def make_parse_result(
config: RuntimeConfig, all_projects: Mapping[str, Project]
) -> ParseResult:
"""Make a ParseResult from the project configuration and the profile."""
# if any of these change, we need to reject the parser
vars_hash = FileHash.from_contents(
'\x00'.join([
getattr(config.args, 'vars', '{}') or '{}',
getattr(config.args, 'profile', '') or '',
getattr(config.args, 'target', '') or '',
__version__
])
)
profile_path = os.path.join(config.args.profiles_dir, 'profiles.yml')
with open(profile_path) as fp:
profile_hash = FileHash.from_contents(fp.read())
project_hashes = {}
for name, project in all_projects.items():
path = os.path.join(project.project_root, 'dbt_project.yml')
with open(path) as fp:
project_hashes[name] = FileHash.from_contents(fp.read())
return ParseResult(
vars_hash=vars_hash,
profile_hash=profile_hash,
project_hashes=project_hashes,
)
def build_manifest_state_check(self):
config = self.root_project
all_projects = self.all_projects
# if any of these change, we need to reject the parser
vars_hash = FileHash.from_contents(
'\x00'.join([
getattr(config.args, 'vars', '{}') or '{}',
getattr(config.args, 'profile', '') or '',
getattr(config.args, 'target', '') or '',
__version__
])
)
profile_path = os.path.join(config.args.profiles_dir, 'profiles.yml')
with open(profile_path) as fp:
profile_hash = FileHash.from_contents(fp.read())
project_hashes = {}
for name, project in all_projects.items():
path = os.path.join(project.project_root, 'dbt_project.yml')
with open(path) as fp:
project_hashes[name] = FileHash.from_contents(fp.read())
state_check = ManifestStateCheck(
vars_hash=vars_hash,
profile_hash=profile_hash,
project_hashes=project_hashes,
)
return state_check
def _mock_state_check(self):
config = self.root_project
all_projects = self.all_projects
return ManifestStateCheck(
vars_hash=FileHash.from_contents('vars'),
project_hashes={name: FileHash.from_contents(name) for name in all_projects},
profile_hash=FileHash.from_contents('profile'),
)
def basic_compiled_schema_test_node():
return CompiledSchemaTestNode(
package_name='test',
root_path='/root/',
path='/root/x/path.sql',
original_file_path='/root/path.sql',
raw_sql='select * from {{ ref("other") }}',
name='foo',
resource_type=NodeType.Test,
unique_id='model.test.foo',
fqn=['test', 'models', 'foo'],
refs=[],
sources=[],
depends_on=DependsOn(),
deferred=False,
description='',
database='test_db',
schema='test_schema',
alias='bar',
tags=[],
config=TestConfig(severity='warn'),
meta={},
compiled=True,
extra_ctes=[InjectedCTE('whatever', 'select * from other')],
extra_ctes_injected=True,
compiled_sql='with whatever as (select * from other) select * from whatever',
column_name='id',
test_metadata=TestMetadata(namespace=None, name='foo', kwargs={}),
checksum=FileHash.from_contents(''),
unrendered_config={
'severity': 'warn',
}
)
def basic_compiled_model():
return CompiledModelNode(
package_name='test',
root_path='/root/',
path='/root/models/foo.sql',
original_file_path='models/foo.sql',
raw_sql='select * from {{ ref("other") }}',
name='foo',
resource_type=NodeType.Model,
unique_id='model.test.foo',
fqn=['test', 'models', 'foo'],
refs=[],
sources=[],
depends_on=DependsOn(),
deferred=True,
description='',
database='test_db',
schema='test_schema',
alias='bar',
tags=[],
config=NodeConfig(),
meta={},
compiled=True,
extra_ctes=[InjectedCTE('whatever', 'select * from other')],
extra_ctes_injected=True,
compiled_sql='with whatever as (select * from other) select * from whatever',
checksum=FileHash.from_contents(''),
unrendered_config={}
)
def test_select_state_changed_seed_checksum_sha_to_sha(manifest,
previous_state, seed):
change_node(manifest,
seed.replace(checksum=FileHash.from_contents('changed')))
method = statemethod(manifest, previous_state)
assert search_manifest_using_method(manifest, method,
'modified') == {'seed'}
assert not search_manifest_using_method(manifest, method, 'new')
def make_model(pkg,
name,
sql,
refs=None,
sources=None,
tags=None,
path=None,
alias=None,
config_kwargs=None,
fqn_extras=None):
if refs is None:
refs = []
if sources is None:
sources = []
if tags is None:
tags = []
if path is None:
path = f'{name}.sql'
if alias is None:
alias = name
if config_kwargs is None:
config_kwargs = {}
if fqn_extras is None:
fqn_extras = []
fqn = [pkg] + fqn_extras + [name]
depends_on_nodes = []
source_values = []
ref_values = []
for ref in refs:
ref_values.append([ref.name])
depends_on_nodes.append(ref.unique_id)
for src in sources:
source_values.append([src.source_name, src.name])
depends_on_nodes.append(src.unique_id)
return ParsedModelNode(
raw_sql=sql,
database='dbt',
schema='dbt_schema',
alias=alias,
name=name,
fqn=fqn,
unique_id=f'model.{pkg}.{name}',
package_name=pkg,
root_path='/usr/dbt/some-project',
path=path,
original_file_path=f'models/{path}',
config=NodeConfig(**config_kwargs),
tags=tags,
refs=ref_values,
sources=source_values,
depends_on=DependsOn(nodes=depends_on_nodes),
resource_type=NodeType.Model,
checksum=FileHash.from_contents(''),
)
def load_source_file(path: FilePath, parse_file_type: ParseFileType,
project_name: str) -> SourceFile:
file_contents = load_file_contents(path.absolute_path, strip=False)
checksum = FileHash.from_contents(file_contents)
source_file = SourceFile(path=path,
checksum=checksum,
parse_file_type=parse_file_type,
project_name=project_name)
source_file.contents = file_contents.strip()
return source_file
def use_models(self, models):
for k, v in models.items():
path = FilePath(
searched_path='models',
project_root=os.path.normcase(os.getcwd()),
relative_path='{}.sql'.format(k),
)
# FileHash can't be empty or 'search_key' will be None
source_file = SourceFile(path=path, checksum=FileHash.from_contents('abc'))
source_file.contents = v
self.mock_models.append(source_file)
def load_seed_source_file(match: FilePath, project_name) -> SourceFile:
if match.seed_too_large():
# We don't want to calculate a hash of this file. Use the path.
source_file = SourceFile.big_seed(match)
else:
file_contents = load_file_contents(match.absolute_path, strip=False)
checksum = FileHash.from_contents(file_contents)
source_file = SourceFile(path=match, checksum=checksum)
source_file.contents = ''
source_file.parse_file_type = ParseFileType.Seed
source_file.project_name = project_name
return source_file
def test_select_state_changed_seed_checksum_path_to_path(
manifest, previous_state, seed):
change_node(
previous_state.manifest,
seed.replace(
checksum=FileHash(name='path', checksum=seed.original_file_path)))
change_node(
manifest,
seed.replace(
checksum=FileHash(name='path', checksum=seed.original_file_path)))
method = statemethod(manifest, previous_state)
with mock.patch(
'dbt.contracts.graph.parsed.warn_or_error') as warn_or_error_patch:
assert not search_manifest_using_method(manifest, method, 'modified')
warn_or_error_patch.assert_called_once()
msg = warn_or_error_patch.call_args[0][0]
assert msg.startswith('Found a seed (pkg.seed) >1MB in size')
with mock.patch(
'dbt.contracts.graph.parsed.warn_or_error') as warn_or_error_patch:
assert not search_manifest_using_method(manifest, method, 'new')
warn_or_error_patch.assert_not_called()
def make_data_test(pkg,
name,
sql,
refs=None,
sources=None,
tags=None,
path=None,
config_kwargs=None):
if refs is None:
refs = []
if sources is None:
sources = []
if tags is None:
tags = ['data']
if path is None:
path = f'{name}.sql'
if config_kwargs is None:
config_kwargs = {}
fqn = ['minimal', 'data_test', name]
depends_on_nodes = []
source_values = []
ref_values = []
for ref in refs:
ref_values.append([ref.name])
depends_on_nodes.append(ref.unique_id)
for src in sources:
source_values.append([src.source_name, src.name])
depends_on_nodes.append(src.unique_id)
return ParsedDataTestNode(
raw_sql=sql,
database='dbt',
schema='dbt_schema',
name=name,
alias=name,
fqn=fqn,
unique_id=f'test.{pkg}.{name}',
package_name=pkg,
root_path='/usr/dbt/some-project',
path=path,
original_file_path=f'tests/{path}',
config=TestConfig(**config_kwargs),
tags=tags,
refs=ref_values,
sources=source_values,
depends_on=DependsOn(nodes=depends_on_nodes),
resource_type=NodeType.Test,
checksum=FileHash.from_contents(''),
)
def make_seed(pkg,
name,
path=None,
loader=None,
alias=None,
tags=None,
fqn_extras=None,
checksum=None):
if alias is None:
alias = name
if tags is None:
tags = []
if path is None:
path = f'{name}.csv'
if fqn_extras is None:
fqn_extras = []
if checksum is None:
checksum = FileHash.from_contents('')
fqn = [pkg] + fqn_extras + [name]
return ParsedSeedNode(
raw_sql='',
database='dbt',
schema='dbt_schema',
alias=alias,
name=name,
fqn=fqn,
unique_id=f'seed.{pkg}.{name}',
package_name=pkg,
root_path='/usr/dbt/some-project',
path=path,
original_file_path=f'data/{path}',
tags=tags,
resource_type=NodeType.Seed,
checksum=FileHash.from_contents(''),
)
def file_block_for(self, data: str, filename: str, searched: str):
root_dir = get_abs_os_path('./dbt_modules/snowplow')
filename = normalize(filename)
path = FilePath(
searched_path=searched,
relative_path=filename,
project_root=root_dir,
)
source_file = SourceFile(
path=path,
checksum=FileHash.from_contents(data),
)
source_file.contents = data
return FileBlock(file=source_file)
def load_file(
self,
path: FilePath,
*,
set_contents: bool = True,
) -> SourceFile:
file_contents = load_file_contents(path.absolute_path, strip=False)
checksum = FileHash.from_contents(file_contents)
source_file = SourceFile(path=path, checksum=checksum)
if set_contents:
source_file.contents = file_contents.strip()
else:
source_file.contents = ''
return source_file
def test_select_state_changed_seed_checksum_path_to_sha(
manifest, previous_state, seed):
change_node(
previous_state.manifest,
seed.replace(
checksum=FileHash(name='path', checksum=seed.original_file_path)))
method = statemethod(manifest, previous_state)
with mock.patch(
'dbt.contracts.graph.parsed.warn_or_error') as warn_or_error_patch:
assert search_manifest_using_method(manifest, method,
'modified') == {'seed'}
warn_or_error_patch.assert_not_called()
with mock.patch(
'dbt.contracts.graph.parsed.warn_or_error') as warn_or_error_patch:
assert not search_manifest_using_method(manifest, method, 'new')
warn_or_error_patch.assert_not_called()
def create_test_node(
self,
target: Union[UnpatchedSourceDefinition, UnparsedNodeUpdate],
path: str,
config: ContextConfig,
tags: List[str],
fqn: List[str],
name: str,
raw_sql: str,
test_metadata: Dict[str, Any],
column_name: Optional[str],
) -> ParsedSchemaTestNode:
dct = {
'alias': name,
'schema': self.default_schema,
'database': self.default_database,
'fqn': fqn,
'name': name,
'root_path': self.project.project_root,
'resource_type': self.resource_type,
'tags': tags,
'path': path,
'original_file_path': target.original_file_path,
'package_name': self.project.project_name,
'raw_sql': raw_sql,
'unique_id': self.generate_unique_id(name),
'config': self.config_dict(config),
'test_metadata': test_metadata,
'column_name': column_name,
'checksum': FileHash.empty().to_dict(omit_none=True),
}
try:
ParsedSchemaTestNode.validate(dct)
return ParsedSchemaTestNode.from_dict(dct)
except ValidationError as exc:
msg = validator_error_message(exc)
# this is a bit silly, but build an UnparsedNode just for error
# message reasons
node = self._create_error_node(
name=target.name,
path=path,
original_file_path=target.original_file_path,
raw_sql=raw_sql,
)
raise CompilationException(msg, node=node) from exc
def setUp(self):
self.model = ParsedModelNode(
alias='model_one',
name='model_one',
database='dbt',
schema='analytics',
resource_type=NodeType.Model,
unique_id='model.root.model_one',
fqn=['root', 'model_one'],
package_name='root',
original_file_path='model_one.sql',
root_path='/usr/src/app',
refs=[],
sources=[],
depends_on=DependsOn(),
config=NodeConfig.from_dict({
'enabled': True,
'materialized': 'view',
'persist_docs': {},
'post-hook': [],
'pre-hook': [],
'vars': {},
'quoting': {},
'column_types': {},
'tags': [],
}),
tags=[],
path='model_one.sql',
raw_sql='',
description='',
columns={},
checksum=FileHash.from_contents(''),
)
self.context = mock.MagicMock()
self.provider = VarProvider({})
self.config = mock.MagicMock(config_version=2,
vars=self.provider,
cli_vars={},
project_name='root')
def test__prepend_ctes__already_has_cte(self):
ephemeral_config = self.model_config.replace(materialized='ephemeral')
manifest = Manifest(
macros={},
nodes={
'model.root.view': CompiledModelNode(
name='view',
database='dbt',
schema='analytics',
alias='view',
resource_type=NodeType.Model,
unique_id='model.root.view',
fqn=['root', 'view'],
package_name='root',
root_path='/usr/src/app',
refs=[],
sources=[],
depends_on=DependsOn(nodes=['model.root.ephemeral']),
config=self.model_config,
tags=[],
path='view.sql',
original_file_path='view.sql',
raw_sql='select * from {{ref("ephemeral")}}',
compiled=True,
extra_ctes_injected=False,
extra_ctes=[InjectedCTE(id='model.root.ephemeral', sql='select * from source_table')],
compiled_sql=(
'with cte as (select * from something_else) '
'select * from __dbt__cte__ephemeral'),
checksum=FileHash.from_contents(''),
),
'model.root.ephemeral': CompiledModelNode(
name='ephemeral',
database='dbt',
schema='analytics',
alias='view',
resource_type=NodeType.Model,
unique_id='model.root.ephemeral',
fqn=['root', 'ephemeral'],
package_name='root',
root_path='/usr/src/app',
refs=[],
sources=[],
depends_on=DependsOn(),
config=ephemeral_config,
tags=[],
path='ephemeral.sql',
original_file_path='ephemeral.sql',
raw_sql='select * from source_table',
compiled=True,
compiled_sql='select * from source_table',
extra_ctes_injected=False,
extra_ctes=[],
checksum=FileHash.from_contents(''),
),
},
sources={},
docs={},
disabled=[],
files={},
exposures={},
selectors={},
)
compiler = dbt.compilation.Compiler(self.config)
result, _ = compiler._recursively_prepend_ctes(
manifest.nodes['model.root.view'],
manifest,
{}
)
self.assertEqual(result, manifest.nodes['model.root.view'])
self.assertEqual(result.extra_ctes_injected, True)
self.assertEqualIgnoreWhitespace(
result.compiled_sql,
('with __dbt__cte__ephemeral as ('
'select * from source_table'
'), cte as (select * from something_else) '
'select * from __dbt__cte__ephemeral'))
self.assertEqual(
manifest.nodes['model.root.ephemeral'].extra_ctes_injected,
False)
def test__prepend_ctes__no_ctes(self):
manifest = Manifest(
macros={},
nodes={
'model.root.view': CompiledModelNode(
name='view',
database='dbt',
schema='analytics',
alias='view',
resource_type=NodeType.Model,
unique_id='model.root.view',
fqn=['root', 'view'],
package_name='root',
root_path='/usr/src/app',
refs=[],
sources=[],
depends_on=DependsOn(),
config=self.model_config,
tags=[],
path='view.sql',
original_file_path='view.sql',
raw_sql=('with cte as (select * from something_else) '
'select * from source_table'),
compiled=True,
extra_ctes_injected=False,
extra_ctes=[],
compiled_sql=('with cte as (select * from something_else) '
'select * from source_table'),
checksum=FileHash.from_contents(''),
),
'model.root.view_no_cte': CompiledModelNode(
name='view_no_cte',
database='dbt',
schema='analytics',
alias='view_no_cte',
resource_type=NodeType.Model,
unique_id='model.root.view_no_cte',
fqn=['root', 'view_no_cte'],
package_name='root',
root_path='/usr/src/app',
refs=[],
sources=[],
depends_on=DependsOn(),
config=self.model_config,
tags=[],
path='view.sql',
original_file_path='view.sql',
raw_sql='select * from source_table',
compiled=True,
extra_ctes_injected=False,
extra_ctes=[],
compiled_sql=('select * from source_table'),
checksum=FileHash.from_contents(''),
),
},
sources={},
docs={},
disabled=[],
files={},
exposures={},
selectors={},
)
compiler = dbt.compilation.Compiler(self.config)
result, _ = compiler._recursively_prepend_ctes(
manifest.nodes['model.root.view'],
manifest,
{}
)
self.assertEqual(
result,
manifest.nodes.get('model.root.view'))
self.assertTrue(result.extra_ctes_injected)
self.assertEqualIgnoreWhitespace(
result.compiled_sql,
manifest.nodes.get('model.root.view').compiled_sql)
compiler = dbt.compilation.Compiler(self.config)
result, _ = compiler._recursively_prepend_ctes(
manifest.nodes.get('model.root.view_no_cte'),
manifest,
{})
self.assertEqual(
result,
manifest.nodes.get('model.root.view_no_cte'))
self.assertTrue(result.extra_ctes_injected)
self.assertEqualIgnoreWhitespace(
result.compiled_sql,
manifest.nodes.get('model.root.view_no_cte').compiled_sql)
def make_schema_test(pkg,
test_name,
test_model,
test_kwargs,
path=None,
refs=None,
sources=None,
tags=None,
column_name=None):
kwargs = test_kwargs.copy()
ref_values = []
source_values = []
# this doesn't really have to be correct
if isinstance(test_model, ParsedSourceDefinition):
kwargs['model'] = "{{ source('" + test_model.source_name + \
"', '" + test_model.name + "') }}"
source_values.append([test_model.source_name, test_model.name])
else:
kwargs['model'] = "{{ ref('" + test_model.name + "')}}"
ref_values.append([test_model.name])
if column_name is not None:
kwargs['column_name'] = column_name
# whatever
args_name = test_model.search_name.replace(".", "_")
if column_name is not None:
args_name += '_' + column_name
node_name = f'{test_name}_{args_name}'
raw_sql = '{{ config(severity="ERROR") }}{{ test_' + \
test_name + '(**dbt_schema_test_kwargs) }}'
name_parts = test_name.split('.')
if len(name_parts) == 2:
namespace, test_name = name_parts
macro_depends = f'model.{namespace}.{test_name}'
elif len(name_parts) == 1:
namespace = None
macro_depends = f'model.dbt.{test_name}'
else:
assert False, f'invalid test name: {test_name}'
if path is None:
path = 'schema.yml'
if tags is None:
tags = ['schema']
if refs is None:
refs = []
if sources is None:
sources = []
depends_on_nodes = []
for ref in refs:
ref_values.append([ref.name])
depends_on_nodes.append(ref.unique_id)
for source in sources:
source_values.append([source.source_name, source.name])
depends_on_nodes.append(source.unique_id)
return ParsedSchemaTestNode(
raw_sql=raw_sql,
test_metadata=TestMetadata(
namespace=namespace,
name=test_name,
kwargs=kwargs,
),
database='dbt',
schema='dbt_postgres',
name=node_name,
alias=node_name,
fqn=['minimal', 'schema_test', node_name],
unique_id=f'test.{pkg}.{node_name}',
package_name=pkg,
root_path='/usr/dbt/some-project',
path=f'schema_test/{node_name}.sql',
original_file_path=f'models/{path}',
resource_type=NodeType.Test,
tags=tags,
refs=ref_values,
sources=[],
depends_on=DependsOn(macros=[macro_depends], nodes=depends_on_nodes),
column_name=column_name,
checksum=FileHash.from_contents(''),
)
def rpc(cls):
# ugh!
return cls(FileHash.empty(), FileHash.empty(), {})
