From 2a1af0b880aa2e1eebb8e0d67ac7aff5d530befc Mon Sep 17 00:00:00 2001
From: Peter Rowlands <peter@pmrowla.com>
Date: Mon, 21 Nov 2022 18:05:18 +0900
Subject: [PATCH] add: merge existing file meta for versioned dirs

---
 dvc/output.py         | 22 ++++++++++++++++++++++
 dvc/repo/add.py       |  2 +-
 dvc/stage/__init__.py | 33 ++++++++++++++++++++++++++++++---
 3 files changed, 53 insertions(+), 4 deletions(-)

diff --git a/dvc/output.py b/dvc/output.py
index 37f763fe6a..00f471778c 100644
--- a/dvc/output.py
+++ b/dvc/output.py
@@ -1203,6 +1203,28 @@ def restore_fields(self, other: "Output"):
         self.remote = other.remote
         self.can_push = other.can_push
 
+    def merge_version_meta(self, other: "Output"):
+        """Merge version meta for files which are unchanged from other."""
+        if not self.hash_info:
+            return
+        if self.hash_info.isdir:
+            return self._merge_dir_version_meta(other)
+        if self.hash_info != other.hash_info:
+            return
+        self.meta = other.meta
+
+    def _merge_dir_version_meta(self, other: "Output"):
+        from dvc_data.hashfile.tree import update_meta
+
+        if not self.obj or not other.hash_info.isdir:
+            return
+        other_obj = other.obj if other.obj is not None else other.get_obj()
+        assert isinstance(self.obj, Tree) and isinstance(other_obj, Tree)
+        updated = update_meta(self.obj, other_obj)
+        assert updated.hash_info == self.obj.hash_info
+        self.obj = updated
+        self.files = updated.as_list(with_meta=True)
+
 
 META_SCHEMA = {
     Meta.PARAM_SIZE: int,
diff --git a/dvc/repo/add.py b/dvc/repo/add.py
index 3c28316d00..be341724c2 100644
--- a/dvc/repo/add.py
+++ b/dvc/repo/add.py
@@ -187,7 +187,7 @@ def add(  # noqa: C901
                 stage.transfer(source, to_remote=to_remote, odb=odb, **kwargs)
             else:
                 try:
-                    stage.save()
+                    stage.save(merge_versioned=True)
                     if not no_commit:
                         stage.commit()
                 except CacheLinkError:
diff --git a/dvc/stage/__init__.py b/dvc/stage/__init__.py
index 1d47840e5c..6bdb5446c6 100644
--- a/dvc/stage/__init__.py
+++ b/dvc/stage/__init__.py
@@ -463,10 +463,12 @@ def compute_md5(self):
         logger.debug("Computed %s md5: '%s'", self, m)
         return m
 
-    def save(self, allow_missing=False):
+    def save(self, allow_missing: bool = False, merge_versioned: bool = False):
         self.save_deps(allow_missing=allow_missing)
 
-        self.save_outs(allow_missing=allow_missing)
+        self.save_outs(
+            allow_missing=allow_missing, merge_versioned=merge_versioned
+        )
         self.md5 = self.compute_md5()
 
         self.repo.stage_cache.save(self)
@@ -481,15 +483,40 @@ def save_deps(self, allow_missing=False):
                 if not allow_missing:
                     raise
 
-    def save_outs(self, allow_missing=False):
+    def save_outs(
+        self, allow_missing: bool = False, merge_versioned: bool = False
+    ):
         from dvc.output import OutputDoesNotExistError
 
+        from .exceptions import StageFileDoesNotExistError, StageNotFound
+
+        if merge_versioned:
+            try:
+                old = self.reload()
+                old_outs = {out.def_path: out for out in old.outs}
+                merge_versioned = any(
+                    (
+                        out.files is not None
+                        or (
+                            out.meta is not None
+                            and out.meta.version_id is not None
+                        )
+                    )
+                    for out in old_outs.values()
+                )
+            except (StageFileDoesNotExistError, StageNotFound):
+                merge_versioned = False
+
         for out in self.outs:
             try:
                 out.save()
             except OutputDoesNotExistError:
                 if not (allow_missing or out.checkpoint):
                     raise
+            if merge_versioned:
+                old_out = old_outs.get(out.def_path)
+                if old_out is not None:
+                    out.merge_version_meta(old_out)
 
     def ignore_outs(self):
         for out in self.outs: