Better way to remove specific characters from a Perl string

With a character class this big it is easier to say what you want to keep. A caret in the first position of a character class inverts its sense, so you can write

$varTemp =~ s/[^"%'+\-0-9<=>a-z_{|}]+//gi

or, using the more efficient tr

$varTemp =~ tr/"%'+\-0-9<=>A-Z_a-z{|}//cd

tr docs

You could use the tr instead:

       $p =~ tr/fo//d;

will delete every f and every o from $p. In your case it should be:

       $p =~ tr/\$#@~!&*()[];.,:?^ `\\\///d

See Perl's tr documentation.

tr/SEARCHLIST/REPLACEMENTLIST/cdsr

Transliterates all occurrences of the characters found (or not found if the /c modifier is specified) in the search list with the positionally corresponding character in the replacement list, possibly deleting some, depending on the modifiers specified.

[…]

If the /d modifier is specified, any characters specified by SEARCHLIST not found in REPLACEMENTLIST are deleted.

You've misunderstood how character classes are used:

$varTemp =~ s/[\$#@~!&*()\[\];.,:?^ `\\\/]+//g;

does the same as your regex (assuming you didn't mean to remove ' characters from your strings).

Edit: The + allows several of those "special characters" to match at once, so it should also be faster.