Commit c295ff64 authored by Sébastien Tourbier's avatar Sébastien Tourbier
Browse files

feat: add dataset and project commands / methods to release a version

parent cc31fc11

datahipy/bids/dataset.py

+2 −1
Original line number Diff line number Diff line
@@ -22,7 +22,6 @@ from datahipy.bids.validation import ( 
    get_bids_validator_output_info,

)

from datahipy.bids.version import determine_bids_schema_version

from datahipy.utils.versioning import get_latest_tag



# Set the number of threads to use for parallel processing

# Modify this value if you want to use more or less threads or
@@ -241,6 +240,8 @@ def get_bidsdataset_content(bids_dir=None): 
    dataset_desc : dict

        Dictionary storing dataset information indexed by the HIP platform.

    """

    # Import here to avoid circular import

    from datahipy.utils.versioning import get_latest_tag

    # Load the dataset_description.json as initial dictionary-based description

    with open(os.path.join(bids_dir, "dataset_description.json"), "r") as f:

        dataset_desc = json.load(f)

datahipy/cli/run.py

+10 −1
Original line number Diff line number Diff line
@@ -9,7 +9,7 @@ from datahipy.bids.dataset import get_all_datasets_content 
from datahipy.handlers.dataset import DatasetHandler

from datahipy.handlers.participants import ParticipantHandler

from datahipy.handlers.project import create_project, import_subject, import_document

from datahipy.utils.versioning import create_tag, get_tags, checkout_tag

from datahipy.utils.versioning import create_tag, get_tags, checkout_tag, release_version



VALID_COMMANDS = [

    "dataset.create",
@@ -18,6 +18,7 @@ VALID_COMMANDS = [ 
    "dataset.get_tags",

    "dataset.checkout_tag",

    "datasets.get",

    "dataset.release_version",

    "sub.get",

    "sub.import",

    "sub.edit.clinical",
@@ -29,6 +30,7 @@ VALID_COMMANDS = [ 
    "project.create_tag",

    "project.get_tags",

    "project.checkout_tag",

    "project.release_version",

]
@@ -69,6 +71,7 @@ def main(): 
    dhdl = DatasetHandler(dataset_path=dataset_path)

    phdl = ParticipantHandler(dataset_path=dataset_path, input_path=input_path)



    # Dataset commands

    if command == "dataset.create":

        return dhdl.dataset_create(input_data=input_data)

    if command == "dataset.get":
@@ -84,6 +87,9 @@ def main(): 
            input_data=input_data,

            output_file=output_file,

        )

    if command == "dataset.release_version":

        return release_version(input_data=input_data, output_file=output_file)

    # Dataset subject / participant-level commands

    if command == "sub.import":

        return phdl.sub_import(input_data=input_data)

    if command == "sub.edit.clinical":
@@ -94,6 +100,7 @@ def main(): 
        return phdl.sub_delete(input_data=input_data)

    if command == "sub.delete.file":

        return phdl.sub_delete_file(input_data=input_data)

    # Project commands

    if command == "project.create":

        create_project(input_data=input_data, output_file=output_file)

    if command == "project.sub.import":
@@ -106,6 +113,8 @@ def main(): 
        return get_tags(input_data=input_data, output_file=output_file)

    if command == "project.checkout_tag":

        return checkout_tag(input_data=input_data)

    if command == "project.release_version":

        return release_version(input_data=input_data, output_file=output_file)





if __name__ == "__main__":

datahipy/utils/versioning.py

+72 −11
Original line number Diff line number Diff line
@@ -12,7 +12,6 @@ from datalad.support.gitrepo import GitRepo 


from datahipy.bids.version import create_bids_changes_tag_entry, update_bids_changes





TAG_EXCEPTIONS = ["master", "main", "HEAD"]
@@ -58,7 +57,7 @@ def create_tag(input_data): 
    Parameters

    ----------

    input_data : dict

        Dictionary containing the input data for the command

        Dictionary or JSON file containing the input data for the command

        in the format::



            {
@@ -69,10 +68,12 @@ def create_tag(input_data): 
            }

    """

    # Load input data

    if isinstance(input_data, str) or isinstance(input_data, os.PathLike):

        with open(input_data, "r") as f:

            input_data = json.load(f)

    elif isinstance(input_data, dict):

        pass

    print(f"Create tag {input_data['tag']} for dataset {input_data['path']}...")



    # Check if the tag is valid

    if not validate_tag(input_data["tag"], discard_exceptions=True):

        raise ValueError(
@@ -101,7 +102,7 @@ def create_tag(input_data): 
        bids_dir=bids_dir,

        changes_tag_entry=changes_tag_entry,

    )

    # Create a tag on the dataset

    # Create a tag on the BIDS dataset or on the nested BIDS dataset of a project

    save_params = {

        "dataset": input_data["path"],

        "message": "".join(changes_tag_entry),
@@ -137,17 +138,14 @@ def get_tags(input_data, output_file): 
    # Load input data

    with open(input_data, "r") as f:

        input_data = json.load(f)



    tags = [tag_dict["name"] for tag_dict in GitRepo(input_data["path"]).get_tags()]

    # Save the tags to a JSON file

    dict_tags = {

        "path": input_data["path"],

        "tags": tags,

    }



    print(f"Tags: {tags}")

    # Save the tags to a JSON file

    with open(output_file, "w") as f:

        json.dump(dict_tags, f)

        json.dump(dict_tags, f, indent=4)

    print(SUCCESS)
@@ -246,3 +244,66 @@ def increment_tag(tag, level): 
    elif level == "patch":

        tag[2] = str(int(tag[2]) + 1)

    return ".".join(tag)





def release_version(input_data, output_file):

    """Release a new version of a project / BIDS dataset managed by Git/Datalad.



    **Note:** The tag is created on a Datalad dataset, not on a specific file.

    If the dataset is a BIDS dataset, the tag will be created only on this dataset.

    If the dataset is a Collaborative Project, the tag will be created recursively on

      the project dataset and on the nested BIDS dataset of the project.



    Parameters

    ----------

    input_data : dict

        Dictionary containing the input data for the command

        in the format::



            {

                "path": "/path/to/dataset",

                "type": "bids",  # or "project"

                "level": "major",  # or "minor" or "patch"

                "changes_list": ["Change 1", "Change 2"]

            }

    

    output_file : str

        Absolute path to the output JSON file containing the dataset summary

        generated by :py:func:`datahipy.bids.dataset.get_bidsdataset_content`.

    """

    # Import here to avoid circular imports

    from datahipy.bids.dataset import get_bidsdataset_content

    # Load input data

    with open(input_data, "r") as f:

        input_data = json.load(f)

    print(f"Release version for dataset {input_data['path']}...")

    # Get the latest tag of the dataset

    latest_tag = get_latest_tag(input_data["path"])

    # Increment the latest tag by the specified level

    new_tag = increment_tag(latest_tag, input_data["level"])

    # Create a tag on the dataset

    create_tag(

        input_data={

            "path": input_data["path"],

            "type": input_data["type"],

            "tag": new_tag,

            "changes_list": input_data["changes_list"],

        }

    )

    # Get the BIDS directory of the dataset if it is

    # a project dataset or a BIDS dataset itself 

    bids_dir  = (

        input_data["path"]

        if input_data["type"] == "bids"

        else os.path.join(

            input_data["path"], "inputs", "bids-dataset"

        )

    )

    # Generate the dataset summary dictionary

    dataset_summary = get_bidsdataset_content(

        bids_dir=bids_dir

    )

    # Save the dataset summary to a JSON file

    with open(output_file, "w") as f:

        json.dump(dataset_summary, f, indent=4)

    print(SUCCESS)