Get path of repository folder

repo_root.py

import os

def isdrive(path):
    """Return `True` if path is the drive root

    Parameters
    ----------
    path : path-like
    """
    path = os.path.abspath(path)
    drive = os.path.splitdrive(path)

    # *nix drive root
    if drive == ('', '/'):
        return True

    # windows drive root
    elif drive[0].__contains__(':'):
        return True

    
def get_repo_path(filepath):
    """Detects root of repo given `.git` dir

    Parameters
    ----------
    filepath : pathlike
        `__file__` from the calling script

    Examples
    --------

    >>> if (__name__ == '__main__') and (__package__ is None):
    >>>     REPO_ROOT = get_repo_root(__file__)
    >>>     sys.path.append(REPO_ROOT + '/src')
    >>>     os.chdir(os.path.join(REPO_ROOT, 'src'))

    """

    if os.path.isdir(filepath):
        path_ = os.path.abspath(filepath)
    else:
        path_ = os.path.abspath(os.path.dirname(filepath))

    while not isdrive(path_):
        dirs_at_path = [os.path.basename(f.path) for f in os.scandir(path_) if f.is_dir()]
        if ".git" in dirs_at_path:
            # parent contains a .git folder
            return path_
        else:
            # get parent dir
            path_ = os.path.abspath(os.path.dirname(path_))

    else:
        logger.warning(
            f"File running from unrecognized/unmanaged path: {os.path.dirname(os.path.abspath(path_))}"
        )

        
def get_project_path(filepath):
    """Detects root of project directory from calling filepath assuming repo contains at least directories ['data', 'notebooks', 'src']
    
    Parameters
    ----------
    filepath : pathlike
        `__file__` from the calling script

    Examples
    --------

    >>> if (__name__ == '__main__') and (__package__ is None):
    >>>     PROJECT_ROOT = get_project_root(__file__)
    >>>     sys.path.append(PROJECT_ROOT + '/src')

    """

    # repo root can be identified by having this minimal set of subdirs
    repo_subdirs = {
        # 'artifacts',
        'data',
        # 'docs',
        'notebooks',
        # 'queries',
        'src',
        # 'tests',
        # 'writeup',
    }

    if os.path.isdir(filepath):
        path_ = os.path.abspath(filepath)
    else:
        path_ = os.path.abspath(os.path.dirname(filepath))
        
    while not isdrive(path_):
        parent = os.path.dirname(path_)
        if repo_subdirs.issubset(set(os.listdir(parent))):
            # parent contains the minimum folders we expect in our repos
            return parent
        else:
            path_ = parent

    else:
        logger.warning(
            f"File running from unrecognized/unmanaged path: {os.path.dirname(os.path.abspath(path_))}"
        )

src_path.py

from pathlib import Path

def src_path():
  """Return path to 'src/' directory."""
  return next(p for p in Path(__file__).parents if str(p).endswith("src"))

SRC_DIR = src_path()
PERSIST_DIR = SRC_DIR.parent
sys.path.insert(1, str(SRC_DIR))
# __package__ = "<PACKAGE_NAME>"