#!/usr/bin/perl -w
#
# webcollage, Copyright © 1999-2019 by Jamie Zawinski <jwz@jwz.org>
# This program decorates the screen with random images from the web.
# One satisfied customer described it as "a nonstop pop culture brainbath."
#
# Permission to use, copy, modify, distribute, and sell this software and its
# documentation for any purpose is hereby granted without fee, provided that
# the above copyright notice appear in all copies and that both that
# copyright notice and this permission notice appear in supporting
# documentation. No representations are made about the suitability of this
# software for any purpose. It is provided "as is" without express or
# implied warranty.
# To run this as a display mode with xscreensaver, add this to `programs':
#
# webcollage --root
# webcollage --root --filter 'vidwhacker --stdin --stdout'
#
#
# You can see this in action at https://www.jwz.org/webcollage/ --
# it auto-reloads about once a minute. To make a page similar to
# that on your own system, do this:
#
# webcollage --size '800x600' --imagemap $HOME/www/webcollage/index
#
#
# Requires that either the provided "webcollage-helper" program or
# ImageMagick's "convert" be available on $PATH.
#
#
# If you have the "driftnet" program installed, webcollage can display a
# collage of images sniffed off your local ethernet, instead of pulled out
# of search engines: in that way, your screensaver can display the images
# that your co-workers are downloading!
#
# Driftnet is available here: http://www.ex-parrot.com/~chris/driftnet/
# Use it like so:
#
# webcollage --root --driftnet
#
# Driftnet is the Unix implementation of the MacOS "EtherPEG" program.
require 5;
use strict;
# We can't "use diagnostics" here, because that library malfunctions if
# you signal and catch alarms: it says "Uncaught exception from user code"
# and exits, even though I damned well AM catching it!
#use diagnostics;
require Time::Local;
require POSIX;
use Fcntl ':flock'; # import LOCK_* constants
use POSIX qw(strftime);
use LWP::UserAgent;
my $progname = $0; $progname =~ s@.*/@@g;
my ($version) = ('$Revision: 1.183 $' =~ m/\s(\d[.\d]+)\s/s);
my $copyright = "WebCollage $version, Copyright (c) 1999-2017" .
" Jamie Zawinski <jwz\@jwz.org>\n" .
" https://www.jwz.org/webcollage/\n";
my @search_methods = (
# Google is rate-limiting us now, so this works ok from
# a short-running screen saver, but not as a batch job.
# I haven't found a workaround.
#
5, "googlephotos", \&pick_from_google_image_photos,
3, "googleimgs", \&pick_from_google_images,
3, "googlenums", \&pick_from_google_image_numbers,
# So let's try Bing instead. No rate limiting yet!
#
13, "bingphotos", \&pick_from_bing_image_photos,
11, "bingimgs", \&pick_from_bing_images,
10, "bingnums", \&pick_from_bing_image_numbers,
20, "flickr_recent", \&pick_from_flickr_recent,
15, "flickr_random", \&pick_from_flickr_random,
6, "livejournal", \&pick_from_livejournal_images,
11, "imgur", \&pick_from_imgur,
# Tumblr doesn't have an "or" search, so this isn't great.
3, "tumblr", \&pick_from_tumblr,
# I ran out of usable access tokens, May 2017
# 0, "instagram", \&pick_from_instagram,
# No longer exists, as of Apr 2014
# 0, "yahoorand", \&pick_from_yahoo_random_link,
# Twitter destroyed their whole API in 2013.
# 0, "twitpic", \&pick_from_twitpic_images,
# 0, "twitter", \&pick_from_twitter_images,
# 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,
# Nonfunctional as of June 2011.
# 0, "altavista", \&pick_from_alta_vista_random_link,
# In Apr 2002, Google asked me to stop searching them.
# I asked them to add a "random link" url. They said
# "that would be easy, we'll think about it" and then
# never wrote back. Booo Google! Booooo! So, screw
# those turkeys, I've turned Google searching back on.
# I'm sure they can take it. (Jan 2005.)
# Jan 2005: Yahoo fucked up their search form so that
# it's no longer possible to do "or" searches on news
# images, so we rarely get any hits there any more.
#
# 0, "yahoonews", \&pick_from_yahoo_news_text,
# Dec 2004: the ircimages guy's server can't take the
# heat, so he started banning the webcollage user agent.
# I tried to convince him to add a lighter-weight page to
# support webcollage better, but he doesn't care.
#
# 0, "ircimages", \&pick_from_ircimages,
# Dec 2002: Alta Vista has a new "random link" URL now.
# They added it specifically to better support webcollage!
# That was super cool of them. This is how we used to do
# it, before:
#
# 0, "avimages", \&pick_from_alta_vista_images,
# 0, "avtext", \&pick_from_alta_vista_text,
# This broke in 2004. Eh, Lycos sucks anyway.
#
# 0, "lycos", \&pick_from_lycos_text,
# This broke in 2003, I think. I suspect Hotbot is
# actually the same search engine data as Lycos.
#
# 0, "hotbot", \&pick_from_hotbot_text,
);
# programs we can use to write to the root window (tried in ascending order.)
#
my @root_displayers = (
"xscreensaver-getimage -root -file",
"chbg -once -xscreensaver -max_size 100",
"xv -root -quit -viewonly +noresetroot -quick24 -rmode 5" .
" -rfg black -rbg black",
"xli -quiet -onroot -center -border black",
"xloadimage -quiet -onroot -center -border black",
# this lame program wasn't built with vroot.h:
# "xsri -scale -keep-aspect -center-horizontal -center-vertical",
);
# Some sites need cookies to work properly. These are they.
#
my %cookies = (
"www.altavista.com" => "AV_ALL=1", # request uncensored searches
"web.altavista.com" => "AV_ALL=1",
"ircimages.com" => 'disclaimer=1',
);
# If this is set, it's a helper program to use for pasting images together:
# this is somewhat faster than using ImageMagick.
#
my $webcollage_helper = undef;
my $convert_cmd = 'convert';
my $opacity = 0.85; # Opacity when pasting images together.
# Some sites have managed to poison the search engines. These are they.
# (We auto-detect sites that have poisoned the search engines via excessive
# keywords or dictionary words, but these are ones that slip through
# anyway.)
#
# This can contain full host names, or 2 or 3 component domains.
#
my %poisoners = (
"die.net" => 1, # 'l33t h4ck3r d00dz.
"genforum.genealogy.com" => 1, # Cluttering avtext with human names.
"rootsweb.com" => 1, # Cluttering avtext with human names.
"akamai.net" => 1, # Lots of sites have their images on Akamai.
"akamaitech.net" => 1, # But those are pretty much all banners.
# Since Akamai is super-expensive, let's
# go out on a limb and assume that all of
# their customers are rich-and-boring.
"bartleby.com" => 1, # Dictionary, cluttering avtext.
"encyclopedia.com" => 1, # Dictionary, cluttering avtext.
"onlinedictionary.datasegment.com" => 1, # Dictionary, cluttering avtext.
"hotlinkpics.com" => 1, # Porn site that has poisoned avimages
# (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.
);
# When verbosity is turned on, we warn about sites that we seem to be hitting
# a lot: usually this means some new poisoner has made it into the search
# engines. But sometimes, the warning is just because that site has a lot
# of stuff on it. So these are the sites that are immune to the "frequent
# site" diagnostic message.
#
my %warningless_sites = (
"home.earthlink.net" => 1,
"www.angelfire.com" => 1,
"members.aol.com" => 1,
"img.photobucket.com" => 1,
"pics.livejournal.com" => 1,
"tinypic.com" => 1,
"flickr.com" => 1,
"staticflickr.com" => 1,
"live.staticflickr.com" => 1,
"pbase.com" => 1,
"blogger.com" => 1,
"multiply.com" => 1,
"wikimedia.org" => 1,
"twitpic.com" => 1,
"amazonaws.com" => 1,
"blogspot.com" => 1,
"photoshelter.com" => 1,
"myspacecdn.com" => 1,
"feedburner.com" => 1,
"wikia.com" => 1,
"ljplus.ru" => 1,
"yandex.ru" => 1,
"imgur.com" => 1,
"tumblr.com" => 1,
"yfrog.com" => 1,
"cdninstagram.com" => 1,
"gstatic.com" => 1,
"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.
"images.quizfarm.com" => 1, # damn those LJ quizzes...
"images.quizilla.com" => 1,
"images.quizdiva.net" => 1,
"driftnet" => 1, # builtin...
"local-directory" => 1, # builtin...
);
# For decoding HTML-encoded character entities to URLs.
# In This Modern World, probably we should use HTML::Entities instead.
#
my %entity_table = (
"apos" => '\'',
"quot" => '"', "amp" => '&', "lt" => '<',
"gt" => '>', "nbsp" => ' ', "iexcl" => '',
"cent" => "\xA2", "pound" => "\xA3", "curren" => "\xA4",
"yen" => "\xA5", "brvbar" => "\xA6", "sect" => "\xA7",
"uml" => "\xA8", "copy" => "\xA9", "ordf" => "\xAA",
"laquo" => "\xAB", "not" => "\xAC", "shy" => "\xAD",
"reg" => "\xAE", "macr" => "\xAF", "deg" => "\xB0",
"plusmn" => "\xB1", "sup2" => "\xB2", "sup3" => "\xB3",
"acute" => "\xB4", "micro" => "\xB5", "para" => "\xB6",
"middot" => "\xB7", "cedil" => "\xB8", "sup1" => "\xB9",
"ordm" => "\xBA", "raquo" => "\xBB", "frac14" => "\xBC",
"frac12" => "\xBD", "frac34" => "\xBE", "iquest" => "\xBF",
"Agrave" => "\xC0", "Aacute" => "\xC1", "Acirc" => "\xC2",
"Atilde" => "\xC3", "Auml" => "\xC4", "Aring" => "\xC5",
"AElig" => "\xC6", "Ccedil" => "\xC7", "Egrave" => "\xC8",
"Eacute" => "\xC9", "Ecirc" => "\xCA", "Euml" => "\xCB",
"Igrave" => "\xCC", "Iacute" => "\xCD", "Icirc" => "\xCE",
"Iuml" => "\xCF", "ETH" => "\xD0", "Ntilde" => "\xD1",
"Ograve" => "\xD2", "Oacute" => "\xD3", "Ocirc" => "\xD4",
"Otilde" => "\xD5", "Ouml" => "\xD6", "times" => "\xD7",
"Oslash" => "\xD8", "Ugrave" => "\xD9", "Uacute" => "\xDA",
"Ucirc" => "\xDB", "Uuml" => "\xDC", "Yacute" => "\xDD",
"THORN" => "\xDE", "szlig" => "\xDF", "agrave" => "\xE0",
"aacute" => "\xE1", "acirc" => "\xE2", "atilde" => "\xE3",
"auml" => "\xE4", "aring" => "\xE5", "aelig" => "\xE6",
"ccedil" => "\xE7", "egrave" => "\xE8", "eacute" => "\xE9",
"ecirc" => "\xEA", "euml" => "\xEB", "igrave" => "\xEC",
"iacute" => "\xED", "icirc" => "\xEE", "iuml" => "\xEF",
"eth" => "\xF0", "ntilde" => "\xF1", "ograve" => "\xF2",
"oacute" => "\xF3", "ocirc" => "\xF4", "otilde" => "\xF5",
"ouml" => "\xF6", "divide" => "\xF7", "oslash" => "\xF8",
"ugrave" => "\xF9", "uacute" => "\xFA", "ucirc" => "\xFB",
"uuml" => "\xFC", "yacute" => "\xFD", "thorn" => "\xFE",
"yuml" => "\xFF",
# HTML 4 entities that do not have 1:1 Latin1 mappings.
"bull" => "*", "hellip"=> "...", "prime" => "'", "Prime" => "\"",
"frasl" => "/", "trade" => "[tm]", "larr" => "<-", "rarr" => "->",
"harr" => "<->", "lArr" => "<=", "rArr" => "=>", "hArr" => "<=>",
"empty" => "\xD8", "minus" => "-", "lowast"=> "*", "sim" => "~",
"cong" => "=~", "asymp" => "~", "ne" => "!=", "equiv" => "==",
"le" => "<=", "ge" => ">=", "lang" => "<", "rang" => ">",
"loz" => "<>", "OElig" => "OE", "oelig" => "oe", "Yuml" => "Y",
"circ" => "^", "tilde" => "~", "ensp" => " ", "emsp" => " ",
"thinsp"=> " ", "ndash" => "-", "mdash" => "--", "lsquo" => "`",
"rsquo" => "'", "sbquo" => "'", "ldquo" => "\"", "rdquo" => "\"",
"bdquo" => "\"", "lsaquo"=> "<", "rsaquo"=> ">",
);
##############################################################################
#
# Various global flags set by command line parameters, or computed
#
##############################################################################
my $current_state = "???"; # for diagnostics
my $load_method;
my $last_search;
my $image_succeeded = -1;
my $suppress_audit = 0;
my $verbose_imgmap = 0; # print out rectangles and URLs only (stdout)
my $verbose_warnings = 0; # print out warnings when things go wrong
my $verbose_load = 0; # diagnostics about loading of URLs
my $verbose_filter = 0; # diagnostics about page selection/rejection
my $verbose_net = 0; # diagnostics about network I/O
my $verbose_decode = 0; # diagnostics about img conversion pipelines
my $verbose_http = 0; # diagnostics about all HTTP activity
my $verbose_exec = 0; # diagnostics about executing programs
my $report_performance_interval = 60 * 15; # print some stats every 15 minutes
my $http_proxy = undef;
my $http_timeout = 20;
my $cvt_timeout = 10;
my $min_width = 50;
my $min_height = 50;
my $min_ratio = 1/5;
my $min_gif_area = (120 * 120);
my $no_output_p = 0;
my $urls_only_p = 0;
my $cocoa_p = 0;
my $imagemap_base = undef;
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 $local_magic = 'local-directory';
my $local_dir = undef;
my $wordlist;
my %rejected_urls;
my @tripwire_words = ("aberrate", "abode", "amorphous", "antioch",
"arrhenius", "arteriole", "blanket", "brainchild",
"burdensome", "carnival", "cherub", "chord", "clever",
"dedicate", "dilogarithm", "dolan", "dryden",
"eggplant");
##############################################################################
#
# Retrieving URLs
#
##############################################################################
# returns three values: the HTTP response line; the document headers;
# and the document body.
#
sub get_document_1($$$) {
my ($url, $referer, $timeout) = @_;
if (!defined($timeout)) { $timeout = $http_timeout; }
if ($timeout > $http_timeout) { $timeout = $http_timeout; }
my $user_agent = "$progname/$version";
if ($url =~ m@^https?://www\.altavista\.com/@s ||
$url =~ m@^https?://random\.yahoo\.com/@s ||
$url =~ m@^https?://[^./]+\.google\.com/@s ||
$url =~ m@^https?://www\.livejournal\.com/@s) {
# block this, you turkeys.
$user_agent = 'Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.7)' .
' Gecko/20070914 Firefox/2.0.0.7';
}
my $ua = LWP::UserAgent->new ( agent => $user_agent,
keep_alive => 0,
env_proxy => 0,
);
$ua->proxy ('http', $http_proxy) if $http_proxy;
$ua->default_header ('Referer' => $referer) if $referer;
$ua->default_header ('Accept' => '*/*');
$ua->timeout($timeout) if $timeout;
if (0) {
$ua->add_handler ("request_send",
sub($$$) {
my ($req, $ua, $h) = @_;
print "\n>>[[\n"; $req->dump; print "\n]]\n";
return;
});
$ua->add_handler ("response_data",
sub($$$$) {
my ($req, $ua, $h, $data) = @_;
#print "\n<<[[\n"; print $data; print "\n]]\n";
return 1;
});
$ua->add_handler ("request_done",
sub($$$) {
my ($req, $ua, $h) = @_;
print "\n<<[[\n"; $req->dump; print "\n]]\n";
return;
});
}
if ($verbose_http) {
LOG (1, " ==> GET $url");
LOG (1, " ==> User-Agent: $user_agent");
LOG (1, " ==> Referer: $referer") if $referer;
}
my $res = $ua->get ($url);
my $http = ($res ? $res->status_line : '') || '';
my $head = ($res ? $res->headers() : '') || '';
$head = $head->as_string() if $head;
my $body = ($res && $res->is_success ? $res->decoded_content : '') || '';
LOG ($verbose_net, "get_document_1 $url " . ($referer ? $referer : ""));
$head =~ s/\r\n/\n/gs;
$head =~ s/\r/\n/gs;
if ($verbose_http) {
foreach (split (/\n/, $head)) {
LOG ($verbose_http, " <== $_");
}
}
my @L = split(/\r\n|\r|\n/, $body);
my $lines = @L;
LOG ($verbose_http,
" <== [ body ]: $lines lines, " . length($body) . " bytes");
if (!$http) {
LOG (($verbose_net || $verbose_load), "null response: $url");
return ();
}
return ( $http, $head, $body );
}
# returns two values: the document headers; and the document body.
# if the given URL did a redirect, returns the redirected-to document.
#
sub get_document($$;$) {
my ($url, $referer, $timeout) = @_;
my $start = time;
if (defined($referer) && $referer eq $driftnet_magic) {
return get_driftnet_file ($url);
}
if (defined($referer) && $referer eq $local_magic) {
return get_local_file ($url);
}
my $orig_url = $url;
my $loop_count = 0;
my $max_loop_count = 4;
do {
if (defined($timeout) && $timeout <= 0) {
LOG (($verbose_net || $verbose_load), "timed out for $url");
$suppress_audit = 1;
return ();
}
my ( $http, $head, $body ) = get_document_1 ($url, $referer, $timeout);
if (defined ($timeout)) {
my $now = time;
my $elapsed = $now - $start;
$timeout -= $elapsed;
$start = $now;
}
return () unless $http; # error message already printed
$http =~ s/[\r\n]+$//s;
if ( $http =~ m@^HTTP/[0-9.]+ 30[123]@ ) {
$_ = $head;
my ( $location ) = m@^location:[ \t]*(.*)$@im;
if ( $location ) {
$location =~ s/[\r\n]$//;
LOG ($verbose_net, "redirect from $url to $location");
$referer = $url;
$url = $location;
if ($url =~ m@^/@) {
$referer =~ m@^(https?://[^/]+)@i;
$url = $1 . $url;
} elsif (! ($url =~ m@^[a-z]+:@i)) {
$_ = $referer;
s@[^/]+$@@g if m@^https?://[^/]+/@i;
$_ .= "/" if m@^https?://[^/]+$@i;
$url = $_ . $url;
}
} else {
LOG ($verbose_net, "no Location with \"$http\"");
return ( $url, $body );
}
if ($loop_count++ > $max_loop_count) {
LOG ($verbose_net,
"too many redirects ($max_loop_count) from $orig_url");
$body = undef;
return ();
}
} elsif ( $http =~ m@^HTTP/[0-9.]+ ([4-9][0-9][0-9].*)$@ ) {
LOG (($verbose_net || $verbose_load), "failed: $1 ($url)");
# http errors -- return nothing.
$body = undef;
return ();
} elsif (!$body) {
LOG (($verbose_net || $verbose_load), "document contains no data: $url");
return ();
} else {
# ok!
return ( $url, $body );
}
} while (1);
}
# If we already have a cookie defined for this site, and the site is trying
# to overwrite that very same cookie, let it do so. This is because nytimes
# expires its cookies - it lets you upgrade to a new cookie without logging
# in again, but you have to present the old cookie to get the new cookie.
# So, by doing this, the built-in cypherpunks cookie will never go "stale".
#
sub set_cookie($$) {
my ($host, $cookie) = @_;
my $oc = $cookies{$host};
return unless $oc;
$_ = $oc;
my ($oc_name, $oc_value) = m@^([^= \t\r\n]+)=(.*)$@;
$_ = $cookie;
my ($nc_name, $nc_value) = m@^([^= \t\r\n]+)=(.*)$@;
if ($oc_name eq $nc_name &&
$oc_value ne $nc_value) {
$cookies{$host} = $cookie;
LOG ($verbose_net, "overwrote ${host}'s $oc_name cookie");
}
}
############################################################################
#
# Extracting image URLs from HTML
#
############################################################################
# given a URL and the body text at that URL, selects and returns a random
# image from it. returns () if no suitable images found.
#
sub pick_image_from_body($$) {
my ($url, $body) = @_;
my $base = $url;
$_ = $url;
# if there's at least one slash after the host, take off the last
# pathname component
if ( m@^https?://[^/]+/@io ) {
$base =~ s@[^/]+$@@go;
}
# if there are no slashes after the host at all, put one on the end.
if ( m@^https?://[^/]+$@io ) {
$base .= "/";
}
$_ = $body;
# strip out newlines, compress whitespace
s/[\r\n\t ]+/ /go;
# nuke comments
s/<!--.*?-->//go;
# There are certain web sites that list huge numbers of dictionary
# words in their bodies or in their <META NAME=KEYWORDS> tags (surprise!
# Porn sites tend not to be reputable!)
#
# I do not want webcollage to filter on content: I want it to select
# randomly from the set of images on the web. All the logic here for
# rejecting some images is really a set of heuristics for rejecting
# images that are not really images: for rejecting *text* that is in
# GIF/JPEG/PNG form. I don't want text, I want pictures, and I want
# the content of the pictures to be randomly selected from among all
# the available content.
#
# So, filtering out "dirty" pictures by looking for "dirty" keywords
# would be wrong: dirty pictures exist, like it or not, so webcollage
# should be able to select them.
#
# However, picking a random URL is a hard thing to do. The mechanism I'm
# using is to search for a selection of random words. This is not
# perfect, but works ok most of the time. The way it breaks down is when
# some URLs get precedence because their pages list *every word* as
# related -- those URLs come up more often than others.
#
# So, after we've retrieved a URL, if it has too many keywords, reject
# it. We reject it not on the basis of what those keywords are, but on
# the basis that by having so many, the page has gotten an unfair
# advantage against our randomizer.
#
my $trip_count = 0;
foreach my $trip (@tripwire_words) {
$trip_count++ if m/$trip/i;
}
if ($trip_count >= $#tripwire_words - 2) {
LOG (($verbose_filter || $verbose_load),
"there is probably a dictionary in \"$url\": rejecting.");
$rejected_urls{$url} = -1;
$body = undef;
$_ = undef;
return ();
}
my @urls;
my %unique_urls;
foreach (split(/ *</)) {
if ( m/^meta.*["']keywords["']/i ) {
# Likewise, reject any web pages that have a KEYWORDS meta tag
# that is too long.
#
my $L = length($_);
if ($L > 1000) {
LOG (($verbose_filter || $verbose_load),
"excessive keywords ($L bytes) in $url: rejecting.");
$rejected_urls{$url} = $L;
$body = undef;
$_ = undef;
return ();
} else {
LOG ($verbose_filter, " keywords ($L bytes) in $url (ok)");
}
} elsif (m/^ (IMG|A) \b .* (SRC|HREF) \s* = \s* ["']? (.*?) [ "'<>] /six ||
m/^ (LINK|META) \b .* (REL|PROPERTY) \s* = \s*
["']? (image_src|og:image) ["']? /six) {
my $was_inline = (lc($1) eq 'img');
my $was_meta = (lc($1) eq 'link' || lc($1) eq 'meta');
my $link = $3;
# For <link rel="image_src" href="...">
# and <meta property="og:image" content="...">
#
if ($was_meta) {
next unless (m/ (HREF|CONTENT) \s* = \s* ["']? (.*?) [ "'<>] /six);
$link = $2;
}
my ( $width ) = m/width ?=[ \"]*(\d+)/oi;
my ( $height ) = m/height ?=[ \"]*(\d+)/oi;
$_ = $link;
if ( m@^/@o ) {
my $site;
( $site = $base ) =~ s@^(https?://[^/]*).*@$1@gio;
$_ = "$site$link";
} elsif ( ! m@^[^/:?]+:@ ) {
$_ = "$base$link";
s@/\./@/@g;
1 while (s@/[^/]+/\.\./@/@g);
}
# skip non-http
if ( ! m@^https?://@io ) {
next;
}
# skip non-image
if ( ! m@[.](gif|jpg|jpeg|pjpg|pjpeg|png)$@io ) {
next;
}
# skip really short or really narrow images
if ( $width && $width < $min_width) {
if (!$height) { $height = "?"; }
LOG ($verbose_filter, " skip narrow image $_ (${width}x$height)");
next;
}
if ( $height && $height < $min_height) {
if (!$width) { $width = "?"; }
LOG ($verbose_filter, " skip short image $_ (${width}x$height)");
next;
}
# skip images with ratios that make them look like banners.
if ($min_ratio && $width && $height &&
($width * $min_ratio ) > $height) {
if (!$height) { $height = "?"; }
LOG ($verbose_filter, " skip bad ratio $_ (${width}x$height)");
next;
}
# skip GIFs with a small number of pixels -- those usually suck.
if ($width && $height &&
m/\.gif$/io &&
($width * $height) < $min_gif_area) {
LOG ($verbose_filter, " skip small GIF $_ (${width}x$height)");
next;
}
# skip images with a URL that indicates a Yahoo thumbnail.
if (m@\.yimg\.com/.*/t/@) {
if (!$width) { $width = "?"; }
if (!$height) { $height = "?"; }
LOG ($verbose_filter, " skip yahoo thumb $_ (${width}x$height)");
next;
}
my $url = $_;
if ($unique_urls{$url}) {
LOG ($verbose_filter, " skip duplicate image $_");
next;
}
LOG ($verbose_filter,
" image $url" .
($width && $height ? " (${width}x${height})" : "") .
($was_meta ? " (meta)" : $was_inline ? " (inline)" : ""));
my $weight = 1;
if ($was_meta) {
$weight = 20; # meta tag images are far preferable to inline images.
} else {
if ($url !~ m@[.](gif|png)$@io ) {
$weight += 2; # JPEGs are preferable to GIFs and PNGs.
}
if (! $was_inline) {
$weight += 4; # pointers to images are preferable to inlined images.
}
}
$unique_urls{$url}++;
for (my $i = 0; $i < $weight; $i++) {
$urls[++$#urls] = $url;
}
}
}
my $fsp = ($body =~ m@<frameset@i);
$_ = undef;
$body = undef;
@urls = depoison (@urls);
if ( $#urls < 0 ) {
LOG ($verbose_load, "no images on $base" . ($fsp ? " (frameset)" : ""));
return ();
}
# pick a random element of the table
my $i = int(rand($#urls+1));
$url = $urls[$i];
LOG ($verbose_load, "picked image " .($i+1) . "/" . ($#urls+1) . ": $url");
return $url;
}
# Given a URL and the RSS feed from that URL, pick a random image from
# the feed. This is a lot simpler than extracting images out of a page:
# we already know we have reasonable images, so we just pick one.
# Returns: the real URL of the page (preferably not the RSS version),
# and the image.
sub pick_image_from_rss($$) {
my ($url, $body) = @_;
my ($base) = ($body =~ m@<link>([^<>]+)</link>@si); # root link
my @items = ($body =~ m@<item\b[^<>]*>(.*?)</item>@gsi);
return unless @items;
my $n = @items;
my $i = int(rand($n));
my $item = $items[$i];
$base = $1 if ($item =~ m@<link>([^<>]+)</link>@si); # item link
$base = $url unless $base;
($url) = ($item =~ m/<enclosure\b[^<>]*\burl="(.*?)"/si);
return unless $url;
LOG ($verbose_load, "picked image $i/$n: $url");
return ($base, $url);
}
############################################################################
#
# Subroutines for getting pages and images out of search engines
#
############################################################################
sub pick_dictionary() {
my @dicts = ("/usr/dict/words",
"/usr/share/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 = $f;
last;
}
}
error ("$dicts[0] does not exist") unless defined($wordlist);
}
# returns a random word from the dictionary
#
sub random_word() {
return undef unless open (my $in, '<', $wordlist);
my $size = (stat($in))[7];
my $word = undef;
my $count = 0;
while (1) {
error ("looping ($count) while reading $wordlist")
if (++$count > 100);
my $pos = int (rand ($size));
if (seek ($in, $pos, 0)) {
$word = <$in>; # toss partial line
$word = <$in>; # keep next line
}
next unless ($word);
next if ($word =~ m/^[-\']/);
$word = lc($word);
$word =~ s/^.*-//s;
$word =~ s/^[^a-z]+//s;
$word =~ s/[^a-z]+$//s;
$word =~ s/\'s$//s;
$word =~ s/ys$/y/s;
$word =~ s/ally$//s;
$word =~ s/ly$//s;
$word =~ s/ies$/y/s;
$word =~ s/ally$/al/s;
$word =~ s/izes$/ize/s;
$word =~ s/esses$/ess/s;
$word =~ s/(.{5})ing$/$1/s;
next if (length ($word) > 14);
last if ($word);
}
close ($in);
if ( $word =~ s/\s/\+/gs ) { # convert intra-word spaces to "+".
$word = "\%22$word\%22"; # And put quotes (%22) around it.
}
return $word;
}
sub random_words($) {
my ($sep) = @_;
return (random_word() . $sep .
random_word() . $sep .
random_word() . $sep .
random_word() . $sep .
random_word());
}
sub url_quote($) {
my ($s) = @_;
$s =~ s|([^-a-zA-Z0-9.\@/_\r\n])|sprintf("%%%02X", ord($1))|ge;
return $s;
}
sub url_unquote($) {
my ($s) = @_;
$s =~ s/[+]/ /g;
$s =~ s/%([a-z0-9]{2})/chr(hex($1))/ige;
return $s;
}
sub html_quote($) {
my ($s) = @_;
$s =~ s/&/&/gi;
$s =~ s/</</gi;
$s =~ s/>/>/gi;
$s =~ s/\"/"/gi;
return $s;
}
sub html_unquote($) {
my ($s) = @_;
$s =~ s/(&([a-z]+);)/{ $entity_table{$2} || $1; }/gexi; # e.g., '
$s =~ s/(&\#(\d+);)/{ chr($2) }/gexi; # e.g., '
return $s;
}
# Loads the given URL (a search on some search engine) and returns:
# - the total number of hits the search engine claimed it had;
# - a list of URLs from the page that the search engine returned;
# Note that this list contains all kinds of internal search engine
# junk URLs too -- caller must prune them.
#
sub pick_from_search_engine($$$) {
my ( $timeout, $search_url, $words ) = @_;
$_ = $words;
s/%20/ /g;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "words: $_");
LOG ($verbose_load, "URL: $search_url");
$last_search = $search_url; # for warnings
my $start = time;
my ( $base, $body ) = get_document ($search_url, undef, $timeout);
if (defined ($timeout)) {
$timeout -= (time - $start);
if ($timeout <= 0) {
$body = undef;
LOG (($verbose_net || $verbose_load),
"timed out (late) for $search_url");
$suppress_audit = 1;
return ();
}
}
return () if (! $body);
my @subpages;
my $search_count = "?";
if ($body =~ m@found (approximately |about )?(<B>)?(\d+)(</B>)? image@) {
$search_count = $3;
} elsif ($body =~ m@<NOBR>((\d{1,3})(,\d{3})*) @i) {
$search_count = $1;
} elsif ($body =~ m@found ((\d{1,3})(,\d{3})*|\d+) Web p@) {
$search_count = $1;
} elsif ($body =~ m@found about ((\d{1,3})(,\d{3})*|\d+) results@) {
$search_count = $1;
} elsif ($body =~ m@\b\d+ - \d+ of (\d+)\b@i) { # avimages
$search_count = $1;
} elsif ($body =~ m@About ((\d{1,3})(,\d{3})*) images@i) { # avimages
$search_count = $1;
} elsif ($body =~ m@We found ((\d{1,3})(,\d{3})*|\d+) results@i) { # *vista
$search_count = $1;
} elsif ($body =~ m@ of about <B>((\d{1,3})(,\d{3})*)<@i) { # googleimages
$search_count = $1;
} elsif ($body =~ m@<B>((\d{1,3})(,\d{3})*)</B> Web sites were found@i) {
$search_count = $1; # lycos
} elsif ($body =~ m@WEB.*?RESULTS.*?\b((\d{1,3})(,\d{3})*)\b.*?Matches@i) {
$search_count = $1; # hotbot
} elsif ($body =~ m@no photos were found containing@i) { # avimages
$search_count = "0";
} elsif ($body =~ m@found no document matching@i) { # avtext
$search_count = "0";
}
1 while ($search_count =~ s/^(\d+)(\d{3})/$1,$2/);
# if ($search_count eq "?" || $search_count eq "0") {
# my $file = "/tmp/wc.html";
# open (my $out, '>', $file) || error ("writing $file: $!");
# print $out $body;
# close $out;
# print STDERR blurb() . "###### wrote $file\n";
# }
my $length = length($body);
my $href_count = 0;
$_ = $body;
s/[\r\n\t ]+/ /g;
s/(<A )/\n$1/gi;
foreach (split(/\n/)) {
$href_count++;
my ($u) = m@<A\s.*?\bHREF\s*=\s*([\"\'][^\"\'<>]+)@i;
next unless $u;
my ($u2) = m@<IMG\s.*\bSRC\s*=\s*[\"\']([^\"\'<>]+)@i;
if (m/\bm="\{(.*?)\}"/s) { # Bing info is inside JSON crud
my $json = html_unquote($1);
my ($href) = ($json =~ m/\b(?:surl|purl)\"?:\s*"(.*?)"/s);
my ($img) = ($json =~ m/\b(?:imgurl|murl)\"?:\s*"(.*?)"/s);
$u = "$img\t$href" if ($img && $href);
} elsif ($u2 && $u2 =~ m@://[^/]*\.gstatic\.com/@s) { $u = $u2;
$u =~ s/^\"|\"$//s;
} elsif ($u =~ m/^\"([^\"]*)\"/) { $u = $1 # quoted string
} elsif ($u =~ m/^([^\s]*)\s/) { $u = $1; # or token
}
if ( $rejected_urls{$u} ) {
LOG ($verbose_filter, " pre-rejecting candidate: $u");
next;
}
LOG ($verbose_http, " HREF: $u");
$subpages[++$#subpages] = $u;
}
if ( $#subpages < 0 ) {
LOG ($verbose_filter,
"found nothing on $base ($length bytes, $href_count links).");
return ();
}
LOG ($verbose_filter, "" . $#subpages+1 . " links on $search_url");
return ($search_count, @subpages);
}
sub depoison(@) {
my (@urls) = @_;
my @urls2 = ();
foreach (@urls) {
my ($h) = m@^https?://([^/: \t\r\n]+)@i;
next unless defined($h);
if ($poisoners{$h}) {
LOG (($verbose_filter), " rejecting poisoner: $_");
next;
}
if ($h =~ m@([^.]+\.[^.]+\.[^.]+)$@ &&
$poisoners{$1}) {
LOG (($verbose_filter), " rejecting poisoner: $_");
next;
}
if ($h =~ m@([^.]+\.[^.]+)$@ &&
$poisoners{$1}) {
LOG (($verbose_filter), " rejecting poisoner: $_");
next;
}
push @urls2, $_;
}
return @urls2;
}
# given a list of URLs, picks one at random; loads it; and returns a
# random image from it.
# returns the url of the page loaded; the url of the image chosen.
#
sub pick_image_from_pages($$$$@) {
my ($base, $total_hit_count, $unfiltered_link_count, $timeout, @pages) = @_;
$total_hit_count = "?" unless defined($total_hit_count);
@pages = depoison (@pages);
LOG ($verbose_load,
"" . ($#pages+1) . " candidates of $unfiltered_link_count links" .
" ($total_hit_count total)");
return () if ($#pages < 0);
my $i = int(rand($#pages+1));
my $page = $pages[$i];
LOG ($verbose_load, "picked page $page");
$suppress_audit = 1;
my ( $base2, $body2 ) = get_document ($page, $base, $timeout);
if (!$base2 || !$body2) {
$body2 = undef;
return ();
}
my $img = pick_image_from_body ($base2, $body2);
$body2 = undef;
if ($img) {
return ($base2, $img);
} else {
return ();
}
}
#############################################################################
##
## Pick images from random pages returned by the Yahoo Random Link
##
#############################################################################
#
## yahoorand
#my $yahoo_random_link = "http://random.yahoo.com/fast/ryl";
#
#
# Picks a random page; picks a random image on that page;
# returns two URLs: the page containing the image, and the image.
# Returns () if nothing found this time.
#
#sub pick_from_yahoo_random_link($) {
# my ($timeout) = @_;
#
# print STDERR "\n\n" if ($verbose_load);
# LOG ($verbose_load, "URL: $yahoo_random_link");
#
# $last_search = $yahoo_random_link; # for warnings
#
# $suppress_audit = 1;
#
# my ( $base, $body ) = get_document ($yahoo_random_link, undef, $timeout);
# if (!$base || !$body) {
# $body = undef;
# return;
# }
#
# LOG ($verbose_load, "redirected to: $base");
#
# my $img = pick_image_from_body ($base, $body);
# $body = undef;
#
# if ($img) {
# return ($base, $img);
# } else {
# return ();
# }
#}
############################################################################
#
# Pick images from random pages returned by the Alta Vista Random Link
# Note: this seems to have gotten a *lot* less random lately (2007).
#
############################################################################
# altavista
my $alta_vista_random_link = "http://www.altavista.com/image/randomlink";
# Picks a random page; picks a random image on that page;
# returns two URLs: the page containing the image, and the image.
# Returns () if nothing found this time.
#
sub pick_from_alta_vista_random_link($) {
my ($timeout) = @_;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "URL: $alta_vista_random_link");
$last_search = $alta_vista_random_link; # for warnings
$suppress_audit = 1;
my ( $base, $body ) = get_document ($alta_vista_random_link,
undef, $timeout);
if (!$base || !$body) {
$body = undef;
return;
}
LOG ($verbose_load, "redirected to: $base");
my $img = pick_image_from_body ($base, $body);
$body = undef;
if ($img) {
return ($base, $img);
} else {
return ();
}
}
############################################################################
#
# Pick images by feeding random words into Alta Vista Image Search
#
############################################################################
my $alta_vista_images_url = "http://www.altavista.com/image/results" .
"?ipht=1" . # photos
"&igrph=1" . # graphics
"&iclr=1" . # color
"&ibw=1" . # b&w
"&micat=1" . # no partner sites
"&sc=on" . # "site collapse"
"&q=";
# avimages
sub pick_from_alta_vista_images($) {
my ($timeout) = @_;
my $words = random_word();
my $page = (int(rand(9)) + 1);
my $search_url = $alta_vista_images_url . $words;
if ($page > 1) {
$search_url .= "&pgno=" . $page; # page number
$search_url .= "&stq=" . (($page-1) * 12); # first hit result on page
}
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
# avimages is encoding their URLs now.
next unless ($u =~ s/^.*\*\*(http%3a.*$)/$1/gsi);
$u = url_unquote($u);
next unless ($u =~ m@^https?://@i); # skip non-HTTP or relative URLs
next if ($u =~ m@[/.]altavista\.com\b@i); # skip altavista builtins
next if ($u =~ m@[/.]yahoo\.com\b@i); # yahoo and av in cahoots?
next if ($u =~ m@[/.]doubleclick\.net\b@i); # you cretins
next if ($u =~ m@[/.]clicktomarket\.com\b@i); # more cretins
next if ($u =~ m@[/.]viewimages\.com\b@i); # stacked deck
next if ($u =~ m@[/.]gettyimages\.com\b@i);
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
$timeout, @candidates);
}
############################################################################
#
# Pick images from Aptix security cameras
# Cribbed liberally from google image search code.
# By Jason Sullivan <jasonsul@us.ibm.com>
#
############################################################################
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);
}
############################################################################
#
# Pick images by feeding random words into Google Image Search.
# By Charles Gales <gales@us.ibm.com>
#
############################################################################
my $google_images_url = 'https://www.google.com/search' .
'?source=lnms&tbm=isch&tbs=isz:l&q=';
# googleimgs
sub pick_from_google_images($;$$) {
my ($timeout, $words, $max_page) = @_;
if (!defined($words)) {
$words = random_word(); # only one word for Google
}
my $off = int(rand(40));
my $search_url = $google_images_url . $words . "&start=" . $off;
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
$u = html_unquote($u);
# next if ($u =~ m@^https?://[^.]*\.(google|youtube)\.com/@s);
next unless ($u =~ m@^https?://[^/]*\.gstatic\.com@s);
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
@candidates = depoison (@candidates);
return () if ($#candidates < 0);
my $i = int(rand($#candidates+1));
my $img = $candidates[$i];
LOG ($verbose_load, "picked image " . ($i+1) . ": $img");
return ($img, $img);
}
############################################################################
#
# Pick images by feeding random numbers into Google Image Search.
# By jwz, suggested by Ian O'Donnell.
#
############################################################################
# googlenums
sub pick_from_google_image_numbers($) {
my ($timeout) = @_;
my $max = 9999;
my $number = int(rand($max));
$number = sprintf("%04d", $number)
if (rand() < 0.3);
pick_from_google_images ($timeout, "$number");
}
############################################################################
#
# Pick images by feeding random digital camera file names into
# Google Image Search.
# By jwz, inspired by the excellent Random Personal Picture Finder
# at http://www.diddly.com/random/
# May 2017: Commented out a bunch of formats that have fallen out of favor.
#
############################################################################
my @photomakers = (
#
# Common digital camera file name formats, as described at
# http://www.diddly.com/random/about.html
#
# sub { sprintf ("dcp%05d.jpg", int(rand(4000))); }, # Kodak
sub { sprintf ("dsc%05d.jpg", int(rand(4000))); }, # Nikon
sub { sprintf ("dscn%04d.jpg", int(rand(4000))); }, # Nikon
# sub { sprintf ("mvc-%03d.jpg", int(rand(999))); }, # Sony Mavica
# sub { sprintf ("mvc%05d.jpg", int(rand(9999))); }, # Sony Mavica
# sub { sprintf ("P101%04d.jpg", int(rand(9999))); }, # Olympus w/ date=101
# sub { sprintf ("P%x%02d%04d.jpg", # Olympus
# int(rand(0xC)), int(rand(30))+1,
# rand(9999)); },
sub { sprintf ("IMG_%03d.jpg", int(rand(999))); }, # ?
# sub { sprintf ("IMAG%04d.jpg", int(rand(9999))); }, # RCA and Samsung
# sub { my $n = int(rand(9999)); # Canon
# sprintf ("1%02d-%04d.jpg", int($n/100), $n); },
# sub { my $n = int(rand(9999)); # Canon
# sprintf ("1%02d-%04d_IMG.jpg",
# int($n/100), $n); },
sub { sprintf ("IMG_%04d.jpg", int(rand(9999))); }, # Canon
sub { sprintf ("dscf%04d.jpg", int(rand(9999))); }, # Fuji Finepix
# sub { sprintf ("pdrm%04d.jpg", int(rand(9999))); }, # Toshiba PDR
# sub { sprintf ("IM%06d.jpg", int(rand(9999))); }, # HP Photosmart
# sub { sprintf ("EX%06d.jpg", int(rand(9999))); }, # HP Photosmart
# sub { my $n = int(rand(3)); # Kodak DC-40,50,120
# sprintf ("DC%04d%s.jpg", int(rand(9999)),
# $n == 0 ? 'S' : $n == 1 ? 'M' : 'L'); },
sub { sprintf ("pict%04d.jpg", int(rand(9999))); }, # Minolta Dimage
# sub { sprintf ("P%07d.jpg", int(rand(9999))); }, # Kodak DC290
# sub { sprintf ("%02d%02d%04d.jpg", # Casio QV3000, QV4000
# int(rand(12))+1, int(rand(31))+1,
# int(rand(999))); },
# sub { sprintf ("%02d%x%02d%04d.jpg", # Casio QV7000
# int(rand(6)), # year
# int(rand(12))+1, int(rand(31))+1,
# int(rand(999))); },
sub { sprintf ("IMGP%04d.jpg", int(rand(9999))); }, # Pentax Optio S
# sub { sprintf ("PANA%04d.jpg", int(rand(9999))); }, # Panasonic vid still
sub { sprintf ("HPIM%04d.jpg", int(rand(9999))); }, # HP Photosmart
# sub { sprintf ("PCDV%04d.jpg", int(rand(9999))); }, # ?
);
# googlephotos
sub pick_from_google_image_photos($) {
my ($timeout) = @_;
my $i = int(rand($#photomakers + 1));
my $fn = $photomakers[$i];
my $file = &$fn;
#$file .= "%20filetype:jpg";
pick_from_google_images ($timeout, $file);
}
############################################################################
#
# Pick images by feeding random words into Google Image Search.
# By the way: fuck Microsoft.
#
############################################################################
my $bing_images_url = "http://www.bing.com/images/async?q=";
# bingimgs
sub pick_from_bing_images($;$$) {
my ($timeout, $words, $max_page) = @_;
if (!defined($words)) {
$words = random_word(); # only one word for Bing
}
my $off = int(rand(300));
my $search_url = $bing_images_url . $words . "&first=" . $off;
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
my %referers;
foreach my $u (@subpages) {
my ($img, $ref) = ($u =~ m/^(.*?)\t(.*)$/s);
next unless $img;
LOG ($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, $img);
}
############################################################################
#
# Pick images by feeding random numbers into Bing Image Search.
#
############################################################################
# bingnums
sub pick_from_bing_image_numbers($) {
my ($timeout) = @_;
my $max = 9999;
my $number = int(rand($max));
$number = sprintf("%04d", $number)
if (rand() < 0.3);
pick_from_bing_images ($timeout, "$number");
}
############################################################################
#
# Pick images by feeding random numbers into Bing Image Search.
#
############################################################################
# bingphotos
sub pick_from_bing_image_photos($) {
my ($timeout) = @_;
my $i = int(rand($#photomakers + 1));
my $fn = $photomakers[$i];
my $file = &$fn;
pick_from_bing_images ($timeout, $file);
}
############################################################################
#
# Pick images by feeding random words into Alta Vista Text Search
#
############################################################################
my $alta_vista_url = "http://www.altavista.com/web/results" .
"?pg=aq" .
"&aqmode=s" .
"&filetype=html" .
"&sc=on" . # "site collapse"
"&nbq=50" .
"&aqo=";
# avtext
sub pick_from_alta_vista_text($) {
my ($timeout) = @_;
my $words = random_words('%20');
my $page = (int(rand(9)) + 1);
my $search_url = $alta_vista_url . $words;
if ($page > 1) {
$search_url .= "&pgno=" . $page;
$search_url .= "&stq=" . (($page-1) * 10);
}
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
# Those altavista fuckers are playing really nasty redirection games
# these days: the filter your clicks through their site, but use
# onMouseOver to make it look like they're not! Well, it makes it
# easier for us to identify search results...
#
next unless ($u =~ s/^.*\*\*(http%3a.*$)/$1/gsi);
$u = url_unquote($u);
next unless ($u =~ m@^https?://@i); # skip non-HTTP or relative URLs
next if ($u =~ m@[/.]altavista\.com\b@i); # skip altavista builtins
next if ($u =~ m@[/.]yahoo\.com\b@i); # yahoo and av in cahoots?
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
$timeout, @candidates);
}
############################################################################
#
# Pick images by feeding random words into Hotbot
#
############################################################################
my $hotbot_search_url =("http://hotbot.lycos.com/default.asp" .
"?ca=w" .
"&descriptiontype=0" .
"&imagetoggle=1" .
"&matchmode=any" .
"&nummod=2" .
"&recordcount=50" .
"&sitegroup=1" .
"&stem=1" .
"&cobrand=undefined" .
"&query=");
sub pick_from_hotbot_text($) {
my ($timeout) = @_;
$last_search = $hotbot_search_url; # for warnings
# lycos seems to always give us back dictionaries and word lists if
# we search for more than one word...
#
my $words = random_word();
my $start = int(rand(8)) * 10 + 1;
my $search_url = $hotbot_search_url . $words . "&first=$start&page=more";
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
# Hotbot plays redirection games too
# (not any more?)
# next unless ($u =~ m@/director.asp\?.*\btarget=([^&]+)@);
# $u = url_decode($1);
next unless ($u =~ m@^https?://@i); # skip non-HTTP or relative URLs
next if ($u =~ m@[/.]hotbot\.com\b@i); # skip hotbot builtins
next if ($u =~ m@[/.]lycos\.com\b@i); # skip hotbot builtins
next if ($u =~ m@[/.]inktomi\.com\b@i); # skip hotbot builtins
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
$timeout, @candidates);
}
############################################################################
#
# Pick images by feeding random words into Lycos
#
############################################################################
my $lycos_search_url = "http://search.lycos.com/default.asp" .
"?lpv=1" .
"&loc=searchhp" .
"&tab=web" .
"&query=";
sub pick_from_lycos_text($) {
my ($timeout) = @_;
$last_search = $lycos_search_url; # for warnings
# lycos seems to always give us back dictionaries and word lists if
# we search for more than one word...
#
my $words = random_word();
my $start = int(rand(8)) * 10 + 1;
my $search_url = $lycos_search_url . $words . "&first=$start&page=more";
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
# Lycos plays redirection games.
# (not any more?)
# next unless ($u =~ m@^https?://click.lycos.com/director.asp
# .*
# \btarget=([^&]+)
# .*
# @x);
# $u = url_decode($1);
next unless ($u =~ m@^https?://@i); # skip non-HTTP or relative URLs
next if ($u =~ m@[/.]hotbot\.com\b@i); # skip lycos builtins
next if ($u =~ m@[/.]lycos\.com\b@i); # skip lycos builtins
next if ($u =~ m@[/.]terralycos\.com\b@i); # skip lycos builtins
next if ($u =~ m@[/.]inktomi\.com\b@i); # skip lycos builtins
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
$timeout, @candidates);
}
############################################################################
#
# Pick images by feeding random words into news.yahoo.com
#
############################################################################
my $yahoo_news_url = "http://news.search.yahoo.com/search/news" .
"?c=news_photos" .
"&p=";
# yahoonews
sub pick_from_yahoo_news_text($) {
my ($timeout) = @_;
$last_search = $yahoo_news_url; # for warnings
my $words = random_word();
my $search_url = $yahoo_news_url . $words;
my ($search_hit_count, @subpages) =
pick_from_search_engine ($timeout, $search_url, $words);
my @candidates = ();
foreach my $u (@subpages) {
# de-redirectize the URLs
$u =~ s@^https?://rds\.yahoo\.com/.*-http%3A@http:@s;
# only accept URLs on Yahoo's news site
next unless ($u =~ m@^https?://dailynews\.yahoo\.com/@i ||
$u =~ m@^https?://story\.news\.yahoo\.com/@i);
next unless ($u =~ m@&u=/@);
LOG ($verbose_filter, " candidate: $u");
push @candidates, $u;
}
return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
$timeout, @candidates);
}
############################################################################
#
# Pick images from LiveJournal's list of recently-posted images.
#
############################################################################
my $livejournal_img_url = "http://www.livejournal.com/stats/latest-img.bml";
# With most of our image sources, we get a random page and then select
# from the images on it. However, in the case of LiveJournal, the page
# of images tends to update slowly; so we'll remember the last N entries
# on it and randomly select from those, to get a wider variety each time.
my $lj_cache_size = 1000;
my @lj_cache = (); # fifo, for ordering by age
my %lj_cache = (); # hash, for detecting dups
# livejournal
sub pick_from_livejournal_images($) {
my ($timeout) = @_;
$last_search = $livejournal_img_url; # for warnings
my ( $base, $body ) = get_document ($livejournal_img_url, undef, $timeout);
# Often the document comes back empty. If so, just use the cache.
# return () unless $body;
$body = '' unless defined($body);
$body =~ s/\n/ /gs;
$body =~ s/(<recent-image)\b/\n$1/gsi;
foreach (split (/\n/, $body)) {
next unless (m/^<recent-image\b/);
next unless (m/\bIMG=[\'\"]([^\'\"]+)[\'\"]/si);
my $img = html_unquote ($1);
next if ($lj_cache{$img}); # already have it
next unless (m/\bURL=[\'\"]([^\'\"]+)[\'\"]/si);
my $page = html_unquote ($1);
my @pair = ($img, $page);
LOG ($verbose_filter, " candidate: $img");
push @lj_cache, \@pair;
$lj_cache{$img} = \@pair;
}
return () if ($#lj_cache == -1);
my $n = $#lj_cache+1;
my $i = int(rand($n));
my ($img, $page) = @{$lj_cache[$i]};
# delete this one from @lj_cache and from %lj_cache.
#
@lj_cache = ( @lj_cache[0 .. $i-1],
@lj_cache[$i+1 .. $#lj_cache] );
delete $lj_cache{$img};
# Keep the size of the cache under the limit by nuking older entries
#
while ($#lj_cache >= $lj_cache_size) {
my $pairP = shift @lj_cache;
my $img = $pairP->[0];
delete $lj_cache{$img};
}
LOG ($verbose_load, "picked image " .($i+1) . "/$n: $img");
return ($page, $img);
}
############################################################################
#
# Pick images from ircimages.com (images that have been in the /topic of
# various IRC channels.)
#
############################################################################
my $ircimages_url = "http://ircimages.com/";
# ircimages
sub pick_from_ircimages($) {
my ($timeout) = @_;
$last_search = $ircimages_url; # for warnings
my $n = int(rand(2900));
my $search_url = $ircimages_url . "page-$n";
my ( $base, $body ) = get_document ($search_url, undef, $timeout);
return () unless $body;
my @candidates = ();
$body =~ s/\n/ /gs;
$body =~ s/(<A)\b/\n$1/gsi;
foreach (split (/\n/, $body)) {
my ($u) = m@<A\s.*\bHREF\s*=\s*([^>]+)>@i;
next unless $u;
if ($u =~ m/^\"([^\"]*)\"/) { $u = $1; } # quoted string
elsif ($u =~ m/^([^\s]*)\s/) { $u = $1; } # or token
next unless ($u =~ m/^https?:/i);
next if ($u =~ m@^https?://(searchirc\.com\|ircimages\.com)@i);
next unless ($u =~ m@[.](gif|jpg|jpeg|pjpg|pjpeg|png)$@i);
LOG ($verbose_http, " HREF: $u");
push @candidates, $u;
}
LOG ($verbose_filter, "" . $#candidates+1 . " links on $search_url");
return () if ($#candidates == -1);
my $i = int(rand($#candidates+1));
my $img = $candidates[$i];
LOG ($verbose_load, "picked image " .($i+1) . "/" . ($#candidates+1) .
": $img");
$search_url = $img; # hmm...
return ($search_url, $img);
}
############################################################################
#
# Pick images from Twitpic's list of recently-posted images.
#
############################################################################
my $twitpic_img_url = "http://twitpic.com/public_timeline/feed.rss";
# With most of our image sources, we get a random page and then select
# from the images on it. However, in the case of Twitpic, the page
# of images tends to update slowly; so we'll remember the last N entries
# on it and randomly select from those, to get a wider variety each time.
my $twitpic_cache_size = 1000;
my @twitpic_cache = (); # fifo, for ordering by age
my %twitpic_cache = (); # hash, for detecting dups
# twitpic
sub pick_from_twitpic_images($) {
my ($timeout) = @_;
$last_search = $twitpic_img_url; # for warnings
my ( $base, $body ) = get_document ($twitpic_img_url, undef, $timeout);
# Update the cache.
if ($body) {
$body =~ s/\n/ /gs;
$body =~ s/(<item)\b/\n$1/gsi;
my @items = split (/\n/, $body);
shift @items;
foreach (@items) {
next unless (m@<link>([^<>]*)</link>@si);
my $page = html_unquote ($1);
$page =~ s@/$@@s;
$page .= '/full';
next if ($twitpic_cache{$page}); # already have it
LOG ($verbose_filter, " candidate: $page");
push @twitpic_cache, $page;
$twitpic_cache{$page} = $page;
}
}
# Pull from the cache.
return () if ($#twitpic_cache == -1);
my $n = $#twitpic_cache+1;
my $i = int(rand($n));
my $page = $twitpic_cache[$i];
# delete this one from @twitpic_cache and from %twitpic_cache.
#
@twitpic_cache = ( @twitpic_cache[0 .. $i-1],
@twitpic_cache[$i+1 .. $#twitpic_cache] );
delete $twitpic_cache{$page};
# Keep the size of the cache under the limit by nuking older entries
#
while ($#twitpic_cache >= $twitpic_cache_size) {
my $page = shift @twitpic_cache;
delete $twitpic_cache{$page};
}
( $base, $body ) = get_document ($page, undef, $timeout);
my $img = undef;
$body = '' unless defined($body);
foreach (split (/<img\s+/, $body)) {
my ($src) = m/\bsrc=[\"\'](.*?)[\"\']/si;
next unless $src;
next if m@/js/@s;
next if m@/images/@s;
$img = $src;
$img = "http:$img" if ($img =~ m@^//@s); # Oh come on
# Sometimes these images are hosted on twitpic, sometimes on Amazon.
if ($img =~ m@^/@) {
$base =~ s@^(https?://[^/]+)/.*@$1@s;
$img = $base . $img;
}
last;
}
if (!$img) {
LOG ($verbose_load, "no matching images on $page\n");
return ();
}
LOG ($verbose_load, "picked image " .($i+1) . "/$n: $img");
return ($page, $img);
}
############################################################################
#
# Pick images from Twitter's list of recently-posted updates.
#
############################################################################
# With most of our image sources, we get a random page and then select
# from the images on it. However, in the case of Twitter, the page
# of images only updates once a minute; so we'll remember the last N entries
# on it and randomly select from those, to get a wider variety each time.
my $twitter_img_url = "http://api.twitter.com/1/statuses/" .
"public_timeline.json" .
"?include_entities=true" .
"&include_rts=true" .
"&count=200";
my $twitter_cache_size = 1000;
my @twitter_cache = (); # fifo, for ordering by age
my %twitter_cache = (); # hash, for detecting dups
# twitter
sub pick_from_twitter_images($) {
my ($timeout) = @_;
$last_search = $twitter_img_url; # for warnings
my ( $base, $body ) = get_document ($twitter_img_url, undef, $timeout);
# Update the cache.
if ($body) {
$body =~ s/[\r\n]+/ /gs;
# Parsing JSON is a pain in the ass. So we halfass it as usual.
$body =~ s/^\[|\]$//s;
$body =~ s/(\[.*?\])/{ $_ = $1; s@\},@\} @gs; $_; }/gsexi;
my @items = split (/\},\{/, $body);
foreach (@items) {
my ($name) = m@"screen_name":"([^\"]+)"@si;
my ($img) = m@"media_url":"([^\"]+)"@si;
my ($page) = m@"display_url":"([^\"]+)"@si;
next unless ($name && $img && $page);
foreach ($img, $page) {
s/\\//gs;
$_ = "http://$_" unless (m/^http/si);
}
next if ($twitter_cache{$page}); # already have it
LOG ($verbose_filter, " candidate: $page - $img");
push @twitter_cache, $page;
$twitter_cache{$page} = $img;
}
}
# Pull from the cache.
return () if ($#twitter_cache == -1);
my $n = $#twitter_cache+1;
my $i = int(rand($n));
my $page = $twitter_cache[$i];
my $url = $twitter_cache{$page};
# delete this one from @twitter_cache and from %twitter_cache.
#
@twitter_cache = ( @twitter_cache[0 .. $i-1],
@twitter_cache[$i+1 .. $#twitter_cache] );
delete $twitter_cache{$page};
# Keep the size of the cache under the limit by nuking older entries
#
while ($#twitter_cache >= $twitter_cache_size) {
my $page = shift @twitter_cache;
delete $twitter_cache{$page};
}
LOG ($verbose_load, "picked page $url");
$suppress_audit = 1;
return ($page, $url);
}
############################################################################
#
# Pick images from Flickr's page of recently-posted photos.
#
############################################################################
my $flickr_img_url = "http://www.flickr.com/explore/";
# Like LiveJournal, the Flickr page of images tends to update slowly,
# so remember the last N entries on it and randomly select from those.
# I know that Flickr has an API (http://www.flickr.com/services/api/)
# but it was easy enough to scrape the HTML, so I didn't bother exploring.
my $flickr_cache_size = 1000;
my @flickr_cache = (); # fifo, for ordering by age
my %flickr_cache = (); # hash, for detecting dups
# flickr_recent
sub pick_from_flickr_recent($) {
my ($timeout) = @_;
my $start = 16 * int(rand(100));
$last_search = $flickr_img_url; # for warnings
$last_search .= "?start=$start" if ($start > 0);
my ( $base, $body ) = get_document ($last_search, undef, $timeout);
# If the document comes back empty. just use the cache.
# return () unless $body;
$body = '' unless defined($body);
my $count = 0;
my $count2 = 0;
if ($body =~ m@{ *"_data": \[ ( .*? \} ) \]@six) {
$body = $1;
} else {
LOG ($verbose_load, "flickr unparsable: $last_search");
return ();
}
$body =~ s/[\r\n]/ /gs;
$body =~ s/(\},) *(\{)/$1\n$2/gs; # "_flickrModelRegistry"
foreach my $chunk (split (/\n/, $body)) {
my ($img) = ($chunk =~ m@"displayUrl": *"(.*?)"@six);
next unless defined ($img);
$img =~ s/\\//gs;
$img = "http:$img" unless ($img =~ m/^http/s);
my ($user) = ($chunk =~ m/"pathAlias": *"(.*?)"/si);
next unless defined ($user);
my ($id) = ($img =~ m@/\d+/(\d+)_([\da-f]+)_@si);
my ($page) = "https://www.flickr.com/photos/$user/$id/";
# $img =~ s/_[a-z](\.[a-z\d]+)$/$1/si; # take off "thumb" suffix
$count++;
next if ($flickr_cache{$img}); # already have it
my @pair = ($img, $page, $start);
LOG ($verbose_filter, " candidate: $img");
push @flickr_cache, \@pair;
$flickr_cache{$img} = \@pair;
$count2++;
}
return () if ($#flickr_cache == -1);
my $n = $#flickr_cache+1;
my $i = int(rand($n));
my ($img, $page) = @{$flickr_cache[$i]};
# delete this one from @flickr_cache and from %flickr_cache.
#
@flickr_cache = ( @flickr_cache[0 .. $i-1],
@flickr_cache[$i+1 .. $#flickr_cache] );
delete $flickr_cache{$img};
# Keep the size of the cache under the limit by nuking older entries
#
while ($#flickr_cache >= $flickr_cache_size) {
my $pairP = shift @flickr_cache;
my $img = $pairP->[0];
delete $flickr_cache{$img};
}
LOG ($verbose_load, "picked image " .($i+1) . "/$n: $img");
return ($page, $img);
}
############################################################################
#
# Pick images from a random RSS feed on Flickr.
#
############################################################################
my $flickr_rss_base = ("http://www.flickr.com/services/feeds/" .
"photos_public.gne" .
"?format=rss_200_enc&tagmode=any&tags=");
# Picks a random RSS feed; picks a random image from that feed;
# returns 2 URLs: the page containing the image, and the image.
# Mostly by Joe Mcmahon <mcmahon@yahoo-inc.com>
#
# flickr_random
sub pick_from_flickr_random($) {
my $timeout = shift;
my $words = random_words(',');
my $rss = $flickr_rss_base . $words;
$last_search = $rss;
$_ = $words;
s/,/ /g;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "words: $_");
LOG ($verbose_load, "URL: $last_search");
$suppress_audit = 1;
my ( $base, $body ) = get_document ($last_search, undef, $timeout);
if (!$base || !$body) {
$body = undef;
return;
}
my $img;
($base, $img) = pick_image_from_rss ($base, $body);
$body = undef;
return () unless defined ($img);
LOG ($verbose_load, "redirected to: $base");
return ($base, $img);
}
############################################################################
#
# Pick random images from Instagram.
#
############################################################################
my $instagram_url_base = "https://api.instagram.com/v1/media/popular";
# instagram_random
sub pick_from_instagram($) {
my $timeout = shift;
# Liberated access tokens.
# jsdo.it search for: instagram client_id
# Google search for: instagram "&client_id=" site:jsfiddle.net
my @tokens = (#'b59fbe4563944b6c88cced13495c0f49', # gramfeed.com
#'fa26679250df49c48a33fbcf30aae989', # instac.at
#'d9494686198d4dfeb954979a3e270e5e', # iconosquare.com
#'793ef48bb18e4197b61afce2d799b81c', # jsdo.it
#'67b8a3e0073449bba70600d0fc68e6cb', # jsdo.it
#'26a098e0df4d4b9ea8b4ce6c505b7742', # jsdo.it
#'2437cbcd906a4c10940f990d283d3cd5', # jsdo.it
#'191c7d7d5312464cbd92134f36ffdab5', # jsdo.it
#'acfec809437b4340b2c38f66503af774', # jsdo.it
#'e9f77604a3a24beba949c12d18130988', # jsdo.it
#'2cd7bcf68ae346529770073d311575b3', # jsdo.it
#'830c600fe8d742e2ab3f3b94f9bb22b7', # jsdo.it
#'55865a0397ad41e5997dd95ef4df8da1', # jsdo.it
#'192a5742f3644ea8bed1d25e439286a8', # jsdo.it
#'38ed1477e7a44595861b8842cdb8ba23', # jsdo.it
#'e52f79f645f54488ad0cc47f6f55ade6', # jsfiddle.net
);
my $tok = $tokens[int(rand($#tokens+1))];
$last_search = $instagram_url_base . "?client_id=" . $tok;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "URL: $last_search");
my ( $base, $body ) = get_document ($last_search, undef, $timeout);
if (!$base || !$body) {
$body = undef;
return;
}
$body =~ s/("link")/\001$1/gs;
my @chunks = split(/\001/, $body);
shift @chunks;
my @urls = ();
foreach (@chunks) {
s/\\//gs;
my ($url) = m/"link":\s*"(.*?)"/s;
my ($img) = m/"standard_resolution":\{"url":\s*"(.*?)"/s;
($img) = m/"url":\s*"(.*?)"/s unless $url;
next unless ($url && $img);
push @urls, [ $url, $img ];
}
if ($#urls < 0) {
LOG ($verbose_load, "no images on $last_search");
return ();
}
my $i = int(rand($#urls+1));
my ($url, $img) = @{$urls[$i]};
LOG ($verbose_load, "picked image " .($i+1) . "/" . ($#urls+1) . ": $url");
return ($url, $img);
}
############################################################################
#
# Pick images from Imgur.
#
############################################################################
my $imgur_base = 'http://imgur.com/search?qs=thumb&q_any=';
sub pick_from_imgur($) {
my $timeout = shift;
my $words = random_words('%20');
$last_search = $imgur_base . $words;
$_ = $words;
s/%20/ /g;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "words: $_");
LOG ($verbose_load, "URL: $last_search");
$suppress_audit = 1;
my ( $base, $body ) = get_document ($last_search, undef, $timeout);
if (!$base || !$body) {
$body = undef;
return;
}
my @imgs = ($body =~ m@\bHREF=[\"\']([^\'\"<>]*/gallery/[^\'\"<>]+)@gsi);
return () unless @imgs;
my $n = @imgs;
my $i = int(rand($n));
my $page = $imgs[$i];
$page =~ s/[?&].*$//s;
$page = "http://imgur.com$page" if ($page =~ m@^/@s);
my ($id) = ($page =~ m@([^/?&]+)$@s);
my $img = "http://i.imgur.com/$id.jpg";
LOG ($verbose_load, "picked image " .($i+1) . "/$n: $img");
return ($page, $img);
}
############################################################################
#
# Pick images from Tumblr.
#
############################################################################
my $tumblr_base = 'https://www.tumblr.com/search/';
sub pick_from_tumblr($) {
my $timeout = shift;
# Tumblr doesn't have an "or" search, which means our vocabulary is
# a bit too extensive to work well...
my $words = random_word();
$last_search = $tumblr_base . $words;
print STDERR "\n\n" if ($verbose_load);
LOG ($verbose_load, "words: $words");
LOG ($verbose_load, "URL: $last_search");
$suppress_audit = 1;
my ( $base, $body ) = get_document ($last_search, undef, $timeout);
if (!$base || !$body) {
$body = undef;
return;
}
my @imgs0 = ($body =~ m@<IMG\b([^<>]*)>@gsi);
return () unless @imgs0;
my @imgs;
foreach my $img (@imgs0) {
my ($src) = ($img =~ m@\bsrc=[\"\'](.*?)[\"\']@si);
my ($href) = ($img =~ m@\bdata-pin-url=[\"\'](.*?)[\"\']@si);
next unless ($src && $href);
next if ($src =~ m/^data:/s);
foreach ($src, $href) { $_ = "http://www.tumblr.com$_" if (m@^/@s); }
push @imgs, [$href, $src];
}
return () unless @imgs;
my $n = @imgs;
my $i = int(rand($n));
my $page = $imgs[$i]->[0];
my $img = $imgs[$i]->[1];
LOG ($verbose_load, "picked image " .($i+1) . "/$n: $img");
return ($page, $img);
}
############################################################################
#
# Pick images by waiting for driftnet to populate a temp dir with files.
# Requires driftnet version 0.1.5 or later.
# (Driftnet is a program by Chris Lightfoot that sniffs your local ethernet
# for images being downloaded by others.)
# Driftnet/webcollage integration by jwz.
#
############################################################################
# driftnet
sub pick_from_driftnet($) {
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) {
opendir (my $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 = $driftnet_magic;
error ("$id: $file not in $driftnet_dir?")
unless ($file =~ m@^\Q$driftnet_dir@o);
open (my $in, '<', $file) || error ("$id: $file: $!");
my $body = '';
local $/ = undef; # read entire file
$body = <$in>;
close ($in) || error ("$id: $file: $!");
unlink ($file) || error ("$id: $file: rm: $!");
return ($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, $pid));
}
# local-directory
sub pick_from_local_dir($) {
my ($timeout) = @_;
my $id = $local_magic;
$last_search = $id;
my $dir = $local_dir;
error ("\$local_dir unset?") unless ($dir);
$dir =~ s@/+$@@;
error ("$dir unreadable") unless (-d "$dir/.");
my $v = ($verbose_exec ? "-v" : "");
my $pick = `xscreensaver-getimage-file $v "$dir"`;
$pick =~ s/\s+$//s;
$pick = "$dir/$pick" unless ($pick =~ m@^/@s); # relative path
LOG ($verbose_load, "picked file $pick ($id)");
return ($id, $pick);
}
sub get_local_file($) {
my ($file) = @_;
error ("\$local_dir unset?") unless ($local_dir);
my $id = $local_magic;
error ("$id: $file not in $local_dir?")
unless ($file =~ m@^\Q$local_dir@o);
open (my $in, '<:raw', $file) || error ("$id: $file: $!");
local $/ = undef; # read entire file
my $body = <$in>;
close ($in) || error ("$id: $file: $!");
return ($id, $body);
}
############################################################################
#
# Pick a random image in a random way
#
############################################################################
# Picks a random image on a random page, and returns two URLs:
# the page containing the image, and the image.
# Returns () if nothing found this time.
#
sub pick_image(;$) {
my ($timeout) = @_;
$current_state = "select";
$load_method = "none";
my $n = int(rand(100));
my $fn = undef;
my $total = 0;
my @rest = @search_methods;
while (@rest) {
my $pct = shift @rest;
my $name = shift @rest;
my $tfn = shift @rest;
$total += $pct;
if ($total > $n && !defined($fn)) {
$fn = $tfn;
$current_state = $name;
$load_method = $current_state;
}
}
if ($total != 100) {
error ("internal error: \@search_methods totals to $total%!");
}
record_attempt ($current_state);
return $fn->($timeout);
}
############################################################################
#
# Statistics and logging
#
############################################################################
sub timestr() {
return strftime ("%H:%M:%S: ", localtime);
}
sub blurb() {
return "$progname: " . timestr() . "$current_state: ";
}
sub error($) {
my ($err) = @_;
print STDERR blurb() . "$err\n";
exit 1;
}
sub stacktrace() {
my $i = 1;
print STDERR "$progname: stack trace:\n";
while (1) {
my ($package, $filename, $line, $subroutine) = caller($i++);
last unless defined($package);
$filename =~ s@^.*/@@;
print STDERR " $filename#$line, $subroutine\n";
}
}
my $lastlog = "";
sub clearlog() {
$lastlog = "";
}
sub showlog() {
my $head = "$progname: DEBUG: ";
foreach (split (/\n/, $lastlog)) {
print STDERR "$head$_\n";
}
$lastlog = "";
}
sub LOG($$) {
my ($print, $msg) = @_;
my $blurb = timestr() . "$current_state: ";
$lastlog .= "$blurb$msg\n";
print STDERR "$progname: $blurb$msg\n" if $print;
}
my %stats_attempts;
my %stats_successes;
my %stats_elapsed;
my $last_state = undef;
sub record_attempt($) {
my ($name) = @_;
if ($last_state) {
record_failure($last_state) unless ($image_succeeded > 0);
}
$last_state = $name;
clearlog();
report_performance();
start_timer($name);
$image_succeeded = 0;
$suppress_audit = 0;
}
sub record_success($$$) {
my ($name, $url, $base) = @_;
if (defined($stats_successes{$name})) {
$stats_successes{$name}++;
} else {
$stats_successes{$name} = 1;
}
stop_timer ($name, 1);
my $o = $current_state;
$current_state = $name;
save_recent_url ($url, $base);
$current_state = $o;
$image_succeeded = 1;
clearlog();
}
sub record_failure($) {
my ($name) = @_;
return if $image_succeeded;
stop_timer ($name, 0);
if ($verbose_load && !$verbose_exec) {
if ($suppress_audit) {
print STDERR "$progname: " . timestr() . "(audit log suppressed)\n";
return;
}
my $o = $current_state;
$current_state = "DEBUG";
my $line = "#" x 78;
print STDERR "\n\n\n";
print STDERR ("#" x 78) . "\n";
print STDERR blurb() . "failed to get an image. Full audit log:\n";
print STDERR "\n";
showlog();
print STDERR ("-" x 78) . "\n";
print STDERR "\n\n";
$current_state = $o;
}
$image_succeeded = 0;
}
sub stats_of($) {
my ($name) = @_;
my $i = $stats_successes{$name};
my $j = $stats_attempts{$name};
$i = 0 unless $i;
$j = 0 unless $j;
return "" . ($j ? int($i * 100 / $j) : "0") . "%";
}
my $current_start_time = 0;
sub start_timer($) {
my ($name) = @_;
$current_start_time = time;
if (defined($stats_attempts{$name})) {
$stats_attempts{$name}++;
} else {
$stats_attempts{$name} = 1;
}
if (!defined($stats_elapsed{$name})) {
$stats_elapsed{$name} = 0;
}
}
sub stop_timer($$) {
my ($name, $success) = @_;
$stats_elapsed{$name} += time - $current_start_time;
}
my $last_report_time = 0;
sub report_performance() {
return unless $verbose_warnings;
my $now = time;
return unless ($now >= $last_report_time + $report_performance_interval);
my $ot = $last_report_time;
$last_report_time = $now;
return if ($ot == 0);
my $blurb = "$progname: " . timestr();
print STDERR "\n";
print STDERR "${blurb}Current standings:\n";
foreach my $name (sort keys (%stats_attempts)) {
my $try = $stats_attempts{$name};
my $suc = $stats_successes{$name} || 0;
my $pct = int($suc * 100 / $try);
my $secs = $stats_elapsed{$name};
my $secs_link = $secs / $try;
print STDERR sprintf ("$blurb %-14s %4s (%d/%d);" .
" \t %.1f secs/link\n",
"$name:", "$pct%", $suc, $try, $secs_link);
}
}
my $max_recent_images = 400;
my $max_recent_sites = 20;
my @recent_images = ();
my @recent_sites = ();
sub save_recent_url($$) {
my ($url, $base) = @_;
return unless ($verbose_warnings);
$_ = $url;
my ($site) = m@^https?://([^ \t\n\r/:]+)@;
return unless defined ($site);
if ($base eq $driftnet_magic || $base eq $local_magic) {
$site = $base;
@recent_images = ();
}
my $done = 0;
foreach (@recent_images) {
if ($_ eq $url) {
print STDERR blurb() . "WARNING: recently-duplicated image: $url" .
" (on $base via $last_search)\n";
$done = 1;
last;
}
}
# suppress "duplicate site" warning via %warningless_sites.
#
if ($warningless_sites{$site}) {
$done = 1;
} elsif ($site =~ m@([^.]+\.[^.]+\.[^.]+)$@ &&
$warningless_sites{$1}) {
$done = 1;
} elsif ($site =~ m@([^.]+\.[^.]+)$@ &&
$warningless_sites{$1}) {
$done = 1;
}
if (!$done) {
foreach (@recent_sites) {
if ($_ eq $site) {
print STDERR blurb() . "WARNING: recently-duplicated site: $site" .
" ($url on $base via $last_search)\n";
last;
}
}
}
push @recent_images, $url;
push @recent_sites, $site;
shift @recent_images if ($#recent_images >= $max_recent_images);
shift @recent_sites if ($#recent_sites >= $max_recent_sites);
}
##############################################################################
#
# other utilities
#
##############################################################################
# Does %-decoding.
#
sub url_decode($) {
($_) = @_;
tr/+/ /;
s/%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))/eg;
return $_;
}
# Given the raw body of a GIF document, returns the dimensions of the image.
#
sub gif_size($) {
my ($body) = @_;
my $type = substr($body, 0, 6);
my $s;
return () unless ($type =~ /GIF8[7,9]a/);
$s = substr ($body, 6, 10);
my ($a,$b,$c,$d) = unpack ("C"x4, $s);
return () unless defined ($d);
return (($b<<8|$a), ($d<<8|$c));
}
# Given the raw body of a JPEG document, returns the dimensions of the image.
#
sub jpeg_size($) {
my ($body) = @_;
my $i = 0;
my $L = length($body);
my $c1 = substr($body, $i, 1); $i++;
my $c2 = substr($body, $i, 1); $i++;
return () unless (ord($c1) == 0xFF && ord($c2) == 0xD8);
my $ch = "0";
while (ord($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++;
}
# $ch contains the value of the marker.
my $marker = ord($ch);
if (($marker >= 0xC0) &&
($marker <= 0xCF) &&
($marker != 0xC4) &&
($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|$b));
} else {
# We must skip variables, since FFs in variable names aren't
# valid JPEG markers.
return () if (length($body) <= $i);
my $s = substr($body, $i, 2); $i += 2;
my ($c1, $c2) = unpack ("C"x2, $s);
my $length = ($c1 << 8) | $c2;
return () if ($length < 2);
$i += $length-2;
}
}
return ();
}
# Given the raw body of a PNG document, returns the dimensions of the image.
#
sub png_size($) {
my ($body) = @_;
return () unless ($body =~ m/^\211PNG\r/);
my ($bits) = ($body =~ m/^.{12}(.{12})/s);
return () unless defined ($bits);
return () unless ($bits =~ /^IHDR/);
my ($ign, $w, $h) = unpack("a4N2", $bits);
return ($w, $h);
}
# Given the raw body of a PNM document, returns the dimensions of the image.
#
sub pnm_size($) {
my ($body) = @_;
return () unless ($body =~ m/^P[1-6]\r?\n(\d+) +(\d+)\r?\n/s);
return ($1, $2);
}
# Given the raw body of a GIF, JPEG, or PNG document, returns the dimensions
# of the image.
#
sub image_size($) {
my ($body) = @_;
my ($w, $h) = gif_size ($body);
if ($w && $h) { return ($w, $h); }
($w, $h) = jpeg_size ($body);
if ($w && $h) { return ($w, $h); }
return png_size ($body);
if ($w && $h) { return ($w, $h); }
return pnm_size ($body);
}
# returns the full path of the named program, or undef.
#
sub which($) {
my ($prog) = @_;
foreach (split (/:/, $ENV{PATH})) {
my $path = "$_/$prog";
if (-x $path) {
return $path;
}
}
return undef;
}
# Like rand(), but chooses numbers with a bell curve distribution.
sub bellrand(;$) {
($_) = @_;
$_ = 1.0 unless defined($_);
$_ /= 3.0;
return (rand($_) + rand($_) + rand($_));
}
sub exit_cleanup() {
x_cleanup();
print STDERR "$progname: exiting\n" if ($verbose_warnings);
if (@pids_to_kill) {
print STDERR blurb() . "killing: " . join(' ', @pids_to_kill) . "\n";
kill ('TERM', @pids_to_kill);
}
}
sub signal_cleanup($) {
my ($sig) = @_;
print STDERR blurb() . (defined($sig)
? "caught signal $sig."
: "exiting.")
. "\n"
if ($verbose_exec || $verbose_warnings);
exit 1;
}
##############################################################################
#
# Generating a list of urls only
#
##############################################################################
sub url_only_output() {
do {
my ($base, $img) = pick_image;
if ($img) {
$base =~ s/ /%20/g;
$img =~ s/ /%20/g;
print "$img $base\n";
}
} while (1);
}
##############################################################################
#
# Running as an xscreensaver module, or as a web page imagemap
#
##############################################################################
my ($image_png, $image_tmp1, $image_tmp2);
{
my $seed = rand(0xFFFFFFFF);
$image_png = sprintf ("%s/webcollage-%08x",
($ENV{TMPDIR} ? $ENV{TMPDIR} : "/tmp"),
$seed);
$image_tmp1 = $image_png . '-1.png';
$image_tmp2 = $image_png . '-2.png';
$image_png .= '.png';
}
my $filter_cmd = undef;
my $post_filter_cmd = undef;
my $background = undef;
my @imagemap_areas = ();
my $imagemap_html_tmp = undef;
my $imagemap_jpg_tmp = undef;
my $img_width; # size of the image being generated.
my $img_height;
my $delay = 2;
sub x_cleanup() {
unlink $image_png, $image_tmp1, $image_tmp2;
unlink $imagemap_html_tmp, $imagemap_jpg_tmp
if (defined ($imagemap_html_tmp));
}
# Like system, but prints status about exit codes, and kills this process
# with whatever signal killed the sub-process, if any.
#
sub nontrapping_system(@) {
$! = 0;
$_ = join(" ", @_);
s/\"[^\"]+\"/\"...\"/g;
LOG ($verbose_exec, "executing \"$_\"");
my $rc = system @_;
if ($rc == 0) {
LOG ($verbose_exec, "subproc exited normally.");
} elsif (($rc & 0xff) == 0) {
$rc >>= 8;
LOG ($verbose_exec, "subproc exited with status $rc.");
} else {
if ($rc & 0x80) {
LOG ($verbose_exec, "subproc dumped core.");
$rc &= ~0x80;
}
LOG ($verbose_exec, "subproc died with signal $rc.");
# die that way ourselves.
kill $rc, $$;
}
return $rc;
}
# Creates a solid-colored PNG.
#
sub pngmake($$$$) {
my ($outfile, $bgcolor, $w, $h) = @_;
my @cmd;
if ($webcollage_helper) {
@cmd = ($webcollage_helper, $bgcolor, $w, $h, $outfile);
} else {
@cmd = ($convert_cmd, '-size', "${w}x${h}", "xc:$bgcolor", $outfile);
}
my $rc = nontrapping_system (@cmd);
if ($rc != 0) {
LOG(0, "failed to create $bgcolor image: \"$outfile\"");
exit(1);
}
}
sub pick_root_displayer() {
my @names = ();
if ($cocoa_p) {
# see "xscreensaver/hacks/webcollage-cocoa.m"
return "echo COCOA LOAD ";
}
foreach my $cmd (@root_displayers) {
$_ = $cmd;
my ($name) = m/^([^ ]+)/;
push @names, "\"$name\"";
LOG ($verbose_exec, "looking for $name...");
foreach my $dir (split (/:/, $ENV{PATH})) {
LOG ($verbose_exec, " checking $dir/$name");
return $cmd if (-x "$dir/$name");
}
}
$names[$#names] = "or " . $names[$#names];
error "none of: " . join (", ", @names) . " were found on \$PATH.";
}
my $png_to_root_window_cmd = undef;
sub x_or_image_output($) {
my ($window_id) = @_;
# Adjust the PATH for OS X 10.10.
#
$_ = $0;
s:/[^/]*$::;
s/([^a-zA-Z0-9._\-+\/])/\\$1/g;
$ENV{PATH} = "$_:$ENV{PATH}";
# Check for our helper program, to see whether we need to use imagemagick.
#
$_ = "webcollage-helper";
if (! defined ($webcollage_helper)) {
$webcollage_helper = which ($_);
}
if (defined ($webcollage_helper)) {
LOG ($verbose_decode, "found \"$webcollage_helper\"");
} else {
LOG (($verbose_decode || $verbose_load), "no $_ program");
}
if ($cocoa_p && !defined ($webcollage_helper)) {
error ("webcollage-helper not found in Cocoa-mode!");
}
if (!$cocoa_p && defined ($webcollage_helper)) {
foreach ($image_png, $image_tmp1, $image_tmp2) {
s/\.png$/.jpg/si;
}
}
# make sure the various programs we execute exist, right up front.
#
my @progs = ();
push @progs, $convert_cmd unless defined($webcollage_helper);
foreach (@progs) {
which ($_) || error "$_ not found on \$PATH.";
}
# find a root-window displayer program.
#
if (!$no_output_p) {
$png_to_root_window_cmd = pick_root_displayer();
}
if (defined ($window_id)) {
error ("-window-id only works if xscreensaver-getimage is installed")
unless ($png_to_root_window_cmd =~ m/^xscreensaver-getimage\b/);
error ("unparsable window id: $window_id")
unless ($window_id =~ m/^\d+$|^0x[\da-f]+$/i);
$png_to_root_window_cmd =~ s/--?root\b/$window_id/ ||
error ("unable to munge displayer: $png_to_root_window_cmd");
}
if (!$img_width || !$img_height) {
if (!defined ($window_id) &&
defined ($ENV{XSCREENSAVER_WINDOW})) {
$window_id = $ENV{XSCREENSAVER_WINDOW};
}
if (!defined ($window_id)) {
$_ = "xdpyinfo";
which ($_) || error "$_ not found on \$PATH.";
$_ = `$_`;
($img_width, $img_height) = m/dimensions: *(\d+)x(\d+) /;
if (!defined($img_height)) {
error "xdpyinfo failed.";
}
} else { # we have a window id
$_ = "xwininfo";
which ($_) || error "$_ not found on \$PATH.";
$_ .= " -id $window_id";
$_ = `$_`;
($img_width, $img_height) = m/^\s*Width:\s*(\d+)\n\s*Height:\s*(\d+)\n/m;
if (!defined($img_height)) {
error "xwininfo failed.";
}
}
}
my $bgcolor = "#000000";
my $bgimage = undef;
if ($background) {
if ($background =~ m/^\#[0-9a-f]+$/i) {
$bgcolor = $background;
} elsif (-r $background) {
$bgimage = $background;
} elsif (! $background =~ m@^[-a-z0-9 ]+$@i) {
error "not a color or readable file: $background";
} else {
# default to assuming it's a color
$bgcolor = $background;
}
}
# Create the sold-colored base image.
#
LOG ($verbose_decode, "creating base image: ${img_width}x${img_height}");
$_ = pngmake ($image_png, $bgcolor, $img_width, $img_height);
# Paste the default background image in the middle of it.
#
if ($bgimage) {
open (my $in, '<:raw', $bgimage) || error ("$bgimage: $!");
local $/ = undef; # read entire file
my $body = <$in>;
close ($in) || error ("$bgimage: $!");
paste_image ('init', $image_png, $body, 'init', 1);
}
clearlog();
while (1) {
my ($base, $img) = pick_image();
my $source = $current_state;
$current_state = "loadimage";
if ($img) {
my ($headers, $body) = get_document ($img, $base);
if ($body) {
paste_image ($base, $img, $body, $source);
$body = undef;
}
}
$current_state = "idle";
$load_method = "none";
unlink $image_tmp1, $image_tmp2;
sleep $delay;
}
}
sub paste_image($$$$;$) {
my ($base, $img, $body, $source, $init_p) = @_;
$current_state = "paste";
$suppress_audit = 0;
LOG ($verbose_decode, "got $img (" . length($body) . ")");
my ($iw, $ih) = image_size ($body);
if (!$iw || !$ih) {
LOG (($verbose_decode || $verbose_load),
"not a GIF, JPG, or PNG" .
(($body =~ m@<(base|html|head|body|script|table|a href)>@i)
? " (looks like HTML)" : "") .
": $img");
$suppress_audit = 1;
$body = undef;
return 0;
}
if ($iw <= 0 || $ih <= 0 || $iw > 9999 || $ih > 9999) {
LOG (($verbose_decode || $verbose_load),
"ludicrous image dimensions: $iw x $ih (" . length($body) .
"): $img");
$body = undef;
return 0;
}
open (my $out, '>:raw', $image_tmp1) || error ("writing $image_tmp1: $!");
(print $out $body) || error ("writing $image_tmp1: $!");
close ($out) || error ("writing $image_tmp1: $!");
record_success ($load_method, $img, $base);
my $ow = $iw; # used only for error messages
my $oh = $ih;
# don't just tack this onto the front of the pipeline -- we want it to
# be able to change the size of the input image.
#
if ($filter_cmd && !$init_p) {
LOG ($verbose_decode, "running $filter_cmd");
# #### Historically, $filter_cmd read and write PPM files.
# This is doing PNG or JPEG now. I'll bet nobody uses this.
my $rc = nontrapping_system "($filter_cmd) < $image_tmp1 >$image_tmp2";
if ($rc != 0) {
LOG(($verbose_decode || $verbose_load),
"failed command: \"$filter_cmd\"");
LOG(($verbose_decode || $verbose_load),
"failed URL: \"$img\" (${ow}x$oh)");
return;
}
rename ($image_tmp2, $image_tmp1);
# re-get the width/height in case the filter resized it.
open (my $imgf, '<:raw', $image_tmp1) || return 0;
my $b = '';
sysread ($imgf, $b, 10240);
close $imgf;
($iw, $ih) = image_size ($b);
return 0 unless ($iw && $ih);
}
my $target_w = $img_width; # max rectangle into which the image must fit
my $target_h = $img_height;
my $scale = 1.0;
my $crop_x = 0; # the sub-rectangle of the image
my $crop_y = 0; # that we will actually paste.
my $crop_w = $iw;
my $crop_h = $ih;
my $x = 0;
my $y = 0;
if (!$init_p) {
# Usually scale the image to fit on the screen -- but sometimes scale it
# to fit on half or a quarter of the screen. (We do this by reducing the
# size of the target rectangle.) Note that the image is not merely scaled
# to fit; we instead cut the image in half repeatedly until it fits in the
# target rectangle -- that gives a wider distribution of sizes.
#
if (rand() < 0.3) { $target_w /= 2; $target_h /= 2; } # reduce target rect
if (rand() < 0.3) { $target_w /= 2; $target_h /= 2; }
if ($iw > $target_w || $ih > $target_h) {
while ($iw > $target_w ||
$ih > $target_h) {
$iw = int($iw / 2);
$ih = int($ih / 2);
$scale /= 2;
}
if ($iw <= 10 || $ih <= 10) {
LOG ($verbose_decode, "scaling ${ow}x${oh} to ${iw}x$ih" .
" would have been bogus.");
return 0;
}
$crop_w = $iw;
$crop_h = $ih;
LOG ($verbose_decode, "scaling ${ow}x${oh} to ${iw}x$ih ($scale)");
}
my $src = $image_tmp1;
# The chance that we will randomly crop out a section of an image starts
# out fairly low, but goes up for images that are very large, or images
# that have ratios that make them look like banners (we try to avoid
# banner images entirely, but they slip through when the IMG tags didn't
# have WIDTH and HEIGHT specified.)
#
my $crop_chance = 0.2;
if ($iw > $img_width * 0.4 || $ih > $img_height * 0.4) {
$crop_chance += 0.2;
}
if ($iw > $img_width * 0.7 || $ih > $img_height * 0.7) {
$crop_chance += 0.2;
}
if ($min_ratio && ($iw * $min_ratio) > $ih) {
$crop_chance += 0.7;
}
if ($crop_chance > 0.1) {
LOG ($verbose_decode, "crop chance: $crop_chance");
}
if (rand() < $crop_chance) {
my $ow = $crop_w;
my $oh = $crop_h;
if ($crop_w > $min_width) {
# if it's a banner, select the width linearly.
# otherwise, select a bell.
my $r = (($min_ratio && ($iw * $min_ratio) > $ih)
? rand()
: bellrand());
$crop_w = $min_width + int ($r * ($crop_w - $min_width));
$crop_x = int (rand() * ($ow - $crop_w));
}
if ($crop_h > $min_height) {
# height always selects as a bell.
$crop_h = $min_height + int (bellrand() * ($crop_h - $min_height));
$crop_y = int (rand() * ($oh - $crop_h));
}
# Clip it to the actual post-scaling image size.
if ($crop_x + $crop_w > $iw) { $crop_w = $iw - $crop_x; }
if ($crop_y + $crop_h > $ih) { $crop_h = $ih - $crop_y; }
if ($crop_x < 0) { $crop_w += $crop_x; $crop_x = 0; }
if ($crop_y < 0) { $crop_h += $crop_y; $crop_y = 0; }
if ($crop_x != 0 || $crop_y != 0 ||
$crop_w != $iw || $crop_h != $ih) {
LOG ($verbose_decode,
"randomly cropping to ${crop_w}x$crop_h \@ $crop_x,$crop_y");
}
}
# Where the image should logically land -- this might be negative.
#
$x = int((rand() * ($img_width + $crop_w/2)) - $crop_w*3/4);
$y = int((rand() * ($img_height + $crop_h/2)) - $crop_h*3/4);
# if we have chosen to paste the image outside of the rectangle of the
# screen, then we need to crop it.
#
if ($x < 0 ||
$y < 0 ||
$x + $crop_w > $img_width ||
$y + $crop_h > $img_height) {
LOG ($verbose_decode,
"cropping for effective paste of ${crop_w}x${crop_h} \@ $x,$y");
if ($x < 0) { $crop_x -= $x; $crop_w += $x; $x = 0; }
if ($y < 0) { $crop_y -= $y; $crop_h += $y; $y = 0; }
if ($x + $crop_w >= $img_width) { $crop_w = $img_width - $x - 1; }
if ($y + $crop_h >= $img_height) { $crop_h = $img_height - $y - 1; }
}
# If any cropping needs to happen, add pnmcut.
#
if ($crop_x != 0 || $crop_y != 0 ||
$crop_w != $iw || $crop_h != $ih) {
$iw = $crop_w;
$ih = $crop_h;
LOG ($verbose_decode, "cropping to ${crop_w}x$crop_h \@ " .
"$crop_x,$crop_y");
}
LOG ($verbose_decode, "pasting ${iw}x$ih \@ $x,$y in $image_png");
}
my @cmd;
if (defined ($webcollage_helper)) {
@cmd = ($webcollage_helper,
$image_tmp1, $image_png,
$scale, $opacity,
$crop_x, $crop_y, $x, $y,
$iw, $ih);
} else {
@cmd = ($convert_cmd,
$image_png,
'(',
$image_tmp1 . '[0]',
'-scale', sprintf("%.2f%%", 100 * $scale),
'-crop', "${iw}x${ih}+${crop_x}+${crop_y}",
'-geometry', "+${x}+${y}",
($init_p ? () :
(
# Blurry edges with rounded corners
'-alpha', 'set',
'-virtual-pixel', 'transparent',
'-channel', 'A',
'-blur', '0x12',
'-level', '50%,100%',
# Overall transparency
'-evaluate', 'multiply', $opacity,
'+channel',
)),
')',
'-composite',
'+repage',
'-strip',
$image_png);
}
#### $verbose_decode should mean 2>/dev/null
my $rc = nontrapping_system (@cmd);
if (-z $image_png) {
LOG (1, "failed command: \"@cmd\"");
print STDERR "\naudit log:\n\n\n";
print STDERR ("#" x 78) . "\n";
print STDERR blurb() . "$image_png has zero size\n";
showlog();
print STDERR "\n\n";
exit (1);
}
if ($rc != 0) {
LOG (($verbose_decode || $verbose_load), "failed command: \"@cmd\"");
LOG (($verbose_decode || $verbose_load),
"failed URL: \"$img\" (${ow}x$oh)");
return;
}
my $target = "$image_png";
# don't just tack this onto the end of the pipeline -- we don't want it
# to end up in $image_png, because we don't want the results to be
# cumulative.
#
if ($post_filter_cmd) {
# #### Historically, $post_filter_cmd read and write PPM files.
# This is doing PNG or JPEG now. I'll bet nobody uses this.
$target = $image_tmp1;
my $cmd = "($post_filter_cmd) < $image_png > $target";
$rc = nontrapping_system ($cmd);
if ($rc != 0) {
LOG ($verbose_decode, "filter failed: \"$post_filter_cmd\"\n");
return;
}
}
if (!$no_output_p) {
my $tsize = (stat($target))[7];
if ($tsize > 200) {
my $cmd = "$png_to_root_window_cmd $target";
# xv seems to hate being killed. it tends to forget to clean
# up after itself, and leaves windows around and colors allocated.
# I had this same problem with vidwhacker, and I'm not entirely
# sure what I did to fix it. But, let's try this: launch xv
# in the background, so that killing this process doesn't kill it.
# it will die of its own accord soon enough. So this means we
# start pumping bits to the root window in parallel with starting
# the next network retrieval, which is probably a better thing
# to do anyway.
#
$cmd .= " &" unless ($cocoa_p);
$rc = nontrapping_system ($cmd);
if ($rc != 0) {
LOG (($verbose_decode || $verbose_load), "display failed: \"$cmd\"");
return;
}
} else {
LOG ($verbose_decode, "$target size is $tsize");
}
}
if (defined($source)) {
$source .= "-" . stats_of($source);
print STDOUT "image: ${iw}x${ih} @ $x,$y $base $source\n"
if ($verbose_imgmap);
if ($imagemap_base) {
update_imagemap ($base, $x, $y, $iw, $ih,
$image_png, $img_width, $img_height);
}
}
clearlog();
return 1;
}
sub update_imagemap($$$$$$$$) {
my ($url, $x, $y, $w, $h, $image_png, $image_width, $image_height) = @_;
$current_state = "imagemap";
my $max_areas = 200;
$url = html_quote ($url);
push @imagemap_areas, [$x, $y, $w, $h, $url];
shift @imagemap_areas if (@imagemap_areas > $max_areas);
LOG ($verbose_decode, "area: $x,$y,$w,$h");
my $map_name = $imagemap_base;
$map_name =~ s@^.*/@@;
$map_name = 'collage' if ($map_name eq '');
my $imagemap_html = $imagemap_base . ".html";
my $imagemap_jpg = $imagemap_base . ".jpg";
my $imagemap_jpg2 = $imagemap_jpg;
$imagemap_jpg2 =~ s@^.*/@@gs;
if (!defined ($imagemap_html_tmp)) {
$imagemap_html_tmp = $imagemap_html . sprintf (".%08x", rand(0xffffffff));
$imagemap_jpg_tmp = $imagemap_jpg . sprintf (".%08x", rand(0xffffffff));
}
# Read the imagemap html file (if any) to get a template.
#
my $template_html = '';
{
if (open (my $in, '<', $imagemap_html)) {
local $/ = undef; # read entire file
$template_html = <$in>;
close $in;
LOG ($verbose_decode, "read template $imagemap_html");
}
if (! ($template_html =~ m/\.webcollage_box\b/s)) { # missing or old
$template_html =
'<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
"http://www.w3.org/TR/html4/loose.dtd">
<HTML>
<HEAD>
<BASE TARGET="_new">
<meta HTTP-EQUIV="Refresh" content="60" />
<TITLE>WebCollage</TITLE>
<STYLE TYPE="text/css">
<!--
body { color: #FFF; max-width: 100%; }
.webcollage_date, .webcollage_size {
display: block; margin-top: 4px; font-size: 7pt; color: #888;
}
.webcollage_date { float: left; }
.webcollage_size { float: right; }
.webcollage_frame {
overflow: hidden;
position: relative;
padding-bottom: 56.25%;
padding-bottom: 75%; /* 4:3 aspect ratio */
border: 1px solid #888;
background: #000;
}
.webcollage_box {
position: absolute; top: 0; left: 0;
border: 0; margin: 0; padding: 0;
width: 100%;
height: 100%;
}
.webcollage_box > img { width: 100%; height: 100%; border: 0; }
.webcollage_frame > a {
display: block;
position: absolute;
border-radius: 16px;
}
.webcollage_frame > a:hover {
background: rgba(1,1,1,.25);
}
-->
</STYLE>
</HEAD>
<BODY>
<DIV CLASS="webcollage_frame">
</DIV>
</BODY>
</HTML>
';
LOG ($verbose_decode, "created dummy template");
}
}
# Write the jpg to a tmp file
#
{
my @cmd;
if (defined($webcollage_helper)) {
@cmd = ('cp', '-p', $image_png, $imagemap_jpg_tmp);
} else {
@cmd = ($convert_cmd, $image_png, 'jpeg:' . $imagemap_jpg_tmp);
}
my $rc = nontrapping_system (@cmd);
if ($rc != 0) {
error ("imagemap jpeg failed: \"@cmd\"\n");
}
}
# Write the html to a tmp file
#
{
my $body = $template_html;
my $img = (" <DIV CLASS=\"webcollage_box\">" .
"<IMG SRC=\"$imagemap_jpg2\">" .
"</DIV>\n");
foreach my $a (@imagemap_areas) {
my ($x, $y, $w, $h, $u) = @$a;
$x /= $img_width / 100;
$y /= $img_height / 100;
$w /= $img_width / 100;
$h /= $img_height / 100;
foreach ($x, $y, $w, $h) { $_ = sprintf("%.1f%%", $_); }
$u = html_quote($u);
$img .= (" <A HREF=\"$u\" STYLE=\"" .
"left:$x;top:$y;width:$w;height:$h\"/>\n");
}
$img = ("<DIV CLASS=\"webcollage_frame\">\n" .
$img .
" </DIV>\n");
$body =~ s@<DIV \s+ CLASS=[\"\']webcollage_frame[\"\']>
.*? </DIV> .*? </DIV>@$img@sex ||
error ("$imagemap_html_tmp: unable to splice image");
# if there are magic webcollage spans in the html, update those too.
#
{
my @st = stat ($imagemap_jpg_tmp);
my $date = strftime("%d-%b-%Y %l:%M:%S %p %Z", localtime($st[9]));
my $size = int(($st[7] / 1024) + 0.5) . "K";
$body =~ s@(<SPAN\s+CLASS=\"webcollage_date\">).*?(</SPAN>)@$1$date$2@si;
$body =~ s@(<SPAN\s+CLASS=\"webcollage_size\">).*?(</SPAN>)@$1$size$2@si;
}
open (my $out, '>', $imagemap_html_tmp) || error("$imagemap_html_tmp: $!");
(print $out $body) || error("$imagemap_html_tmp: $!");
close ($out) || error("$imagemap_html_tmp: $!");
LOG ($verbose_decode, "wrote $imagemap_html_tmp");
}
# Rename the two tmp files to the real files
#
rename ($imagemap_html_tmp, $imagemap_html) ||
error "renaming $imagemap_html_tmp to $imagemap_html";
LOG ($verbose_decode, "wrote $imagemap_html");
if ($imagemap_jpg_tmp ne $image_png) {
rename ($imagemap_jpg_tmp, $imagemap_jpg) ||
error "renaming $imagemap_jpg_tmp to $imagemap_jpg";
LOG ($verbose_decode, "wrote $imagemap_jpg");
}
}
# Figure out what the proxy server should be, either from environment
# variables or by parsing the output of the (MacOS) program "scutil",
# which tells us what the system-wide proxy settings are.
#
sub set_proxy() {
if (! defined($http_proxy)) {
# historical suckage: the environment variable name is lower case.
$http_proxy = $ENV{http_proxy} || $ENV{HTTP_PROXY};
}
if (defined ($http_proxy)) {
if ($http_proxy && $http_proxy =~ m@^https?://([^/]*)/?$@ ) {
# historical suckage: allow "http://host:port" as well as "host:port".
$http_proxy = $1;
}
} else {
my $proxy_data = `scutil --proxy 2>/dev/null`;
my ($server) = ($proxy_data =~ m/\bHTTPProxy\s*:\s*([^\s]+)/s);
my ($port) = ($proxy_data =~ m/\bHTTPPort\s*:\s*([^\s]+)/s);
# Note: this ignores the "ExceptionsList".
if ($server) {
$http_proxy = $server;
$http_proxy .= ":$port" if $port;
}
}
delete $ENV{http_proxy};
delete $ENV{HTTP_PROXY};
delete $ENV{https_proxy};
delete $ENV{HTTPS_PROXY};
delete $ENV{PERL_LWP_ENV_PROXY};
if ($http_proxy) {
$http_proxy = 'http://' . $http_proxy;
LOG ($verbose_net, "proxy server: $http_proxy");
} else {
$http_proxy = undef; # for --proxy ''
}
}
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 subprocess filters die, we don't die.
$SIG{PIPE} = 'IGNORE';
}
END { exit_cleanup(); }
sub main() {
$| = 1;
srand(time ^ $$);
my $verbose = 0;
my $dict;
my $driftnet_cmd = 0;
$current_state = "init";
$load_method = "none";
my $root_p = 0;
my $window_id = undef;
while ($#ARGV >= 0) {
$_ = shift @ARGV;
if (m/^--?d(i(s(p(l(a(y)?)?)?)?)?)?$/s) {
$ENV{DISPLAY} = shift @ARGV;
} elsif (m/^--?root$/s) {
$root_p = 1;
} elsif (m/^--?window-id$/s) {
$window_id = shift @ARGV;
$root_p = 1;
} elsif (m/^--?no-output$/s) {
$no_output_p = 1;
} elsif (m/^--?urls(-only)?$/s) {
$urls_only_p = 1;
$no_output_p = 1;
} elsif (m/^--?cocoa$/s) {
$cocoa_p = 1;
} elsif (m/^--?imagemap$/s) {
$imagemap_base = shift @ARGV;
$no_output_p = 1;
} elsif (m/^--?verbose$/s) {
$verbose++;
} elsif (m/^-v+$/) {
$verbose += length($_)-1;
} elsif (m/^--?delay$/s) {
$delay = shift @ARGV;
} elsif (m/^--?timeout$/s) {
$http_timeout = shift @ARGV;
} elsif (m/^--?filter$/s) {
$filter_cmd = shift @ARGV;
} elsif (m/^--?filter2$/s) {
$post_filter_cmd = shift @ARGV;
} elsif (m/^--?(background|bg)$/s) {
$background = shift @ARGV;
} elsif (m/^--?size$/s) {
$_ = shift @ARGV;
if (m@^(\d+)x(\d+)$@) {
$img_width = $1;
$img_height = $2;
} else {
error "argument to \"--size\" must be of the form \"640x400\"";
}
} elsif (m/^--?(http-)?proxy$/s) {
$http_proxy = shift @ARGV;
} elsif (m/^--?dict(ionary)?$/s) {
$dict = shift @ARGV;
} elsif (m/^--?opacity$/s) {
$opacity = shift @ARGV;
error ("opacity must be between 0.0 and 1.0")
if ($opacity <= 0 || $opacity > 1);
} elsif (m/^--?driftnet$/s) {
@search_methods = ( 100, "driftnet", \&pick_from_driftnet );
if (! ($ARGV[0] =~ m/^-/)) {
$driftnet_cmd = shift @ARGV;
} else {
$driftnet_cmd = $default_driftnet_cmd;
}
} elsif (m/^--?dir(ectory)?$/s) {
@search_methods = ( 100, "local", \&pick_from_local_dir );
if (! ($ARGV[0] =~ m/^-/)) {
$local_dir = shift @ARGV;
} else {
error ("local directory path must be set")
}
} elsif (m/^--?fps$/s) {
# -fps only works on MacOS, via "webcollage-cocoa.m".
# Ignore it if passed to this script in an X11 context.
} elsif (m/^--?debug$/s) {
my $which = shift @ARGV;
my @rest = @search_methods;
my $ok = 0;
while (@rest) {
my $pct = shift @rest;
my $name = shift @rest;
my $tfn = shift @rest;
if ($name eq $which) {
@search_methods = (100, $name, $tfn);
$ok = 1;
last;
}
}
error "no such search method as \"$which\"" unless ($ok);
LOG (1, "DEBUG: using only \"$which\"");
$report_performance_interval = 30;
} else {
print STDERR "unknown option: $_\n\n";
print STDERR "$copyright\nusage: $progname " .
"[--root] [--display dpy] [--verbose] [--debug which]\n" .
"\t\t [--timeout secs] [--delay secs] [--size WxH]\n" .
"\t\t [--no-output] [--urls-only] [--imagemap filename]\n" .
"\t\t [--background color] [--opacity f]\n" .
"\t\t [--filter cmd] [--filter2 cmd]\n" .
"\t\t [--dictionary dictionary-file] [--http-proxy host[:port]]\n" .
"\t\t [--driftnet [driftnet-program-and-args]]\n" .
"\t\t [--directory local-image-directory]\n" .
"\n";
exit 1;
}
}
if (!$root_p && !$no_output_p && !$cocoa_p) {
print STDERR $copyright;
error "the --root argument is mandatory (for now.)";
}
if (!$no_output_p && !$cocoa_p && !$ENV{DISPLAY}) {
error "\$DISPLAY is not set.";
}
if ($verbose == 1) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
} elsif ($verbose == 2) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
} elsif ($verbose == 3) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
$verbose_filter = 1;
} elsif ($verbose == 4) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
$verbose_filter = 1;
$verbose_net = 1;
} elsif ($verbose == 5) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
$verbose_filter = 1;
$verbose_net = 1;
$verbose_decode = 1;
} elsif ($verbose == 6) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
$verbose_filter = 1;
$verbose_net = 1;
$verbose_decode = 1;
$verbose_http = 1;
} elsif ($verbose >= 7) {
$verbose_imgmap = 1;
$verbose_warnings = 1;
$verbose_load = 1;
$verbose_filter = 1;
$verbose_net = 1;
$verbose_decode = 1;
$verbose_http = 1;
$verbose_exec = 1;
}
if ($dict) {
error ("$dict does not exist") unless (-f $dict);
$wordlist = $dict;
} else {
pick_dictionary();
}
if ($imagemap_base && !($img_width && $img_height)) {
error ("--size WxH is required with --imagemap");
}
if (defined ($local_dir)) {
$_ = "xscreensaver-getimage-file";
which ($_) || error "$_ not found on \$PATH.";
}
init_signals();
set_proxy();
spawn_driftnet ($driftnet_cmd) if ($driftnet_cmd);
if ($urls_only_p) {
url_only_output ();
} else {
x_or_image_output ($window_id);
}
}
main();
exit (0);
