--- rxvt-unicode/src/perl/selection	2006/01/03 21:08:39	1.9
+++ rxvt-unicode/src/perl/selection	2006/01/23 23:13:03	1.34
@@ -9,38 +9,122 @@
    ()
 }
 
-my @patterns = (
-   # urls
-   qr{ ([a-z0-9.+\-]+://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),]+) }x,
+sub on_init {
+   my ($self) = @_;
+
+   for (my $idx = 0; defined (my $res = $self->x_resource ("selection.pattern-$idx")); $idx++) {
+      $res = $self->locale_decode ($res);
+      utf8::encode $res;
+      push @{ $self->{patterns} }, qr/$res/;
+   }
+
+   ()
+}
+
+# "find interesting things"-patterns
+my @mark_patterns = (
+   # common types of "parentheses"
+   qr{ (?<![^[:space:]]) ‘ ([^‘’]+) ’ (?![^[:space]]) }x,
+   qr{ (?<![^[:space:]]) ` ([^`']+) ' (?![^[:space]]) }x,
+   qr{ (?<![^[:space:]]) (" [^[:space:]] [^"]* ")                 }x,
+   qr{                   (" [^"]* [^[:space:]] ") (?![^[:space]]) }x,
+   qr{ \< ([^<>[:space:]]+) \> }x,
+   qr{ \{ ([^{}[:space:]]+) \} }x,
+   qr{ \[ ([^{}[:space:]]+) \] }x,
+   qr{ \( ([^()[:space:]]+) \) }x,
+
+   # urls, just a heuristic
+   qr{(
+      (?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+
+      [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~]   # exclude some trailing characters (heuristic)
+   )}x,
 
-   # shell-like argument quoting
-   qr{\G\s*(
+   # shell-like argument quoting, basically always matches
+   qr{\G [\ \t|&;<>()]* (
       (?:
-         [^"'\\ \t]+
+         [^\\"'\ \t|&;<>()]+
          | \\.
-         | " ([^\\"]+ | \\. )* "
+         | " (?: [^\\"]+ | \\. )* "
          | ' [^']* '
       )+
-   )}xs,
+   )}x,
+);
+
+# "correct obvious? crap"-patterns
+my @simplify_patterns = (
+   qr{^"([^\\"'\ \t|&;<>()*?]+)"$}, # "simple" => simple
+   qr{^(.*)[,\-]$},                 # strip off trailing , and -
 );
 
 sub on_sel_extend {
-   my ($self) = @_;
+   my ($self, $time) = @_;
 
    my ($row, $col) = $self->selection_mark;
    my $line = $self->line ($row);
-   my $offset = $line->offset_of ($row, $col);
    my $text = $line->t;
+   my $markofs = $line->offset_of ($row, $col);
+   my $curlen  = $line->offset_of ($self->selection_end)
+               - $line->offset_of ($self->selection_beg);
+
+   my @matches;
+
+   if ($markofs < $line->l) {
+      # convert markofs form character to UTF-8 offset space
+      {
+         my $prefix = substr $text, 0, $markofs;
+         utf8::encode $prefix;
+         $markofs = length $prefix;
+      }
+
+      # not doing matches in unicode mode helps speed
+      # enourmously here. working in utf-8 should be
+      # equivalent due to the magic of utf-8 encoding.
+      utf8::encode $text;
+      study $text; # _really_ helps, too :)
+
+      for my $regex (@mark_patterns, @{ $self->{patterns} }) {
+         while ($text =~ /$regex/g) {
+            if ($-[1] <= $markofs and $markofs <= $+[1]) {
+               my $ofs = $-[1];
+               my $match = $1;
+
+               for my $regex (@simplify_patterns) {
+                  if ($match =~ $regex) {
+                     $match = $1;
+                     $ofs += $-[1];
+                  }
+               }
 
-   for my $regex (@patterns) {
-      while ($text =~ /$regex/g) {
-         if ($-[1] <= $offset and $offset <= $+[1]) {
-            $self->selection_beg ($line->coord_of ($-[1]));
-            $self->selection_end ($line->coord_of ($+[1]));
-            return 1;
+               push @matches, [$ofs, length $match];
+            }
          }
       }
    }
 
+   # whole line
+   push @matches, [0, ($line->end - $line->beg + 1) * $self->ncol];
+
+   for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) {
+      my ($ofs, $len) = @$_;
+
+      next if $len <= $curlen;
+
+      # convert back from UTF-8 offset space to character space
+      {
+         my $length = substr "$text ", $ofs, $len;
+         utf8::decode $length;
+         $len = length $length;
+      }
+      {
+         my $prefix = substr $text, 0, $ofs;
+         utf8::decode $prefix;
+         $ofs = length $prefix;
+      }
+
+      $self->selection_beg ($line->coord_of ($ofs));
+      $self->selection_end ($line->coord_of ($ofs + $len));
+      return 1;
+   }
+
    ()
 }