From bc81d96fe100e5b40f38efdbc7dc5ace0835d321 Mon Sep 17 00:00:00 2001
From: pawel <pawelredzynski@gmail.com>
Date: Mon, 10 Feb 2020 15:39:28 +0100
Subject: [PATCH] add: meaningful message upon adding overlapping paths

---
 dvc/exceptions.py      | 15 ++++-----------
 dvc/repo/__init__.py   | 12 +++++++++++-
 dvc/repo/add.py        | 22 ++++++++++++++++++++--
 tests/func/test_add.py | 16 +++++++++++++++-
 4 files changed, 50 insertions(+), 15 deletions(-)

diff --git a/dvc/exceptions.py b/dvc/exceptions.py
index 4af6111080..4459121e92 100644
--- a/dvc/exceptions.py
+++ b/dvc/exceptions.py
@@ -188,17 +188,10 @@ def __init__(self):
 
 
 class OverlappingOutputPathsError(DvcException):
-    def __init__(self, out_1, out_2):
-        super().__init__(
-            "Paths for outs:\n'{}'('{}')\n'{}'('{}')\noverlap. To avoid "
-            "unpredictable behaviour, rerun command with non overlapping outs "
-            "paths.".format(
-                str(out_1),
-                out_1.stage.relpath,
-                str(out_2),
-                out_2.stage.relpath,
-            )
-        )
+    def __init__(self, parent, overlapping_out, message):
+        self.parent = parent
+        self.overlapping_out = overlapping_out
+        super().__init__(message)
 
 
 class CheckoutErrorSuggestGit(DvcException):
diff --git a/dvc/repo/__init__.py b/dvc/repo/__init__.py
index 16062042fd..f46ea73803 100644
--- a/dvc/repo/__init__.py
+++ b/dvc/repo/__init__.py
@@ -334,7 +334,17 @@ def _collect_graph(self, stages=None):
             for out in stage.outs:
                 for p in out.path_info.parents:
                     if p in outs:
-                        raise OverlappingOutputPathsError(outs[p], out)
+                        msg = (
+                            "Paths for outs:\n'{}'('{}')\n'{}'('{}')\n"
+                            "overlap. To avoid unpredictable behaviour, "
+                            "rerun command with non overlapping outs paths."
+                        ).format(
+                            str(outs[p]),
+                            outs[p].stage.relpath,
+                            str(out),
+                            out.stage.relpath,
+                        )
+                        raise OverlappingOutputPathsError(outs[p], out, msg)
 
         for stage in stages:
             stage_path_info = PathInfo(stage.path)
diff --git a/dvc/repo/add.py b/dvc/repo/add.py
index e469119b93..66fbea0c7f 100644
--- a/dvc/repo/add.py
+++ b/dvc/repo/add.py
@@ -4,7 +4,10 @@
 import colorama
 
 from . import locked
-from ..exceptions import RecursiveAddingWhileUsingFilename
+from ..exceptions import (
+    RecursiveAddingWhileUsingFilename,
+    OverlappingOutputPathsError,
+)
 from ..output.base import OutputDoesNotExistError
 from ..progress import Tqdm
 from ..repo.scm_context import scm_context
@@ -49,7 +52,22 @@ def add(repo, targets, recursive=False, no_commit=False, fname=None):
 
             stages = _create_stages(repo, sub_targets, fname, pbar=pbar)
 
-            repo.check_modified_graph(stages)
+            try:
+                repo.check_modified_graph(stages)
+            except OverlappingOutputPathsError as exc:
+                msg = (
+                    "Cannot add '{out}', because it is overlapping with other "
+                    "DVC tracked output: '{parent}'.\n"
+                    "To include '{out}' in '{parent}', run "
+                    "'dvc commit {parent_stage}'"
+                ).format(
+                    out=exc.overlapping_out.path_info,
+                    parent=exc.parent.path_info,
+                    parent_stage=exc.parent.stage.relpath,
+                )
+                raise OverlappingOutputPathsError(
+                    exc.parent, exc.overlapping_out, msg
+                )
 
             with Tqdm(
                 total=len(stages),
diff --git a/tests/func/test_add.py b/tests/func/test_add.py
index 002722f52b..b2e6dfb1b4 100644
--- a/tests/func/test_add.py
+++ b/tests/func/test_add.py
@@ -11,7 +11,7 @@
 
 import dvc as dvc_module
 from dvc.cache import Cache
-from dvc.exceptions import DvcException
+from dvc.exceptions import DvcException, OverlappingOutputPathsError
 from dvc.exceptions import RecursiveAddingWhileUsingFilename
 from dvc.exceptions import StageFileCorruptedError
 from dvc.main import main
@@ -641,3 +641,17 @@ def test_escape_gitignore_entries(tmp_dir, scm, dvc):
 
     tmp_dir.dvc_gen(fname, "...")
     assert ignored_fname in get_gitignore_content()
+
+
+def test_add_from_data_dir(tmp_dir, scm, dvc):
+    tmp_dir.dvc_gen({"dir": {"file1": "file1 content"}})
+
+    tmp_dir.gen({"dir": {"file2": "file2 content"}})
+
+    with pytest.raises(OverlappingOutputPathsError) as e:
+        dvc.add(os.path.join("dir", "file2"))
+    assert str(e.value) == (
+        "Cannot add '{out}', because it is overlapping with other DVC "
+        "tracked output: 'dir'.\n"
+        "To include '{out}' in 'dir', run 'dvc commit dir.dvc'"
+    ).format(out=os.path.join("dir", "file2"))