--- trunk/spider/filter.pm 2004/01/20 18:40:06 51 +++ trunk/spider/filter.pm 2004/01/23 13:10:40 56 @@ -37,6 +37,55 @@ } + # is second argument collection? + my $collection = shift || return $contents; + + # construct new title (from various parts of DocBook if available) + my $new_title; + + if ($contents =~ m,,is) { + $new_title = $1; + } elsif ($contents =~ m,]*>([^<]+),is) { + $new_title = $1; + } elsif ($contents =~ m,]*>([^<]+),is) { + $new_title = $1; + } elsif ($contents =~ m,]*>([^<]+),is) { + $new_title = $1; + } else { + if ($contents =~ m,([^<]+),is) { + $new_title = $1; + } elsif ($contents =~ m,]*>([^<]+),is) { + $new_title = $1; + } + } + + if ($new_title) { + # check if new title is same as collection name + my ($a,$b) = ($new_title,$collection); + $a =~ s/([^a-zA-Z])+/ /gs; + $b =~ s/([^a-zA-Z])+/ /gs; + if ($a =~ m/$b/i) { + $new_title = $collection; + } else { + $new_title = "$collection :: $new_title"; + } + } else { + # fall-back to collection title + $new_title = $collection; + } + + $new_title =~ s/\s\s+/ /g; + + if ($contents =~ s,(.*),$new_title,is) { + print STDERR "replace title '$1' with '$new_title'\n"; + } else { + print STDERR "adding new title '$new_title'\n"; + # try to insert after , or at top + $contents =~ s,(),$1$new_title,is || + $contents =~ s,(),$1$new_title,is || + $contents =~ s,^,$new_title,; + } + return $contents; }