reintroduce dvcignore collection test

dvc/ignore.py

@@ -77,10 +77,9 @@ class DvcIgnoreFilter(object):
     def __init__(self, tree):
         self.tree = tree
         self.ignores = {DvcIgnoreDirs([".git", ".hg", ".dvc"])}
-        self._update(self.tree.tree_root)
-        for root, dirs, _ in self.tree.walk(self.tree.tree_root):
-            for d in dirs:
-                self._update(os.path.join(root, d))
+        for root, dirs, files in self.tree.walk(self.tree.tree_root):
+            self._update(root)
+            dirs[:], files[:] = self(root, dirs, files)
 
     def _update(self, dirname):
         ignore_file_path = os.path.join(dirname, DvcIgnore.DVCIGNORE_FILE)

dvc/repo/brancher.py

@@ -59,7 +59,7 @@ def brancher(  # noqa: E302
     # code which might expect the tree on which exception was raised to
     # stay in place. This behavior is a subject to change.
     for sha, names in group_by(scm.resolve_rev, revs).items():
-        self.tree = scm.get_tree(sha)
+        self.tree = CleanTree(scm.get_tree(sha))
         yield ", ".join(names)
 
     self.tree = saved_tree

dvc/scm/git/__init__.py

@@ -8,7 +8,6 @@ from funcy import cached_property
 from pathspec.patterns import GitWildMatchPattern
 
 from dvc.exceptions import GitHookAlreadyExistsError
-from dvc.ignore import CleanTree
 from dvc.scm.base import Base
 from dvc.scm.base import CloneError
 from dvc.scm.base import FileNotInRepoError
@@ -308,7 +307,7 @@ class Git(Base):
         return basename == self.ignore_file or Git.GIT_DIR in path_parts
 
     def get_tree(self, rev):
-        return CleanTree(GitTree(self.repo, rev))
+        return GitTree(self.repo, rev)
 
     def _get_diff_trees(self, a_ref, b_ref):
         """Private method for getting the trees and commit hashes of 2 git
@@ -322,6 +321,7 @@ class Git(Base):
             tuple: tuple with elements: (trees, commits)
         """
         from gitdb.exc import BadObject, BadName
+        from dvc.ignore import CleanTree
 
         trees = {DIFF_A_TREE: None, DIFF_B_TREE: None}
         commits = []
@@ -334,8 +334,8 @@ class Git(Base):
             # /en/2.1.11/reference.html#git.objects.base.Object.__str__
             commits.append(a_commit)
             commits.append(b_commit)
-            trees[DIFF_A_TREE] = self.get_tree(commits[0])
-            trees[DIFF_B_TREE] = self.get_tree(commits[1])
+            trees[DIFF_A_TREE] = CleanTree(self.get_tree(commits[0]))
+            trees[DIFF_B_TREE] = CleanTree(self.get_tree(commits[1]))
         except (BadName, BadObject) as e:
             raise SCMError("git problem", cause=e)
         return trees, commits

dvc/scm/tree.py

@@ -1,6 +1,6 @@
 import os
 
-from dvc.utils.compat import open
+from dvc.utils.compat import open, fspath
 
 
 class BaseTree(object):
@@ -65,6 +65,8 @@ class WorkingTree(BaseTree):
         - it could raise exceptions, there is no onerror argument
         """
 
+        top = fspath(top)
+
         def onerror(e):
             raise e
 

dvc/utils/fs.py

@@ -34,8 +34,6 @@ def get_inode(path):
     return inode
 
 
-# FIXME: There is no use case of get_mtime_and_size without tree, we should
-# move it to tree
 def get_mtime_and_size(path, tree):
     if os.path.isdir(fspath_py35(path)):
         size = 0

tests/func/test_ignore.py

@@ -5,7 +5,8 @@ import shutil
 import pytest
 
 from dvc.exceptions import DvcIgnoreInCollectedDirError
-from dvc.ignore import DvcIgnore
+from dvc.ignore import DvcIgnore, DvcIgnoreDirs, DvcIgnorePatterns, CleanTree
+from dvc.scm.tree import WorkingTree
 from dvc.utils import relpath
 from dvc.utils.compat import fspath_py35
 from dvc.utils.compat import fspath
@@ -89,6 +90,22 @@ def test_dvcignore_in_out_dir(tmp_dir, dvc):
         dvc.add("dir")
 
 
+@pytest.mark.parametrize("dname", ["dir", "dir/subdir"])
+def test_ignore_collecting_dvcignores(tmp_dir, dvc, dname):
+    tmp_dir.gen({"dir": {"subdir": {}}})
+
+    top_ignore_file = (tmp_dir / dname).with_name(DvcIgnore.DVCIGNORE_FILE)
+    top_ignore_file.write_text(os.path.basename(dname))
+
+    ignore_file = tmp_dir / dname / DvcIgnore.DVCIGNORE_FILE
+    ignore_file.write_text("foo")
+
+    assert dvc.tree.dvcignore.ignores == {
+        DvcIgnoreDirs([".git", ".hg", ".dvc"]),
+        DvcIgnorePatterns(fspath(top_ignore_file), WorkingTree(dvc.root_dir)),
+    }
+
+
 def test_ignore_on_branch(tmp_dir, scm, dvc):
     tmp_dir.scm_gen({"foo": "foo", "bar": "bar"}, commit="add files")
 
@@ -98,7 +115,7 @@ def test_ignore_on_branch(tmp_dir, scm, dvc):
     scm.checkout("master")
     assert _files_set(".", dvc.tree) == {"./foo", "./bar"}
 
-    tree = scm.get_tree("branch")
+    tree = CleanTree(scm.get_tree("branch"))
     files_on_branch = {
         f for r, ds, fs in tree.walk(tree.tree_root) for f in fs
     }

tests/func/test_repo.py

@@ -1,6 +1,5 @@
 import os
 
-from dvc.ignore import CleanTree
 from dvc.scm.git.tree import GitTree
 from dvc.cache import Cache
 from dvc.repo import Repo
@@ -51,7 +50,7 @@ def test_collect(tmp_dir, scm, dvc, run_copy):
 
     assert collect_outs("bar.dvc", with_deps=True) == {"foo", "bar"}
 
-    dvc.tree = CleanTree(GitTree(scm.repo, "new-branch"))
+    dvc.tree = GitTree(scm.repo, "new-branch")
 
     assert collect_outs("buzz.dvc", with_deps=True) == {"foo", "bar", "buzz"}
     assert collect_outs("buzz.dvc", with_deps=False) == {"buzz"}

tests/func/test_tree.py

@@ -82,14 +82,14 @@ class TestGitTree(TestGit, GitTreeTests):
     def setUp(self):
         super(TestGitTree, self).setUp()
         self.scm = SCM(self._root_dir)
-        self.tree = CleanTree(GitTree(self.git, "master"))
+        self.tree = GitTree(self.git, "master")
 
 
 class TestGitSubmoduleTree(TestGitSubmodule, GitTreeTests):
     def setUp(self):
         super(TestGitSubmoduleTree, self).setUp()
         self.scm = SCM(self._root_dir)
-        self.tree = CleanTree(GitTree(self.git, "master"))
+        self.tree = GitTree(self.git, "master")
         self._pushd(self._root_dir)
 
 
@@ -108,7 +108,7 @@ class AssertWalkEqualMixin(object):
 
 class TestWalkInNoSCM(AssertWalkEqualMixin, TestDir):
     def test(self):
-        tree = CleanTree(WorkingTree(self._root_dir))
+        tree = WorkingTree(self._root_dir)
         self.assertWalkEqual(
             tree.walk(self._root_dir),
             [
@@ -127,7 +127,7 @@ class TestWalkInNoSCM(AssertWalkEqualMixin, TestDir):
         )
 
     def test_subdir(self):
-        tree = CleanTree(WorkingTree(self._root_dir))
+        tree = WorkingTree(self._root_dir)
         self.assertWalkEqual(
             tree.walk(join("data_dir", "data_sub_dir")),
             [(join("data_dir", "data_sub_dir"), [], ["data_sub"])],
@@ -154,7 +154,7 @@ class TestWalkInGit(AssertWalkEqualMixin, TestGit):
         scm = SCM(self._root_dir)
         scm.add([self.DATA_SUB_DIR])
         scm.commit("add data_dir/data_sub_dir/data_sub")
-        tree = CleanTree(GitTree(self.git, "master"))
+        tree = GitTree(self.git, "master")
         self.assertWalkEqual(
             tree.walk("."),
             [