sub fix_11 { # HTML named entities
# valamint 20: Symbol for dead
# valamint 50: en dash or em dash
my ($error_code, $title_str, $text_str) = @_;
my %HTML_entitiy = (
quot => "\x{0022}",
sect => "\x{00a7}",
copy => "\x{00a9}",
middot => "\x{00b7}",
Agrave => "\x{00c0}",
Aring => "\x{00c5}",
Auml => "\x{00c4}",
AElig => "\x{00c6}",
Ccedil => "\x{00c7}",
Ntilde => "\x{00d1}",
Uuml => "\x{00dc}",
agrave => "\x{00e0}",
auml => "\x{00e4}",
aring => "\x{00e5}",
aelig => "\x{00e6}",
ccedil => "\x{00e7}",
ntilde => "\x{00f1}",
uuml => "\x{00fc}",
minus => "\x{2012}",
ndash => "\x{2013}",
mdash => "\x{2014}",
dagger => "\x{2020}",
Dagger => "\x{2021}",
bull => "\x{2022}",
hellip => "\x{2026}",
trade => "\x{2122}",
larr => "\x{2190}",
uarr => "\x{2191}",
rarr => "\x{2192}",
darr => "\x{2193}",
harr => "\x{2194}",
lArr => "\x{21d0}",
uArr => "\x{21d1}",
rArr => "\x{21d2}",
dArr => "\x{21d3}",
hArr => "\x{21d4}",
);
my $excluded = search_tags($text_str);
my @entities;
my $count = 0;
while ($text_str =~ m|\&(\w+);|gs) {
my ($pos, $entity, $len) =
(pos($text_str), $1, length($&));
$pos -= $len;
next unless exists $HTML_entitiy{$entity};
next if $excluded->contains($pos);
unshift(@entities, {
'pos' => $pos,
len => $len,
entity => $entity,
replacement => $HTML_entitiy{$entity},
});
$count++;
}
$count or return ('Nincs HTML entitas');
foreach (@entities) {
substr($text_str, $_->{pos}, $_->{len}) = $_->{replacement};
}
my $summary_str = $latin2->decode(
"Bot: $count HTML entitás lecserélve. (Hibakód: $error_code)"
);
return ($summary_str, $count, $text_str);
}
