def __init__(self, uuid_columns: Set[str]) -> None:
self.__unique_uuid_columns = uuid_columns
self.__uuid_column_match = Or([String(u_col) for u_col in uuid_columns])
self.uuid_in_condition = FunctionCallMatch(
Or((String(ConditionFunctions.IN), String(ConditionFunctions.NOT_IN))),
(
self.formatted_uuid_pattern(),
Param("params", FunctionCallMatch(String("tuple"), None)),
),
)
self.uuid_condition = FunctionCallMatch(
Or(
[
String(op)
for op in FUNCTION_TO_OPERATOR
if op not in (ConditionFunctions.IN, ConditionFunctions.NOT_IN)
]
),
(
Or(
(
Param("literal_0", LiteralMatch(AnyOptionalString())),
self.formatted_uuid_pattern("_0"),
)
),
Or(
(
Param("literal_1", LiteralMatch(AnyOptionalString())),
self.formatted_uuid_pattern("_1"),
)
),
),
)
self.formatted: Optional[str] = None
def extract_granularity_from_query(query: Query, column: str) -> Optional[int]:
"""
This extracts the `granularity` from the `groupby` statement of the query.
The matches are essentially the reverse of `TimeSeriesProcessor.__group_time_function`.
"""
groupby = query.get_groupby()
column_match = ColumnMatch(None, String(column))
fn_match = FunctionCallMatch(
Param(
"time_fn",
Or(
[
String("toStartOfHour"),
String("toStartOfMinute"),
String("toStartOfDay"),
String("toDate"),
]
),
),
(column_match,),
with_optionals=True,
)
expr_match = FunctionCallMatch(
String("toDateTime"),
(
FunctionCallMatch(
String("multiply"),
(
FunctionCallMatch(
String("intDiv"),
(
FunctionCallMatch(String("toUInt32"), (column_match,)),
LiteralMatch(Param("granularity", Any(int))),
),
),
LiteralMatch(Param("granularity", Any(int))),
),
),
LiteralMatch(Any(str)),
),
)
for top_expr in groupby:
for expr in top_expr:
result = fn_match.match(expr)
if result is not None:
return GRANULARITY_MAPPING[result.string("time_fn")]
result = expr_match.match(expr)
if result is not None:
return result.integer("granularity")
return None
class DefaultIfNullFunctionMapper(FunctionCallMapper):
"""
If a function is being called on a column that doesn't exist, or is being
called on NULL, change the entire function to be NULL.
"""
function_match = FunctionCallMatch(StringMatch("identity"),
(LiteralMatch(value=Any(type(None))), ))
def attempt_map(
self,
expression: FunctionCall,
children_translator: SnubaClickhouseStrictTranslator,
) -> Optional[FunctionCall]:
# HACK: Quick fix to avoid this function dropping important conditions from the query
logical_functions = {"and", "or", "xor"}
if expression.function_name in logical_functions:
return None
parameters = tuple(
p.accept(children_translator) for p in expression.parameters)
for param in parameters:
# All impossible columns will have been converted to the identity function.
# So we know that if a function has the identity function as a parameter, we can
# collapse the entire expression.
fmatch = self.function_match.match(param)
if fmatch is not None:
return identity(Literal(None, None), expression.alias)
return None
class DefaultIfNullCurriedFunctionMapper(CurriedFunctionCallMapper):
"""
If a curried function is being called on a column that doesn't exist, or is being
called on NULL, change the entire function to be NULL.
"""
function_match = FunctionCallMatch(StringMatch("identity"),
(LiteralMatch(), ))
def attempt_map(
self,
expression: CurriedFunctionCall,
children_translator: SnubaClickhouseStrictTranslator,
) -> Optional[Union[CurriedFunctionCall, FunctionCall]]:
internal_function = expression.internal_function.accept(
children_translator)
assert isinstance(internal_function, FunctionCall) # mypy
parameters = tuple(
p.accept(children_translator) for p in expression.parameters)
for param in parameters:
# All impossible columns that have been converted to NULL will be the identity function.
# So we know that if a function has the identity function as a parameter, we can
# collapse the entire expression.
fmatch = self.function_match.match(param)
if fmatch is not None:
return identity(Literal(None, None), expression.alias)
return None
def process_query(self, query: Query,
query_settings: QuerySettings) -> None:
having_clause = query.get_having()
if not having_clause:
return None
selected_columns = query.get_selected_columns()
uniq_matcher = Param("function", FunctionCallMatch(String("uniq")))
found_functions = []
for exp in having_clause:
match = uniq_matcher.match(exp)
if match is not None:
found_functions.append(match.expression("function"))
if found_functions is not None:
matcher = _ExpressionOrAliasMatcher(found_functions)
for col in selected_columns:
col.expression.accept(matcher)
if not all(matcher.found_expressions):
should_throw = get_config("throw_on_uniq_select_and_having",
False)
error = MismatchedAggregationException(
"Aggregation is in HAVING clause but not SELECT",
query=str(query))
if should_throw:
raise error
else:
logging.warning(
"Aggregation is in HAVING clause but not SELECT",
exc_info=True,
extra=cast(Dict[str, Any], error.to_dict()),
)
def formatted_uuid_pattern(self, suffix: str = "") -> FunctionCallMatch:
return FunctionCallMatch(
String("replaceAll"),
(
FunctionCallMatch(
String("toString"),
(
Param(
"formatted_uuid_column" + suffix,
ColumnMatch(None, self.__uuid_column_match),
),
),
),
),
with_optionals=True,
)
def __init__(self, columns: Set[str]):
self.columns = columns
column_match = Or([String(col) for col in columns])
literal = Param("literal", LiteralMatch(AnyMatch(str)))
operator = Param(
"operator",
Or(
[
String(op)
for op in FUNCTION_TO_OPERATOR
if op not in (ConditionFunctions.IN, ConditionFunctions.NOT_IN)
]
),
)
in_operators = Param(
"operator",
Or((String(ConditionFunctions.IN), String(ConditionFunctions.NOT_IN))),
)
col = Param("col", ColumnMatch(None, column_match))
self.__condition_matcher = Or(
[
FunctionCallMatch(operator, (literal, col)),
FunctionCallMatch(operator, (col, literal)),
FunctionCallMatch(Param("operator", String("has")), (col, literal)),
]
)
self.__in_condition_matcher = FunctionCallMatch(
in_operators,
(
col,
Param(
"tuple",
FunctionCallMatch(String("tuple"), all_parameters=LiteralMatch()),
),
),
)
def build_match(
col: str, ops: Sequence[str], param_type: Any
) -> Or[Expression]:
# The IN condition has to be checked separately since each parameter
# has to be checked individually.
column_match = ColumnMatch(alias_match, StringMatch(col))
return Or(
[
FunctionCallMatch(
Or([StringMatch(op) for op in ops]),
(column_match, LiteralMatch(AnyMatch(param_type))),
),
FunctionCallMatch(
StringMatch(ConditionFunctions.IN),
(
column_match,
FunctionCallMatch(
Or([StringMatch("array"), StringMatch("tuple")]),
all_parameters=LiteralMatch(AnyMatch(param_type)),
),
),
),
]
)
class DefaultIfNullCurriedFunctionMapper(CurriedFunctionCallMapper):
"""
If a curried function is being called on a column that doesn't exist, or is being
called on NULL, change the entire function to be NULL.
"""
function_match = FunctionCallMatch(StringMatch("ifNull"),
(LiteralMatch(), LiteralMatch()))
def attempt_map(
self,
expression: CurriedFunctionCall,
children_translator: SnubaClickhouseStrictTranslator,
) -> Optional[CurriedFunctionCall]:
internal_function = expression.internal_function.accept(
children_translator)
assert isinstance(internal_function, FunctionCall) # mypy
parameters = tuple(
p.accept(children_translator) for p in expression.parameters)
all_null = True
for param in parameters:
# Handle wrapped functions that have been converted to ifNull(NULL, NULL)
fmatch = self.function_match.match(param)
if fmatch is None:
if isinstance(param, Literal):
if param.value is not None:
all_null = False
break
else:
all_null = False
break
if all_null and len(parameters) > 0:
# Currently curried function mappers require returning other curried functions.
# So return this to keep the mapper happy.
return CurriedFunctionCall(
alias=expression.alias,
internal_function=FunctionCall(
None,
f"{internal_function.function_name}OrNull",
internal_function.parameters,
),
parameters=tuple(Literal(None, None) for p in parameters),
)
return None
class DefaultIfNullFunctionMapper(FunctionCallMapper):
"""
If a function is being called on a column that doesn't exist, or is being
called on NULL, change the entire function to be NULL.
"""
function_match = FunctionCallMatch(StringMatch("ifNull"),
(LiteralMatch(), LiteralMatch()))
def attempt_map(
self,
expression: FunctionCall,
children_translator: SnubaClickhouseStrictTranslator,
) -> Optional[FunctionCall]:
parameters = tuple(
p.accept(children_translator) for p in expression.parameters)
all_null = True
for param in parameters:
# Handle wrapped functions that have been converted to ifNull(NULL, NULL)
fmatch = self.function_match.match(param)
if fmatch is None:
if isinstance(param, Literal):
if param.value is not None:
all_null = False
break
else:
all_null = False
break
if all_null and len(parameters) > 0:
# Currently function mappers require returning other functions. So return this
# to keep the mapper happy.
return FunctionCall(expression.alias, "ifNull",
(Literal(None, None), Literal(None, None)))
return None
# {"a": f(x), "x": g(k)} -> {"a": f(g(k)), "x": g(k)}
aliased_expressions = {
exp.alias: exp
for exp in query.get_all_expressions() if exp.alias is not None
}
fully_resolved_aliases = {
alias: exp.accept(
AliasExpanderVisitor(aliased_expressions, [], expand_nested=True))
for alias, exp in aliased_expressions.items()
}
visitor = AliasExpanderVisitor(fully_resolved_aliases, [])
query.transform(visitor)
ARRAYJOIN_FUNCTION_MATCH = FunctionCallMatch(StringMatch("arrayJoin"), None)
def _validate_arrayjoin(query: Query) -> None:
# TODO: Actually validate arrayjoin. For now log how it is used.
body_arrayjoin = ""
arrayjoin = query.get_arrayjoin_from_ast()
if arrayjoin is not None:
if isinstance(arrayjoin, Column):
body_arrayjoin = arrayjoin.column_name
array_joins = set()
if body_arrayjoin:
array_joins.add(body_arrayjoin)
for exp in query.get_all_expressions():
match = ARRAYJOIN_FUNCTION_MATCH.match(exp)
selected_entity = match_query_to_entity(query, EVENTS_COLUMNS,
TRANSACTIONS_COLUMNS)
track_bad_query(query, selected_entity, EVENTS_COLUMNS,
TRANSACTIONS_COLUMNS)
return selected_entity
metrics = MetricsWrapper(environment.metrics, "api.discover")
logger = logging.getLogger(__name__)
EVENT_CONDITION = FunctionCallMatch(
Param("function", Or([StringMatch(op) for op in BINARY_OPERATORS])),
(
Or([ColumnMatch(None, StringMatch("type")),
LiteralMatch(None)]),
Param("event_type", Or([ColumnMatch(), LiteralMatch()])),
),
)
TRANSACTION_FUNCTIONS = FunctionCallMatch(
Or([StringMatch("apdex"),
StringMatch("failure_rate")]), None)
EVENT_FUNCTIONS = FunctionCallMatch(
Or([StringMatch("isHandled"),
StringMatch("notHandled")]), None)
def match_query_to_entity(
query: Query,
def __init__(self, time_group_columns: Mapping[str, str],
time_parse_columns: Sequence[str]) -> None:
# Column names that should be mapped to different columns.
self.__time_replace_columns = time_group_columns
# time_parse_columns is a list of columns that, if used in a condition, should be compared with datetimes.
# The columns here might overlap with the columns that get replaced, so we have to search for transformed
# columns.
column_match = ColumnMatch(
None,
Param(
"column_name",
Or([String(tc) for tc in time_parse_columns]),
),
)
self.condition_match = FunctionCallMatch(
Or([
String(ConditionFunctions.GT),
String(ConditionFunctions.GTE),
String(ConditionFunctions.LT),
String(ConditionFunctions.LTE),
String(ConditionFunctions.EQ),
String(ConditionFunctions.NEQ),
]),
(
Or([
column_match,
FunctionCallMatch(
Or([
String("toStartOfHour"),
String("toStartOfMinute"),
String("toStartOfDay"),
String("toDate"),
]),
(column_match, ),
with_optionals=True,
),
FunctionCallMatch(
String("toDateTime"),
(
FunctionCallMatch(
String("multiply"),
(
FunctionCallMatch(
String("intDiv"),
(
FunctionCallMatch(
String("toUInt32"),
(column_match, ),
),
LiteralMatch(Any(int)),
),
),
LiteralMatch(Any(int)),
),
),
LiteralMatch(Any(str)),
),
),
]),
Param("literal", LiteralMatch(Any(str))),
),
)
from snuba.utils.metrics.wrapper import MetricsWrapper
EVENTS = EntityKey.EVENTS
TRANSACTIONS = EntityKey.TRANSACTIONS
EVENTS_AND_TRANSACTIONS = "events_and_transactions"
metrics = MetricsWrapper(environment.metrics, "api.discover")
logger = logging.getLogger(__name__)
EVENT_CONDITION = FunctionCallMatch(
None,
Param("function", Or([StringMatch(op) for op in BINARY_OPERATORS])),
(
Or(
[
ColumnMatch(None, None, StringMatch("type")),
LiteralMatch(StringMatch("type"), None),
]
),
Param("event_type", Or([ColumnMatch(), LiteralMatch()])),
),
)
def match_query_to_table(
query: Query, events_only_columns: ColumnSet, transactions_only_columns: ColumnSet
) -> Union[EntityKey, str]:
# First check for a top level condition on the event type
condition = query.get_condition_from_ast()
event_types = set()
if condition:
def __init__(self, columns: Set[str], optimize_ordering: bool = False):
self.columns = columns
self.optimize_ordering = optimize_ordering
column_match = Or([String(col) for col in columns])
literal = Param("literal", LiteralMatch(AnyMatch(str)))
ordering_operators = (
ConditionFunctions.GT,
ConditionFunctions.GTE,
ConditionFunctions.LT,
ConditionFunctions.LTE,
)
operator = Param(
"operator",
Or([
String(op) for op in (
ConditionFunctions.EQ,
ConditionFunctions.NEQ,
ConditionFunctions.IS_NULL,
ConditionFunctions.IS_NOT_NULL,
*(ordering_operators if self.optimize_ordering else ()),
)
]),
)
unoptimizable_operator = Param(
"operator",
Or([
String(op) for op in (
ConditionFunctions.LIKE,
ConditionFunctions.NOT_LIKE,
*(() if self.optimize_ordering else ordering_operators),
)
]),
)
in_operators = Param(
"operator",
Or((String(ConditionFunctions.IN),
String(ConditionFunctions.NOT_IN))),
)
col = Param("col", ColumnMatch(None, column_match))
self.__condition_matcher = Or([
FunctionCallMatch(operator, (literal, col)),
FunctionCallMatch(operator, (col, literal)),
FunctionCallMatch(Param("operator", String("has")),
(col, literal)),
])
self.__in_condition_matcher = FunctionCallMatch(
in_operators,
(
col,
Param(
"tuple",
FunctionCallMatch(String("tuple"),
all_parameters=LiteralMatch()),
),
),
)
self.__unoptimizable_condition_matcher = Or([
FunctionCallMatch(unoptimizable_operator, (literal, col)),
FunctionCallMatch(unoptimizable_operator, (col, literal)),
])
def __post_init__(self) -> None:
self.function_match = FunctionCallMatch(
Or([StringMatch(func) for func in self.function_names]))
if not isinstance(exp, Column):
return exp
parts = exp.column_name.split(".", 1)
if len(parts) != 2 or parts[0] not in aliases:
raise ParsingException(
f"column {exp.column_name} must be qualified in a join query"
)
return Column(exp.alias, parts[0], parts[1])
query.transform_expressions(transform)
DATETIME_MATCH = FunctionCallMatch(
StringMatch("toDateTime"), (Param("date_string", LiteralMatch(AnyMatch(str))),)
)
def _parse_datetime_literals(
query: Union[CompositeQuery[QueryEntity], LogicalQuery]
) -> None:
def parse(exp: Expression) -> Expression:
result = DATETIME_MATCH.match(exp)
if result is not None:
date_string = result.expression("date_string")
assert isinstance(date_string, Literal) # mypy
assert isinstance(date_string.value, str) # mypy
return Literal(exp.alias, parse_datetime(date_string.value))
return exp
