normalize insert query more efficiently

[eurekaka]

Feature Request

Is your feature request related to a problem? Please describe:

For an insert query containing large strings in the VALUES part, the normalize procedure is not that efficient since it processes the query from beginning to the end.

Describe the feature you'd like:

For insert ... values (xx), we can specially normalize it to be insert ... values(...), i.e, we can skip processing the possible large string after values.

Since normalize only involves lex analysis, with no grammar analysis, we have to specially check the insert and values keywords, which makes it pretty tricky.

Besides, there may be ON DUPLICATE UPDATE in the end of the query, we need to specially handle them as well.

Describe alternatives you've considered:

N/A

Teachability, Documentation, Adoption, Migration Strategy:

Side effect of this improvement is that, both insert into t values(1,1) and insert into t values(1,1),(2,2) would be normalized to insert into t values(...).

[kennytm]

this should also recognize replace into. the keyword values can also be value.