Re^2: problem with optional capture group

The problem with using the + quantifier is that the entire regex will not match a line that has an opening <div tag but no closing </div tag on the same line. For example, your modification will not match the following:

my $line = "<div id=\"roguebin-response-35911\" class=\"bin-response\"
+>";
[download]

I was hoping to write a single regex that will handle both cases, i.e. an opening <div tag with a closing </div tag on the same line, and an opening <div tag with no closing </div tag on the same line. I understand a built in parser would make this task easier, but I would still like to understand how to write a single regex that would capture both of these cases.

Comment on Re^2: problem with optional capture group Download Code

Replies are listed 'Best First'.
Re^3: problem with optional capture group by AnomalousMonk (Archbishop) on Dec 22, 2020 at 22:57 UTC
Win8 Strawberry 5.8.9.5 (32) Tue 12/22/2020 16:43:09 C:\@Work\Perl\monks >perl -Mstrict -Mwarnings for my $line ( '<div id="foo-bar-321" class="bin-boff"></div>', '<div id="foo-bar-321" class="bin-boff"> </div>', '<div id="foo-bar-321" class="bin-boff">foo</div>', '<div id="foo-bar-321" class="bin-boff"> foo </div>', '<div id="foo-bar-321" class="bin-boff">', '<div id="foo-bar-321" class="bin-boff"> ', '<div id="foo-bar-321" class="bin-boff">foo', '<div id="foo-bar-321" class="bin-boff"> foo', ) { if ($line =~ m{ <div (?: (?! </div) .)+ (</div)? }xms) { print "line matched \n '$&' \n"; if (defined $1) { print " right after match, \$1 is defined '$1' \n"; } } } ^Z line matched '<div id="foo-bar-321" class="bin-boff"></div' right after match, $1 is defined '</div' line matched '<div id="foo-bar-321" class="bin-boff"> </div' right after match, $1 is defined '</div' line matched '<div id="foo-bar-321" class="bin-boff">foo</div' right after match, $1 is defined '</div' line matched '<div id="foo-bar-321" class="bin-boff"> foo </div' right after match, $1 is defined '</div' line matched '<div id="foo-bar-321" class="bin-boff">' line matched '<div id="foo-bar-321" class="bin-boff"> ' line matched '<div id="foo-bar-321" class="bin-boff">foo' line matched '<div id="foo-bar-321" class="bin-boff"> foo' [download] Give a man a fish: `<%-{-{-{-<`	[reply] [d/l] [select]
Re^4: problem with optional capture group by Special_K (Pilgrim) on Dec 23, 2020 at 16:31 UTC
`m{ <div (?: (?! </div) .)+ (</div)? }xms)` [download] Can you please give a brief explanation regarding how the above regex works? It seems to use a few constructs I've never seen before and searching Google for regex symbols doesn't work very well. In particular, is enclosing a regex in 'm()', as you have done above, equivalent to enclosing it in '//'? What is the trailing xms doing?	[reply] [d/l]
Re^5: problem with optional capture group by AnomalousMonk (Archbishop) on Dec 23, 2020 at 20:54 UTC
... enclosing a regex in 'm()' ... The `m` `open-delimiter` `pattern` `close-delimiter` form is what I think of as the "canonical" form of the `m//` operator, where the delimiters can be a wide variety of characters including `{} () <> []` matching braces. The `//` match form is a special case. Likewise the `qr// s///` operators. This alleviates a lot of escape-ology connected with the `/` character in regexes. See perlop. (Note that `q// qq// qx// qw// tr/// y///` and maybe some others also use this delimiter convention.) What is the trailing xms doing? I use the `/ms` modifiers as part of a standard "tail" on all my `qr// m// s///` expressions to give the `. ^ $` operators a ~~standard~~ \| fixed behavior. This eliminates some degrees of freedom in regex behavior and makes them slightly easier to understand. The `/x` modifier in the standard tail enables the use of whitespace to help clarify a regex. See Modifiers in perlre. `(?: (?! </div) .)+` This has already been covered by GrandFather here. This expression just steps forward grabbing one character after another as long as that character is not a part of whatever matches the `(?!...)` negative lookahead expression, a closing `div` tag fragment in this case. A bit slow perhaps, but effective and flexible (update: flexible in that the lookahead expression can be of any complexity). See Lookaround Assertions in perlre; see also perlretut, perlreref and perlrequick. `(</div)?` Optionally capture a literal character sequence if it is present. The capture variable `$1` (in this case) will hold the captured sequence if it was present, otherwise `$1` will be undefined. See perlre, etc., as above. Give a man a fish: `<%-{-{-{-<`	[reply] [d/l] [select]