sed和grep中单词的开头和结尾

Question

I don't understand the difference between \\b and \\< in GNU sed and GNU grep. It seems to me \\b can always replace \\< and \\\\> without changing the set of matching strings.

More specifically, I am trying to find examples in which \\bsomething and \\\\< something do not match exactly the same strings.

Same question for something\\b and something\\\\> .

Thank you

Answer 1

I suspect that it very rarely makes a difference whether you use (the more common) \\b or (the more specific) \\< and \\> , but I can think of an example where it would. This is quite contrived, and I suspect that in most real-world regex use it wouldn't make a difference, but this should demonstrate that it at least could make a difference in some cases.

If I have the following text:

this is his pig

and I want to know if /\\bis\\b/ matches, it wouldn't matter if I instead used /\\<is\\>/ or I instead used /\\>is\\</

But what if my text was instead

is this his pig

There's no longer a word-final boundary before the 'is', only a word-initial boundary. Using /\\bis\\b/ matches, and of course /\\<is\\>/ does too, but /\\>is\\</ does not.

In real life, though, I think it is not common that you really need to be able to make this distinction, which is why (at least outside of sed) \\b is the normal word boundary marker for regular expressions.

Answer 2

\\< matches the transition from non-word to word.

\\> matches the transition from word to non-word.

\\b is equivalent to (\\<|\\>) in extended regex.

So I won't say \\b and \\< are the same. I'd say \\b is a superset of \\< . Vice versa for \\b and \\> .

Answer 3

I stumbled upon such an example earlier.
\\<.\\> matches a one letter word.
Using \\b you would need to put something like \\b[^ ]\\b, because \\b.\\b matches a space between two words.

Answer 4

According to LinuxTopia the only difference between the two type of word boundaries is that whilst \\< and \\> work in most sed versions; the latter \\b works only if your system is using gsed

And a quotation from the wiki:

These symbols include '\\<' and '>' (gsed, ssed, sed15, sed16, sedmod) and '\\b' and '\\B' (gsed only).

Other than that the two are identical. Also here is a table that explains all possible scenarios that use word boundaries:

  Match position      Possible word boundaries   HHsed   GNU sed
  ---------------------------------------------------------------
  start of word    [nonword char]^[word char]      \<    \< or \b
  end of word         [word char]^[nonword char]   \>    \> or \b
  middle of word      [word char]^[word char]     none      \B
  outside of word  [nonword char]^[nonword char]  none      \B
  ---------------------------------------------------------------