ${only.partition}` }
254
- }
255
270
  if (candidates.length > 1) {
256
271
  const best = candidates[0]
257
272
  return {
@@ -260,7 +275,10 @@ async function precheckQueue(target, args) {
260
275
  `按你要的资源建议 ${best.partition}(空闲 ${best.free_nodes} 节点)。`,
261
276
  }
262
277
  }
263
- return {}
278
+ if (candidates.length === 1) {
279
+ return { error: `请确认 queue:当前唯一可用队列是 ${candidates[0].partition}(不会自动选择计费队列)。` }
280
+ }
281
+ return { error: '没有找到匹配当前资源需求的可用队列。' }
264
282
  }
265
283
 
266
284
  const hit = queues.find((item) => String(item?.partition) === queue)
@@ -348,6 +366,51 @@ export function apply(ctx) {
348
366
  }),
349
367
  )
350
368
 
369
+ ctx.tools.register(
370
+ defineTool({
371
+ name: 'scnet_account_summary',
372
+ description:
373
+ '只读查询当前 SCNet 账户摘要,包括用户名、账户状态、余额、默认区域和区域信息;不需要额外参数。',
374
+ parameters: {},
375
+ output: {
376
+ schema: { type: 'string' },
377
+ render: (_args, value) => textBlock(value),
378
+ },
379
+ async execute() {
380
+ return scnetResult(
381
+ await runScnet(['--backend', 'openapi', 'account']),
382
+ '查询账户摘要',
383
+ )
384
+ },
385
+ }),
386
+ )
387
+
388
+ ctx.tools.register(
389
+ defineTool({
390
+ name: 'scnet_resource_summary',
391
+ description:
392
+ '只读汇总目标区域的队列、空闲资源和用户资源限制;region 可省略,后端会使用默认区域。',
393
+ parameters: {
394
+ region: { type: 'string', description: '区域名称或 ID;省略时使用默认区域' },
395
+ scheduler_id: { type: 'string', description: '调度器 ID;多调度器时填写' },
396
+ },
397
+ output: {
398
+ schema: { type: 'string' },
399
+ render: (_args, value) => textBlock(value),
400
+ },
401
+ async execute(args) {
402
+ const argv = [
403
+ '--backend',
404
+ 'openapi',
405
+ ...regionArgs(args.region),
406
+ ...(args.scheduler_id ? ['--scheduler-id', String(args.scheduler_id)] : []),
407
+ 'resource-summary',
408
+ ]
409
+ return scnetResult(await runScnet(argv), '查询资源摘要')
410
+ },
411
+ }),
412
+ )
413
+
351
414
  ctx.tools.register(
352
415
  defineTool({
353
416
  name: 'scnet_job_queues',
@@ -602,6 +665,43 @@ export function apply(ctx) {
602
665
  }),
603
666
  )
604
667
 
668
+ ctx.tools.register(
669
+ defineTool({
670
+ name: 'scnet_job_wait',
671
+ description:
672
+ '只读等待一个作业进入终态。默认最多等待 300 秒、每 10 秒查询一次;不会提交、取消或自动重试作业。',
673
+ parameters: {
674
+ job_id: { type: 'string', required: true, description: '作业号' },
675
+ backend: { type: 'string', description: 'openapi(默认)或 ssh' },
676
+ region: { type: 'string', description: 'OpenAPI 区域;省略时使用默认区域' },
677
+ scheduler_id: { type: 'string', description: 'OpenAPI 调度器 ID;多调度器时填写' },
678
+ cluster: { type: 'string', description: 'SSH 集群短名;多个 profile 时必须指定' },
679
+ wait_timeout: { type: 'number', description: '最长等待秒数,默认 300,最大 600' },
680
+ interval: { type: 'number', description: '查询间隔秒数,默认 10,范围 1-60' },
681
+ },
682
+ timeoutMs: 660000,
683
+ output: {
684
+ schema: { type: 'string' },
685
+ render: (_args, value) => textBlock(value),
686
+ },
687
+ async execute(args) {
688
+ const jobId = String(args.job_id ?? '').trim()
689
+ if (!jobId) return MISSING_HINTS.job_id
690
+ const target = await resolveTarget(args)
691
+ if (target.error) return target.error
692
+ const argv = [...target.argv, 'wait', jobId]
693
+ for (const [flag, key] of [['--wait-timeout', 'wait_timeout'], ['--interval', 'interval']]) {
694
+ if (args[key] !== undefined) {
695
+ const checked = positiveInt(args[key], key)
696
+ if (checked?.error) return checked.error
697
+ argv.push(flag, checked.value)
698
+ }
699
+ }
700
+ return scnetJobResult(await runScnet(argv, 660000), '等待作业')
701
+ },
702
+ }),
703
+ )
704
+
605
705
  ctx.tools.register(
606
706
  defineTool({
607
707
  name: 'scnet_job_logs',
@@ -706,7 +806,7 @@ export function apply(ctx) {
706
806
  '在本地与 SCNet 区域共享存储之间传输一个文件。direction=upload 或 download;upload 的 remote_dir 是目录而不是文件名。默认不覆盖,执行前应确认区域和路径,dry_run=true 只预览。',
707
807
  parameters: {
708
808
  direction: { type: 'string', required: true, description: 'upload 或 download' },
709
- local_path: { type: 'string', required: true, description: '本地文件路径' },
809
+ local_path: { type: 'string', description: '本地文件路径;download 省略时使用远端文件名' },
710
810
  remote_dir: { type: 'string', description: 'upload 时必填:区域共享存储中的绝对目录;文件名由 local_path 保留' },
711
811
  remote_path: { type: 'string', description: 'download:区域共享存储中的绝对文件路径' },
712
812
  backend: { type: 'string', description: 'openapi(默认)或 ssh' },
@@ -727,7 +827,7 @@ export function apply(ctx) {
727
827
  const localPath = String(args.local_path ?? '').trim()
728
828
  const remoteDir = String(args.remote_dir ?? '').trim()
729
829
  const remotePath = String(args.remote_path ?? '').trim()
730
- if (!localPath) return '缺少必填参数 local_path。'
830
+ if (direction === 'upload' && !localPath) return '缺少必填参数 local_path。'
731
831
  if (direction === 'upload' && !remoteDir) return '缺少必填参数 remote_dir(目录,不是文件名)。'
732
832
  if (direction === 'download' && !remotePath) return '缺少必填参数 remote_path。'
733
833
  const target = await resolveTarget(args)
@@ -737,7 +837,8 @@ export function apply(ctx) {
737
837
  if (direction === 'upload') {
738
838
  argv.push('upload', localPath, remoteDir)
739
839
  } else {
740
- argv.push('download', remotePath, localPath)
840
+ argv.push('download', remotePath)
841
+ if (localPath) argv.push(localPath)
741
842
  }
742
843
  if (args.overwrite === true) argv.push('--cover')
743
844
  return scnetResult(
@@ -1067,6 +1168,7 @@ export const __testables = {
1067
1168
  MISSING_HINTS,
1068
1169
  SUBMIT_REQUIRED_ORDER,
1069
1170
  errorTextFrom,
1171
+ errorCodeFrom,
1070
1172
  firstMissingRequired,
1071
1173
  parseWalltimeSeconds,
1072
1174
  pickBackend,
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "dsh-scnet",
3
- "version": "0.6.2",
3
+ "version": "0.6.3",
4
4
  "description": "Community DeepSeek Harness bundle for SCNet SSH and OpenAPI jobs, files, Notebook resources, Slurm diagnostics, and Hygon DCU/DTK guidance.",
5
5
  "type": "module",
6
6
  "main": "./index.mjs",
@@ -1 +1 @@
1
- 0.6.2
1
+ 0.6.3
@@ -128,6 +128,20 @@ python3 scripts/scnet.py --backend openapi --region <id> \
128
128
  `job <job-id>` checks the realtime endpoint first and automatically falls back to the filtered
129
129
  history-list endpoint. The caller does not need to provide `acctTime`.
130
130
 
131
+ Account and resource summaries are read-only:
132
+
133
+ ```bash
134
+ python3 scripts/scnet.py --backend openapi --region <id> account
135
+ python3 scripts/scnet.py --backend openapi --region <id> resource-summary
136
+ ```
137
+
138
+ Wait for a job with a bounded timeout:
139
+
140
+ ```bash
141
+ python3 scripts/scnet.py --backend openapi --region <id> \
142
+ wait <job-id> --wait-timeout 300 --interval 10
143
+ ```
144
+
131
145
  Region tokens are cached under `${XDG_CACHE_HOME:-~/.cache}/scnet-hpc` when that location is
132
146
  writable. Cache locking or writes are best-effort; an unavailable cache never blocks an OpenAPI
133
147
  operation.
@@ -15,6 +15,7 @@ import shlex
15
15
  import shutil
16
16
  import subprocess
17
17
  import sys
18
+ import time
18
19
  from pathlib import Path
19
20
  from typing import Any, Mapping
20
21
 
@@ -120,9 +121,19 @@ def build_parser() -> argparse.ArgumentParser:
120
121
  limits = subparsers.add_parser("limits", help="show scheduler or user resource limits")
121
122
  limits.add_argument("--partition")
122
123
 
124
+ account = subparsers.add_parser("account", help="show account summary")
125
+ resource_summary = subparsers.add_parser(
126
+ "resource-summary", help="show region resources and limits"
127
+ )
128
+
123
129
  job = subparsers.add_parser("job", help="show one job")
124
130
  job.add_argument("job_id")
125
131
 
132
+ wait = subparsers.add_parser("wait", help="wait for one job to reach a terminal state")
133
+ wait.add_argument("job_id")
134
+ wait.add_argument("--wait-timeout", type=int, default=300)
135
+ wait.add_argument("--interval", type=int, default=10)
136
+
126
137
  jobs = subparsers.add_parser("jobs", help="list active or historical jobs")
127
138
  jobs.add_argument("--scope", choices=("active", "history"), default="active")
128
139
  jobs.add_argument("--limit", type=int, default=20)
@@ -192,7 +203,7 @@ def build_parser() -> argparse.ArgumentParser:
192
203
 
193
204
  download = subparsers.add_parser("download", help="download one file or folder")
194
205
  download.add_argument("remote_path")
195
- download.add_argument("local_path")
206
+ download.add_argument("local_path", nargs="?")
196
207
  download.add_argument("--cover", action="store_true")
197
208
 
198
209
  execute = subparsers.add_parser("exec", help="run an SSH command")
@@ -225,6 +236,68 @@ def compact_output(operation: str, data: Any) -> str:
225
236
  return json.dumps(data, ensure_ascii=False, indent=2, sort_keys=True)
226
237
 
227
238
 
239
+ def _job_terminal(state: Any) -> bool:
240
+ value = str(state or "").upper()
241
+ return value in {
242
+ "COMPLETED",
243
+ "FAILED",
244
+ "CANCELLED",
245
+ "TIMEOUT",
246
+ "NODE_ERROR",
247
+ "ABNORMAL",
248
+ "EXITING",
249
+ "EXITED",
250
+ "DONE",
251
+ "OUT_OF_MEMORY",
252
+ "NODE_FAIL",
253
+ }
254
+
255
+
256
+ def wait_for_job(backend, options: Mapping[str, Any]) -> dict[str, Any]:
257
+ timeout = int(options.get("wait_timeout") or 300)
258
+ interval = int(options.get("interval") or 10)
259
+ if timeout < 1 or timeout > 600:
260
+ raise BackendError("wait_timeout must be between 1 and 600 seconds")
261
+ if interval < 1 or interval > 60:
262
+ raise BackendError("interval must be between 1 and 60 seconds")
263
+ started = time.monotonic()
264
+ while True:
265
+ latest = backend.execute("job", options)
266
+ if isinstance(latest, dict) and _job_terminal(latest.get("state")):
267
+ latest["waited_seconds"] = round(time.monotonic() - started, 1)
268
+ latest["timed_out"] = False
269
+ return latest
270
+ elapsed = time.monotonic() - started
271
+ if elapsed >= timeout:
272
+ return {
273
+ "job": latest,
274
+ "waited_seconds": round(elapsed, 1),
275
+ "timed_out": True,
276
+ }
277
+ time.sleep(min(interval, timeout - elapsed))
278
+
279
+
280
+ def error_code_for(text: str) -> str:
281
+ value = str(text).lower()
282
+ if "credentials are not configured" in value or "token" in value and "expired" in value:
283
+ return "AUTHENTICATION_FAILED"
284
+ if "region" in value and ("not available" in value or "missing" in value):
285
+ return "REGION_NOT_FOUND"
286
+ if "scheduler" in value and ("not available" in value or "multiple" in value):
287
+ return "SCHEDULER_NOT_FOUND"
288
+ if "queue" in value and ("not found" in value or "no available" in value):
289
+ return "QUEUE_QUERY_FAILED"
290
+ if "job" in value and ("not found" in value or "no job record" in value):
291
+ return "JOB_NOT_FOUND"
292
+ if "already exists" in value:
293
+ return "FILE_TARGET_EXISTS"
294
+ if "does not exist" in value or "not found" in value and "file" in value:
295
+ return "FILE_NOT_FOUND"
296
+ if "timed out" in value or "timeout" in value:
297
+ return "NETWORK_TIMEOUT"
298
+ return "BACKEND_ERROR"
299
+
300
+
228
301
  def _ask(prompt: str, default: str = "") -> str:
229
302
  suffix = f" [{default}]" if default else ""
230
303
  answer = input(f"{prompt}{suffix}: ").strip()
@@ -1506,6 +1579,9 @@ def main(argv: list[str] | None = None) -> int:
1506
1579
  openapi_config = user_config.get("openapi", {})
1507
1580
  if isinstance(openapi_config, dict):
1508
1581
  apply_openapi_defaults(options, openapi_config)
1582
+ if args.operation == "download" and not options.get("local_path"):
1583
+ remote_path = str(options.get("remote_path") or "").rstrip("/")
1584
+ options["local_path"] = remote_path.rsplit("/", 1)[-1] or "download"
1509
1585
  if args.operation == "notebook":
1510
1586
  if backend_name != "openapi":
1511
1587
  raise BackendError(
@@ -1532,6 +1608,8 @@ def main(argv: list[str] | None = None) -> int:
1532
1608
  "name": backend.name,
1533
1609
  "capabilities": sorted(backend.capabilities),
1534
1610
  }
1611
+ elif args.operation == "wait":
1612
+ data = wait_for_job(backend, options)
1535
1613
  elif args.operation == "exec":
1536
1614
  options["command"] = " ".join(args.command).strip()
1537
1615
  data = backend.execute(args.operation, options)
@@ -1564,6 +1642,7 @@ def main(argv: list[str] | None = None) -> int:
1564
1642
  {
1565
1643
  "ok": False,
1566
1644
  "operation": args.operation,
1645
+ "error_code": error_code_for(str(exc)),
1567
1646
  "error": str(exc),
1568
1647
  },
1569
1648
  ensure_ascii=False,
@@ -92,6 +92,8 @@ class OpenAPIBackend(Backend):
92
92
  "clusters",
93
93
  "queues",
94
94
  "limits",
95
+ "account",
96
+ "resource-summary",
95
97
  "job",
96
98
  "jobs",
97
99
  "logs",
@@ -351,6 +353,37 @@ class OpenAPIBackend(Backend):
351
353
  result = {target: data.get(source) for source, target in key_map.items()}
352
354
  return {"data": result, "raw": data} if options.get("raw") else result
353
355
 
356
+ def op_account(self, options: Mapping[str, Any]) -> Any:
357
+ _, token, region = self._center(options)
358
+ url = self._env(
359
+ "SCNET_OPENAPI_USER_URL",
360
+ "https://www.scnet.cn/ac/openapi/v2/user",
361
+ )
362
+ data = self._json_request("GET", url, token=token)
363
+ if not isinstance(data, dict):
364
+ raise BackendError("user endpoint returned an unexpected data shape")
365
+ result = {
366
+ "user": data.get("userName"),
367
+ "account_name": data.get("accountName"),
368
+ "account_status": data.get("accountStatus"),
369
+ "balance": data.get("accountBalance"),
370
+ "computer_center": data.get("computerCenter"),
371
+ "region_id": region.get("clusterId"),
372
+ "region_name": region.get("clusterName"),
373
+ }
374
+ return {"data": result, "raw": data} if options.get("raw") else result
375
+
376
+ def op_resource_summary(self, options: Mapping[str, Any]) -> Any:
377
+ _, _, region = self._center(options)
378
+ queues = self.op_queues(options)
379
+ limits = self.op_limits(options)
380
+ return {
381
+ "region_id": region.get("clusterId"),
382
+ "region_name": region.get("clusterName"),
383
+ "queues": queues,
384
+ "limits": limits,
385
+ }
386
+
354
387
  def op_job(self, options: Mapping[str, Any]) -> Any:
355
388
  job_id = str(require_option(options, "job_id"))
356
389
  hpc_url, token, scheduler_id, _, _ = self._hpc_context(options)