The slash is a basic literal. They wanted to work with roughly the same example for the three cases, and they built a pattern whose beginning matches both opening tags '<t' and closing tags '</t'. They coded this as '</?t'. It has no real purpose in the first, greedy case. The purpose appears in the second case, where you see that it allows extracting both opening and closing tags whose name start with t.
They could have explained it as follows: if you wanted to extract all tags from the following string:
str = '<tr><td><p>text</p></td>' ;
your first attempt would probably be something like
>> regexp( str, '<.*>', 'match' )
ans =
'<tr><td><p>text</p></td>'
which doesn't work, because, by default, REGEXP is greedy: it maximizes the match. One way to overcome this could be to match not any character with . but any character different from > :
>> regexp( str, '<[^>]*>', 'match' )
ans =
'<tr>' '<td>' '<p>' '</p>' '</td>'
Yet, it is not always possible to specify a single character which should not be matched, and we sometimes have to make the greedy * quantifier lazy (as little as necessary) by appending a ?:
>> regexp( str, '<.*?>', 'match' )
ans =
'<tr>' '<td>' '<p>' '</p>' '</td>'
Etc. Now the rest is extra material for making the example more specific, but it has nothing to do with greedy/lazy quantifiers.
Best Answer