From 470825aec507c1c348367f5b87234d607fbd94b4 Mon Sep 17 00:00:00 2001
From: Ruslan Kuprieiev <ruslan@iterative.ai>
Date: Sat, 15 Sep 2018 07:21:58 +0300
Subject: [PATCH] dvc: don't scan output directories for dvc files

`dvc status` for directory with 100K files sped up from ~6 sec to ~2
sec.

Fixes #1121

Signed-off-by: Ruslan Kuprieiev <ruslan@iterative.ai>
---
 dvc/project.py | 18 +++++++++++++++++-
 dvc/scm.py     |  4 ++++
 2 files changed, 21 insertions(+), 1 deletion(-)

diff --git a/dvc/project.py b/dvc/project.py
index 98e58f59a..eb5dc80ff 100644
--- a/dvc/project.py
+++ b/dvc/project.py
@@ -746,12 +746,28 @@ class Project(object):
 
     def stages(self):
         stages = []
+        outs = []
         for root, dirs, files in os.walk(self.root_dir):
             for fname in files:
                 path = os.path.join(root, fname)
                 if not Stage.is_stage_file(path):
                     continue
-                stages.append(Stage.load(self, path))
+                stage = Stage.load(self, path)
+                for out in stage.outs:
+                    outs.append(out.path + out.sep)
+                stages.append(stage)
+
+            def filter_dirs(dname):
+                path = os.path.join(root, dname)
+                if path == self.dvc_dir or path == self.scm.dir:
+                    return False
+                for out in outs:
+                    if path == os.path.normpath(out) or path.startswith(out):
+                        return False
+                return True
+
+            dirs[:] = list(filter(filter_dirs, dirs))
+
         return stages
 
     def active_stages(self):
diff --git a/dvc/scm.py b/dvc/scm.py
index eadaf9943..5ac1d16ca 100644
--- a/dvc/scm.py
+++ b/dvc/scm.py
@@ -99,6 +99,10 @@ class Git(Base):
         git_dir = os.path.join(root_dir, Git.GIT_DIR)
         return os.path.isdir(git_dir)
 
+    @property
+    def dir(self):
+        return self.repo.git_dir
+
     def ignore_file(self):
         return self.GITIGNORE