Merge pull request #668 from splitgraph/feature/writeable-lq-splitfiles

Add support for writeable LQ to Splitfiles

splitgraph/commandline/cloud.py

@@ -968,7 +968,9 @@ def upload_c(remote, file_format, repository, files):
 
     wait_for_load(client, repository.namespace, repository.repository, task_id)
 
-    web_url = _construct_repo_url(gql_endpoint=client.endpoint, full_repo=repository) + "/-/tables"
+    web_url = (
+        _construct_repo_url(gql_endpoint=client.endpoint, full_repo=repository) + "/latest/-/tables"
+    )
     click.echo()
     click.echo(
         "Success. See the repository at " + Color.BLUE + web_url + Color.END + " or query it with:"

splitgraph/commandline/splitfile.py

@@ -21,7 +21,13 @@
 @click.option(
     "-o", "--output-repository", help="Repository to store the result in.", type=RepositoryType()
 )
-def build_c(splitfile, args, output_repository):
+@click.option(
+    "-l",
+    "--layered-querying",
+    help="Use writeable layered querying when checking images out. Experimental.",
+    is_flag=True,
+)
+def build_c(splitfile, args, output_repository, layered_querying):
     """
     Build Splitgraph images.
 
@@ -54,7 +60,9 @@ def build_c(splitfile, args, output_repository):
         file_name = os.path.splitext(os.path.basename(splitfile.name))[0]
         output_repository = Repository.from_schema(file_name)
 
-    execute_commands(splitfile.read(), args, output=output_repository)
+    execute_commands(
+        splitfile.read(), args, output=output_repository, use_writeable_lq=layered_querying
+    )
 
 
 @click.command(name="provenance")
@@ -205,7 +213,13 @@ def dependents_c(image_spec, source_on, dependents_on):
     help="Images to substitute into the reconstructed Splitfile, of the form"
     " [NAMESPACE/]REPOSITORY[:HASH_OR_TAG]. Default tag is 'latest'.",
 )
-def rebuild_c(image_spec, update, against):
+@click.option(
+    "-l",
+    "--layered-querying",
+    help="Use writeable layered querying when checking images out. Experimental.",
+    is_flag=True,
+)
+def rebuild_c(image_spec, update, against, layered_querying):
     """
     Rebuild images against different dependencies.
 
@@ -244,4 +258,4 @@ def rebuild_c(image_spec, update, against):
 
     from splitgraph.splitfile.execution import rebuild_image
 
-    rebuild_image(image, new_images)
+    rebuild_image(image, new_images, use_writeable_lq=layered_querying)

splitgraph/splitfile/execution.py

@@ -38,13 +38,15 @@ def _combine_hashes(hashes: List[str]) -> str:
     return sha256("".join(hashes).encode("ascii")).hexdigest()
 
 
-def _checkout_or_calculate_layer(output: Repository, image_hash: str, calc_func: Callable) -> None:
+def _checkout_or_calculate_layer(
+    output: Repository, image_hash: str, calc_func: Callable, use_writeable_lq: bool = False
+) -> None:
     # Future optimization here: don't actually check the layer out if it exists -- only do it at Splitfile execution
     # end or when a command needs it.
 
     # Have we already calculated this hash?
     try:
-        output.images.by_hash(image_hash).checkout()
+        output.images.by_hash(image_hash).checkout(layered=use_writeable_lq)
         logging.info(" ---> Using cache")
     except ImageNotFoundError:
         try:
@@ -64,6 +66,7 @@ def execute_commands(
     params: Optional[Dict[str, str]] = None,
     output: Optional[Repository] = None,
     output_base: str = "0" * 32,
+    use_writeable_lq: bool = False,
 ) -> None:
     """
     Executes a series of Splitfile commands.
@@ -74,6 +77,7 @@ def execute_commands(
     :param output: Output repository to execute the Splitfile against.
     :param output_base: If not None, a revision that gets checked out for all Splitfile actions to be committed
         on top of it.
+    :param use_writeable_lq: Use writeable LQ to execute commands
     """
     if params is None:
         params = {}
@@ -152,16 +156,18 @@ def _initialize_output(output):
         raise
 
 
-def checkout_if_changed(repository: Repository, image_hash: str) -> None:
+def checkout_if_changed(
+    repository: Repository, image_hash: str, use_writeable_lq: bool = False
+) -> None:
     if (
         repository.head is None
         or (repository.head.image_hash != image_hash)
         or repository.has_pending_changes()
     ):
-        repository.images.by_hash(image_hash).checkout()
+        repository.images.by_hash(image_hash).checkout(layered=use_writeable_lq)
     else:
         logging.info(
-            "Skipping checkout of %s as %s has it checked out " "and there have been no changes",
+            "Skipping checkout of %s as %s has it checked out and there have been no changes",
             image_hash,
             repository,
         )
@@ -379,7 +385,9 @@ def _calc():
             source_mountpoint.delete()
 
 
-def _execute_custom(node: Node, output: Repository) -> ProvenanceLine:
+def _execute_custom(
+    node: Node, output: Repository, use_writeable_lq: bool = False
+) -> ProvenanceLine:
     assert output.head is not None
     command, args = parse_custom_command(node)
 
@@ -410,7 +418,7 @@ def _execute_custom(node: Node, output: Repository) -> ProvenanceLine:
     if command_hash is not None:
         image_hash = _combine_hashes([output_head, command_hash])
         try:
-            output.images.by_hash(image_hash).checkout()
+            output.images.by_hash(image_hash).checkout(layered=use_writeable_lq)
             logging.info(" ---> Using cache")
             return {"type": "CUSTOM"}
         except ImageNotFoundError:
@@ -425,14 +433,16 @@ def _execute_custom(node: Node, output: Repository) -> ProvenanceLine:
 
     # Check just in case if the new hash produced by the command already exists.
     try:
-        output.images.by_hash(image_hash).checkout()
+        output.images.by_hash(image_hash).checkout(layered=use_writeable_lq)
     except ImageNotFoundError:
         # Full command as a commit comment
         output.commit(image_hash, comment=node.text)
     return {"type": "CUSTOM"}
 
 
-def rebuild_image(image: Image, source_replacement: Dict[Repository, str]) -> None:
+def rebuild_image(
+    image: Image, source_replacement: Dict[Repository, str], use_writeable_lq: bool = False
+) -> None:
     """
     Recreates the Splitfile used to create a given image and reruns it, replacing its dependencies with a different
     set of versions.
@@ -444,4 +454,6 @@ def rebuild_image(image: Image, source_replacement: Dict[Repository, str]) -> No
         ignore_irreproducible=False, source_replacement=source_replacement
     )
     # Params are supposed to be stored in the commands already (baked in) -- what if there's sensitive data there?
-    execute_commands("\n".join(splitfile_commands), output=image.repository)
+    execute_commands(
+        "\n".join(splitfile_commands), output=image.repository, use_writeable_lq=use_writeable_lq
+    )

test/splitgraph/commandline/snapshots/test_cloud_jobs/test_csv_upload/True/sgr_cloud_upload_success.txt

@@ -3,5 +3,5 @@ Uploading the files...
  (SUCCESS) Waiting for task ID ingest_task
 
 
-Success. See the repository at http://www.example.com/someuser/somerepo_1/-/tables or query it with:
+Success. See the repository at http://www.example.com/someuser/somerepo_1/latest/-/tables or query it with:
     sgr cloud sql 'SELECT * FROM "someuser/somerepo_1"."base_df"'

test/splitgraph/commandline/test_splitfile.py

@@ -2,6 +2,7 @@
 from unittest import mock
 from unittest.mock import call
 
+import pytest
 from click.testing import CliRunner
 
 from splitgraph.commandline import build_c, dependents_c, provenance_c, rebuild_c
@@ -17,11 +18,17 @@ def test_splitfile_default():
             [RESOURCES + "import_remote_multiple.splitfile", "-a", "TAG", "latest"],
         )
     assert ec.mock_calls == [
-        call(mock.ANY, {"TAG": "latest"}, output=Repository("", "import_remote_multiple"))
+        call(
+            mock.ANY,
+            {"TAG": "latest"},
+            output=Repository("", "import_remote_multiple"),
+            use_writeable_lq=False,
+        )
     ]
 
 
-def test_splitfile(local_engine_empty, pg_repo_remote):
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_splitfile(local_engine_empty, pg_repo_remote, use_writeable_lq):
     runner = CliRunner()
 
     result = runner.invoke(
@@ -33,7 +40,8 @@ def test_splitfile(local_engine_empty, pg_repo_remote):
             "latest",
             "-o",
             "output",
-        ],
+        ]
+        + (["-l"] if use_writeable_lq else []),
     )
     assert result.exit_code == 0
     assert OUTPUT.run_sql("SELECT id, fruit, vegetable FROM join_table") == [
@@ -78,17 +86,22 @@ def test_splitfile(local_engine_empty, pg_repo_remote):
     assert "%s:%s" % (OUTPUT, OUTPUT.head.image_hash) in result.output
 
 
-def test_splitfile_rebuild_update(local_engine_empty, pg_repo_remote_multitag):
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_splitfile_rebuild_update(local_engine_empty, pg_repo_remote_multitag, use_writeable_lq):
     runner = CliRunner()
 
     result = runner.invoke(
         build_c,
-        [RESOURCES + "import_remote_multiple.splitfile", "-a", "TAG", "v1", "-o", "output"],
+        [RESOURCES + "import_remote_multiple.splitfile", "-a", "TAG", "v1", "-o", "output"]
+        + (["-l"] if use_writeable_lq else []),
     )
     assert result.exit_code == 0
 
     # Rerun the output:latest against v2 of the test/pg_mount
-    result = runner.invoke(rebuild_c, ["output:latest", "--against", "test/pg_mount:v2"])
+    result = runner.invoke(
+        rebuild_c,
+        ["output:latest", "--against", "test/pg_mount:v2"] + (["-l"] if use_writeable_lq else []),
+    )
     output_v2 = OUTPUT.head
     assert result.exit_code == 0
     v2 = pg_repo_remote_multitag.images["v2"]
@@ -98,7 +111,10 @@ def test_splitfile_rebuild_update(local_engine_empty, pg_repo_remote_multitag):
     # In this case, this should all resolve to the same version of test/pg_mount (v2) and not produce
     # any extra commits.
     curr_commits = OUTPUT.images()
-    result = runner.invoke(rebuild_c, ["output:latest", "-u"])
+    result = runner.invoke(
+        rebuild_c,
+        ["output:latest", "-u"] + (["-l"] if use_writeable_lq else []),
+    )
     assert result.exit_code == 0
     assert output_v2 == OUTPUT.head
     assert OUTPUT.images() == curr_commits

test/splitgraph/splitfile/test_execution.py

@@ -109,8 +109,13 @@ def test_local_import_splitfile(pg_repo_local):
     assert not OUTPUT.engine.table_exists(OUTPUT.to_schema(), "fruits")
 
 
-def test_advanced_splitfile(pg_repo_local):
-    execute_commands(load_splitfile("import_local_multiple_with_queries.splitfile"), output=OUTPUT)
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_advanced_splitfile(pg_repo_local, use_writeable_lq):
+    execute_commands(
+        load_splitfile("import_local_multiple_with_queries.splitfile"),
+        output=OUTPUT,
+        use_writeable_lq=use_writeable_lq,
+    )
 
     assert OUTPUT.engine.table_exists(OUTPUT.to_schema(), "my_fruits")
     assert OUTPUT.engine.table_exists(OUTPUT.to_schema(), "vegetables")
@@ -128,13 +133,22 @@ def test_advanced_splitfile(pg_repo_local):
     assert OUTPUT.run_sql("SELECT * FROM my_fruits") == [(2, "orange")]
 
 
-def test_splitfile_cached(pg_repo_local):
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_splitfile_cached(pg_repo_local, use_writeable_lq):
     # Check that no new commits/snaps are created if we rerun the same splitfile
-    execute_commands(load_splitfile("import_local_multiple_with_queries.splitfile"), output=OUTPUT)
+    execute_commands(
+        load_splitfile("import_local_multiple_with_queries.splitfile"),
+        output=OUTPUT,
+        use_writeable_lq=use_writeable_lq,
+    )
     images = OUTPUT.images()
     assert len(images) == 4
 
-    execute_commands(load_splitfile("import_local_multiple_with_queries.splitfile"), output=OUTPUT)
+    execute_commands(
+        load_splitfile("import_local_multiple_with_queries.splitfile"),
+        output=OUTPUT,
+        use_writeable_lq=use_writeable_lq,
+    )
     new_images = OUTPUT.images()
     assert new_images == images
 
@@ -205,8 +219,14 @@ def test_import_updating_splitfile_with_uploading(
 
 
 @pytest.mark.mounting
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
 def test_splitfile_end_to_end_with_uploading(
-    local_engine_empty, remote_engine, pg_repo_remote_multitag, mg_repo_remote, clean_minio
+    local_engine_empty,
+    remote_engine,
+    pg_repo_remote_multitag,
+    mg_repo_remote,
+    clean_minio,
+    use_writeable_lq,
 ):
     # An end-to-end test:
     #   * Create a derived dataset from some tables imported from the remote engine
@@ -217,7 +237,10 @@ def test_splitfile_end_to_end_with_uploading(
 
     # Do the same setting up first and run the splitfile against the remote data.
     execute_commands(
-        load_splitfile("import_remote_multiple.splitfile"), params={"TAG": "v1"}, output=OUTPUT
+        load_splitfile("import_remote_multiple.splitfile"),
+        params={"TAG": "v1"},
+        output=OUTPUT,
+        use_writeable_lq=use_writeable_lq,
     )
 
     remote_output = Repository(OUTPUT.namespace, OUTPUT.repository, remote_engine)
@@ -230,22 +253,31 @@ def test_splitfile_end_to_end_with_uploading(
     OUTPUT.objects.cleanup()
 
     stage_2 = R("output_stage_2")
-    execute_commands(load_splitfile("import_from_preuploaded_remote.splitfile"), output=stage_2)
+    execute_commands(
+        load_splitfile("import_from_preuploaded_remote.splitfile"),
+        output=stage_2,
+        use_writeable_lq=use_writeable_lq,
+    )
 
     assert stage_2.run_sql("SELECT id, name, fruit, vegetable FROM diet") == [
         (2, "James", "orange", "carrot")
     ]
 
 
 @pytest.mark.mounting
-def test_splitfile_schema_changes(pg_repo_local, mg_repo_local):
-    execute_commands(load_splitfile("schema_changes.splitfile"), output=OUTPUT)
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_splitfile_schema_changes(pg_repo_local, mg_repo_local, use_writeable_lq):
+    execute_commands(
+        load_splitfile("schema_changes.splitfile"), output=OUTPUT, use_writeable_lq=use_writeable_lq
+    )
     old_output_head = OUTPUT.head
 
     # Then, alter the dataset and rerun the splitfile.
     pg_repo_local.run_sql("INSERT INTO fruits VALUES (12, 'mayonnaise')")
     pg_repo_local.commit()
-    execute_commands(load_splitfile("schema_changes.splitfile"), output=OUTPUT)
+    execute_commands(
+        load_splitfile("schema_changes.splitfile"), output=OUTPUT, use_writeable_lq=use_writeable_lq
+    )
     new_output_head = OUTPUT.head
 
     old_output_head.checkout()
@@ -438,16 +470,16 @@ def test_splitfile_with_external_sql(readonly_pg_repo):
 
 
 @pytest.mark.registry
-def test_splitfile_inline_sql(readonly_pg_repo, pg_repo_local):
+@pytest.mark.parametrize("use_writeable_lq", [True, False])
+def test_splitfile_inline_sql(readonly_pg_repo, pg_repo_local, use_writeable_lq):
     # Test SQL commands accessing repos directly -- join a remote repo with
     # some local data.
 
     prepare_lq_repo(pg_repo_local, commit_after_every=False, include_pk=True)
     pg_repo_local.head.tag("v2")
 
     execute_commands(
-        load_splitfile("inline_sql.splitfile"),
-        output=OUTPUT,
+        load_splitfile("inline_sql.splitfile"), output=OUTPUT, use_writeable_lq=use_writeable_lq
     )
 
     new_head = OUTPUT.head