Added parameter ignore_device to constructor of DirectorySnapshotDi…

…ff (gorakhargosh#597)

* Added parameter `ignore_device` to constructor of DirectorySnapshotDiff.

* Fixed E704.

* Added docstring.

* Added test for `ignore_device` parameter on DirectorySnapshotDiff constructor.

src/watchdog/utils/dirsnapshot.py

@@ -68,15 +68,32 @@ class DirectorySnapshotDiff(object):
         with the reference snapshot.
     :type snapshot:
         :class:`DirectorySnapshot`
+    :param ignore_device:
+        A boolean indicating whether to ignore the device id or not.
+        By default, a file may be uniquely identified by a combination of its first
+        inode and its device id. The problem is that the device id may (or may not)
+        change between system boots. This problem would cause the DirectorySnapshotDiff
+        to think a file has been deleted and created again but it would be the
+        exact same file.
+        Set to True only if you are sure you will always use the same device.
+    :type ignore_device:
+        :class:`bool`
     """
 
-    def __init__(self, ref, snapshot):
+    def __init__(self, ref, snapshot, ignore_device=False):
         created = snapshot.paths - ref.paths
         deleted = ref.paths - snapshot.paths
 
+        if ignore_device:
+            def get_inode(directory, full_path):
+                return directory.inode(full_path)[0]
+        else:
+            def get_inode(directory, full_path):
+                return directory.inode(full_path)
+
         # check that all unchanged paths have the same inode
         for path in ref.paths & snapshot.paths:
-            if ref.inode(path) != snapshot.inode(path):
+            if get_inode(ref, path) != get_inode(snapshot, path):
                 created.add(path)
                 deleted.add(path)
 
@@ -101,7 +118,7 @@ def __init__(self, ref, snapshot):
         # first check paths that have not moved
         modified = set()
         for path in ref.paths & snapshot.paths:
-            if ref.inode(path) == snapshot.inode(path):
+            if get_inode(ref, path) == get_inode(snapshot, path):
                 if ref.mtime(path) != snapshot.mtime(path) or ref.size(path) != snapshot.size(path):
                     modified.add(path)
 

tests/test_snapshot_diff.py

@@ -156,3 +156,37 @@ def walk(self, root):
 
     # Children of a/b/ are no more accessible and so removed in the new snapshot
     assert diff.dirs_deleted == [(p('a', 'b', 'c'))]
+
+
+def test_ignore_device(monkeypatch, p):
+    # Create a file and take a snapshot.
+    touch(p('file'))
+    ref = DirectorySnapshot(p(''))
+    wait()
+
+    def inode(self, path):
+        # This function will always return a different device_id,
+        # even for the same file.
+        result = inode_orig(self, path)
+        inode.times += 1
+        return result[0], result[1] + inode.times
+    inode.times = 0
+
+    # Set the custom inode function.
+    inode_orig = DirectorySnapshot.inode
+    monkeypatch.setattr(DirectorySnapshot, 'inode', inode)
+
+    # If we make the diff of the same directory, since by default the
+    # DirectorySnapshotDiff compares the snapshots using the device_id (and it will
+    # be different), it thinks that the same file has been deleted and created again.
+    snapshot = DirectorySnapshot(p(''))
+    diff_with_device = DirectorySnapshotDiff(ref, snapshot)
+    assert diff_with_device.files_deleted == [(p('file'))]
+    assert diff_with_device.files_created == [(p('file'))]
+
+    # Otherwise, if we choose to ignore the device, the file will not be detected as
+    # deleted and re-created.
+    snapshot = DirectorySnapshot(p(''))
+    diff_without_device = DirectorySnapshotDiff(ref, snapshot, ignore_device=True)
+    assert diff_without_device.files_deleted == []
+    assert diff_without_device.files_created == []