How do I match exact word boundary, but excluding special characters in front and back?

Question

I want the regex in tcl to match my specified word boundary, without any special characters such as +-.() in front or back.

Here are the things I tried and it just doesn't match properly:

Let's say I have the following string:

hello world +hello world -hello world hello+ hello

I want it to only match hello, not hello+ or -hello.

\bhello\b

• hello
• +hello
• -hello
• hello+

[^+-]\bhello\b[^+-]

• no matches

[^+-]\bhello\b

• (doesn't match the first hello even though it should've matched)
• hello+
• hello

(?![+-])\bhello\b(?![+-])

• hello
• +hello
• -hello

Answer 1

As documented, Tcl uses \y to match a word boundary, not \b (which is a backspace character for compatibility with the escapes used by general Tcl code). This means you need an RE something like this:

(?:^|[^-+])\yhello\y(?:$|[^-+])

The middle piece is \yhello\y which matches the word, and then we need ^|[^-+] at the beginning to match either the beginning of the string or a character other than - or +, and equivalently $|[^-+] for the end. (I put those in (?:…) just to limit the scope of the | RE operator.)

Demonstrating from an interactive session:

% set RE {(?:^|[^-+])\yhello\y(?:$|[^-+])}
(?:^|[^-+])\yhello\y(?:$|[^-+])
% regexp $RE "hello"
1
% regexp $RE "ahello"
0
% regexp $RE "+hello"
0
% regexp $RE "+ hello"
1
% regexp $RE "hello+"
0
% regexp $RE "hello-"
0
% regexp $RE "hello.-"
1

Answer 2

Another option instead of using a word boundary, is to exclude matching a word char in the negated character class together with the other characters that should not be matched:

(?:^|[^-+.()\w])(hello)(?![-+.()\w])

• (?: Non capture group
  • ^ Start of string
  • | Or
  • [^-+.()\w] Negated character class [^, match any char except the listed
• ) Close non capture group
• (hello) Capture group 1, match hello
• (?! Negative lookahead, assert what is directly to the right is not
  • [-+.()\w] Match any of the listed
• ) Close lokahead

Regex demo

Answer 3

This regex matches the word hello, allows spaces before and after, but doesn't allow anything else. Creating a boundary of word characters AND special characters.

(?<!\S)hello(?!\S)

This uses the "negative look-ahead" and "negative look-behind" syntax.

(?<!\S): Look behind and make sure there aren't any non-whitespace characters.

(?!\S): Look ahead and make sure there aren't any non-whitespace characters.