def _dialect_specific_segment_parses(dialect, segmentref, raw, caplog):
"""Test that specific segments parse as expected.
NB: We're testing the PARSE function not the MATCH function
although this will be a recursive parse and so the match
function of SUBSECTIONS will be tested if present. The match
function of the parent will not be tested.
"""
config = FluffConfig(overrides=dict(dialect=dialect))
seg_list = lex(raw, config=config)
Seg = validate_segment(segmentref, config=config)
# Most segments won't handle the end of file marker. We should strip it.
if seg_list[-1].is_type("end_of_file"):
seg_list = seg_list[:-1]
# This test is different if we're working with RawSegment
# derivatives or not.
if isinstance(Seg, Matchable) or issubclass(Seg, RawSegment):
print("Raw/Parser route...")
with RootParseContext.from_config(config) as ctx:
with caplog.at_level(logging.DEBUG):
parsed = Seg.match(segments=seg_list, parse_context=ctx)
assert isinstance(parsed, MatchResult)
assert len(parsed.matched_segments) == 1
print(parsed)
parsed = parsed.matched_segments[0]
print(parsed)
else:
print("Base route...")
# Construct an unparsed segment
seg = Seg(seg_list, pos_marker=seg_list[0].pos_marker)
# Perform the match (THIS IS THE MEAT OF THE TEST)
with RootParseContext.from_config(config) as ctx:
with caplog.at_level(logging.DEBUG):
parsed = seg.parse(parse_context=ctx)
print(parsed)
assert isinstance(parsed, Seg)
# Check we get a good response
print(parsed)
print(type(parsed))
print(type(parsed.raw))
# Check we're all there.
assert parsed.raw == raw
# Check that there's nothing un parsable
typs = parsed.type_set()
assert "unparsable" not in typs
def test__parser__grammar__base__bracket_fail_with_unexpected_end_bracket(
generate_test_segments, fresh_ansi_dialect):
"""Test _bracket_sensitive_look_ahead_match edge case.
Should fail gracefully and stop matching if we find a trailing unmatched.
"""
fs = StringParser("foo", KeywordSegment)
# We need a dialect here to do bracket matching
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
_, match, _ = BaseGrammar._bracket_sensitive_look_ahead_match(
generate_test_segments([
"bar",
"(", # This bracket pair should be mutated
")",
" ",
")", # This is the unmatched bracket
" ",
"foo",
]),
[fs],
ctx,
)
# Check we don't match (even though there's a foo at the end)
assert not match
# Check the first bracket pair have been mutated.
segs = match.unmatched_segments
assert segs[1].is_type("bracketed")
assert segs[1].raw == "()"
assert len(segs[1].segments) == 2
# Check the trailing foo hasn't been mutated
assert segs[5].raw == "foo"
assert not isinstance(segs[5], KeywordSegment)
def test__parser__grammar__base__bracket_sensitive_look_ahead_match(
bracket_seg_list, fresh_ansi_dialect):
"""Test the _bracket_sensitive_look_ahead_match method of the BaseGrammar."""
bs = StringParser("bar", KeywordSegment)
fs = StringParser("foo", KeywordSegment)
# We need a dialect here to do bracket matching
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Basic version, we should find bar first
pre_section, match, matcher = BaseGrammar._bracket_sensitive_look_ahead_match(
bracket_seg_list, [fs, bs], ctx)
assert pre_section == ()
assert matcher == bs
# NB the middle element is a match object
assert match.matched_segments == (KeywordSegment(
"bar", bracket_seg_list[0].pos_marker), )
# Look ahead for foo, we should find the one AFTER the brackets, not the
# on IN the brackets.
pre_section, match, matcher = BaseGrammar._bracket_sensitive_look_ahead_match(
bracket_seg_list, [fs], ctx)
# NB: The bracket segments will have been mutated, so we can't directly compare.
# Make sure we've got a bracketed section in there.
assert len(pre_section) == 5
assert pre_section[2].is_type("bracketed")
assert len(pre_section[2].segments) == 4
assert matcher == fs
# We shouldn't match the whitespace with the keyword
assert match.matched_segments == (KeywordSegment(
"foo", bracket_seg_list[8].pos_marker), )
# Check that the unmatched segments are nothing.
assert not match.unmatched_segments
def test__parser__grammar_sequence_indent_conditional(seg_list, caplog):
"""Test the Sequence grammar with indents."""
bs = StringParser("bar", KeywordSegment)
fs = StringParser("foo", KeywordSegment)
# We will assume the default config has indented_joins = False.
# We're testing without explictly setting the `config_type` because
# that's the assumed way of using the grammar in practice.
g = Sequence(
Conditional(Indent, indented_joins=False),
bs,
Conditional(Indent, indented_joins=True),
fs,
)
with RootParseContext(dialect=None) as ctx:
with caplog.at_level(logging.DEBUG, logger="sqlfluff.parser"):
m = g.match(seg_list, parse_context=ctx)
assert m
# Check we get an Indent.
assert isinstance(m.matched_segments[0], Indent)
assert isinstance(m.matched_segments[1], KeywordSegment)
# check the whitespace is still there
assert isinstance(m.matched_segments[2], WhitespaceSegment)
# Check the second Indent does not appear
assert not isinstance(m.matched_segments[3], Indent)
assert isinstance(m.matched_segments[3], KeywordSegment)
def test__parser__grammar_greedyuntil_bracketed(bracket_seg_list, fresh_ansi_dialect):
"""Test the GreedyUntil grammar with brackets."""
fs = KeywordSegment.make("foo")
g = GreedyUntil(fs)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Check that we can make it past the brackets
assert len(g.match(bracket_seg_list, parse_context=ctx)) == 7
def test__parser__grammar__base__bracket_sensitive_look_ahead_match(
bracket_seg_list, fresh_ansi_dialect
):
"""Test the _bracket_sensitive_look_ahead_match method of the BaseGrammar."""
fs = KeywordSegment.make("foo")
bs = KeywordSegment.make("bar")
# We need a dialect here to do bracket matching
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Basic version, we should find bar first
pre_section, match, matcher = BaseGrammar._bracket_sensitive_look_ahead_match(
bracket_seg_list, [fs, bs], ctx
)
assert pre_section == ()
assert matcher == bs
# NB the middle element is a match object
assert match.matched_segments == (bs("bar", bracket_seg_list[0].pos_marker),)
# Look ahead for foo, we should find the one AFTER the brackets, not the
# on IN the brackets.
pre_section, match, matcher = BaseGrammar._bracket_sensitive_look_ahead_match(
bracket_seg_list, [fs], ctx
)
# NB: The bracket segments will have been mutated, so we can't directly compare
assert len(pre_section) == 8
assert matcher == fs
# We shouldn't match the whitespace with the keyword
assert match.matched_segments == (fs("foo", bracket_seg_list[8].pos_marker),)
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_sequence(seg_list, caplog):
"""Test the Sequence grammar."""
bs = StringParser("bar", KeywordSegment)
fs = StringParser("foo", KeywordSegment)
g = Sequence(bs, fs)
# If running in the test environment, assert that Sequence recognises this
if getenv("SQLFLUFF_TESTENV", ""):
assert g.test_env
gc = Sequence(bs, fs, allow_gaps=False)
with RootParseContext(dialect=None) as ctx:
with caplog.at_level(logging.DEBUG, logger="sqlfluff.parser"):
# Should be able to match the list using the normal matcher
logging.info("#### TEST 1")
m = g.match(seg_list, parse_context=ctx)
assert m
assert len(m) == 3
assert m.matched_segments == (
KeywordSegment("bar", seg_list[0].pos_marker),
seg_list[1], # This will be the whitespace segment
KeywordSegment("foo", seg_list[2].pos_marker),
)
# Shouldn't with the allow_gaps matcher
logging.info("#### TEST 2")
assert not gc.match(seg_list, parse_context=ctx)
# Shouldn't match even on the normal one if we don't start at the beginning
logging.info("#### TEST 2")
assert not g.match(seg_list[1:], parse_context=ctx)
def test__parser__core_keyword(raw_seg_list):
"""Test the Mystical KeywordSegment."""
# First make a keyword
FooKeyword = StringParser("foo", KeywordSegment, type="bar")
# Check it looks as expected
assert FooKeyword.template.upper() == "FOO"
with RootParseContext(dialect=None) as ctx:
# Match it against a list and check it doesn't match
assert not FooKeyword.match(raw_seg_list, parse_context=ctx)
# Match it against a the first element and check it doesn't match
assert not FooKeyword.match(raw_seg_list[0], parse_context=ctx)
# Match it against a the first element as a list and check it doesn't match
assert not FooKeyword.match([raw_seg_list[0]], parse_context=ctx)
# Match it against the final element (returns tuple)
m = FooKeyword.match(raw_seg_list[1], parse_context=ctx)
assert m
assert m.matched_segments[0].raw == "foo"
assert isinstance(m.matched_segments[0], KeywordSegment)
# Match it against the final element as a list
assert FooKeyword.match([raw_seg_list[1]], parse_context=ctx)
# Match it against a list slice and check it still works
assert FooKeyword.match(raw_seg_list[1:], parse_context=ctx)
# Check that the types work right. Importantly that the "bar"
# type makes it in.
assert m.matched_segments[0].class_types == {"base", "keyword", "raw", "bar"}
def test__parser__grammar__base__bracket_fail_with_open_paren_close_square_mismatch(
generate_test_segments, fresh_ansi_dialect
):
"""Test _bracket_sensitive_look_ahead_match failure case.
Should fail when the type of a close bracket doesn't match the type of the
corresponding open bracket, but both are "definite" brackets.
"""
fs = StringParser("foo", KeywordSegment)
# We need a dialect here to do bracket matching
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Basic version, we should find bar first
with pytest.raises(SQLParseError) as sql_parse_error:
BaseGrammar._bracket_sensitive_look_ahead_match(
generate_test_segments(
[
"select",
" ",
"*",
" ",
"from",
"(",
"foo",
"]", # Bracket types don't match (parens vs square)
]
),
[fs],
ctx,
)
assert sql_parse_error.match("Found unexpected end bracket")
def test__parser__base_segments_base(raw_seg_list, fresh_ansi_dialect):
"""Test base segments behave as expected."""
base_seg = DummySegment(raw_seg_list)
# Check we assume the position correctly
assert (
base_seg.pos_marker.start_point_marker()
== raw_seg_list[0].pos_marker.start_point_marker()
)
assert (
base_seg.pos_marker.end_point_marker()
== raw_seg_list[-1].pos_marker.end_point_marker()
)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Expand and given we don't have a grammar we should get the same thing
assert base_seg.parse(parse_context=ctx) == base_seg
# Check that we correctly reconstruct the raw
assert base_seg.raw == "foobar.barfoo"
# Check tuple
assert base_seg.to_tuple() == (
"dummy",
(raw_seg_list[0].to_tuple(), raw_seg_list[1].to_tuple()),
)
# Check Formatting and Stringification
assert str(base_seg) == repr(base_seg) == "<DummySegment: ([L: 1, P: 1])>"
assert base_seg.stringify(ident=1, tabsize=2) == (
"[L: 1, P: 1] | dummy:\n"
"[L: 1, P: 1] | raw: "
" 'foobar'\n"
"[L: 1, P: 7] | raw: "
" '.barfoo'\n"
)
def test__parser__grammar_noncode(seg_list, fresh_ansi_dialect):
"""Test the NonCodeMatcher."""
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
m = NonCodeMatcher().match(seg_list[1:], parse_context=ctx)
# NonCode Matcher doesn't work with simple
assert NonCodeMatcher().simple(ctx) is None
# We should match one and only one segment
assert len(m) == 1
def test__parser__grammar_startswith_a(keyword, match_truthy, seg_list,
fresh_ansi_dialect, caplog):
"""Test the StartsWith grammar simply."""
Keyword = StringParser(keyword, KeywordSegment)
grammar = StartsWith(Keyword)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
with caplog.at_level(logging.DEBUG, logger="sqlfluff.parser"):
m = grammar.match(seg_list, parse_context=ctx)
assert bool(m) is match_truthy
def test__parser__grammar_ref_exclude(generate_test_segments,
fresh_ansi_dialect):
"""Test the Ref grammar exclude option."""
ni = Ref("NakedIdentifierSegment", exclude=Ref.keyword("ABS"))
ts = generate_test_segments(["ABS", "ABSOLUTE"])
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Asset ABS does not match, due to the exclude
assert not ni.match([ts[0]], parse_context=ctx)
# Asset ABSOLUTE does match
assert ni.match([ts[1]], parse_context=ctx)
def test__parser__parse_match(seg_list):
"""Test match method on a real segment."""
with RootParseContext(dialect=None) as ctx:
# This should match and have consumed everything, which should
# now be part of a BasicSegment.
m = BasicSegment.match(seg_list[:1], parse_context=ctx)
assert m
assert len(m.matched_segments) == 1
assert isinstance(m.matched_segments[0], BasicSegment)
assert m.matched_segments[0].segments[0].type == "raw"
def test__parser__grammar_oneof_exclude(seg_list):
"""Test the OneOf grammar exclude option."""
fs = KeywordSegment.make("foo")
bs = KeywordSegment.make("bar")
g = OneOf(bs, exclude=Sequence(bs, fs))
with RootParseContext(dialect=None) as ctx:
# Just against the first alone
assert g.match(seg_list[:1], parse_context=ctx)
# Now with the bit to exclude included
assert not g.match(seg_list, parse_context=ctx)
def test__parser__grammar_greedyuntil(keyword, seg_list, enforce_ws, slice_len,
fresh_ansi_dialect):
"""Test the GreedyUntil grammar."""
grammar = GreedyUntil(
StringParser(keyword, KeywordSegment),
enforce_whitespace_preceding_terminator=enforce_ws,
)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
assert (grammar.match(
seg_list,
parse_context=ctx).matched_segments == seg_list[:slice_len])
def test__parser__grammar_startswith_b(
include_terminator, match_length, seg_list, fresh_ansi_dialect, caplog
):
"""Test the StartsWith grammar with a terminator (included & exluded)."""
baar = KeywordSegment.make("baar")
bar = KeywordSegment.make("bar")
grammar = StartsWith(bar, terminator=baar, include_terminator=include_terminator)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
with caplog.at_level(logging.DEBUG, logger="sqlfluff.parser"):
m = grammar.match(seg_list, parse_context=ctx)
assert len(m) == match_length
def parse(
self, segments: Sequence["BaseSegment"], recurse=True, fname: str = None
) -> Optional["BaseSegment"]:
"""Parse a series of lexed tokens using the current dialect."""
if not segments:
return None
# Instantiate the root segment
root_segment = self.RootSegment(segments=segments, fname=fname)
# Call .parse() on that segment
with RootParseContext.from_config(config=self.config, recurse=recurse) as ctx:
parsed = root_segment.parse(parse_context=ctx)
return parsed
def test__parser__grammar_sequence_indent(seg_list, caplog):
"""Test the Sequence grammar with indents."""
bs = StringParser("bar", KeywordSegment)
fs = StringParser("foo", KeywordSegment)
g = Sequence(Indent, bs, fs)
with RootParseContext(dialect=None) as ctx:
with caplog.at_level(logging.DEBUG, logger="sqlfluff.parser"):
m = g.match(seg_list, parse_context=ctx)
assert m
# check we get an indent.
assert isinstance(m.matched_segments[0], Indent)
assert isinstance(m.matched_segments[1], KeywordSegment)
def test__parser__grammar_greedyuntil_bracketed(bracket_seg_list, fresh_ansi_dialect):
"""Test the GreedyUntil grammar with brackets."""
fs = StringParser("foo", KeywordSegment)
g = GreedyUntil(fs)
with RootParseContext(dialect=fresh_ansi_dialect) as ctx:
# Check that we can make it past the brackets
match = g.match(bracket_seg_list, parse_context=ctx)
assert len(match) == 4
# Check we successfully constructed a bracketed segment
assert match.matched_segments[2].is_type("bracketed")
assert match.matched_segments[2].raw == "(foo )"
# Check that the unmatched segments is foo AND the whitespace
assert len(match.unmatched_segments) == 2
def test__parser__grammar_oneof_templated(seg_list):
"""Test the OneOf grammar.
NB: Should behave the same regardless of code_only.
"""
bs = StringParser("bar", KeywordSegment)
fs = StringParser("foo", KeywordSegment)
g = OneOf(fs, bs)
with RootParseContext(dialect=None) as ctx:
# This shouldn't match, but it *ALSO* shouldn't raise an exception.
# https://github.com/sqlfluff/sqlfluff/issues/780
assert not g.match(seg_list[5:], parse_context=ctx)
def test__parser__multistringparser__match(generate_test_segments):
"""Test the MultiStringParser matchable."""
parser = MultiStringParser(["foo", "bar"], KeywordSegment)
with RootParseContext(dialect=None) as ctx:
# Check directly
seg_list = generate_test_segments(["foo", "fo"])
# Matches when it should
assert parser.match(
seg_list[:1],
parse_context=ctx).matched_segments == (KeywordSegment(
"foo", seg_list[0].pos_marker), )
# Doesn't match when it shouldn't
assert parser.match(seg_list[1:],
parse_context=ctx).matched_segments == tuple()
def test__parser__parse_expand(seg_list):
"""Test expand method on a real segment."""
with RootParseContext(dialect=None) as ctx:
# Match the segment, and get the matched segments
segments = BasicSegment.match(seg_list[:1], parse_context=ctx).matched_segments
# Remind ourselves that this should be tuple containing a BasicSegment
assert isinstance(segments[0], BasicSegment)
# Now expand those segments, using the base class version (not that it should matter)
res = BasicSegment.expand(segments, parse_context=ctx)
# Check we get an iterable containing a BasicSegment
assert isinstance(res[0], BasicSegment)
# Check that we now have a keyword inside
assert isinstance(res[0].segments[0], KeywordSegment)
def test__parser__grammar_oneof(seg_list, allow_gaps):
"""Test the OneOf grammar.
NB: Should behave the same regardless of code_only.
"""
fs = KeywordSegment.make("foo")
bs = KeywordSegment.make("bar")
g = OneOf(fs, bs, allow_gaps=allow_gaps)
with RootParseContext(dialect=None) as ctx:
# Check directly
assert g.match(seg_list, parse_context=ctx).matched_segments == (bs(
"bar", seg_list[0].pos_marker), )
# Check with a bit of whitespace
assert not g.match(seg_list[1:], parse_context=ctx)
def _dialect_specific_segment_not_match(dialect, segmentref, raw, caplog):
"""Test that specific segments do not match.
NB: We're testing the MATCH function not the PARSE function.
This is the opposite to the above.
"""
config = FluffConfig(overrides=dict(dialect=dialect))
seg_list = lex(raw, config=config)
Seg = validate_segment(segmentref, config=config)
with RootParseContext.from_config(config) as ctx:
with caplog.at_level(logging.DEBUG):
match = Seg.match(segments=seg_list, parse_context=ctx)
assert not match
def test__parser__parse_parse(seg_list, caplog):
"""Test parse method on a real segment."""
with RootParseContext(dialect=None) as ctx:
# Match the segment, and get the inner segment
seg = BasicSegment.match(seg_list[:1], parse_context=ctx).matched_segments[0]
# Remind ourselves that this should be an unparsed BasicSegment
assert isinstance(seg, BasicSegment)
# Now parse that segment, with debugging because this is
# where we'll need to debug if things fail.
with caplog.at_level(logging.DEBUG):
res = seg.parse(parse_context=ctx)
# Check it's still a BasicSegment
assert isinstance(res, BasicSegment)
# Check that we now have a keyword inside
assert isinstance(res.segments[0], KeywordSegment)
def test__parser__grammar__base__ephemeral_segment(seg_list):
"""Test the ephemeral features BaseGrammar.
Normally you cant call .match() on a BaseGrammar, but
if things are set up right, then it should be possible
in the case that the ephemeral_name is set.
"""
g = BaseGrammar(ephemeral_name="TestGrammar")
with RootParseContext(dialect=None) as ctx:
m = g.match(seg_list, ctx)
# Check we get an ephemeral segment
assert isinstance(m.matched_segments[0], EphemeralSegment)
chkpoint = m.matched_segments[0]
# Check it's got the same content.
assert chkpoint.segments == seg_list
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 = RegexParser(r"fo{2}", KeywordSegment)
foo = StringParser("foo", KeywordSegment)
# 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 == (
KeywordSegment("foo", seg_list[2].pos_marker),
)
assert g2.match(seg_list[2:], parse_context=ctx).matched_segments == (
KeywordSegment("foo", seg_list[2].pos_marker),
)
def test__parser__grammar__base__look_ahead_match(
seg_list_slice,
matcher_keywords,
result_slice,
winning_matcher,
pre_match_slice,
seg_list,
):
"""Test the _look_ahead_match method of the BaseGrammar."""
# Make the matcher keywords
matchers = [
StringParser(keyword, KeywordSegment) for keyword in matcher_keywords
]
# Fetch the matching keyword from above by index
winning_matcher = matchers[matcher_keywords.index(winning_matcher)]
with RootParseContext(dialect=None) as ctx:
m = BaseGrammar._look_ahead_match(
seg_list[seg_list_slice],
matchers,
ctx,
)
# Check structure of the response.
assert isinstance(m, tuple)
assert len(m) == 3
# Unpack
result_pre_match, result_match, result_matcher = m
# Check the right matcher won
assert result_matcher == winning_matcher
# Make check tuple for the pre-match section
if pre_match_slice:
pre_match_slice = seg_list[pre_match_slice]
else:
pre_match_slice = ()
assert result_pre_match == pre_match_slice
# Make the check tuple
expected_result = make_result_tuple(
result_slice=result_slice,
matcher_keywords=matcher_keywords,
seg_list=seg_list,
)
assert result_match.matched_segments == expected_result
