Normalizing token is broken
A few feedbacks received from HN (https://news.ycombinator.com/item?id=39745940) show a regression in the lexer normalizing step.
$ transqlate --pretty
Type SQL snippet and terminate with ^D
select TRUNC(SYSDATE, 'month') from dual
17:26:43 INFO Translating SQL. engine=oracle
17:26:43 WARN Rewrite failure. rule="replace trunc() by date_trunc()" err="unsupported date format: \"month\"" line=1 column=8
-- TRANSLATION ERROR at +1:8: unsupported date format: "month" rule="replace trunc() by date_trunc()"
SELECT date_trunc('month', localtimestamp)
The oracle/trunc.go:translateDateFormat()
method compares uppercased strings only.
Edited by Florent Jardin