Import only import one source

[AbdelsalamHaa]

System:
Windows 11 with WSL
python 3.8

##Issue:
importing the dataset imports only one source of the project, although it might have many sources.

Should be

The import function should import the full project not only one source

details
here is my tree/config.yml file
build_targets: project: parents: [] stages: - hash: '' kind: '' name: root params: {} type: project task_113: parents: [] stages: - hash: '' kind: '' name: root params: {} type: source - hash: '' kind: rename name: stage-1 params: regex: '|frame_|task_113_' type: transform - hash: '' kind: random_split name: stage-2 params: seed: null splits: - - train - 0.7 - - test - 0.3 type: transform task_117: parents: [] stages: - hash: '' kind: '' name: root params: {} type: source - hash: '' kind: rename name: stage-1 params: regex: '|frame_|task_117_' type: transform - hash: '' kind: random_split name: stage-2 params: seed: null splits: - - train - 0.7 - - test - 0.3 type: transform format_version: 2 sources: task_113: format: yolo hash: '' options: {} path: '' url: /home/iradar/aiv_pipeline/datasets/tuna/task_113 task_117: format: yolo hash: '' options: {} path: '' url: /home/iradar/aiv_pipeline/datasets/tuna/task_117

but only task 117 is imported

My code to import
`dataset_path = '/home/iradar/aiv_pipeline/datasets/tuna'

dataset = Dataset.import_from(dataset_path, 'yolo') `

[AbdelsalamHaa]

@vinnamkim
I just would like to add, when I used version 1.0.0, the import function worked well and everything seemed fine.

[vinnamkim]

Hi @AbdelsalamHaa,
Sorry for the late response. It takes time to find the root cause for this. Until we solve this thing, you can merge your datasets manually by code. For example,

from datumaro import Dataset, HLOps

dataset_1 = Dataset.import_from("/home/iradar/aiv_pipeline/datasets/tuna/task_113", 'yolo')
dataset_2 = Dataset.import_from("/home/iradar/aiv_pipeline/datasets/tuna/task_117", 'yolo')
dataset = HLOps.merge([dataset_1, dataset_2])

Thank you.

[wonjuleee]

Thank you @vinnamkim for the kind guidance of Datumaro. I will close this issue.