I would like to check source code scripts according to such a pattern:
'DELETE' or 'FROM' or 'DELETE FROM' followed by a space followed by any word not followed by another word separated by a dot, except for the word 'DUAL'
Something like
'(DELETE|FROM|DELETE FROM) \w+', where \w+ != DUAL
but not
'(DELETE|FROM|DELETE FROM) \w+\.\w+'
Examples
| text fragment | desired result |
|---|---|
| begin DELETE tbl1; | DELETE tbl1 |
| select FROM tbl2) loop | FROM tbl2 |
| fnc(); DELETE FROM tbl3 where | DELETE FROM tbl3 |
| qqq DELETE DUAL; www | |
| eee FROM DUAL rrr | |
| ttt DELETE FROM DUAL where | |
| yy DELETE sch1.tbl1; uuu | |
| iii FROM sch2.tbl2 ooo | |
| ppp DELETE FROM sch3.tbl3 aaa |
My guess
(FROM|DELETE( FROM)?) (?!DUAL)(?!\w+\.)\w+
matches too much by the 1st part. Is it correct in its 2nd part (after space)?
You might use
\b(?>DELETE(?: FROM)?|FROM) (?!DUAL\b)\w+\b(?!\.)
The pattern matches:
\b Word boundary to prevent a partial word match(?>DELETE(?: FROM)?|FROM) Atomic group (No backtracking) to match either DELETE with optional FROM or only FROM(?!DUAL\b) Negative lookahead to assert not the word DUAL directly to the right followed by a word boundary\w+\b Math 1+ word characters and a word boundary(?!\.) Negative lookahead, assert not a dot directly to the right of the current position