X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=hacks%2Fwebcollage;h=8e5b78f6795fcd38e3d5712e2899b21cd5e83f8c;hb=de460e831dc8578acfa8b72251ab9346c99c1f96;hp=cd3fa4319bbf911a96d56e7323ed3320f84eb98c;hpb=c494fd2e6b3b25582375d62e40f4f5cc984ca424;p=xscreensaver diff --git a/hacks/webcollage b/hacks/webcollage index cd3fa431..8e5b78f6 100755 --- a/hacks/webcollage +++ b/hacks/webcollage @@ -60,21 +60,28 @@ use bytes; # Larry can take Unicode and shove it up his ass sideways. my $progname = $0; $progname =~ s@.*/@@g; -my $version = q{ $Revision: 1.133 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/; +my $version = q{ $Revision: 1.142 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/; my $copyright = "WebCollage $version, Copyright (c) 1999-2005" . " Jamie Zawinski \n" . " http://www.jwz.org/webcollage/\n"; -my @search_methods = ( 56, "altavista", \&pick_from_alta_vista_random_link, - 11, "livejournal", \&pick_from_livejournal_images, - 5, "yahoorand", \&pick_from_yahoo_random_link, - 10, "googlephotos", \&pick_from_google_image_photos, - 5, "googleimgs", \&pick_from_google_images, - 3, "googlenums", \&pick_from_google_image_numbers, - 2, "flickr_recent", \&pick_from_flickr_recent, - 8, "flickr_random", \&pick_from_flickr_random, +my @search_methods = ( 20, "googlephotos", \&pick_from_google_image_photos, + 11, "googleimgs", \&pick_from_google_images, + 11, "googlenums", \&pick_from_google_image_numbers, + 20, "altavista", \&pick_from_alta_vista_random_link, + 13, "flickr_recent", \&pick_from_flickr_recent, + 10, "flickr_random", \&pick_from_flickr_random, + 9, "livejournal", \&pick_from_livejournal_images, + 6, "yahoorand", \&pick_from_yahoo_random_link, + + # This is a cute way to search for a certain webcams. + # Not included in default methods, since these images + # aren't terribly interesting by themselves. + # See also "SurveillanceSaver". + # + 0, "securitycam", \&pick_from_security_camera, # In Apr 2002, Google asked me to stop searching them. # I asked them to add a "random link" url. They said @@ -181,6 +188,11 @@ my %poisoners = ( # (I don't see how they did it, though!) "alwayshotels.com" => 1, # Poisoned Lycos pretty heavily. "nextag.com" => 1, # Poisoned Alta Vista real good. + "ghettodriveby.com" => 1, # Poisoned Google Images. + "crosswordsolver.org" => 1, # Poisoned Google Images. + "xona.com" => 1, # Poisoned Google Images. + "freepatentsonline.com" => 1, # Poisoned Google Images. + "herbdatanz.com" => 1, # Poisoned Google Images. ); @@ -199,6 +211,10 @@ my %warningless_sites = ( "pics.livejournal.com" => 1, "tinypic.com" => 1, "flickr.com" => 1, + "pbase.com" => 1, + "blogger.com" => 1, + "multiply.com" => 1, + "wikimedia.org" => 1, "yimg.com" => 1, # This is where dailynews.yahoo.com stores "eimg.com" => 1, # its images, so pick_from_yahoo_news_text() @@ -396,10 +412,17 @@ sub get_document_1($$$) { if ($url =~ m@^http://www\.altavista\.com/@ || $url =~ m@^http://random\.yahoo\.com/@ || - $url =~ m@^http://images\.google\.com/@) { + $url =~ m@^http://images\.google\.com/@ || + $url =~ m@^http://www\.google\.com/@) { # block this, you turkeys. - $user_agent = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.5)" . - " Gecko/20041111 Firefox/1.0"; + $user_agent = "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.7)" . + " Gecko/20070914 Firefox/2.0.0.7"; + + # 28-Jun-2007: Google Images now emits the entire page in JS if + # you claim to be Gecko. They also still block "webcollage". + # They serve non-JS for unrecognised agents, so let's try this... + $user_agent = "NoJavascriptPlease/1.0" + if ($url =~ m@^http://[a-z]+\.google\.com/@); } my $hdrs = "GET " . ($http_proxy ? $url : "/$path") . " HTTP/1.0\r\n" . @@ -848,7 +871,10 @@ sub pick_image_from_rss($$) { sub pick_dictionary() { my @dicts = ("/usr/dict/words", "/usr/share/dict/words", - "/usr/share/lib/dict/words"); + "/usr/share/lib/dict/words", + "/usr/share/dict/cracklib-small", + "/usr/share/dict/cracklib-words" + ); foreach my $f (@dicts) { if (-f $f) { $wordlist = $f; @@ -1183,6 +1209,7 @@ sub pick_from_yahoo_random_link($) { ############################################################################ # # Pick images from random pages returned by the Alta Vista Random Link +# Note: this seems to have gotten a *lot* less random lately (2007). # ############################################################################ @@ -1281,6 +1308,55 @@ sub pick_from_alta_vista_images($) { } + +############################################################################ +# +# Pick images from Aptix security cameras +# Cribbed liberally from google image search code. +# By Jason Sullivan +# +############################################################################ + +my $aptix_images_url = ("http://www.google.com/search" . + "?q=inurl:%22jpg/image.jpg%3Fr%3D%22"); + +# securitycam +sub pick_from_security_camera($) { + my ($timeout) = @_; + + my $page = (int(rand(9)) + 1); + my $num = 20; # 20 images per page + my $search_url = $aptix_images_url; + + if ($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, ''); + + my @candidates = (); + my %referers; + foreach my $u (@subpages) { + next if ($u =~ m@[/.]google\.com\b@i); # skip google builtins (most links) + next unless ($u =~ m@jpg/image.jpg\?r=@i); # All pics contain this + + LOG ($verbose_filter, " candidate: $u"); + push @candidates, $u; + $referers{$u} = $u; + } + + @candidates = depoison (@candidates); + return () if ($#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, $img); +} + ############################################################################ # @@ -1820,7 +1896,7 @@ sub pick_from_flickr_recent($) { $page = html_unquote ($page); $thumb = html_unquote ($thumb); - next unless ($thumb =~ m@^http://photos\d*\.flickr\.com/@); + next unless ($thumb =~ m@^http://farm\d*\.static\.flickr\.com/@); my $base = "http://www.flickr.com/"; $page =~ s@^/@$base@; @@ -2277,8 +2353,9 @@ sub report_performance() { my $suc = $stats_successes{$name} || 0; my $pct = int($suc * 100 / $try); my $secs = $stats_elapsed{$name}; - my $secs_link = int($secs / $try); - print STDERR sprintf ("$blurb %-12s %4s (%d/%d);\t %2d secs/link\n", + my $secs_link = $secs / $try; + print STDERR sprintf ("$blurb %-14s %4s (%d/%d);" . + " \t %.1f secs/link\n", "$name:", "$pct%", $suc, $try, $secs_link); } } @@ -2765,7 +2842,9 @@ sub x_or_pbm_output($) { # find a root-window displayer program. # - $ppm_to_root_window_cmd = pick_root_displayer(); + if (!$no_output_p) { + $ppm_to_root_window_cmd = pick_root_displayer(); + } if (defined ($window_id)) { error ("-window-id only works if xscreensaver-getimage is installed")