URI::Find - Find URIs in arbitrary text

  require URI::Find;

  my $finder = URI::Find->new(\&callback);

  $how_many_found = $finder->find(\$text);

This module does one thing: Finds URIs and URLs in plain text. It finds them quickly and it finds them all (or what URI::URL considers a URI to be.) It only finds URIs which include a scheme (http:// or the like), for something a bit less strict have a look at URI::Find::Schemeless.

For a command-line interface, see Darren Chamberlain's CWurifind script. It's available from his CPAN directory, <http://www.cpan.org/authors/id/D/DA/DARREN/>.

new

  my $finder = URI::Find->new(\&callback);

Creates a new URI::Find object. &callback is a function which is called on each URI found. It is passed two arguments, the first is a URI::URL object representing the URI found. The second is the original text of the URI found. The return value of the callback will replace the original URI in the text.

find

  my $how_many_found = $finder->find(\$text);

$text is a string to search and possibly modify with your callback.

I got a bunch of mail from people asking if I'd add certain features to URI::Find. Most wanted the search to be less restrictive, do more heuristics, etc... Since many of the requests were contradictory, I'm letting people create their own custom subclasses to do what they want.

The following are methods internal to URI::Find which a subclass can override to change the way URI::Find acts. They are only to be called inside a URI::Find subclass. Users of this module are NOT to use these methods.

uri_re

  my $uri_re = $self->uri_re;

Returns the regex for finding absolute, schemed URIs (http://www.foo.com and such). This, combined with schemeless_uri_re() is what finds candidate URIs. Usually this method does not have to be overridden.

schemeless_uri_re

  my $schemeless_re = $self->schemeless_uri_re;

Returns the regex for finding schemeless URIs (www.foo.com and such) and other things which might be URIs. By default this will match nothing (though it used to try to find schemeless URIs which started with CWwww and CWftp). Many people will want to override this method. See URI::Find::Schemeless for a subclass does a reasonable job of finding URIs which might be missing the scheme.

uric_set

  my $uric_set = $self->uric_set;

Returns a set matching the 'uric' set defined in RFC 2396 suitable for putting into a character set ([]) in a regex. You almost never have to override this.

cruft_set

  my $cruft_set = $self->cruft_set;

Returns a set of characters which are considered garbage. Used by decruft().

decruft

  my $uri = $self->decruft($uri);

Sometimes garbage characters like periods and parenthesis get accidentally matched along with the URI. In order for the URI to be properly identified, it must sometimes be decrufted, the garbage characters stripped. This method takes a candidate URI and strips off any cruft it finds.

recruft

  my $uri = $self->recruft($uri);

This method puts back the cruft taken off with decruft(). This is necessary because the cruft is destructively removed from the string before invoking the user's callback, so it has to be put back afterwards.

schemeless_to_schemed

  my $schemed_uri = $self->schemeless_to_schemed($schemeless_uri);

This takes a schemeless URI and returns an absolute, schemed URI. The standard implementation supplies ftp:// for URIs which start with ftp., and http:// otherwise.

is_schemed

  $obj->is_schemed($uri);

Returns whether or not the given URI is schemed or schemeless. True for schemed, false for schemeless.

badinvo

  __PACKAGE__->badinvo($extra_levels, $msg)

This is used to complain about bogus subroutine/method invocations. The args are optional.

The old find_uri() function is still around and it works, but its deprecated.

Simply print the original URI text found and the normalized representation.

  my $finder = URI::Find->new(
                      sub {
                          my($uri, $orig_uri) = @_;
                          print "The text '$orig_uri' represents '$uri'\n";
                          return $orig_uri;
                      });
  $finder->find(\$text);

Check each URI in document to see if it exists.

  use LWP::Simple;

  my $finder = URI::Find->new(sub {
                                  my($uri, $orig_uri) = @_;
                                  if( head $uri ) {
                                      print "$orig_uri is okay\n";
                                  }
                                  else {
                                      print "$orig_uri cannot be found\n";
                                  }
                                  return $orig_uri;
                              });
  $finder->find(\$text);

Turn plain text into HTML, with each URI found wrapped in an HTML anchor.

  use CGI qw(escapeHTML);

  $text = "<pre>\n" . escapeHTML($text) . "</pre>\n";
  my $finder = URI::Find->new(
                              sub {
                                  my($uri, $orig_uri) = @_;
                                  return qq|<a href="$uri">$orig_uri</a>|;
                              });
  $finder->find(\$text);

RFC 2396 Appendix E suggests using the form '<http://www.foo.com>' or '<URL:http://www.foo.com>' when putting URLs in plain text. URI::Find accomidates this suggestion and considers the entire thing (brackets and all) to be part of the URL found. This means that when find_uris() sees '<URL:http://www.foo.com>' it will hand that entire string to your callback, not just the URL.

NOTE: The prototype on find_uris() is already getting annoying to me. I might remove it in a future version.

  L<URI::Find::Schemeless>, L<URI::URL>, L<URI>,
  RFC 2396 (especially Appendix E)

Michael G Schwern <schwern@pobox.com> with insight from Uri Gutman, Greg Bacon, Jeff Pinyan, Roderick Schertler and others.

Currently maintained by Roderick Schertler <roderick@argon.org>.