How can I find the contents of an HTML tag?

[Anonymous Monk]

Given some HTML like

<TR><TD>Foo:</TD><TD> bar </TD></TR>

I want to extract the string "bar" into a variable.

Note the space on each side will be there, even if the value string is "".

Originally posted as a Categorized Question.

[mikfire]

Assuming that the distinguishing characteristic of the TD entry you want to extract is the leading and following space, I'd suggested a regex something like this:

my( $var ) = $html =~ m#<TD> (.*?) </TD>#;

print "We found it: $var\n" if defined $var;

The part inside the capturing parens (.*?) says to save any characters found, possibly none. It says to take the fewest possible characters to complete the match — i.e., be non-greedy.

The only way you will be able to know if the match succeeded is to test for definedness. Testing for true/false will fail on the empty case because perl treats the empty string as false.

If the <TD>Foo:</TD> part will always occur immediately in front of the <TD> instances you're interested in, we can make the regex more robust:

m#<TD>Foo:</TD><TD> (.*?) </TD>#

[chromatic]

One obvious regex-based solution is:

if ( $html =~ m#<TR><TD>Foo:</TD><TD> (.*) </TD></TR># )
{
    $var = $1;
}

[nuance]

You could use a regular expression with the /g modifier:

m#<TD>(.*?)</TD>#g

This will return a list of all the matches.

You can then select whichever term you want from the list:

my $html = "<TR><TD>Foo:</TD><TD> bar </TD></TR>";

my $var = ( $html =~ m#<TD>(.*?)</TD>#g )[1];

Note that you have to modify the asterisk with the question mark to specify non-greedy matching. If you don't, you will get just one big match, like

Foo:</TD><TD> bar

— probably not what you wanted!

You should also add the /s modifier, if the string you need to extract breaks across multiple lines. Specifically, /s allows the dot to match newline characters along with all the other characters.

[davorg]

Parsing HTML using regular expressions can be a complete minefield if you don't know exactly what your data will look like.

In situations like this, you're far better off using a real parser, such as HTML::Parser or one of its sub-classes.