Use system independent paths for test comparisons

Uses pathlib's PurePath to represent every multi-component path
across the unit tests, allowing them to pass on both Linux/MacOS
and Windows

Author: rpschoenburg

oxen/tests/conftest.py

@@ -3,6 +3,7 @@
 
 import uuid
 import os
+from pathlib import PurePath
 
 from oxen import Repo, RemoteRepo
 
@@ -151,7 +152,9 @@ def house_prices_local_repo_fully_committed(house_prices_local_repo_no_commits):
 
 @pytest.fixture
 def empty_remote_repo():
-    repo_name = f"py-ox/test_repo_{str(uuid.uuid4())}"
+
+    repo_str = str(PurePath("py-ox", "test_repo"))
+    repo_name = f"{repo_str}_{str(uuid.uuid4())}"
     repo = RemoteRepo(repo_name, host=TEST_HOST, scheme=TEST_SCHEME)
     repo.create(empty=True)
     yield repo
@@ -162,7 +165,7 @@ def empty_remote_repo():
 def celeba_local_repo_one_image_committed(celeba_local_repo_no_commits):
     repo = celeba_local_repo_no_commits
 
-    image_file = "images/1.jpg"
+    image_file = str(PurePath("images", "1.jpg"))
     full_path = os.path.join(repo.path, image_file)
     repo.add(full_path)
     repo.commit("Adding first image")

oxen/tests/test_add.py

@@ -1,7 +1,7 @@
 import os
 
 from oxen import Repo
-
+from pathlib import PurePath
 
 def test_add(shared_datadir):
     repo_dir = os.path.join(shared_datadir, "CelebA")
@@ -15,8 +15,12 @@ def test_add(shared_datadir):
     added_files = staged_data.added_files()
     added_files.sort()
 
+    test_path = PurePath("annotations", "test.csv")
+    train_path = PurePath("annotations", "train.csv")
+    labels_path = PurePath("annotations", "labels.txt")
+
     assert set(added_files) == {
-        "annotations/test.csv",
-        "annotations/train.csv",
-        "annotations/labels.txt",
-    }
+        str(test_path),
+        str(train_path),
+        str(labels_path),
+    }

oxen/tests/test_commit.py

@@ -1,11 +1,12 @@
 import os
+from pathlib import PurePath
 
 
 def test_commit_one_file(celeba_local_repo_no_commits):
     repo = celeba_local_repo_no_commits
 
     # oxen add
-    image_file = "images/1.jpg"
+    image_file = str(PurePath("images", "1.jpg"))
     full_path = os.path.join(repo.path, image_file)
     repo.add(full_path)
 

oxen/tests/test_data_frame.py

@@ -1,18 +1,21 @@
 import os
 
 import pandas as pd
+from pathlib import PurePath
 
 from oxen import DataFrame, RemoteRepo, Workspace
 
 
 def test_data_frame_crud(celeba_remote_repo_fully_pushed):
     _, remote_repo = celeba_remote_repo_fully_pushed
 
-    new_row = {"file": "images/123456.png", "hair_color": "purple"}
+    image = str(PurePath("images", "123456.png"))
+    new_row = {"file": str(image), "hair_color": "purple"}
 
+    train_path = str(PurePath("annotations", "train.csv"))
     df = DataFrame(
         remote_repo.identifier,
-        "annotations/train.csv",
+        train_path,
         host="localhost:3000",
         scheme="http",
     )
@@ -41,11 +44,13 @@ def test_data_frame_crud(celeba_remote_repo_fully_pushed):
 def test_data_frame_commit(celeba_remote_repo_fully_pushed):
     _, remote_repo = celeba_remote_repo_fully_pushed
 
-    new_row = {"file": "images/123456.png", "hair_color": "purple"}
+    image = str(PurePath("images", "123456.png"))
+    new_row = {"file": image, "hair_color": "purple"}
 
+    train_path = str(PurePath("annotations", "train.csv"))
     df = DataFrame(
         remote_repo.identifier,
-        "annotations/train.csv",
+        train_path,
         host="localhost:3000",
         scheme="http",
     )
@@ -72,14 +77,16 @@ def test_remove_data_frame_row(
     workspace = Workspace(remote_repo, "main")
     print("Created workspace ", workspace)
 
-    new_row = {"file": "images/123456.png", "hair_color": "purple"}
+    image = str(PurePath("images", "123456.png"))
+    new_row = {"file": image, "hair_color": "purple"}
 
     assert len(workspace.status().added_files()) == 0
     workspace.add(file_path, "csvs")
     workspace.commit("add train.csv")
 
+    train_path = str(PurePath("csvs", "train.csv"))
     workspace = Workspace(remote_repo, "main")
-    df = DataFrame(workspace, "csvs/train.csv")
+    df = DataFrame(workspace, train_path)
     _width, og_height = df.size()
 
     row_id = df.insert_row(new_row)

oxen/tests/test_fsspec_backend.py

@@ -2,6 +2,7 @@
 import fsspec
 import pandas as pd
 
+from pathlib import PurePath
 from oxen import RemoteRepo
 
 
@@ -17,10 +18,13 @@ def test_fsspec_read_file(
     fs = fsspec.filesystem(
         "oxen", namespace=namespace, repo=repo_name, host=host, scheme=scheme
     )
-    with fs.open("images/1.jpg", mode="rb") as f:
+
+    image = str(PurePath("images", "1.jpg"))
+    with fs.open(image, mode="rb") as f:
         remote_image_file = f.read()
 
-    with open(os.path.join(shared_datadir, "CelebA/images/1.jpg"), "rb") as f:
+    celeb_path = str(PurePath("CelebA", "images", "1.jpg"))
+    with open(os.path.join(shared_datadir, celeb_path), "rb") as f:
         assert remote_image_file == f.read()
 
 

oxen/tests/test_remote_repo.py

@@ -1,5 +1,5 @@
 import os
-
+from pathlib import PurePath
 
 def test_remote_repo_exists(empty_remote_repo):
     exists = True
@@ -8,7 +8,8 @@ def test_remote_repo_exists(empty_remote_repo):
 
 def test_remote_repo_add(celeba_remote_repo_one_image_pushed, shared_datadir):
     _local_repo, remote_repo = celeba_remote_repo_one_image_pushed
-    full_path = os.path.join(shared_datadir, "ChatBot/examples.tsv")
+    examples_path = str(PurePath("ChatBot", "examples.tsv"))
+    full_path = os.path.join(shared_datadir, examples_path)
     remote_repo.add(full_path)
     status = remote_repo.status()
     added_files = status.added_files()

oxen/tests/test_workspace_add.py

@@ -1,12 +1,14 @@
 import os
 from oxen import RemoteRepo
 from oxen import Workspace
+from pathlib import PurePath
 
 
 def test_workspace_add_single_file(
     celeba_remote_repo_one_image_pushed: RemoteRepo, shared_datadir
 ):
-    full_path = os.path.join(shared_datadir, "CelebA/images/1.jpg")
+    images_path = str(PurePath("CelebA", "images", "1.jpg"))
+    full_path = os.path.join(shared_datadir, images_path)
 
     _, remote_repo = celeba_remote_repo_one_image_pushed
     workspace = Workspace(remote_repo, "main", "test-workspace")
@@ -15,13 +17,15 @@ def test_workspace_add_single_file(
     status = workspace.status()
     added_files = status.added_files()
 
-    assert added_files == ["a-folder/1.jpg"]
+    added_path = str(PurePath("a-folder", "1.jpg"))
+    assert added_files == [added_path]
 
 
 def test_workspace_add_root_dir(
     celeba_remote_repo_one_image_pushed: RemoteRepo, shared_datadir
 ):
-    full_path = os.path.join(shared_datadir, "CelebA/images/3.jpg")
+    images_path = str(PurePath("CelebA", "images", "3.jpg"))
+    full_path = os.path.join(shared_datadir, images_path)
 
     _, remote_repo = celeba_remote_repo_one_image_pushed
     workspace = Workspace(remote_repo, "main", "test-workspace")

oxen/tests/test_workspace_add_to_branch.py

@@ -1,5 +1,6 @@
 import os
 from oxen import Workspace
+from pathlib import PurePath
 
 
 def test_workspace_add_to_branch(celeba_remote_repo_one_image_pushed, shared_datadir):
@@ -9,7 +10,9 @@ def test_workspace_add_to_branch(celeba_remote_repo_one_image_pushed, shared_dat
 
     workspace = Workspace(remote_repo, "newbranch")
 
-    full_path = os.path.join(shared_datadir, "CelebA/images/1.jpg")
+    image = str(PurePath("CelebA", "images", "1.jpg"))
+
+    full_path = os.path.join(shared_datadir, image)
     workspace.add(full_path)
     status = workspace.status()
 

oxen/tests/test_workspace_commit.py

@@ -1,6 +1,7 @@
 import os
 import pytest
 from oxen import RemoteRepo, Workspace
+from pathlib import PurePath
 
 
 def test_commit_one_file(
@@ -9,7 +10,8 @@ def test_commit_one_file(
     _, remote_repo = celeba_remote_repo_one_image_pushed
     # 1 commit pushed in setup
     assert len(remote_repo.log()) == 1
-    full_path = os.path.join(shared_datadir, "CelebA/images/1.jpg")
+    images_path = str(PurePath("CelebA", "images", "1.jpg"))
+    full_path = os.path.join(shared_datadir, images_path)
     workspace = Workspace(remote_repo, "main")
     workspace.add(full_path)
     workspace.commit("a commit message!")

oxen/tests/test_workspace_data_frame_add_row.py

@@ -2,6 +2,7 @@
 import pytest
 import os
 from oxen import Workspace, DataFrame
+from pathlib import PurePath
 
 
 def test_workspace_df_add_row_success(
@@ -15,13 +16,16 @@ def test_workspace_df_add_row_success(
     workspace.add(file_path, "csvs")
     workspace.commit("add train.csv")
 
-    new_row = {"file": "images/123456.png", "hair_color": "purple"}
-    remote_df = DataFrame(workspace, "csvs/train.csv")
+    images_path = str(PurePath("images", "123456.png"))
+    new_row = {"file": images_path, "hair_color": "purple"}
+
+    train_path = str(PurePath("csvs", "train.csv"))
+    remote_df = DataFrame(workspace, train_path)
     remote_df.insert_row(new_row)
     workspace.commit("add row to train.csv")
 
     # Download the file
-    remote_repo.download("csvs/train.csv", file_path)
+    remote_repo.download(train_path, file_path)
 
     # Check the new file
     new_df = pd.read_csv(file_path)
@@ -42,10 +46,13 @@ def test_remote_df_add_row_invalid_schema(
     file_path = os.path.join(shared_datadir, "CelebA", "annotations", "train.csv")
     # df = pd.read_csv(file_path)
 
-    new_row = {"gahfile": "images/123456.png", "hair_color": "purple"}
+    images_path = str(PurePath("images", "123456.png"))
+    new_row = {"gahfile": images_path, "hair_color": "purple"}
 
     workspace.add(file_path, "csvs")
     workspace.commit("add train.csv")
-    remote_df = DataFrame(workspace, "csvs/train.csv")
+
+    train_path = str(PurePath("csvs", "train.csv"))
+    remote_df = DataFrame(workspace, train_path)
     with pytest.raises(ValueError):
         remote_df.insert_row(new_row)