If some string literals are used in DDL, they might have `Character Set Introducers`
https://dev.mysql.com/doc/refman/8.0/en/charset-introducer.html
f.e. _utf8mb4'1' which is not parsable by current ParserStringLiteral.
Since we use utf8 by default, suggesting to automatically convert the string literals to utf8
before executing the query and avoid any charset introducers there.
Conversion from utf8 to utf8 is not needed and skipped.
Also it might convert double quotes to single quotes if any
which might solve issues with COMMENT and empty string literals "" in DEFAULT expressions.
SELECT _latin1"abc"; -- might be also valid for MySQL
... DEFAULT "",
... COMMENT "abc"
Currently no DEFAULT expressions are supported, not parsable at all,
but using this MR it allows to parse the expressions together with double quotes as string literals.
Since ClickHouse does not support unquoted utf-8 strings but MySQL does.
Instead of fixing Lexer to recognize utf-8 chars as TokenType::BareWord,
suggesting to quote all unrecognized tokens before applying any DDL.
Actual parsing and validating the syntax will be done by particular Parser.
If there is any TokenType::Error, the query is unable to be parsed anyway.
Quoting such tokens can provide the support of utf-8 names.
See `tryQuoteUnrecognizedTokens` and `QuoteUnrecognizedTokensTest`.
mysql> CREATE TABLE 道.渠(...
is converted to
CREATE TABLE `道`.`渠`(...
Also fixed the bug with missing * while doing SELECT in full sync because db or table name are back quoted when not needed.
