diff: change structure to be indexed by path

dvc/command/diff.py

@@ -11,7 +11,9 @@ logger = logging.getLogger(__name__)
 class CmdDiff(CmdBase):
     def run(self):
         try:
-            self.repo.diff(self.args.a_ref, self.args.b_ref, self.args.target)
+            self.repo.diff(
+                self.args.a_ref, self.args.b_ref, target=self.args.target
+            )
         except DvcException:
             logger.exception("failed to get 'diff {}'")
             return 1

dvc/repo/diff.py

@@ -7,6 +7,7 @@ from dvc.compat import fspath
 
 Diffable = collections.namedtuple("Diffable", "filename, checksum, size")
 Diffable.__doc__ = "Common interface for comparable entries."
+Diffable._asdict = lambda x: {"checksum": x.checksum, "size": x.size}
 
 
 def diffable_from_output(output):
@@ -18,6 +19,21 @@ def diffable_from_output(output):
     return Diffable(filename=str(output), checksum=output.checksum, size=size)
 
 
+def compare_states(old, new):
+    if old and new:
+        try:
+            size = new["size"] - old["size"]
+        except KeyError:
+            size = "unknown"
+        return {"status": "modified", "size": size}
+
+    if old and not new:
+        return {"status": "deleted", "size": old.get("size", "unknown")}
+
+    if not old and new:
+        return {"status": "added", "size": new.get("size", "unknown")}
+
+
 @locked
 def diff(self, a_ref="HEAD", b_ref=None, *, target=None):
     """
@@ -44,10 +60,15 @@ def diff(self, a_ref="HEAD", b_ref=None, *, target=None):
     if not delta:
         return
 
-    old &= delta
-    new &= delta
+    result = {}
 
-    return {
-        "old": [entry._asdict() for entry in old],
-        "new": [entry._asdict() for entry in new],
-    }
+    for entry in delta:
+        result[entry.filename] = {
+            "old": entry._asdict() if entry in old else {},
+            "new": entry._asdict() if entry in new else {},
+        }
+
+    for filename, entry in result.items():
+        entry.update({"diff": compare_states(entry["old"], entry["new"])})
+
+    return result

tests/func/test_diff.py

@@ -11,14 +11,11 @@ def test_added(tmp_dir, scm, dvc):
     tmp_dir.dvc_gen("file", "text")
 
     result = {
-        "old": [],
-        "new": [
-            {
-                "filename": "file",
-                "checksum": "1cb251ec0d568de6a929b520c4aed8d1",
-                "size": 4,
-            }
-        ],
+        "file": {
+            "old": {},
+            "new": {"checksum": "1cb251ec0d568de6a929b520c4aed8d1", "size": 4},
+            "diff": {"status": "added", "size": 4},
+        }
     }
 
     assert result == dvc.diff()
@@ -26,17 +23,14 @@ def test_added(tmp_dir, scm, dvc):
 
 def test_deleted(tmp_dir, scm, dvc):
     tmp_dir.dvc_gen("file", "text", commit="add file")
-    (tmp_dir / "file").unlink()
+    (tmp_dir / "file.dvc").unlink()
 
     result = {
-        "old": [
-            {
-                "filename": "file",
-                "checksum": "1cb251ec0d568de6a929b520c4aed8d1",
-                "size": 4,
-            }
-        ],
-        "new": [],
+        "file": {
+            "old": {"checksum": "1cb251ec0d568de6a929b520c4aed8d1", "size": 4},
+            "new": {},
+            "diff": {"status": "deleted", "size": 4},
+        }
     }
 
     assert result == dvc.diff()
@@ -47,20 +41,11 @@ def test_modified(tmp_dir, scm, dvc):
     tmp_dir.dvc_gen("file", "second")
 
     result = {
-        "old": [
-            {
-                "filename": "file",
-                "checksum": "8b04d5e3775d298e78455efc5ca404d5",
-                "size": 6,
-            }
-        ],
-        "new": [
-            {
-                "filename": "file",
-                "checksum": "a9f0e61a137d86aa9db53465e0801612",
-                "size": 6,
-            }
-        ],
+        "file": {
+            "old": {"checksum": "8b04d5e3775d298e78455efc5ca404d5", "size": 6},
+            "new": {},
+            "diff": {"status": "deleted", "size": 6},
+        }
     }
 
     assert result == dvc.diff()