repo_path_benchmark.py

import os

from funcy import log_durations
from rich.console import Console

import dvc.api
from dvc._debug import profile
from dvc.external_repo import clean_repos
from dvc.fs.pool import close_pools

repo = "https://github.com/iterative/example-get-started"
status = Console().status


with status("old api"), log_durations(print, "old api"), profile("old_api.prof"):
    dvc.api.read("model.pkl", repo=repo, mode="rb")
    dvc.api.read("scores.json", repo=repo, encoding="utf-8")
    dvc.api.get_url(os.path.join("data", "data.xml"), repo=repo)
    dvc.api.get_url(os.path.join("data", "features", "test.pkl"), repo=repo)
    dvc.api.get_url(os.path.join("data", "features", "train.pkl"), repo=repo)


clean_repos()
close_pools()


with status("files api"), log_durations(print, "files api"), profile("repo_path.prof"):
    with dvc.api.files(repo=repo) as root:
        pkl_file = root / "model.pkl"
        pkl_file.read_bytes()

        scores_file = root / "scores.json"
        scores_file.read_text(encoding="utf-8")

        data_dir = root / "data"
        (data_dir / "data.xml").url()

        features = data_dir / "features"

        test_pkl = features / "test.pkl"
        test_pkl.url()

        train_pkl = features / "train.pkl"
        train_pkl.url()