NOT LIKE has the same precedence as the LIKE operator. The parser was
previously assigning it the precedence of the unary NOT operator. NOT
BETWEEN and NOT IN are treated similarly, as they are equivalent, from a
precedence perspective, to NOT LIKE.
The fix for this requires associating precedences with sequences of
tokens, rather than single tokens, so that "NOT LIKE" and "NOT <expr>"
can have different preferences. Perhaps surprisingly, this change is not
very invasive.
An alternative I considered involved adjusting the tokenizer to lex
NOT, NOT LIKE, NOT BETWEEN, and NOT IN as separate tokens. This broke
symmetry in strange ways, though, as NotLike, NotBetween, and NotIn
gained dedicated tokens, while LIKE, BETWEEN, and IN remained as
stringly identifiers.
Fixes#81.
`BETWEEN <thing> AND <thing>` allows <thing> to be any expr that doesn't
contain boolean operators. (Allowing boolean operators would wreak
havoc, because of the repurposing of AND as both a boolean operation
and part of the syntax of BETWEEN.)
Roughly the <character factor> production - https://jakewheat.github.io/sql-overview/sql-2011-foundation-grammar.html#character-factor
If an expression is followed by the keyword `COLLATE`, it must be
followed by the collation name, which is optionally schema-qualified
identifier.
The `COLLATE` keyword is not a regular binary operator in that it can't
be "nested": `foo COLLATE bar COLLATE baz` is not valid. If you prefer
to think of it as an operator, you might say it has the highest
precedence (judging from the spec), i.e. it binds to the smallest valid
expression to the left of it (so in `foo < bar COLLATE c`, the COLLATE
is applied first).
The `@@version` test is MS' dialect of SQL, it seems, so test it with
its own dialect.
Update the rules for identifiers in Postresql dialect per documentation,
while we're at it. The current identifier rules in Postgresql dialect
were introduced in this commit - as a copy of generic rules, it seems:
810cd8e6cf (diff-2808df0fba0aed85f9d35c167bd6a5f1L138)
