Why does the following regular expression to find all instances EXCEPT abd require a dot at the end of the given pattern to function as intended:
>> regexp({'abc';'abd'},'^(?!abd$)')
ans =
{
[1,1] = [](1x0)
[2,1] = [](1x0)
}
>> regexp({'abc';'abd'},'^(?!abd$).')
ans =
{
[1,1] = 1
[2,1] = [](1x0)
}
This applies in the simpler case without use of a cell as well:
>> regexp('abc','^(?!abd$)')
ans = [](1x0)
>> regexp('abc','^(?!abd$).')
ans = 1
This is clear after you check the regexp reference page and see that by default, empty matches are not returned.
Note that ^(?!abd$) pattern is a sequence of two non-consuming patterns, a start of string anchor (^) and a negative lookahead ((?!abd$)) that matches a location in a string that is not immediately followed with abd and end of string.
When you add a . after the lookahead, you introduce a consuming part to the regex, and now it actually matches and consumes any single char, and then the regexp returns a non-empty match.
All you need is to override the default behavior by adding the emptymatch option:
>> regexp({'abc';'abd'},'^(?!abd$)', 'emptymatch'
)
ans =
{
[1,1] = 1
[2,1] = [](1x0)
}