From 4897a6a44ff0df7a1e2a7a321035ee672179b3b1 Mon Sep 17 00:00:00 2001
From: pared <pawelredzynski@gmail.com>
Date: Fri, 8 Mar 2019 11:09:25 +0100
Subject: [PATCH] add: prevent adding git files

---
 dvc/repo/__init__.py  |  4 ++++
 dvc/repo/add.py       | 24 ++++++++++++++----------
 dvc/scm/base.py       |  3 +++
 dvc/scm/git.py        |  5 +++++
 tests/test_add.py     | 13 +++++++++++++
 tests/unit/repo.py    | 13 +++++++++++++
 tests/unit/scm/git.py | 17 +++++++++++++++++
 7 files changed, 69 insertions(+), 10 deletions(-)
 create mode 100644 tests/unit/repo.py
 create mode 100644 tests/unit/scm/git.py

diff --git a/dvc/repo/__init__.py b/dvc/repo/__init__.py
index 3831d695c..c667f9f0f 100644
--- a/dvc/repo/__init__.py
+++ b/dvc/repo/__init__.py
@@ -415,3 +415,7 @@ class Repo(object):
             raise OutputNotFoundError(path)
 
         return matched
+
+    def is_dvc_internal(self, path):
+        path_parts = os.path.normpath(path).split(os.path.sep)
+        return self.DVC_DIR in path_parts
diff --git a/dvc/repo/add.py b/dvc/repo/add.py
index 27caffc27..2a7182b75 100644
--- a/dvc/repo/add.py
+++ b/dvc/repo/add.py
@@ -11,16 +11,7 @@ def add(repo, target, recursive=False, no_commit=False, fname=None):
     if recursive and fname:
         raise RecursiveAddingWhileUsingFilename()
 
-    targets = [target]
-
-    if os.path.isdir(target) and recursive:
-        targets = [
-            file
-            for file in walk_files(target)
-            if not Stage.is_stage_file(file)
-            if os.path.basename(file) != repo.scm.ignore_file
-            if not repo.scm.is_tracked(file)
-        ]
+    targets = _find_all_targets(repo, target, recursive)
 
     stages = _create_stages(repo, targets, fname, no_commit)
 
@@ -31,6 +22,19 @@ def add(repo, target, recursive=False, no_commit=False, fname=None):
     return stages
 
 
+def _find_all_targets(repo, target, recursive):
+    if os.path.isdir(target) and recursive:
+        return [
+            file
+            for file in walk_files(target)
+            if not repo.is_dvc_internal(file)
+            if not Stage.is_stage_file(file)
+            if not repo.scm.belongs_to_scm(file)
+            if not repo.scm.is_tracked(file)
+        ]
+    return [target]
+
+
 def _create_stages(repo, targets, fname, no_commit):
     stages = []
 
diff --git a/dvc/scm/base.py b/dvc/scm/base.py
index 07611853a..48031ec3e 100644
--- a/dvc/scm/base.py
+++ b/dvc/scm/base.py
@@ -162,3 +162,6 @@ class Base(object):
         Method to add file to mechanism that will remind user
         to track new files
         """
+
+    def belongs_to_scm(self, path):
+        """Return boolean whether file belongs to scm"""
diff --git a/dvc/scm/git.py b/dvc/scm/git.py
index b2bb76cca..f60c2cfb9 100644
--- a/dvc/scm/git.py
+++ b/dvc/scm/git.py
@@ -198,3 +198,8 @@ class Git(Base):
 
     def track_file(self, path):
         self.files_to_track.append(path)
+
+    def belongs_to_scm(self, path):
+        basename = os.path.basename(path)
+        path_parts = os.path.normpath(path).split(os.path.sep)
+        return basename == self.ignore_file or Git.GIT_DIR in path_parts
diff --git a/tests/test_add.py b/tests/test_add.py
index 35ddddd07..b8e8866db 100644
--- a/tests/test_add.py
+++ b/tests/test_add.py
@@ -477,3 +477,16 @@ class TestShouldCleanUpAfterFailedAdd(TestDvc):
 
         gitignore_content = get_gitignore_content()
         self.assertFalse(any(self.BAR in line for line in gitignore_content))
+
+
+class TestShouldNotTrackGitInternalFiles(TestDvc):
+    def test(self):
+        stage_creator_spy = spy(dvc.repo.add._create_stages)
+
+        with patch.object(dvc.repo.add, "_create_stages", stage_creator_spy):
+            ret = main(["add", "-R", self.dvc.root_dir])
+            self.assertEqual(0, ret)
+
+        created_stages_filenames = stage_creator_spy.mock.call_args[0][0]
+        for fname in created_stages_filenames:
+            self.assertNotIn(".git", fname)
diff --git a/tests/unit/repo.py b/tests/unit/repo.py
new file mode 100644
index 000000000..ad7426154
--- /dev/null
+++ b/tests/unit/repo.py
@@ -0,0 +1,13 @@
+import os
+
+from tests.basic_env import TestDvc
+
+
+class TestIsDvcInternal(TestDvc):
+    def test_return_false_on_non_dvc_file(self):
+        path = os.path.join("path", "to-non-.dvc", "file")
+        self.assertFalse(self.dvc.is_dvc_internal(path))
+
+    def test_return_true_on_dvc_internal_file(self):
+        path = os.path.join("path", "to", ".dvc", "file")
+        self.assertTrue(path)
diff --git a/tests/unit/scm/git.py b/tests/unit/scm/git.py
new file mode 100644
index 000000000..b9ad27615
--- /dev/null
+++ b/tests/unit/scm/git.py
@@ -0,0 +1,17 @@
+import os
+
+from tests.basic_env import TestDvc
+
+
+class TestGit(TestDvc):
+    def test_belongs_to_scm_true_on_gitignore(self):
+        path = os.path.join("path", "to", ".gitignore")
+        self.assertTrue(self.dvc.scm.belongs_to_scm(path))
+
+    def test_belongs_to_scm_true_on_git_internal(self):
+        path = os.path.join("path", "to", ".git", "internal", "file")
+        self.assertTrue(self.dvc.scm.belongs_to_scm(path))
+
+    def test_belongs_to_scm_false(self):
+        path = os.path.join("some", "non-.git", "file")
+        self.assertFalse(self.dvc.scm.belongs_to_scm(path))