--- comics/fetch.pl.new 2022/10/04 12:02:03 1.30 +++ comics/fetch.pl.new 2024/12/13 16:03:49 1.31 @@ -2,6 +2,9 @@ ############################################################################### # $Log: fetch.pl.new,v $ +# Revision 1.31 2024/12/13 16:03:49 nick +# This adds the ability to specify a comic as a link only with a default splash image. +# # Revision 1.30 2022/10/04 12:02:03 nick # Added --no-check-certificate for wget calls as arcamax was failing its cert check. Meh, whatever. It's just comics. # @@ -69,10 +72,12 @@ use JSON::Create 'create_json'; use Date::Calc qw/Date_to_Text_Long Today Day_of_Week Day_of_Week_to_Text/; use Data::Dumper; +print("Running"); + ## ## Some default values ## -my $ver = '$Id: fetch.pl.new,v 1.30 2022/10/04 12:02:03 nick Exp $'; +my $ver = '$Id: fetch.pl.new,v 1.31 2024/12/13 16:03:49 nick Exp $'; my $comicFile = "comics.conf"; my $comicConfigVer = "Unknown"; my $reportFile = "/home/httpd/html/daily/comics/status_report.json"; @@ -99,6 +104,8 @@ print STDOUT "Starting comic fetch at $D foreach my $comic ( sort keys %comics ) { + print("Checking Comic $comic\n"); + ## Skip if this is Sunday and the comic is weekdays only next if ( $comic =~ m/config/ ); if (($dates{'wday'} eq "Sunday") && @@ -120,18 +127,20 @@ foreach my $comic ( sort keys %comics ) my $file = "$imageDir/$comic-$dates{'day2'}.$comics{$comic}{'ext'}"; my $size = 0; - my $cmd = "/usr/bin/identify -verbose $file|"; - open(IMG, $cmd) || die ("Can't open: $!\n"); - while() { - if ($_ =~ m/^\s+geometry:\s+(\d+)x\d+.*/i) { - $size = $1 if ( $size == 0); - } - } - close(IMG); - - - system( "/usr/bin/convert -resize 800 $file $file" ) - if ( $size > 800 ) + ## Resize downloaded images + if($comics{$comic}{'mode'} != 3) { + my $cmd = "/usr/bin/identify -verbose $file|"; + open(IMG, $cmd) || die ("Can't open: $!\n"); + while() { + if ($_ =~ m/^\s+geometry:\s+(\d+)x\d+.*/i) { + $size = $1 if ( $size == 0); + } + } + close(IMG); + + system( "/usr/bin/convert -resize 800 $file $file" ) + if ( $size > 800 ) + } } ## &writeMainIndex ( \%dates ); @@ -168,6 +177,10 @@ sub downloadComic ($$) { return directDownload ( \%comics, $comic, $date ); last SWITCH; } + if ( $comics->{$comic}{'mode'} eq 3 ) { + return 0; + last SWITCH; + } } return "ERROR: Unknown download method specified for $comics->{$comic}{'fullName'}."; @@ -257,7 +270,8 @@ sub writeComic ($$) { my $indexFile = $indexDir . "/index-" . $date->{'year2'} . $date->{'mon2'} . $date->{'day2'} . "-" . $sd . ".html"; - $comics->{$comic}{'fullName'} =~ s/&/&/g; + $comics->{$comic}{'fullName'} =~ s/&/&/g; + my $content = <{$comic}{'fullName'}) ******* --> @@ -265,11 +279,22 @@ sub writeComic ($$) { $comics->{$comic}{'fullName'}     - + $comics->{$comic}{'url'}
+EOF + if ( $comics->{$comic}{'mode'} == 3 ) { + print("Mode 3\n"); + $content .= < +EOF + } else { + $content .= < +EOF + } + $content .= <
@@ -395,6 +420,13 @@ sub directDownload ($$) { ####################################################################### ####################################################################### +sub linkOnly ($$) { + my ( $comics, $comic, $date ) = @_; + + return 0; +} +####################################################################### +####################################################################### sub indexDownload ($$) { my ( $comics, $comic, $date ) = @_; my ( @lines, $comicLine, $mainURL ); @@ -437,9 +469,13 @@ sub indexDownload ($$) { ## Find the comic strip URL based on the specified regex in the search ## + print "Using search $comics->{$comic}{'search'}\n"; + foreach my $line (@lines) { if ( $line =~ m/$comics->{$comic}{'search'}/i ) { + print "Found match:\n"; $comicLine = $1; chomp $comicLine; + print "+ $comicLine\n"; } } @@ -450,8 +486,10 @@ sub indexDownload ($$) { my $cDate = $date->{'day2'}; if ( $comicLine ) { + print "Downloading Comic\n"; if ( $comicLine =~ m/(gif|jpg|png)/i ) { $comics->{$comic}{'ext'} = $1; } my $comicURL = ( $comicLine =~ m/http/ ) ? $comicLine : $mainURL . $comicLine; + print "Final URL: $comicURL\n"; # Strip & $comicURL =~ s/\&\;/&/g; my $cmd = "wget --no-check-certificate --user-agent=\"$USER_AGENT\" --referer='" . $comics->{$comic}{'url'} . "' -q '$comicURL' -O images/$cDir/$comic-$cDate.$comics->{$comic}{'ext'}";