def parse(tokens):
expression, next_token = ExpressionParser.parse(
tokens,
can_alias=False,
until_one_of=("group", "having", "order", "limit", "offset", ";"),
)
return WhereClause(expression), next_token
def test_where_clause_keeps_generator_intact_is_null_condition():
tokens = to_tokens("col is null group by col")
actual, next_token = WhereClauseParser.parse(tokens)
expected = WhereClause(Condition(Column("col"), "is", Null()))
assert actual == expected
assert next_token == "group"
assert list(tokens) == ["by", "col"]
def test_nested_parenthesis_boolean():
actual = WhereClauseParser.parse(
to_tokens("(col = 1 and col2=4) or (col = 2 and (col =6 or col=9))"))
expected = WhereClause(
BooleanCondition(
"or",
Parenthesis(
BooleanCondition(
"and",
Condition(Column("col"), "=", Integer(1)),
Condition(Column("col2"), "=", Integer(4)),
)),
Parenthesis(
BooleanCondition(
"and",
Condition(Column("col"), "=", Integer(2)),
Parenthesis(
BooleanCondition(
"or",
Condition(Column("col"), "=", Integer(6)),
Condition(Column("col"), "=", Integer(9)),
)),
)),
))
assert actual == expected
def test_where_clause_keeps_generator_intact():
tokens = to_tokens("col = 3 group by col")
actual, next_token = WhereClauseParser.parse(tokens)
expected = WhereClause(Condition(Column("col"), "=", Integer(3)))
assert actual == expected
assert next_token == "group"
assert list(tokens) == ["by", "col"]
def test_subquery():
actual = SQLStatementParser.parse(
to_tokens(
"SELECT col"
" from (select count(*) col"
" from table group by x) WHERE col > 10 ORDER BY col DESC;"
)
)
expected = SelectStatement(
expressions=[Column("col")],
from_statement=Parenthesis(
SelectStatement(
expressions=[
Alias(CountFunctionCall(Column("*")), "col", with_as=False)
],
from_statement=Table(Column("table")),
group_by_clause=GroupByClause(Column("x")),
semi_colon=False,
)
),
where_clause=WhereClause(Condition(Column("col"), ">", Integer(10))),
order_by_clause=OrderByClause(OrderByItem(Column("col"), has_desc=True)),
semi_colon=True,
)
assert actual == expected
def test_boolean_where_clause():
actual = WhereClauseParser.parse(to_tokens("col = 3 and field = 5"))
expected = WhereClause(
BooleanCondition(
"and",
Condition(Column("col"), "=", Integer(3)),
Condition(Column("field"), "=", Integer(5)),
))
assert actual == expected
def test_between_where_clause():
actual = WhereClauseParser.parse(to_tokens("col between 3 and 5"))
expected = WhereClause(
Condition(
Column("col"),
"between",
BooleanCondition(
"and",
Integer(3),
Integer(5),
),
))
assert actual == expected
def test_multiple_args_boolean_condition():
actual = WhereClauseParser.parse(
to_tokens("(col = 1 and col2=4 and col3=4)"))
expected = WhereClause(
Parenthesis(
BooleanCondition(
"and",
Condition(Column("col"), "=", Integer(1)),
BooleanCondition(
"and",
Condition(Column("col2"), "=", Integer(4)),
Condition(Column("col3"), "=", Integer(4)),
),
), ))
assert actual == expected
def test_parenthesis_boolean_where_clause():
actual = WhereClauseParser.parse(
to_tokens("(col = 3 and field = 5) or (f2 or f3)"))
expected = WhereClause(
BooleanCondition(
"or",
Parenthesis(
BooleanCondition(
"and",
Condition(Column("col"), "=", Integer(3)),
Condition(Column("field"), "=", Integer(5)),
)),
Parenthesis(BooleanCondition("or", Column("f2"), Column("f3"))),
))
assert actual == expected
def parse(tokens):
expression = ExpressionParser.parse(tokens)
return WhereClause(expression)
def test_nested_joins():
sql = """
SELECT COALESCE(sq_1.col, sq_2.col) f0_
FROM (SELECT ANY_VALUE(col) col,
LAST_VALUE(ANY_VALUE(col2)) OVER (PARTITION BY ANY_VALUE(col) ORDER BY SUM(clicks) ASC, SUM(metric) ASC RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) last,
hash
FROM (SELECT *
FROM `events`
WHERE _TABLE_SUFFIX BETWEEN '20200410' AND '20200510')
JOIN
(SELECT * EXCEPT (hash)
FROM (SELECT *,
ROW_NUMBER() OVER (PARTITION BY hash) AS rn
FROM `test-table`
WHERE _TABLE_SUFFIX BETWEEN '20200401' AND '20200501')
WHERE rn = 1)
USING (hash)
GROUP BY hash) sq_1
FULL OUTER JOIN
(SELECT ANY_VALUE(col) col,
hash
FROM (SELECT *
FROM `events`
WHERE _TABLE_SUFFIX BETWEEN '20200310' AND '20200410')
JOIN
(SELECT * EXCEPT (hash),
FROM (SELECT *,
ROW_NUMBER() OVER (PARTITION BY hash) AS rn
FROM `test-table`
WHERE _TABLE_SUFFIX BETWEEN '20200301' AND '20200401')
WHERE rn = 1)
USING (hash)
GROUP BY hash) sq_2
ON sq_1.hash = sq_2.hash
WHERE sq_1.last = 1
GROUP BY f0_
""" # noqa
actual = SQLStatementParser.parse(to_tokens(sql))
expected = SelectStatement(
expressions=[
Alias(
FunctionCall("coalesce", Column("sq_1.col"),
Column("sq_2.col")),
"f0_",
with_as=False,
)
],
from_statement=Join(
"FULL OUTER JOIN",
left_from=Alias(
Parenthesis(
SelectStatement(
expressions=[
Alias(
FunctionCall("any_value", Column("col")),
"col",
with_as=False,
),
Alias(
AnalyticsClause(
FunctionCall(
"last_value",
FunctionCall("any_value",
Column("col2")),
),
partition_by=[
FunctionCall("any_value",
Column("col"))
],
order_by=OrderByClause(
OrderByItem(
FunctionCall(
"sum", Column("clicks")),
has_asc=True,
),
OrderByItem(
FunctionCall(
"sum", Column("metric")),
has_asc=True,
),
),
frame_clause=WindowFrameClause(
"range",
"between unbounded preceding and unbounded following", # noqa
),
),
"last",
with_as=False,
),
Column("hash"),
],
from_statement=Join(
join_type="JOIN",
left_from=Parenthesis(
SelectStatement(
expressions=[Column("*")],
from_statement=Table(
String("events", quotes="`")),
where_clause=WhereClause(
Condition(
Column("_table_suffix"),
"between",
BooleanCondition(
"and",
String("20200410", quotes="'"),
String("20200510", quotes="'"),
),
)),
semi_colon=False,
)),
right_from=Parenthesis(
SelectStatement(
expressions=[
ExceptClause(Column("*"),
[Column("hash")])
],
from_statement=Parenthesis(
SelectStatement(
expressions=[
Column("*"),
Alias(
AnalyticsClause(
FunctionCall(
"row_number"),
partition_by=[
Column("hash")
],
order_by=None,
frame_clause=None,
),
Column("rn"),
with_as=True,
),
],
from_statement=Table(
String("test-table",
quotes="`")),
where_clause=WhereClause(
Condition(
Column("_table_suffix"),
"between",
BooleanCondition(
"and",
String("20200401",
quotes="'"),
String("20200501",
quotes="'"),
),
)),
semi_colon=False,
)),
where_clause=WhereClause(
Condition(Column("rn"), "=",
Integer(1))),
semi_colon=False,
)),
on=None,
using=UsingClause(Parenthesis(Column("hash"))),
),
group_by_clause=GroupByClause(Column("hash")),
semi_colon=False,
)),
"sq_1",
with_as=False,
),
right_from=Alias(
Parenthesis(
SelectStatement(
expressions=[
Alias(
FunctionCall("any_value", Column("col")),
"col",
with_as=False,
),
Column("hash"),
],
from_statement=Join(
join_type="JOIN",
left_from=Parenthesis(
SelectStatement(
expressions=[Column("*")],
from_statement=Table(
String("events", quotes="`")),
where_clause=WhereClause(
Condition(
Column("_table_suffix"),
"between",
BooleanCondition(
"and",
String("20200310", quotes="'"),
String("20200410", quotes="'"),
),
)),
semi_colon=False,
)),
right_from=Parenthesis(
SelectStatement(
expressions=[
ExceptClause(Column("*"),
[Column("hash")])
],
from_statement=Parenthesis(
SelectStatement(
expressions=[
Column("*"),
Alias(
AnalyticsClause(
FunctionCall(
"row_number"),
partition_by=[
Column("hash")
],
order_by=None,
frame_clause=None,
),
Column("rn"),
with_as=True,
),
],
from_statement=Table(
String("test-table",
quotes="`")),
where_clause=WhereClause(
Condition(
Column("_table_suffix"),
"between",
BooleanCondition(
"and",
String("20200301",
quotes="'"),
String("20200401",
quotes="'"),
),
)),
semi_colon=False,
), ),
where_clause=WhereClause(
Condition(Column("rn"), "=",
Integer(1))),
semi_colon=False,
)),
on=None,
using=UsingClause(Parenthesis(Column("hash"))),
),
group_by_clause=GroupByClause(Column("hash")),
semi_colon=False,
)),
"sq_2",
with_as=False,
),
on=OnClause(
Condition(Column("sq_1.hash"), "=", Column("sq_2.hash"))),
using=None,
),
where_clause=WhereClause(
Condition(Column("sq_1.last"), "=", Integer(1))),
group_by_clause=GroupByClause(Column("f0_")),
semi_colon=False,
)
assert actual == expected
def test_parenthesis_expression_where_clause():
actual = WhereClauseParser.parse(to_tokens("(col + 1) = col2"))
expected = WhereClause(
Condition(Parenthesis(Addition(Column("col"), Integer(1))), "=",
Column("col2")))
assert actual == expected
def test_where_different_predicate():
actual = WhereClauseParser.parse(to_tokens("col <> 3"))
expected = WhereClause(Condition(Column("col"), "<>", Integer(3)), )
assert actual == expected
def test_where_clause():
actual = WhereClauseParser.parse(to_tokens("col = 3"))
expected = WhereClause(Condition(Column("col"), "=", Integer(3)))
assert actual == expected
