def test_format_clickhouse_specific_query() -> None:
"""
Adds a few of the Clickhosue specific fields to the query.
"""
query = Query(
{
"sample": 0.1,
"totals": True,
"limitby": (10, "environment")
},
TableSource("my_table", ColumnSet([])),
selected_columns=[
SelectedExpression("column1", Column(None, None, "column1")),
SelectedExpression("column2", Column(None, "table1", "column2")),
],
condition=binary_condition(
None,
"eq",
lhs=Column(None, None, "column1"),
rhs=Literal(None, "blabla"),
),
groupby=[
Column(None, None, "column1"),
Column(None, "table1", "column2")
],
having=binary_condition(
None,
"eq",
lhs=Column(None, None, "column1"),
rhs=Literal(None, 123),
),
order_by=[
OrderBy(OrderByDirection.ASC, Column(None, None, "column1"))
],
array_join=Column(None, None, "column1"),
)
query.set_final(True)
query.set_offset(50)
query.set_limit(100)
request_settings = HTTPRequestSettings()
clickhouse_query = AstSqlQuery(query, request_settings)
expected = {
"from": "FROM my_table FINAL SAMPLE 0.1",
"group": "GROUP BY (column1, table1.column2) WITH TOTALS",
"having": "HAVING eq(column1, 123)",
"array_join": "ARRAY JOIN column1",
"limit": "LIMIT 100 OFFSET 50",
"limitby": "LIMIT 10 BY environment",
"order": "ORDER BY column1 ASC",
"select": "SELECT column1, table1.column2",
"where": "WHERE eq(column1, 'blabla')",
}
assert clickhouse_query.sql_data() == expected
def _format_storage_query_and_run(
timer: Timer,
query_metadata: SnubaQueryMetadata,
from_date: datetime,
to_date: datetime,
referrer: str,
clickhouse_query: Query,
request_settings: RequestSettings,
reader: Reader[SqlQuery],
) -> QueryResult:
"""
Formats the Storage Query and pass it to the DB specific code for execution.
"""
# TODO: This function (well, it will be a wrapper of this function)
# where we will transform the result according to the SelectedExpression
# object in the query to ensure the fields in the QueryResult have
# the same name the user expects.
source = clickhouse_query.get_data_source().format_from()
with sentry_sdk.start_span(description="create_query", op="db") as span:
formatted_query = AstSqlQuery(clickhouse_query, request_settings)
span.set_data("query", formatted_query.sql_data())
metrics.increment("execute")
timer.mark("prepare_query")
stats = {
"clickhouse_table": source,
"final": clickhouse_query.get_final(),
"referrer": referrer,
"num_days": (to_date - from_date).days,
"sample": clickhouse_query.get_sample(),
}
with sentry_sdk.start_span(
description=formatted_query.format_sql(), op="db"
) as span:
span.set_tag("table", source)
return raw_query(
clickhouse_query,
request_settings,
formatted_query,
reader,
timer,
query_metadata,
stats,
span.trace_id,
)
def _format_storage_query_and_run(
timer: Timer,
query_metadata: SnubaQueryMetadata,
from_date: datetime,
to_date: datetime,
referrer: str,
clickhouse_query: Query,
request_settings: RequestSettings,
reader: Reader[SqlQuery],
) -> QueryResult:
"""
Formats the Storage Query and pass it to the DB specific code for execution.
"""
source = clickhouse_query.get_from_clause().format_from()
with sentry_sdk.start_span(description="create_query", op="db") as span:
formatted_query = AstSqlQuery(clickhouse_query, request_settings)
span.set_data("query", formatted_query.sql_data())
metrics.increment("execute")
timer.mark("prepare_query")
stats = {
"clickhouse_table": source,
"final": clickhouse_query.get_final(),
"referrer": referrer,
"num_days": (to_date - from_date).days,
"sample": clickhouse_query.get_sample(),
}
with sentry_sdk.start_span(description=formatted_query.format_sql(),
op="db") as span:
span.set_tag("table", source)
return raw_query(
clickhouse_query,
request_settings,
formatted_query,
reader,
timer,
query_metadata,
stats,
span.trace_id,
)
def test_aliasing() -> None:
"""
Validates aliasing works properly when the query contains both tags_key
and tags_value.
"""
processed = parse_and_process({
"aggregations": [],
"groupby": [],
"selected_columns": ["tags_value"],
"conditions": [["tags_key", "IN", ["t1", "t2"]]],
})
sql = AstSqlQuery(processed, HTTPRequestSettings()).format_sql()
assert sql == (
"SELECT (tupleElement((arrayJoin(arrayMap((x, y -> tuple(x, y)), "
"tags.key, tags.value)) AS snuba_all_tags), 2) AS tags_value) "
"FROM transactions_local "
"WHERE in((tupleElement(snuba_all_tags, 1) AS tags_key), tuple('t1', 't2'))"
)
def test_format_expressions(query: Query, data: List[Tuple[str, str]]) -> None:
request_settings = HTTPRequestSettings()
clickhouse_query = AstSqlQuery(query, request_settings)
assert clickhouse_query.sql_data() == data
def test_format_expressions(query: Query, formatted: str) -> None:
request_settings = HTTPRequestSettings()
clickhouse_query = AstSqlQuery(query, request_settings)
assert clickhouse_query.format_sql() == formatted