From baeeee4bcee45489afe9e9989bde019adcce968e Mon Sep 17 00:00:00 2001
From: Fariz Rahman <farizrahman4u@gmail.com>
Date: Tue, 28 Mar 2023 20:58:54 +0400
Subject: [PATCH] View updates (#2259)

* mp updates

* show commit id

* fix

* fix

* fmt
---
 deeplake/api/tests/test_views.py    | 19 ++++++++++++++
 deeplake/core/dataset/dataset.py    | 40 ++++++++++++++---------------
 deeplake/core/dataset/view_entry.py |  8 +++++-
 deeplake/core/storage/memory.py     |  6 +++++
 4 files changed, 51 insertions(+), 22 deletions(-)

diff --git a/deeplake/api/tests/test_views.py b/deeplake/api/tests/test_views.py
index af3d65f737..5f2e979e34 100644
--- a/deeplake/api/tests/test_views.py
+++ b/deeplake/api/tests/test_views.py
@@ -120,3 +120,22 @@ def test_vds_read_only(hub_cloud_path, hub_cloud_dev_token):
 
     assert view.base_storage.read_only == True
     assert view._vds.base_storage.read_only == True
+
+
+def test_view_from_different_commit(local_ds):
+    with local_ds as ds:
+        ds.create_tensor("x")
+        ds.x.extend(list(range(10)))
+        cid = ds.commit()
+        view = ds[4:9]
+        view.save_view(id="abcd")
+        ds.x.extend(list(range(10, 20)))
+        cid2 = ds.commit()
+        view2 = ds.load_view("abcd")
+        assert view2.commit_id == cid
+        assert ds.commit_id == cid2
+        assert not view2.is_optimized
+        view2.save_view(id="efg", optimize=True)
+        view3 = ds.load_view("efg")
+        assert ds.commit_id == cid2
+        assert view3.is_optimized
diff --git a/deeplake/core/dataset/dataset.py b/deeplake/core/dataset/dataset.py
index cf5bb3b9b2..0dee4cbb81 100644
--- a/deeplake/core/dataset/dataset.py
+++ b/deeplake/core/dataset/dataset.py
@@ -3077,6 +3077,7 @@ def _get_view(self, inherit_creds=True, creds: Optional[Dict] = None):
         )
 
         ds.index = Index()
+        ds.version_state = ds.version_state.copy()
         ds._checkout(commit_id, verbose=False)
         first_index_subscriptable = self.info.get("first-index-subscriptable", True)
         if first_index_subscriptable:
@@ -3117,21 +3118,17 @@ def get_views(self, commit_id: Optional[str] = None) -> List[ViewEntry]:
 
         Args:
             commit_id (str, optional): - Commit from which views should be returned.
-                - If not specified, views from current commit is returned.
-                - If not specified, views from the currently checked out commit will be returned.
+                - If not specified, views from all commits are returned.
 
         Returns:
             List[ViewEntry]: List of :class:`ViewEntry` instances.
         """
-        commit_id = commit_id or self.commit_id
         queries = self._read_queries_json()
-        f = lambda x: x["source-dataset-version"] == commit_id
-        ret = map(
-            partial(ViewEntry, dataset=self),
-            filter(f, queries),
-        )
-
-        return list(ret)
+        if commit_id is not None:
+            queries = filter(
+                lambda x: x["source-dataset-version"] == commit_id, queries
+            )
+        return list(map(partial(ViewEntry, dataset=self), queries))
 
     def get_view(self, id: str) -> ViewEntry:
         """Returns the dataset view corresponding to ``id``.
@@ -3189,18 +3186,15 @@ def load_view(
         Raises:
             KeyError: if view with given id does not exist.
         """
+        view = self.get_view(id)
         if optimize:
-            return (
-                self.get_view(id)
-                .optimize(
-                    tensors=tensors,
-                    num_workers=num_workers,
-                    scheduler=scheduler,
-                    progressbar=progressbar,
-                )
-                .load()
-            )
-        return self.get_view(id).load()
+            return view.optimize(
+                tensors=tensors,
+                num_workers=num_workers,
+                scheduler=scheduler,
+                progressbar=progressbar,
+            ).load()
+        return view.load()
 
     def delete_view(self, id: str):
         """Deletes the view with given view id.
@@ -3887,6 +3881,10 @@ def is_view(self) -> bool:
             or hasattr(self, "_view_entry")
         )
 
+    @property
+    def is_optimized(self) -> bool:
+        return not getattr(getattr(self, "_view_entry", None), "virtual", True)
+
     @property
     def min_view(self):
         """Returns a view of the dataset in which all tensors are sliced to have the same length as
diff --git a/deeplake/core/dataset/view_entry.py b/deeplake/core/dataset/view_entry.py
index b3004f165e..1be02b5cce 100644
--- a/deeplake/core/dataset/view_entry.py
+++ b/deeplake/core/dataset/view_entry.py
@@ -33,8 +33,12 @@ def message(self) -> str:
         """Returns the message with which the view was saved."""
         return self.info.get("message", "")
 
+    @property
+    def commit_id(self) -> str:
+        return self.info["source-dataset-version"]
+
     def __str__(self):
-        return f"View(id='{self.id}', message='{self.message}', virtual={self.virtual})"
+        return f"View(id='{self.id}', message='{self.message}', virtual={self.virtual}, commit_id={self.commit_id})"
 
     __repr__ = __str__
 
@@ -51,6 +55,8 @@ def load(self, verbose=True):
         Returns:
             Dataset: Loaded dataset view.
         """
+        if self.commit_id != self._ds.commit_id:
+            print(f"Loading view from commit id {self.commit_id}.")
         ds = self._ds._sub_ds(
             ".queries/" + (self.info.get("path") or self.info["id"]),
             lock=False,
diff --git a/deeplake/core/storage/memory.py b/deeplake/core/storage/memory.py
index ef7436d714..f15cb1f690 100644
--- a/deeplake/core/storage/memory.py
+++ b/deeplake/core/storage/memory.py
@@ -1,6 +1,7 @@
 from typing import Any, Dict
 from deeplake.core.storage.lru_cache import _get_nbytes
 from deeplake.core.storage.provider import StorageProvider
+import os
 
 
 class MemoryProvider(StorageProvider):
@@ -124,3 +125,8 @@ def __setstate__(self, state: str):
 
     def get_object_size(self, key: str) -> int:
         return _get_nbytes(self[key])
+
+    def subdir(self, path: str, read_only: bool = False):
+        sd = self.__class__(os.path.join(self.root, path))
+        sd.read_only = read_only
+        return sd