From d0d738c99b6fb427e5c0be1da5d41d11b64480df Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Saugat=20Pachhai=20=28=E0=A4=B8=E0=A5=8C=E0=A4=97=E0=A4=BE?=
 =?UTF-8?q?=E0=A4=A4=29?= <suagatchhetri@outlook.com>
Date: Thu, 25 Aug 2022 18:30:09 +0545
Subject: [PATCH] data status: make untracked files to be relative paths

---
 dvc/repo/data.py               | 31 +++++++++++++++++++++++-------
 tests/func/test_data_status.py | 35 ++++++++++++++++++++++++++++++++++
 2 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/dvc/repo/data.py b/dvc/repo/data.py
index 2d84361eec..d77d2fedf9 100644
--- a/dvc/repo/data.py
+++ b/dvc/repo/data.py
@@ -1,6 +1,15 @@
 import os
 from collections import defaultdict
-from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypedDict, cast
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Iterable,
+    List,
+    Optional,
+    TypedDict,
+    cast,
+)
 
 from dvc.ui import ui
 
@@ -250,15 +259,23 @@ class Status(TypedDict):
     git: GitInfo
 
 
-def _transform_git_paths_to_dvc(repo: "Repo", files: List[str]):
+def _transform_git_paths_to_dvc(repo: "Repo", files: Iterable[str]):
     """Transform files rel. to Git root to DVC root, and drop outside files."""
     rel = repo.fs.path.relpath(repo.root_dir, repo.scm.root_dir).rstrip("/")
-    if rel in (os.curdir, ""):
-        return files
 
-    prefix = rel + os.sep
-    length = len(prefix)
-    return [file[length:] for file in files if file.startswith(prefix)]
+    # if we have repo root in a different location than scm's root,
+    # i.e. subdir repo, all git_paths need to be transformed rel. to the DVC
+    # repo root and anything outside need to be filtered out.
+    if rel not in (os.curdir, ""):
+        prefix = rel + os.sep
+        length = len(prefix)
+        files = (file[length:] for file in files if file.startswith(prefix))
+
+    start = repo.fs.path.relpath(repo.fs.path.getcwd(), repo.root_dir)
+    if start in (os.curdir, ""):
+        return list(files)
+    # we need to convert repo relative paths to curdir relative.
+    return [repo.fs.path.relpath(file, start) for file in files]
 
 
 def status(repo: "Repo", untracked_files: str = "no", **kwargs: Any) -> Status:
diff --git a/tests/func/test_data_status.py b/tests/func/test_data_status.py
index 47508e23b1..5ea8913ee9 100644
--- a/tests/func/test_data_status.py
+++ b/tests/func/test_data_status.py
@@ -1,6 +1,9 @@
 from os.path import join
 
+import pytest
+
 from dvc.repo import Repo
+from dvc.repo.data import _transform_git_paths_to_dvc
 from dvc.testing.tmp_dir import make_subrepo
 from dvc.utils.fs import remove
 
@@ -14,6 +17,38 @@
 }
 
 
+@pytest.mark.parametrize("path", [None, ("sub", "repo")])
+def test_git_to_dvc_path_wdir_transformation(tmp_dir, scm, path):
+    struct = {"dir": {"foo": "foo", "bar": "bar"}, "file": "file", "dir2": {}}
+    tmp_dir.gen(struct)
+
+    subdir = tmp_dir.joinpath(*path) if path else tmp_dir
+    make_subrepo(subdir, scm)
+    dvc = subdir.dvc
+
+    with subdir.chdir():
+        subdir.gen(struct)
+        _, _, untracked = scm.status(untracked_files="all")
+        untracked = sorted(untracked, reverse=True)
+        assert _transform_git_paths_to_dvc(dvc, untracked) == [
+            "file",
+            join("dir", "foo"),
+            join("dir", "bar"),
+        ]
+        with (subdir / "dir").chdir():
+            assert _transform_git_paths_to_dvc(dvc, untracked) == [
+                join("..", "file"),
+                "foo",
+                "bar",
+            ]
+        with (subdir / "dir2").chdir():
+            assert _transform_git_paths_to_dvc(dvc, untracked) == [
+                join("..", "file"),
+                join("..", "dir", "foo"),
+                join("..", "dir", "bar"),
+            ]
+
+
 def test_file(M, tmp_dir, dvc, scm):
     tmp_dir.dvc_gen("foo", "foo", commit="add foo")
     tmp_dir.dvc_gen("foo", "foobar")