fix issue on finding subtree

Author: skshetry

dvc/repo/tree.py

@@ -2,7 +2,7 @@
 import os
 from typing import Optional
 
-from funcy import cached_property, post_processing
+from funcy import cached_property
 
 from dvc.dvcfile import is_valid_filename
 from dvc.exceptions import OutputNotFoundError
@@ -277,8 +277,12 @@ def _find_subtree(self, path_prefix) -> Optional[DvcTree]:
     def _find_subtree_with_prefix(self, path):
         # dvctrees is already ordered from low to high
         path_prefix = os.path.abspath(path)
+        exact_match = self._dvctrees.get(path_prefix)
+        if exact_match:
+            return path_prefix, exact_match
+
         for pref, tree in self._dvctrees.items():
-            if path_prefix.startswith(pref):
+            if path_prefix.startswith(pref + os.sep):
                 return pref, tree
         return "", None
 
@@ -487,21 +491,3 @@ def hash_jobs(self):
 
     def in_subtree(self, path):
         return self._find_subtree(path)
-
-    @post_processing("\r\n".join)
-    def _visualize(self, top, **kwargs):
-        """`tree`-like output, useful for debugging/visualizing"""
-        indent = 4
-        spacing = " " * indent
-        tee = "├── "
-        last = "└── "
-        for root, _, files in self.walk(top, **kwargs):
-            level = root.replace(top, "").count(os.sep)
-            indent = spacing * level
-            yield "{}{}/".format(indent, os.path.basename(root))
-            sub_indent = spacing * (level + 1)
-            length = len(files)
-            for i, f in enumerate(files):
-                yield "{}{}{}".format(
-                    sub_indent, tee if i + 1 != length else last, f
-                )

tests/func/test_get.py

@@ -238,12 +238,13 @@ def test_get_pipeline_tracked_outs(
         assert (git_dir / "baz").read_text() == "foo"
 
 
-def make_subrepo(dir_, scm, config):
+def make_subrepo(dir_, scm, config=None):
     dir_.mkdir(parents=True)
     with dir_.chdir():
         dir_.scm = scm
         dir_.init(dvc=True, subdir=True)
-        dir_.add_remote(config=config)
+        if config:
+            dir_.add_remote(config=config)
 
 
 @pytest.mark.parametrize(

tests/unit/repo/test_repo_tree.py

@@ -177,3 +177,36 @@ def test_isdvc(tmp_dir, dvc):
     assert tree.isdvc("dir")
     assert not tree.isdvc("dir/baz")
     assert tree.isdvc("dir/baz", recursive=True, strict=False)
+
+
+def test_in_subtree(tmp_dir, scm, dvc):
+    from tests.func.test_get import make_subrepo
+
+    subrepo1 = tmp_dir / "dir" / "repo"
+    subrepo2 = tmp_dir / "dir" / "repo2"
+
+    for repo in [subrepo1, subrepo2]:
+        make_subrepo(repo, scm)
+
+    (tmp_dir / "dir" / "repotxt").write_text("file to confuse RepoTree")
+    subrepo1.dvc_gen({"foo": "foo"}, commit="FOO")
+    subrepo2.dvc_gen({"bar": "bar"}, commit="BAR")
+
+    # dvc.tree ignores subrepos by default,
+    # but we just want to test `in_subtree()`, which is purely lexical
+    tree = RepoTree(dvc.tree, [dvc, subrepo1.dvc, subrepo2.dvc])
+
+    assert tree.in_subtree(str(tmp_dir / "dir")).repo == dvc
+    assert tree.in_subtree(str(tmp_dir / "dir" / "re")).repo == dvc
+    assert tree.in_subtree(str(tmp_dir / "dir" / "repo")).repo == subrepo1.dvc
+    assert tree.in_subtree(str(tmp_dir / "dir" / "repotxt")).repo == dvc
+    assert tree.in_subtree(str(tmp_dir / "dir" / "repo2")).repo == subrepo2.dvc
+
+    for repo in [tmp_dir, subrepo1, subrepo2]:
+        for path in ["", "foo", "something-that-does-not-exist"]:
+            p = os.path.join(repo, path)
+            subtree = tree.in_subtree(p)
+            assert subtree, f"subtree not found for path '{p}'"
+            assert (
+                subtree.repo == repo.dvc
+            ), f"repo did not match for path '{p}'"