X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=hacks%2Fwebcollage;h=260230abba83d01c820af5991d726579efbcc9f9;hb=13dbc569cdc6e29019722c0ef9b932a925efbcad;hp=bb3c66ef6ded01d0434310ae4c87528fb52b21b9;hpb=a94197e76a5dea5cb60542840809d6c20d0abbf3;p=xscreensaver diff --git a/hacks/webcollage b/hacks/webcollage index bb3c66ef..260230ab 100755 --- a/hacks/webcollage +++ b/hacks/webcollage @@ -35,32 +35,36 @@ use POSIX qw(strftime); my $progname = $0; $progname =~ s@.*/@@g; -my $version = q{ $Revision: 1.87 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/; +my $version = q{ $Revision: 1.93 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/; my $copyright = "WebCollage $version, Copyright (c) 1999-2002" . " Jamie Zawinski \n" . " http://www.jwz.org/xscreensaver/\n"; -my @search_methods = ( 30, "imagevista", \&pick_from_alta_vista_images, - 28, "altavista", \&pick_from_alta_vista_text, - 18, "yahoorand", \&pick_from_yahoo_random_link, - 14, "googleimgs", \&pick_from_google_images, +my @search_methods = ( 40, "imagevista", \&pick_from_alta_vista_images, + 30, "altavista", \&pick_from_alta_vista_text, + 19, "yahoorand", \&pick_from_yahoo_random_link, + 9, "lycos", \&pick_from_lycos_text, 2, "yahoonews", \&pick_from_yahoo_news_text, - 8, "lycos", \&pick_from_lycos_text, # Hotbot gives me "no matches" just about every time. # Then I try the same URL again, and it works. I guess # it caches searches, and webcollage always busts its # cache and time out? Or it just sucks. # 0, "hotbot", \&pick_from_hotbot_text, + + # Google asked (nicely) for me to stop searching them. + # 0, "googlenums", \&pick_from_google_image_numbers, + # 0, "googleimgs", \&pick_from_google_images, + ); # programs we can use to write to the root window (tried in ascending order.) # my @root_displayers = ( "chbg -once -xscreensaver -max_size 100", - "xv -root -quit -viewonly -noresetroot -quick24 -rmode 5" . + "xv -root -quit -viewonly +noresetroot -quick24 -rmode 5" . " -rfg black -rbg black", "xli -quiet -onroot -center -border black", "xloadimage -quiet -onroot -center -border black", @@ -137,6 +141,8 @@ my %warningless_sites = ( "yimg.com" => 1, # This is where dailynews.yahoo.com stores "eimg.com" => 1, # its images, so pick_from_yahoo_news_text() # hits this every time. + + "driftnet" => 1, # builtin... ); @@ -178,6 +184,12 @@ my $min_gif_area = (120 * 120); my $no_output_p = 0; my $urls_only_p = 0; +my @pids_to_kill = (); # forked pids we should kill when we exit, if any. + +my $driftnet_magic = 'driftnet'; +my $driftnet_dir = undef; +my $default_driftnet_cmd = "driftnet -a -m 100"; + my $wordlist; my %rejected_urls; @@ -348,6 +360,10 @@ sub get_document { my ( $url, $referer, $timeout ) = @_; my $start = time; + if (defined($referer) && $referer eq $driftnet_magic) { + return get_driftnet_file ($url); + } + my $orig_url = $url; my $loop_count = 0; my $max_loop_count = 4; @@ -1009,9 +1025,8 @@ sub pick_from_alta_vista_images { foreach my $u (@subpages) { # altavista is encoding their URLs now. - next unless ($u =~ - m@^/r\?ck_sm=[a-zA-Z0-9]+\&ref=[a-zA-Z0-9]+(\&uid=[a-zA-Z0-9]+)?\&r=(.*)@); - $u = url_unquote($2); + next unless ($u =~ m@^/r.*\&r=([^&]+).*@); + $u = url_unquote($1); next unless ($u =~ m@^http://@i); # skip non-HTTP or relative URLs next if ($u =~ m@[/.]altavista\.com\b@i); # skip altavista builtins @@ -1033,7 +1048,7 @@ sub pick_from_alta_vista_images { ############################################################################ # -# Pick images by feeding random words into Google Image Search +# Pick images by feeding random words into Google Image Search. # By Charles Gales # ############################################################################ @@ -1080,6 +1095,65 @@ sub pick_from_google_images { } + +############################################################################ +# +# Pick images by feeding random *numbers* into Google Image Search. +# By jwz, suggested by from Ian O'Donnell. +# +############################################################################ + + +# googlenums +sub pick_from_google_image_numbers { + my ( $timeout ) = @_; + + my $max = 9999; + my $number = int(rand($max)); + + $number = sprintf("%04d", $number) + if (rand() < 0.3); + + my $words = "$number"; + my $page = (int(rand(40)) + 1); + my $num = 20; # 20 images per page + my $search_url = $google_images_url . $words; + + if ($page > 1) { + $search_url .= "&start=" . $page*$num; # page number + $search_url .= "&num=" . $num; #images per page + } + + my ($search_hit_count, @subpages) = + pick_from_search_engine ($timeout, $search_url, $words); + + my @candidates = (); + my %referers; + foreach my $u (@subpages) { + next unless ($u =~ m@imgres\?imgurl@i); # All pics start with this + next if ($u =~ m@[/.]google\.com\b@i); # skip google builtins + + if ($u =~ m@^/imgres\?imgurl=(.*?)\&imgrefurl=(.*?)\&@) { + my $ref = $2; + my $img = "http://$1"; + + LOG ($verbose_filter, " candidate: $ref"); + push @candidates, $img; + $referers{$img} = $ref; + } + } + + @candidates = depoison (@candidates); + return () if ($#candidates < 0); + my $i = int(rand($#candidates+1)); + my $img = $candidates[$i]; + my $ref = $referers{$img}; + + LOG ($verbose_load, "picked image " . ($i+1) . ": $img (on $ref)"); + return ($ref, $img); +} + + ############################################################################ # @@ -1119,8 +1193,7 @@ sub pick_from_alta_vista_text { # onMouseOver to make it look like they're not! Well, it makes it # easier for us to identify search results... # - next unless ($u =~ - m@^/r\?ck_sm=[a-zA-Z0-9]+\&ref=[a-zA-Z0-9]+\&uid=[a-zA-Z0-9]+\&r=(.*)@); + next unless ($u =~ m@^/r.*\&r=([^&]+).*@); $u = url_unquote($1); LOG ($verbose_filter, " candidate: $u"); @@ -1204,7 +1277,13 @@ sub pick_from_lycos_text { # Lycos plays exact the same redirection game as hotbot. # Note that "id=0" is used for internal advertising links, # and 1+ are used for search results. - next unless ($u =~ m@^http://click.hotbot.com/director.asp\?id=[1-9]\d*&target=([^&]+)@); + next unless ($u =~ m@^http://click.hotbot.com/director.asp + .* + [?&]id=[1-9]\d* + .* + \&target=([^&]+) + .* + @x); $u = url_decode($1); LOG ($verbose_filter, " candidate: $u"); @@ -1252,6 +1331,113 @@ sub pick_from_yahoo_news_text { + +############################################################################ +# +# Pick images by waiting for driftnet to populate a temp dir with files. +# Requires driftnet version 0.1.5 or later. +# (Driftnet is a program by Chris Lightfoot that sniffs your local ethernet +# for images being downloaded by others.) +# Driftnet/webcollage integration by jwz. +# +############################################################################ + +# driftnet +sub pick_from_driftnet { + my ( $timeout ) = @_; + + my $id = $driftnet_magic; + my $dir = $driftnet_dir; + my $start = time; + my $now; + + error ("\$driftnet_dir unset?") unless ($dir); + $dir =~ s@/+$@@; + + error ("$dir unreadable") unless (-d "$dir/."); + + $timeout = $http_timeout unless ($timeout); + $last_search = $id; + + while ($now = time, $now < $start + $timeout) { + local *DIR; + opendir (DIR, $dir) || error ("$dir: $!"); + while (my $file = readdir(DIR)) { + next if ($file =~ m/^\./); + $file = "$dir/$file"; + closedir DIR; + LOG ($verbose_load, "picked file $file ($id)"); + return ($id, $file); + } + closedir DIR; + } + LOG (($verbose_net || $verbose_load), "timed out for $id"); + return (); +} + + +sub get_driftnet_file { + my ($file) = @_; + + error ("\$driftnet_dir unset?") unless ($driftnet_dir); + + my $id = $driftnet_magic; + my $re = qr/$driftnet_dir/; + error ("$id: $file not in $driftnet_dir?") + unless ($file =~ m@^$re@o); + + local *IN; + open (IN, $file) || error ("$id: $file: $!"); + my $body = ''; + while () { $body .= $_; } + close IN; + unlink ($file); + return ($id, $body); +} + + +sub spawn_driftnet { + my ($cmd) = @_; + + # make a directory to use. + while (1) { + my $tmp = $ENV{TEMPDIR} || "/tmp"; + $driftnet_dir = sprintf ("$tmp/driftcollage-%08x", rand(0xffffffff)); + LOG ($verbose_exec, "mkdir $driftnet_dir"); + last if mkdir ($driftnet_dir, 0700); + } + + if (! ($cmd =~ m/\s/)) { + # if the command didn't have any arguments in it, then it must be just + # a pointer to the executable. Append the default args to it. + my $dargs = $default_driftnet_cmd; + $dargs =~ s/^[^\s]+//; + $cmd .= $dargs; + } + + # point the driftnet command at our newly-minted private directory. + # + $cmd .= " -d $driftnet_dir"; + $cmd .= ">/dev/null" unless ($verbose_exec); + + my $pid = fork(); + if ($pid < 0) { error ("fork: $!\n"); } + if ($pid) { + # parent fork + push @pids_to_kill, $pid; + LOG ($verbose_exec, "forked for \"$cmd\""); + } else { + # child fork + nontrapping_system ($cmd) || error ("exec: $!"); + } + + # wait a bit, then make sure the process actually started up. + # + sleep (1); + error ("pid $pid failed to start \"$cmd\"") + unless (1 == kill (0, $pid)); +} + ############################################################################ # @@ -1263,7 +1449,6 @@ sub pick_from_yahoo_news_text { # Picks a random image on a random page, and returns two URLs: # the page containing the image, and the image. # Returns () if nothing found this time. -# Uses the url-randomizer 1 time in 5, else the image randomizer. # sub pick_image { @@ -1488,6 +1673,11 @@ sub save_recent_url { $_ = $url; my ($site) = m@^http://([^ \t\n\r/:]+)@; + if ($base eq $driftnet_magic) { + $site = $driftnet_magic; + @recent_images = (); + } + my $done = 0; foreach (@recent_images) { if ($_ eq $url) { @@ -1571,10 +1761,12 @@ sub jpeg_size { while (ord($ch) != 0xDA && $i < $L) { # Find next marker, beginning with 0xFF. while (ord($ch) != 0xFF) { + return () if (length($body) <= $i); $ch = substr($body, $i, 1); $i++; } # markers can be padded with any number of 0xFF. while (ord($ch) == 0xFF) { + return () if (length($body) <= $i); $ch = substr($body, $i, 1); $i++; } @@ -1586,6 +1778,7 @@ sub jpeg_size { ($marker != 0xC4) && ($marker != 0xCC)) { # it's a SOFn marker $i += 3; + return () if (length($body) <= $i); my $s = substr($body, $i, 4); $i += 4; my ($a,$b,$c,$d) = unpack("C"x4, $s); return (($c<<8|$d), ($a<<8|$b)); @@ -1593,6 +1786,7 @@ sub jpeg_size { } else { # We must skip variables, since FFs in variable names aren't # valid JPEG markers. + return () if (length($body) <= $i); my $s = substr($body, $i, 2); $i += 2; my ($c1, $c2) = unpack ("C"x2, $s); my $length = ($c1 << 8) | $c2; @@ -1636,6 +1830,24 @@ sub bellrand { } +sub signal_cleanup { + my ($sig) = @_; + print STDERR blurb() . (defined($sig) + ? "caught signal $sig." + : "exiting.") + . "\n" + if ($verbose_exec); + + x_cleanup(); + + if (@pids_to_kill) { + print STDERR blurb() . "killing: " . join(' ', @pids_to_kill) . "\n"; + kill ('TERM', @pids_to_kill); + } + + exit 1; +} + ############################################################################## # # Generating a list of urls only @@ -1670,14 +1882,10 @@ my $background = undef; my $img_width; # size of the image being generated. my $img_height; -my $delay = 0; - +my $delay = 2; sub x_cleanup { - my ($sig) = @_; - print STDERR blurb() . "caught signal $sig.\n" if ($verbose_exec); unlink $image_ppm, $image_tmp1, $image_tmp2; - exit 1; } @@ -1856,17 +2064,6 @@ sub x_or_pbm_output { # $ppm_to_root_window_cmd = pick_root_displayer(); - - $SIG{HUP} = \&x_cleanup; - $SIG{INT} = \&x_cleanup; - $SIG{QUIT} = \&x_cleanup; - $SIG{ABRT} = \&x_cleanup; - $SIG{KILL} = \&x_cleanup; - $SIG{TERM} = \&x_cleanup; - - # Need this so that if giftopnm dies, we don't die. - $SIG{PIPE} = 'IGNORE'; - if (!$img_width || !$img_height) { $_ = "xdpyinfo"; which ($_) || error "$_ not found on \$PATH."; @@ -1978,7 +2175,13 @@ sub paste_image { my ($iw, $ih); - if (defined ($webcollage_helper)) { + # If we are using the webcollage-helper, then we do not need to convert this + # image to a PPM. But, if we're using a filter command, we still must, since + # that's what the filters expect (webcollage-helper can read PPMs, so that's + # fine.) + # + if (defined ($webcollage_helper) && + !defined ($filter_cmd)) { ($iw, $ih) = image_size ($body); if (!$iw || !$ih) { @@ -2210,8 +2413,20 @@ sub paste_image { # cumulative. # if ($post_filter_cmd) { + + my $cmd; + $target = $image_tmp1; - $rc = nontrapping_system "($post_filter_cmd) < $image_ppm > $target"; + if (!defined ($webcollage_helper)) { + $cmd = "($post_filter_cmd) < $image_ppm > $target"; + } else { + # Blah, my scripts need the JPEG data, but some other folks need + # the PPM data -- what to do? Ignore the problem, that's what! +# $cmd = "djpeg < $image_ppm | ($post_filter_cmd) > $target"; + $cmd = "($post_filter_cmd) < $image_ppm > $target"; + } + + $rc = nontrapping_system ($cmd); if ($rc != 0) { LOG ($verbose_pbm, "filter failed: \"$post_filter_cmd\"\n"); return; @@ -2257,12 +2472,29 @@ sub paste_image { } +sub init_signals { + + $SIG{HUP} = \&signal_cleanup; + $SIG{INT} = \&signal_cleanup; + $SIG{QUIT} = \&signal_cleanup; + $SIG{ABRT} = \&signal_cleanup; + $SIG{KILL} = \&signal_cleanup; + $SIG{TERM} = \&signal_cleanup; + + # Need this so that if giftopnm dies, we don't die. + $SIG{PIPE} = 'IGNORE'; +} + +END { signal_cleanup(); } + + sub main { $| = 1; srand(time ^ $$); my $verbose = 0; my $dict; + my $driftnet_cmd = 0; $current_state = "init"; $load_method = "none"; @@ -2314,6 +2546,13 @@ sub main { $http_proxy = shift @ARGV; } elsif ($_ eq "-dictionary" || $_ eq "-dict") { $dict = shift @ARGV; + } elsif ($_ eq "-driftnet" || $_ eq "--driftnet") { + @search_methods = ( 100, "driftnet", \&pick_from_driftnet ); + if (! ($ARGV[0] =~ m/^-/)) { + $driftnet_cmd = shift @ARGV; + } else { + $driftnet_cmd = $default_driftnet_cmd; + } } elsif ($_ eq "-debug" || $_ eq "--debug") { my $which = shift @ARGV; my @rest = @search_methods; @@ -2333,11 +2572,13 @@ sub main { LOG (1, "DEBUG: using only \"$which\""); } else { - print STDERR "$copyright\nusage: $progname [-root]" . - " [-display dpy] [-root] [-verbose] [-timeout secs]\n" . - "\t\t [-delay secs] [-filter cmd] [-filter2 cmd]\n" . - "\t\t [-dictionary dictionary-file]\n" . - "\t\t [-http-proxy host[:port]]\n"; + print STDERR "$copyright\nusage: $progname " . + "[-root] [-display dpy] [-verbose] [-debug which]\n" . + "\t\t [-timeout secs] [-delay secs] [-filter cmd] [-filter2 cmd]\n" . + "\t\t [-no-output] [-urls-only] [-background color] [-size WxH]\n" . + "\t\t [-dictionary dictionary-file] [-http-proxy host[:port]]\n" . + "\t\t [-driftnet [driftnet-program-and-args]]\n" . + "\n"; exit 1; } } @@ -2417,6 +2658,10 @@ sub main { pick_dictionary(); } + init_signals(); + + spawn_driftnet ($driftnet_cmd) if ($driftnet_cmd); + if ($urls_only_p) { url_only_output; } else {