Example code for converting to- and from- pairtree and stubbytree directory structures for the HTRC Extracted Features dataset

extracted_features_path_converters.py

#!/usr/bin/env python3
# Note: depends on `pairtree` package (`pip install pairtree`)
import os
import pairtree.pairtree_path as ppath


def stubby_to_pairtree(path: str, ef_ext: str = '.json.bz2') -> str:
    assert path.endswith(ef_ext)
    d, f = os.path.split(path)
    assert len(d.split(os.sep)) >= 2
    lib_id, clean_volid = f[:-len(ef_ext)].split('.', 1)
    volid = ppath.id_decode(clean_volid)
    pairtree_root = os.path.join(os.sep.join(d.split(os.sep)[:-2]), lib_id, 'pairtree_root')
    pairtree_dir = ppath.id_to_dirpath(volid, pairtree_root)
    return os.path.join(pairtree_dir, clean_volid, f)


def pairtree_to_stubby(path: str, ef_ext: str = '.json.bz2') -> str:
    assert path.endswith(ef_ext)
    assert 'pairtree_root' in path
    d, f = os.path.split(path)
    lib_id, clean_volid = f[:-len(ef_ext)].split('.', 1)
    root = d[:d.find(lib_id + os.sep + 'pairtree_root' + os.sep)]
    stubby_path = os.path.join(lib_id, clean_volid[::3])
    return os.path.join(root, stubby_path, f)

s = 'loc/a+30795/loc.ark+=13960=t70v90g5f.json.bz2'
p = 'loc/pairtree_root/ar/k+/=1/39/60/=t/70/v9/0g/5f/ark+=13960=t70v90g5f/loc.ark+=13960=t70v90g5f.json.bz2'

pair = stubby_to_pairtree(s)
print(pair)
assert pair == p

stubby = pairtree_to_stubby(p)
print(stubby)
assert stubby == s