onescience_run_manifest.yaml

onescience_manifest_version: "0.1"
resource_type: dataset

resource:
  id: OneScience/cfdbench
  name: cfdbench
  domain: cfd
  domain_tags: [cfd, fluid_dynamics, benchmark]
  task: cfd_surrogate_benchmark_dataset
  task_tags: [train_data, eval_data, inference_input, dataset_validation]
  modalities: [gridded_field, velocity_field, case_parameters]
  input_formats: [npy, json, png]
  output_formats: [npy, json, png]
  summary: CFDBench 插值数据集，包含 cavity、tube、dam、cylinder 四类 CFD 问题及 bc/geo/prop 子集。

platform_resource:
  primary:
    platform: modelscope
    repo_id: OneScience/cfdbench
    repo_type: dataset
    url: https://modelscope.cn/datasets/OneScience/cfdbench
    revision: main
    readme_path: README.md
    manifest_path: onescience_run_manifest.yaml
  mirrors: []
  access:
    visibility: public
    license: unknown

website_integration:
  enabled: true
  click_target:
    platform: modelscope
    resource_url: https://modelscope.cn/datasets/OneScience/cfdbench
  llm_handoff:
    readme_required: true
    manifest_required: true
    download_readme_first: true
    resolve_related_models: true
    default_run_goal: dataset_validation
    cwd_note: 如果使用 modelscope download --cache_dir 下载数据集，请先 cd 到实际下载后的数据集仓库根目录；模型侧将 ONESCIENCE_CFDBENCH_DATA_DIR 指向本仓库 data 目录。

runtime:
  enabled: true
  onescience_domain: cfd
  min_onescience_version: null
  supported_execution: [local_dataset_validation, model_training_input, model_inference_input]
  environment:
    exported_env:
      ONESCIENCE_CFDBENCH_DATA_DIR: <dataset_repo_root>/data
  dependencies:
    python: ">=3.10"
    python_packages: [numpy, pyyaml]

onescience:
  repo: https://gitee.com/onescience-ai/onescience
  official_links:
    gitee:
      doc: https://gitee.com/onescience-ai/onescience-doc
      onescience: https://gitee.com/onescience-ai/onescience
      skills: https://gitee.com/onescience-ai/oneskills
    github:
      doc: https://github.com/onescience-ai/OneScience-doc
      onescience: https://github.com/onescience-ai/OneScience
      skills: https://github.com/onescience-ai/oneskills
  install:
    required_by_default: false
    command: bash install.sh cfd
  source_paths:
    - onescience/src/onescience/datapipes/cfd/cfdbench.py
  compatibility:
    examples_path: onescience/examples/cfd/CFDBench
    status: examples_compatible
  datapipe: onescience.datapipes.cfd.CFDBenchDatapipe

runtime_package:
  kind: standard_runtime_package
  package_root: .
  standard_layout:
    workdir: .
    data_dir: data
    metadata_dir: metadata
    output_dir: validation_outputs
  apply_policy:
    mode: direct_use
    target: session_workdir
    overwrite: false
    protect_installed_onescience: true
  entry_files:
    - data/tube/prop/case0050/u.npy
    - data/tube/prop/case0050/v.npy
    - data/tube/prop/case0050/case.json
    - metadata/cfdbench_schema.json
    - scripts/validate_cfdbench_dataset.py
    - files_sha256.jsonl
  entrypoints:
    preflight: scripts/validate_cfdbench_dataset.py
    validate: scripts/validate_cfdbench_dataset.py
    inference: null
    train: null
    finetune: null
    evaluate: null
    visualize: null
    deploy: null

files:
  model_files: []
  dataset_files:
    - path: data/<problem>/<subset>/case*/case.json
      role: case_metadata
      description_zh: 每个 CFD case 的物理参数、边界条件和几何参数元数据。
      required: true
      required_for: [dataset_validation, train, inference, evaluate]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: data/<problem>/<subset>/case*/case.json
    - path: data/<problem>/<subset>/case*/u.npy
      role: velocity_u_field
      description_zh: 64x64 网格上的 u 速度时间序列数组。
      required: true
      required_for: [dataset_validation, train, inference, evaluate]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: data/<problem>/<subset>/case*/u.npy
    - path: data/<problem>/<subset>/case*/v.npy
      role: velocity_v_field
      description_zh: 64x64 网格上的 v 速度时间序列数组。
      required: true
      required_for: [dataset_validation, train, inference, evaluate]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: data/<problem>/<subset>/case*/v.npy
    - path: files_sha256.jsonl
      role: file_size_sha256_inventory
      description_zh: 整理后数据文件的大小和 SHA256 清单，用于完整性校验。
      required: true
      required_for: [dataset_validation]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: files_sha256.jsonl
    - path: data_integrity_summary.json
      role: source_to_standardized_integrity_summary
      description_zh: 原始数据与整理后数据的文件名、大小、SHA256 对照摘要。
      required: true
      required_for: [dataset_validation]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: data_integrity_summary.json
  config_files:
    - path: metadata/cfdbench_schema.json
      role: dataset_schema
      description_zh: 描述 CFDBench 目录结构、case 文件、数组格式、子集统计和默认模型子集。
      required: true
      required_for: [dataset_validation, train, inference]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: metadata/cfdbench_schema.json
  sample_files:
    - path: scripts/validate_cfdbench_dataset.py
      role: dataset_validation_entry
      description_zh: 数据集读取验证脚本，检查目录结构、case.json、u/v NPY shape、dtype、可读性和可选 SHA256。
      required: true
      required_for: [dataset_validation]
      source: {platform: modelscope, repo_id: OneScience/cfdbench, repo_type: dataset, download_method: command_ref, command_ref: commands.download.download_dataset}
      local_path: scripts/validate_cfdbench_dataset.py

relations:
  required_datasets: []
  optional_datasets: []
  compatible_models:
    - id: OneScience/CFDBench
      role: train_data
      required_for: [train, inference, evaluate, visualize]
      resource_ref:
        platform: modelscope
        repo_id: OneScience/CFDBench
        repo_type: model
        url: https://modelscope.cn/models/OneScience/CFDBench
        revision: main
        readme_path: README.md
        manifest_path: onescience_run_manifest.yaml
      expected_local_path: session_workdir/data
      expected_env: ONESCIENCE_CFDBENCH_DATA_DIR

run_matrix:
  scenarios:
    - name: validate_structure
      capability: dataset_validation
      default: true
      required_datasets:
        - id: OneScience/cfdbench
          role: train_data
          local_path: data
      required_model_files: []
      required_dataset_files:
        - data/<problem>/<subset>/case*/case.json
        - data/<problem>/<subset>/case*/u.npy
        - data/<problem>/<subset>/case*/v.npy
        - metadata/cfdbench_schema.json
      preconditions: [当前工作目录为数据集仓库根目录。]
      command_refs: [commands.preflight.validate_dataset]
      outputs:
        - "控制台输出 [OK] dataset validation completed"
    - name: validate_integrity_full_hash
      capability: dataset_validation
      default: false
      required_datasets:
        - id: OneScience/cfdbench
          role: train_data
          local_path: data
      required_model_files: []
      required_dataset_files: [files_sha256.jsonl, data/**]
      preconditions: [当前工作目录为数据集仓库根目录。]
      command_refs: [commands.preflight.validate_dataset_full_hash]
      outputs:
        - "控制台输出 checksum manifest verified in size+sha256 mode"
    - name: provide_to_cfdbench_model
      capability: train
      default: true
      required_datasets:
        - id: OneScience/cfdbench
          role: train_data
          local_path: data
      required_model_files: [OneScience/CFDBench:conf/cfdbench.yaml]
      required_dataset_files:
        - data/tube/prop/case*/u.npy
        - data/tube/bc/case*/u.npy
        - data/tube/geo/case*/u.npy
      preconditions:
        - 模型仓库 OneScience/CFDBench 已下载。
        - 设置 ONESCIENCE_CFDBENCH_DATA_DIR 为本仓库 data 目录。
      command_refs: [commands.prepare.export_dataset_env]
      outputs:
        - 模型侧 scripts/preflight_check.py 可以读取数据集。

capabilities:
  dataset_validation: true
  train_input: true
  inference_input: true
  evaluation_input: true
  visualization_input: true
  preflight: true
  inference: false
  train: false
  finetune: false
  evaluate: false
  visualize: false
  deploy: false

commands:
  download:
    - name: download_dataset
      command: modelscope download --dataset OneScience/cfdbench
      cwd: session_workdir
    - name: download_model
      command: modelscope download --model OneScience/CFDBench
      cwd: session_workdir
  prepare:
    - name: export_dataset_env
      command: export ONESCIENCE_CFDBENCH_DATA_DIR=/path/to/OneScience_cfdbench/data
      cwd: .
  preflight:
    - name: validate_dataset
      command: python scripts/validate_cfdbench_dataset.py
      cwd: .
    - name: validate_dataset_full_hash
      command: python scripts/validate_cfdbench_dataset.py --full-hash
      cwd: .
  inference: []
  train: []
  finetune: []
  evaluate: []
  visualize: []
  deploy: []

expected_outputs:
  - name: validate_dataset
    paths: []
    success_criteria:
      - "控制台输出 [OK] dataset validation completed"
  - name: validate_dataset_full_hash
    paths: [files_sha256.jsonl]
    success_criteria:
      - "控制台输出 checksum manifest verified in size+sha256 mode: 2681 files"
  - name: model_consumption
    paths: []
    success_criteria: [模型包 scripts/preflight_check.py 通过数据路径和 schema 检查。]

diagnostics:
  - name: missing_data_root
    symptom: dataset data root does not exist
    action: 确认当前目录为 OneScience/cfdbench 数据集仓库根目录，且 data 目录存在。
  - name: missing_case_file
    symptom: missing required case file
    action: 对照 files_sha256.jsonl 重新上传或重新下载缺失 case。
  - name: shape_mismatch
    symptom: u/v shape mismatch
    action: 确认同一 case 下 u.npy 和 v.npy 来自同一数据版本。
  - name: checksum_mismatch
    symptom: sha256 mismatch
    action: 删除损坏文件后重新下载 OneScience/cfdbench。
  - name: model_cannot_find_data
    symptom: 模型侧报告 ONESCIENCE_CFDBENCH_DATA_DIR is not set 或 Problem path not found
    action: 将 ONESCIENCE_CFDBENCH_DATA_DIR 设置为本仓库 data 目录，而不是仓库根目录。

domain_extension:
  cfd:
    dataset_family: CFDBench
    problems: [cavity, tube, dam, cylinder]
    subsets: [bc, geo, prop]
    file_count: 2681
    total_size_bytes: 16017183719
    grid_resolution: 64x64
    variables: [u_velocity, v_velocity]
    case_metadata: case.json