RegEx: Find pattern but exclude words

Question

I want to find all connected words except for specific ones. For example:

0827banana82/+wine22green-729

green and wine should match, but banana not.

I tried the following regular expression with a negative lookahead:

(?!banana)([a-zA-Z]+)

but it excludes only the first letter of banana because anana is still a match for the second pattern. I have no idea how to get rid of that.

Answer 1

You may add a negative lookbehind in your regex to make it work:

(?!banana)(?<![a-zA-Z])[a-zA-Z]+

RegEx Demo

RegEx Details:

• (?!banana): Negative lookahead to assert that we don't have string banana ahead of the current position
• (?<![a-zA-Z]): Negative lookbehind to assert that we don't have a letter before current position
• [a-zA-Z]+: Match 1+ letters

PS: If you want to allow words like bananas then use:

(?!banana(?![a-zA-Z]))(?<![a-zA-Z])[a-zA-Z]+

Answer 2

Well you can use this one:

(banana)|([a-zA-Z]+)

Which will capture banana in 1st group and all the other words in 2nd.

Answer 3

Another variation might be matching the characters a-zA-Z until there are no more. Then assert that banana is not directly to the left.

[a-zA-Z]+(?![a-zA-Z])(?<!banana)

The pattern matches

• [a-zA-Z]+ Match 1+ chars a-zA-Z
• (?![a-zA-Z]) Negative lookahead, assert not a-zA-Z directly to the right
• (?<!banana) Negative lookbehind, assert banana not directly to the left

Regex demo

---

If you want to match bananas or straigtbanana you can assert that on the left is not banana preceded by a char a-zA-Z

[a-zA-Z]+(?![a-zA-Z])(?<!(?<![a-zA-Z])banana)

Regex demo

---

As suggested by @bobble bubble in the comments, if possessive quantifiers are supported and shortening the pattern using a case insensitive match:

[a-z]++(?<!(?<![a-z])banana)

• [a-z]++ Match 1+ chars in the range of a-z (possessive, do not backtrack)
• (?<! Negative lookbehind, assert what is directly to the left is not
  • (?<![a-z])banana Negative lookbehind, match banana not preceded by a-z
• ) Close lookbedhind

Regex demo

Answer 4

My two cents, assuming you do want to match words like "bananas":

(\b|\d)(?:banana|([a-zA-Z]+))(?1)

Your matches are in group 2, see an online demo

• (\b|\d) - A 1st capture group to hold a word-boundary or a digit.
• (?:banana|([a-zA-Z]+)) - A non-capture group with the alternation of either exactly "banana" or a 2nd capture group of 1+ alpha-chars.
• (?1) - Repeat the subpattern of the 1st capture group.

---

EDIT: If the backreference is not supported, you can try

(?:\b|\d)(?:banana|([a-zA-Z]+))(?:\b|\d)

Or, using lookarounds:

(?i)(?<![a-z])(?:banana|([a-z]+))(?![a-z])