Wiktor Stribiżew
2020-11-30 06:17:22
With GNU grep, you can use a PCRE regex like
grep -P '\b(\w+)\b(.*\b\1\b){3}'
See the regex demo.
Test in Ubuntu 18.04.4 LTS:
Details
\b(\w+)\b- a whole word (captured in Group 1) (\bis a word boundary and\wmatches letters, digits or underscores)(.*\b\1\b){3}- three occurrences ({3}) of any text followed with the same value as in Group 1 (as\1is an inline backreference to Group 1 value) as a whole word (again,\bword boundaries are used.)
热门帖子
热门github
6
You can simplify by putting the word bounds in the group:
grep -E '(\b\w\b)(.*\1){3}'. Unless there's an edge case I haven't thought of.@wjandrea The capturing group only keeps the value, not the pattern. So,
\1is unaware of the fact if the string it holds was captured as a whole word or not. We need all the word boundaries I used in the pattern.Ah, I see, if you use input like
o do to moe, mine matches it (false positive), yours doesn't.I am also getting false positives with yours solution @WiktorStribiżew. It most likely has to do with the first word boundary, I would say.
en vrouwen gelijk voor de wet en maken we geenis one of my results, but the word en does not pop up separately 4 times. It does if you count the en inside of the words.Ah yes this seems to work. Thank you for the solution to my problem.