def test_uninitialized_engine_error(local_engine_empty):
# Test things like the audit triggers/splitgraph meta schema missing raise
# uninitialized engine errors rather than generic SQL errors.
try:
local_engine_empty.run_sql("DROP SCHEMA splitgraph_meta CASCADE")
with pytest.raises(EngineInitializationError) as e:
lookup_repository("some/repo", include_local=True)
assert "splitgraph_meta" in str(e.value)
local_engine_empty.initialize()
local_engine_empty.commit()
local_engine_empty.run_sql("DROP SCHEMA splitgraph_api CASCADE")
with pytest.raises(EngineInitializationError) as e:
ObjectManager(local_engine_empty).get_downloaded_objects()
assert "splitgraph_api" in str(e.value)
local_engine_empty.initialize()
local_engine_empty.commit()
local_engine_empty.run_sql("DROP SCHEMA splitgraph_audit CASCADE")
with pytest.raises(EngineInitializationError) as e:
local_engine_empty.discard_pending_changes("some/repo")
assert "Audit triggers" in str(e.value)
finally:
local_engine_empty.initialize()
local_engine_empty.commit()
def test_repo_lookup_override(remote_engine):
test_repo = Repository("overridden", "repo", engine=remote_engine)
try:
test_repo.init()
assert lookup_repository("overridden/repo") == test_repo
finally:
test_repo.delete(unregister=True, uncheckout=True)
def _get_local_image_for_import(hash_or_tag: str,
repository: Repository) -> Tuple[Image, bool]:
"""
Converts a remote repository and tag into an Image object that exists on the engine,
optionally pulling the repository or cloning it into a temporary location.
:param hash_or_tag: Hash/tag
:param repository: Name of the repository (doesn't need to be local)
:return: Image object and a boolean flag showing whether the repository should be deleted
when the image is no longer needed.
"""
tmp_repo = Repository(repository.namespace,
repository.repository + "_tmp_clone")
repo_is_temporary = False
logging.info("Resolving repository %s", repository)
source_repo = lookup_repository(repository.to_schema(), include_local=True)
if source_repo.engine.name != "LOCAL":
clone(source_repo, local_repository=tmp_repo, download_all=False)
source_image = tmp_repo.images[hash_or_tag]
repo_is_temporary = True
else:
# For local repositories, first try to pull them to see if they are clones of a remote.
if source_repo.upstream:
source_repo.pull()
source_image = source_repo.images[hash_or_tag]
return source_image, repo_is_temporary
def _execute_from(
node: Node,
output: Repository) -> Tuple[Repository, Optional[ProvenanceLine]]:
interesting_nodes = extract_nodes(node, ["repo_source", "repository"])
repo_source = get_first_or_none(interesting_nodes, "repo_source")
output_node = get_first_or_none(interesting_nodes, "repository")
provenance: Optional[ProvenanceLine] = None
if output_node:
# AS (output) detected, change the current output repository to it.
output = Repository.from_schema(output_node.match.group(0))
logging.info("Changed output repository to %s" % str(output))
# NB this destroys all data in the case where we ran some commands in the Splitfile and then
# did FROM (...) without AS repository
if repository_exists(output):
logging.info("Clearing all output from %s" % str(output))
output.delete()
if not repository_exists(output):
output.init()
if repo_source:
repository, tag_or_hash = parse_image_spec(repo_source)
source_repo = lookup_repository(repository.to_schema(),
include_local=True)
if source_repo.engine.name == "LOCAL":
# For local repositories, make sure to update them if they've an upstream
if source_repo.upstream:
source_repo.pull()
# Get the target image hash from the source repo: otherwise, if the tag is, say, 'latest' and
# the output has just had the base commit (000...) created in it, that commit will be the latest.
clone(source_repo, local_repository=output, download_all=False)
source_hash = source_repo.images[tag_or_hash].image_hash
output.images.by_hash(source_hash).checkout()
provenance = {
"type": "FROM",
"source_namespace": source_repo.namespace,
"source": source_repo.repository,
"source_hash": source_hash,
}
else:
# FROM EMPTY AS repository -- initializes an empty repository (say to create a table or import
# the results of a previous stage in a multistage build.
# In this case, if AS repository has been specified, it's already been initialized. If not, this command
# literally does nothing
if not output_node:
raise SplitfileError(
"FROM EMPTY without AS (repository) does nothing!")
return output, provenance
def test_repo_lookup_override_fail():
with pytest.raises(RepositoryNotFoundError) as e:
lookup_repository("does/not_exist")
assert "Unknown repository" in str(e.value)
