def test__parser__grammar_oneof_take_longest_match(seg_list):
"""Test that the OneOf grammar takes the longest match."""
fooRegex = ReSegment.make(r"fo{2}")
baar = KeywordSegment.make(
"baar",
)
foo = KeywordSegment.make(
"foo",
)
fooBaar = Sequence(
foo,
baar,
)
# Even if fooRegex comes first, fooBaar
# is a longer match and should be taken
g = OneOf(fooRegex, fooBaar)
with RootParseContext(dialect=None) as ctx:
assert fooRegex.match(seg_list[2:], parse_context=ctx).matched_segments == (
fooRegex("foo", seg_list[2].pos_marker),
)
assert g.match(seg_list[2:], parse_context=ctx).matched_segments == (
foo("foo", seg_list[2].pos_marker),
baar("baar", seg_list[3].pos_marker),
)
def test__parser__grammar_oneof_take_first(seg_list):
"""Test that the OneOf grammar takes first match in case they are of same length."""
fooRegex = ReSegment.make(r"fo{2}")
foo = KeywordSegment.make("foo", )
# Both segments would match "foo"
# so we test that order matters
g1 = OneOf(fooRegex, foo)
g2 = OneOf(foo, fooRegex)
with RootParseContext(dialect=None) as ctx:
assert g1.match(seg_list[2:],
parse_context=ctx).matched_segments == (fooRegex(
"foo", seg_list[2].pos_marker), )
assert g2.match(seg_list[2:],
parse_context=ctx).matched_segments == (foo(
"foo", seg_list[2].pos_marker), )
),
),
]
+ exasol_fs_dialect.get_lexer_struct()
)
exasol_fs_dialect.add(
FunctionScriptTerminatorSegment=NamedSegment.make(
"function_script_terminator", type="statement_terminator"
),
WalrusOperatorSegment=NamedSegment.make(
"walrus_operator", type="assignment_operator"
),
VariableNameSegment=ReSegment.make(
r"[A-Z][A-Z0-9_]*",
name="function_variable",
type="variable",
),
)
exasol_fs_dialect.replace(
SemicolonSegment=SymbolSegment.make(";", name="semicolon", type="semicolon"),
)
@exasol_fs_dialect.segment(replace=True)
class StatementSegment(BaseSegment):
"""A generic segment, to any of its child subsegments."""
type = "statement"
"PIVOT",
"SAMPLE",
"TABLESAMPLE",
"UNPIVOT",
])
snowflake_dialect.add(
# In snowflake, these are case sensitive even though they're not quoted
# so they need a different `name` and `type` so they're not picked up
# by other rules.
ParameterAssignerSegment=SymbolSegment.make("=>",
name="parameter_assigner",
type="parameter_assigner"),
NakedSemiStructuredElementSegment=ReSegment.make(
r"[A-Z0-9_]*",
name="naked_semi_structured_element",
type="semi_structured_element",
),
QuotedSemiStructuredElementSegment=NamedSegment.make(
"double_quote",
name="quoted_semi_structured_element",
type="semi_structured_element",
),
ColumnIndexIdentifierSegment=ReSegment.make(
r"\$[0-9]+", name="column_index_identifier_segment",
type="identifier"),
)
snowflake_dialect.replace(
Accessor_Grammar=AnyNumberOf(
Ref("ArrayAccessorSegment"),
Sequence("FOR",
"SYSTEM_TIME",
"AS",
"OF",
Ref("ExpressionSegment"),
optional=True),
Sequence("WITH",
"OFFSET",
"AS",
Ref("SingleIdentifierGrammar"),
optional=True),
),
FunctionNameSegment=ReSegment.make(
# In BigQuery struct() has a special syntax, so we don't treat it as a function
r"[A-Z][A-Z0-9_]*",
name="function_name",
type="function_name",
_anti_template=r"STRUCT",
),
SelectTargetElementSegment=SelectTargetElementSegment,
SelectClauseSegment=SelectClauseSegment,
)
@bigquery_dialect.segment(replace=True)
class FunctionDefinitionGrammar(BaseSegment):
"""This is the body of a `CREATE FUNCTION AS` statement."""
match_grammar = Sequence(
AnyNumberOf(
Sequence(