我的负前瞻不起作用 - 为什么？

Question

I have a text scattered with various strings, dates, tab characters and language codes. I want to extract the strings that follow a date+tab combination, and which are followed by a language code like '[en]', a tab character, and after which we don't have the string "BAD THINGS" (eg "2020-01-12\tSTRING WE NEED[en]\tGOOD THINGS" , as opposed to "2020-01-12\tSTRING WE DON'T NEED[en]\tBAD THINGS" ).

Here is a short example text of what I'm working with:

\n2021-01-12\tThis string is not needed [it]\tBad things\tBad things\n2021-01-12\tThis string is also not needed [en]\tBad things\tBad things\n2021-01-11\tString 1 that is needed. [it]\tString 1 that is needed. is repeated here\tNot interesting here\n2021-01-11\tString 2 that is needed [fr]\tString 2 that is needed is repeated here\tUnnecessary string\n2021-01-11\tString 3 that is needed... [ru]\tString 3 that is needed... is repeated here\tAnother part we're not interested in

I made this regex to capture all strings between a date and a language code:

(\d{4}-\d{2}-\d{2}\\t)(.*?)(\[\w{2}\]\\t)

This works fine (see here ). However, when I add a negative lookahead to exclude those followed by "Bad things", all my regex goes south:

(\d{4}-\d{2}-\d{2}\\t)(.*?)(\[\w{2}\]\\t)(?!Bad things)

You can see the result here . I understand my lookahead somehow makes the regex greedy, but I have no idea how to avoid this, adding a? after it doesn't work. Can you help me out here?

Answer 1

Not sure if this will cover all the cases but this seems to work:

(\d{4}-\d{2}-\d{2}\\t)([^][]*)(\[\w{2}\]\\t)(?!Bad things)

Demo here .

Explanation:

(\d{4}-\d{2}-\d{2}\\t)   date and tab
([^][]*)                 collect only things that do not contain chars `[` and `]`   
(\[\w{2}\]\\t)           follow up [<tag>]
(?!Bad things)           Negative Lookahead