def _format_groupby(query: AbstractQuery,
formatter: ExpressionVisitor[str]) -> Optional[StringNode]:
group_clause: Optional[StringNode] = None
ast_groupby = query.get_groupby()
if ast_groupby:
groupby_expressions = [e.accept(formatter) for e in ast_groupby]
group_clause_str = f"{', '.join(groupby_expressions)}"
if query.has_totals():
group_clause_str = f"{group_clause_str} WITH TOTALS"
group_clause = StringNode(f"GROUP BY {group_clause_str}")
return group_clause
def _validate_groupby_fields_have_matching_conditions(
query: Query, alias: Optional[str] = None) -> None:
"""
Method that insures that for every field in the group by clause, there should be a
matching a condition. For example, if we had in our groupby clause [project_id, tags[3]],
we should have the following conditions in the where clause `project_id = 3 AND tags[3]
IN array(1,2,3)`. This is necessary because we want to avoid the case where an
unspecified number of buckets is returned.
"""
condition = query.get_condition()
top_level = get_first_level_and_conditions(
condition) if condition else []
for exp in query.get_groupby():
key: Optional[str] = None
if isinstance(exp, SubscriptableReferenceExpr):
column_name = str(exp.column.column_name)
key = str(exp.key.value)
elif isinstance(exp, Column):
column_name = exp.column_name
else:
raise InvalidQueryException(
"Unhandled column type in group by validation")
match = build_match(
col=column_name,
ops=[ConditionFunctions.EQ],
param_type=int,
alias=alias,
key=key,
)
found = any(match.match(cond) for cond in top_level)
if not found:
raise InvalidQueryException(
f"Every field in groupby must have a corresponding condition in "
f"where clause. missing condition for field {exp}")