Re: Mixing greediness in regexp_matches

Поиск
Список
Период
Сортировка
Искать
От
Tom Lane
Тема
Re: Mixing greediness in regexp_matches
Дата
Msg-id
28143.1577114140@sss.pgh.pa.us
Ответ на
Список
Дерево обсуждения
Mixing greediness in regexp_matches "Daniel Verite" <daniel@manitou-mail.org>
Re: Mixing greediness in regexp_matches Tom Lane <tgl@sss.pgh.pa.us>
Re: Mixing greediness in regexp_matches Tom Lane <tgl@sss.pgh.pa.us>
Re: Mixing greediness in regexp_matches "Daniel Verite" <daniel@manitou-mail.org>
Re: Mixing greediness in regexp_matches Tom Lane <tgl@sss.pgh.pa.us>
Re: Mixing greediness in regexp_matches "Daniel Verite" <daniel@manitou-mail.org>
"Daniel Verite"  writes:
> The basic idea is to iterate on the rows produced by
>    regexp_matches(string, '(.*?)(foo|bar|foobar)', 'g')
> to break down the string into pairs of (non-matching segment,
> matching segment) so that a final result can be assembled
> from that (setting aside the last non-matching segment, that
> can be retrieved in a final step).
> The difficulty is that the longest strings in the alternation
> should be prioritized, but the starting (.*?) makes the RE
> non-greedy so "foo" is choosen over "foobar". 

I'd try forcing the match to be the whole string, ie

	^(.*?)(foo|bar|foobar)(.*)$

which would also save some work for restarting the iteration,
since you'd have already captured the all-the-rest substring.

With the endpoints forced, it doesn't really matter whether
the engine deems the RE-as-a-whole to be greedy or not.
I think this would work without needing any explicit greediness
marking for the second and third parts, but I might be wrong
about that detail.

			regards, tom lane


В списке pgsql-general по дате отправления
От: Daniel Verite
Дата:
От: Tom Lane
Дата:
FAQ