Merge pull request #418 from efiop/master

state: don't forget to track inode

dvc/stage.py

@@ -94,6 +94,9 @@ class Dependency(object):
     def mtime(self):
         return os.path.getmtime(self.path)
 
+    def inode(self):
+        return os.stat(self.path).st_ino
+
     def update(self):
         if not os.path.exists(self.path):
             raise CmdOutputDoesNotExistError(self.rel_path)
@@ -101,14 +104,14 @@ class Dependency(object):
             raise CmdOutputIsNotFileError(self.path)
 
         state = self.project.state.get(self.path)
-        if state and state.mtime == self.mtime():
+        if state and state.mtime == self.mtime() and state.inode == self.inode():
             md5 = state.md5
             msg = '{} using md5 {} from state file'
             self.project.logger.debug(msg.format(self.path, md5))
             self.md5 = md5
         else:
             self.md5 = file_md5(self.path)[0]
-            self.project.state.update(self.path, self.md5, self.mtime())
+            self.project.state.update(self.path, self.md5, self.mtime(), self.inode())
 
     def dumpd(self, cwd):
         return {

dvc/state.py

@@ -8,30 +8,35 @@ class StateEntry(object):
     PARAM_PATH = 'path'
     PARAM_MTIME = 'mtime'
     PARAM_MD5 = 'md5'
+    PARAM_INODE = 'inode'
 
-    def __init__(self, root_dir, path, md5, mtime):
+    def __init__(self, root_dir, path, md5, mtime, inode):
         self.root_dir = root_dir
         self.path = path
         self.mtime = mtime
         self.md5 = md5
+        self.inode = inode
         self.dvc_path = os.path.relpath(self.path, self.root_dir)
 
-    def update(self, md5, mtime):
+    def update(self, md5, mtime, inode):
         self.mtime = mtime
         self.md5 = md5
+        self.inode = inode
 
     @staticmethod
     def loadd(root_dir, d):
         path = os.path.join(root_dir, d[StateEntry.PARAM_PATH])
         mtime = d[StateEntry.PARAM_MTIME]
         md5 = d[StateEntry.PARAM_MD5]
-        return StateEntry(root_dir, path, md5, mtime)
+        inode = d[StateEntry.PARAM_INODE]
+        return StateEntry(root_dir, path, md5, mtime, inode)
 
     def dumpd(self):
         return {
             self.PARAM_PATH: self.dvc_path,
             self.PARAM_MD5: self.md5,
-            self.PARAM_MTIME: self.mtime
+            self.PARAM_MTIME: self.mtime,
+            self.PARAM_INODE: self.inode
         }
 
 
@@ -53,18 +58,18 @@ class State(object):
     def init(root_dir, dvc_dir):
         return State(root_dir, dvc_dir)
 
-    def update(self, path, md5, mtime):
+    def update(self, path, md5, mtime, inode):
         existing = self.get(path)
         if not existing:
-            return self.add(path, md5, mtime)
+            return self.add(path, md5, mtime, inode)
 
-        state = StateEntry(self.root_dir, path, md5, mtime)
+        state = StateEntry(self.root_dir, path, md5, mtime, inode)
         self._db.update(state.dumpd(), self._q.path == state.dvc_path)
 
         return state
 
-    def add(self, path, md5, mtime):
-        entry = StateEntry(self.root_dir, path, md5, mtime)
+    def add(self, path, md5, mtime, inode):
+        entry = StateEntry(self.root_dir, path, md5, mtime, inode)
         self._db.insert(entry.dumpd())
         return entry
 

tests/test_state.py

@@ -12,20 +12,24 @@ class TestState(TestDvc):
         md5_2 = '2'
         mtime_1 = 1
         mtime_2 = 2
+        inode_1 = 1
+        inode_2 = 2
 
         state = State(self.dvc.root_dir, self.dvc.dvc_dir)
         self.assertIsNone(state.get(path))
 
-        state.add(path, md5_1, mtime_1)
+        state.add(path, md5_1, mtime_1, inode_1)
         entry = state.get(path)
         self.assertIsInstance(entry, StateEntry)
         self.assertEqual(entry.path, path)
         self.assertEqual(entry.md5, md5_1)
         self.assertEqual(entry.mtime, mtime_1)
+        self.assertEqual(entry.inode, inode_1)
 
-        state.update(path, md5_2, mtime_2)
+        state.update(path, md5_2, mtime_2, inode_2)
         entry = state.get(path)
         self.assertIsInstance(entry, StateEntry)
         self.assertEqual(entry.path, path)
         self.assertEqual(entry.md5, md5_2)
         self.assertEqual(entry.mtime, mtime_2)
+        self.assertEqual(entry.inode, inode_2)