#!/usr/bin/perl -w
#
-# webcollage, Copyright (c) 1999, 2000 by Jamie Zawinski <jwz@jwz.org>
+# webcollage, Copyright (c) 1999-2004 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."
#
# 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
+# 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':
#
-# default-n: webcollage -root \n\
-# default-n: webcollage -root -filter 'vidwhacker -stdin -stdout' \n\
+# webcollage -root
+# webcollage -root -filter 'vidwhacker -stdin -stdout'
+#
+#
+# You can see this in action at http://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
+#
+#
+# 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 diagnostics;
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;
+
+
use Socket;
require Time::Local;
require POSIX;
use Fcntl ':flock'; # import LOCK_* constants
use POSIX qw(strftime);
+use bytes; # Larry can take Unicode and shove it up his ass sideways.
+ # Perl 5.8.0 causes us to start getting incomprehensible
+ # errors about UTF-8 all over the place without this.
-my $version = q{ $Revision: 1.65 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
-my $copyright = "WebCollage $version, Copyright (c) 1999" .
+
+my $progname = $0; $progname =~ s@.*/@@g;
+my $version = q{ $Revision: 1.117 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
+my $copyright = "WebCollage $version, Copyright (c) 1999-2004" .
" Jamie Zawinski <jwz\@jwz.org>\n" .
- " http://www.jwz.org/xscreensaver/\n";
-
-my $argv0 = $0;
-my $progname = $argv0; $progname =~ s@.*/@@g;
-
-my $random_redirector = "http://random.yahoo.com/bin/ryl";
-my $image_randomizer_1 = "http://www.altavista.com/query" .
- "?mmdo=3" .
- "&nbq=12" .
- "&stype=simage" .
- "&iclr=1" .
- "&ibw=1" .
- "&iexc=1" .
- "&what=web" .
- "&q=";
-my $image_randomizer_2 = "http://www.hotbot.com/?clickSrc=search" .
- "&submit=SEARCH&SM=SC&LG=any" .
- "&AM0=MC&AT0=words&AW0=" .
- "&AM1=MN&AT1=words&AW1=" .
- "&savenummod=2&date=within" .
- "&DV=0&DR=newer&DM=1&DD=1&DY=99&FVI=1&FS=&RD=RG" .
- "&RG=all&Domain=&PS=A&PD=&STEM=1&DC=50&DE=0&_v=2" .
- "&OPs=MDRTP&NUMMOD=2" .
- "&MT=";
-my $image_randomizer_3 = "http://www.altavista.com/cgi-bin/query?pg=q" .
- "&text=yes&kl=XX&stype=stext&q=";
-my $image_randomizer_4 = "http://search.news.yahoo.com/search/news_photos?" .
- "&z=&n=100&o=o&2=&3=&p=";
-
-# I guess Photopoint got wise to me, because now they are doing error
-# checking on the user ("u=") and album ("a=") parameters. Oh well.
-#
-#my $photo_randomizer = "http://albums.photopoint.com/j/View?u=1&a=1&p=";
-#my $photo_randomizer_lo = 10000001;
-#my $photo_randomizer_hi = 12400000;
-
-my $image_ppm = ($ENV{TMPDIR} ? $ENV{TMPDIR} : "/tmp") . "/webcollage." . $$;
-my $image_tmp1 = $image_ppm . "-1";
-my $image_tmp2 = $image_ppm . "-2";
-
-my $img_width; # size of the image being generated.
-my $img_height;
+ " http://www.jwz.org/webcollage/\n";
-my $http_proxy = undef;
-my $http_timeout = 30;
-my $cvt_timeout = 10;
+
+
+my @search_methods = ( 71, "altavista", \&pick_from_alta_vista_random_link,
+ 10, "livejournal", \&pick_from_livejournal_images,
+ 8, "yahoorand", \&pick_from_yahoo_random_link,
+ 6, "yahoonews", \&pick_from_yahoo_news_text,
+ 5, "ircimages", \&pick_from_ircimages,
+
+ # 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,
+
+ # Google asked (nicely) for 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!
+ #
+ # 0, "googlenums", \&pick_from_google_image_numbers,
+ # 0, "googleimgs", \&pick_from_google_images,
+
+ # I suspect Hotbot is actually the same search engine
+ # data as Lycos.
+ #
+ # 0, "hotbot", \&pick_from_hotbot_text,
+
+ # Eh, Lycos sucks anyway.
+ # 0, "lycos", \&pick_from_lycos_text,
+ );
# programs we can use to write to the root window (tried in ascending order.)
-my $ppm_to_root_window_cmd_1 = "xloadimage -onroot -quiet %%PPM%%";
-my $ppm_to_root_window_cmd_2 = "xli -quiet -onroot -center" .
- " -border black %%PPM%%";
-my $ppm_to_root_window_cmd_3 = "xv -root -rmode 5 -viewonly" .
- " +noresetroot %%PPM%% -quit";
+#
+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",
-my $ppm_to_root_window_cmd = undef; # initialized by x_output()
+# this lame program wasn't built with vroot.h:
+# "xsri -scale -keep-aspect -center-horizontal -center-vertical",
+);
-my $filter_cmd = undef;
-my $post_filter_cmd = undef;
-my $background = undef;
-my $no_output_p = 0;
-my $urls_only_p = 0;
-my $delay = 0;
-my $wordlist = "/usr/dict/words";
+# 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",
+
+ # log in as "cipherpunk"
+ "www.nytimes.com" => 'NYT-S=18cHMIlJOn2Y1bu5xvEG3Ufuk6E1oJ.' .
+ 'FMxWaQV0igaB5Yi/Q/guDnLeoL.pe7i1oakSb' .
+ '/VqfdUdb2Uo27Vzt1jmPn3cpYRlTw9',
-if (!-r $wordlist) {
- $wordlist = "/usr/share/dict/words"; # BSD
-}
-if (!-r $wordlist) {
- $wordlist = "/usr/share/lib/dict/words"; # Irix
-}
-die "$wordlist doesn't exist!\n" unless (-r $wordlist);
+ "ircimages.com" => 'disclaimer=1',
+);
+
+
+# If this is set, it's a helper program to use for pasting images together:
+# this is a lot faster and more efficient than using PPM pipelines, which is
+# what we do if this program doesn't exist. (We check for "webcollage-helper"
+# on $PATH at startup, and set this variable appropriately.)
+#
+my $webcollage_helper = undef;
+
+
+# If we have the webcollage-helper program, then it will paste the images
+# together with transparency! 0.0 is invisible, 1.0 is totally opaque.
+#
+my $opacity = 0.85;
+
+
+# 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.
+);
+
+
+# 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, # Lots of home pages here.
+ "www.geocities.com" => 1,
+ "www.angelfire.com" => 1,
+ "members.aol.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.
+
+ "driftnet" => 1, # builtin...
+);
+##############################################################################
+#
+# 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_pbm = 0; # diagnostics about PBM 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 = 30;
+my $cvt_timeout = 10;
+
my $min_width = 50;
my $min_height = 50;
my $min_ratio = 1/5;
-my $verbose = 0;
+my $min_gif_area = (120 * 120);
+
+
+my $no_output_p = 0;
+my $urls_only_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 $wordlist;
my %rejected_urls;
my @tripwire_words = ("aberrate", "abode", "amorphous", "antioch",
"eggplant");
-sub blurb {
- return "$progname: " . strftime ("%H:%M:%S: ", localtime);
-}
-
-
##############################################################################
#
# Retrieving URLs
# and the document body.
#
sub get_document_1 {
- my ( $url, $referer, $timeout ) = @_;
+ my ( $url, $referer, $timeout ) = @_;
- if (!defined($timeout)) { $timeout = $http_timeout; }
- if ($timeout <= 0) { return (); }
- if ($timeout > $http_timeout) { $timeout = $http_timeout; }
+ if (!defined($timeout)) { $timeout = $http_timeout; }
+ if ($timeout > $http_timeout) { $timeout = $http_timeout; }
- if ( $verbose > 3 ) {
- print STDERR blurb() . "get_document_1 $url " .
- ($referer ? $referer : "") . "\n";
- }
-
- my($url_proto, $dummy, $serverstring, $path) = split(/\//, $url, 4);
- if (! ($url_proto && $url_proto =~ m/^http:$/i)) {
- if ($verbose) { print STDERR blurb() . "not an HTTP URL: $url\n"; }
- return ();
- }
+ if ($timeout <= 0) {
+ LOG (($verbose_net || $verbose_load), "timed out for $url");
+ return ();
+ }
- $path = "" unless $path;
+ LOG ($verbose_net, "get_document_1 $url " . ($referer ? $referer : ""));
- my($them,$port) = split(/:/, $serverstring);
- $port = 80 unless $port;
+ if (! ($url =~ m@^http://@i)) {
+ LOG ($verbose_net, "not an HTTP URL: $url");
+ return ();
+ }
- my $them2 = $them;
- my $port2 = $port;
- if ($http_proxy) {
- $serverstring = $http_proxy if $http_proxy;
- ($them2,$port2) = split(/:/, $serverstring);
- $port2 = 80 unless $port2;
- }
+ my ($url_proto, $dummy, $serverstring, $path) = split(/\//, $url, 4);
+ $path = "" unless $path;
- my ($remote, $iaddr, $paddr, $proto, $line);
- $remote = $them2;
- if ($port2 =~ /\D/) { $port2 = getservbyname($port2, 'tcp') }
- return unless $port2;
- $iaddr = inet_aton($remote) || return;
- $paddr = sockaddr_in($port2, $iaddr);
+ if (!$url_proto || !$serverstring) {
+ LOG (($verbose_net || $verbose_load), "unparsable URL: $url");
+ return ();
+ }
+
+ my ($them,$port) = split(/:/, $serverstring);
+ $port = 80 unless $port;
+
+ my $them2 = $them;
+ my $port2 = $port;
+ if ($http_proxy) {
+ $serverstring = $http_proxy if $http_proxy;
+ $serverstring =~ s@^[a-z]+://@@;
+ ($them2,$port2) = split(/:/, $serverstring);
+ $port2 = 80 unless $port2;
+ }
+
+ my ($remote, $iaddr, $paddr, $proto, $line);
+ $remote = $them2;
+ if ($port2 =~ /\D/) { $port2 = getservbyname($port2, 'tcp') }
+ if (!$port2) {
+ LOG (($verbose_net || $verbose_load), "unrecognised port in $url");
+ return ();
+ }
+ $iaddr = inet_aton($remote);
+ if (!$iaddr) {
+ LOG (($verbose_net || $verbose_load), "host not found: $remote");
+ return ();
+ }
+ $paddr = sockaddr_in($port2, $iaddr);
- my $head = "";
- my $body = "";
+ my $head = "";
+ my $body = "";
- @_ =
+ @_ =
eval {
- local $SIG{ALRM} = sub {
- if ($verbose > 0) {
- print STDERR blurb() . "timed out ($timeout) for $url\n";
- }
- die "alarm\n"
- };
- alarm $timeout;
-
- $proto = getprotobyname('tcp');
- if (!socket(S, PF_INET, SOCK_STREAM, $proto)) {
- print STDERR blurb() . "socket: $!\n" if ($verbose);
- return;
- }
- if (!connect(S, $paddr)) {
- print STDERR blurb() . "connect($serverstring): $!\n"
- if ($verbose);
- return;
- }
+ local $SIG{ALRM} = sub {
+ LOG (($verbose_net || $verbose_load), "timed out ($timeout) for $url");
+ die "alarm\n";
+ };
+ alarm $timeout;
+
+ $proto = getprotobyname('tcp');
+ if (!socket(S, PF_INET, SOCK_STREAM, $proto)) {
+ LOG (($verbose_net || $verbose_load), "socket: $!");
+ return ();
+ }
+ if (!connect(S, $paddr)) {
+ LOG (($verbose_net || $verbose_load), "connect($serverstring): $!");
+ return ();
+ }
- select(S); $| = 1; select(STDOUT);
+ select(S); $| = 1; select(STDOUT);
- my $cookie;
- if ($remote =~ m/\baltavista\.com$/i) {
- # kludge to tell the various altavista sites to be uncensored.
- $cookie = "AV_ALL=1";
- }
+ my $cookie = $cookies{$them};
- print S ("GET " . ($http_proxy ? $url : "/$path") . " HTTP/1.0\r\n" .
+ my $user_agent = "$progname/$version";
+
+ if ($url =~ m@^http://www\.altavista\.com/@ ||
+ $url =~ m@^http://random\.yahoo\.com/@ ||
+ $url =~ m@^http://images\.google\.com/@) {
+ # block this, you turkeys.
+ $user_agent = "Mozilla/4.76 [en] (X11; U; Linux 2.2.16-22 i686; Nav)";
+ }
+
+ my $hdrs = "GET " . ($http_proxy ? $url : "/$path") . " HTTP/1.0\r\n" .
"Host: $them\r\n" .
- "User-Agent: $progname/$version\r\n" .
- ($referer ? "Referer: $referer\r\n" : "") .
- ($cookie ? "Cookie: $cookie\r\n" : "") .
- "\r\n");
- my $http = <S>;
-
- while (<S>) {
- $head .= $_;
- last if m@^[\r\n]@;
- }
- while (<S>) {
- $body .= $_;
+ "User-Agent: $user_agent\r\n";
+ if ($referer) {
+ $hdrs .= "Referer: $referer\r\n";
+ }
+ if ($cookie) {
+ my @cc = split(/\r?\n/, $cookie);
+ $hdrs .= "Cookie: " . join('; ', @cc) . "\r\n";
+ }
+ $hdrs .= "\r\n";
+
+ foreach (split('\r?\n', $hdrs)) {
+ LOG ($verbose_http, " ==> $_");
+ }
+ print S $hdrs;
+ my $http = <S> || "";
+
+ # Kludge: the Yahoo Random Link is now returning as its first
+ # line "Status: 301" instead of "HTTP/1.0 301 Found". Fix it...
+ #
+ $http =~ s@^Status:\s+(\d+)\b@HTTP/1.0 $1@i;
+
+ $_ = $http;
+ s/[\r\n]+$//s;
+ LOG ($verbose_http, " <== $_");
+
+ while (<S>) {
+ $head .= $_;
+ s/[\r\n]+$//s;
+ last if m@^$@;
+ LOG ($verbose_http, " <== $_");
+
+ if (m@^Set-cookie:\s*([^;\r\n]+)@i) {
+ set_cookie($them, $1)
}
+ }
- close S;
+ my $lines = 0;
+ while (<S>) {
+ $body .= $_;
+ $lines++;
+ }
- if ( $verbose > 3 ) {
- print STDERR blurb() . " ==> $http\n";
- }
+ LOG ($verbose_http,
+ " <== [ body ]: $lines lines, " . length($body) . " bytes");
- return ( $http, $head, $body );
- };
- die if ($@ && $@ ne "alarm\n"); # propagate errors
- if ($@) {
- # timed out
- $head = undef;
- $body = undef;
+ close S;
+
+ if (!$http) {
+ LOG (($verbose_net || $verbose_load), "null response: $url");
return ();
- } else {
- # didn't
- alarm 0;
- return @_;
- }
+ }
+
+ return ( $http, $head, $body );
+ };
+ die if ($@ && $@ ne "alarm\n"); # propagate errors
+ if ($@) {
+ # timed out
+ $head = undef;
+ $body = undef;
+ $suppress_audit = 1;
+ return ();
+ } else {
+ # didn't
+ alarm 0;
+ return @_;
+ }
}
# if the given URL did a redirect, returns the redirected-to document.
#
sub get_document {
- my ( $url, $referer, $timeout ) = @_;
- my $start = time;
-
- my $orig_url = $url;
- my $loop_count = 0;
- my $max_loop_count = 4;
-
- do {
- if (defined($timeout) && $timeout <= 0) { return (); }
-
- my ( $http, $head, $body ) = get_document_1 ($url, $referer, $timeout);
+ my ( $url, $referer, $timeout ) = @_;
+ my $start = time;
- if (defined ($timeout)) {
- my $now = time;
- my $elapsed = $now - $start;
- $timeout -= $elapsed;
- $start = $now;
- }
-
- return () if ( ! $body );
-
- if ( $http =~ m@HTTP/[0-9.]+ 30[23]@ ) {
- $_ = $head;
- my ( $location ) = m@^location:[ \t]*(.*)$@im;
- if ( $location ) {
- $location =~ s/[\r\n]$//;
-
- if ( $verbose > 3 ) {
- print STDERR blurb() . "redirect from " .
- "$url to $location\n";
- }
- $referer = $url;
- $url = $location;
-
- if ($url =~ m@^/@) {
- $referer =~ m@^(http://[^/]+)@i;
- $url = $1 . $url;
- } elsif (! ($url =~ m@^[a-z]+:@i)) {
- $_ = $referer;
- s@[^/]+$@@g if m@^http://[^/]+/@i;
- $_ .= "/" if m@^http://[^/]+$@i;
- $url = $_ . $url;
- }
-
- } else {
- return ( $url, $body );
- }
-
- if ($loop_count++ > $max_loop_count) {
- if ( $verbose > 1 ) {
- print STDERR blurb() . "too many redirects " .
- "($max_loop_count) from $orig_url\n";
- }
- $body = undef;
- return ();
- }
-
- } elsif ( $http =~ m@HTTP/[0-9.]+ [4-9][0-9][0-9]@ ) {
- # http errors -- return nothing.
- $body = undef;
- return ();
-
- } else {
-
- return ( $url, $body );
- }
-
- } while (1);
-}
-
-
-# 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 ) = @_;
+ if (defined($referer) && $referer eq $driftnet_magic) {
+ return get_driftnet_file ($url);
+ }
- my $base = $url;
- $_ = $url;
+ my $orig_url = $url;
+ my $loop_count = 0;
+ my $max_loop_count = 4;
- # if there's at least one slash after the host, take off the last
- # pathname component
- if ( m@^http://[^/]+/@io ) {
- $base =~ s@[^/]+$@@go;
+ do {
+ if (defined($timeout) && $timeout <= 0) {
+ LOG (($verbose_net || $verbose_load), "timed out for $url");
+ $suppress_audit = 1;
+ return ();
}
- # if there are no slashes after the host at all, put one on the end.
- if ( m@^http://[^/]+$@io ) {
- $base .= "/";
- }
+ my ( $http, $head, $body ) = get_document_1 ($url, $referer, $timeout);
- if ( $verbose > 3 ) {
- print STDERR blurb() . "base is $base\n";
+ if (defined ($timeout)) {
+ my $now = time;
+ my $elapsed = $now - $start;
+ $timeout -= $elapsed;
+ $start = $now;
}
+ return () unless $http; # error message already printed
- $_ = $body;
+ $http =~ s/[\r\n]+$//s;
- # strip out newlines, compress whitespace
- s/[\r\n\t ]+/ /go;
+ if ( $http =~ m@^HTTP/[0-9.]+ 30[123]@ ) {
+ $_ = $head;
- # nuke comments
- s/<!--.*?-->//go;
+ my ( $location ) = m@^location:[ \t]*(.*)$@im;
+ if ( $location ) {
+ $location =~ s/[\r\n]$//;
+ LOG ($verbose_net, "redirect from $url to $location");
+ $referer = $url;
+ $url = $location;
- # 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 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) {
- if ($verbose > 1) {
- print STDERR blurb() . "there is probably a dictionary in" .
- " \"$url\": rejecting.\n";
+ if ($url =~ m@^/@) {
+ $referer =~ m@^(http://[^/]+)@i;
+ $url = $1 . $url;
+ } elsif (! ($url =~ m@^[a-z]+:@i)) {
+ $_ = $referer;
+ s@[^/]+$@@g if m@^http://[^/]+/@i;
+ $_ .= "/" if m@^http://[^/]+$@i;
+ $url = $_ . $url;
}
- $rejected_urls{$url} = -1;
+
+ } 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;
- $_ = undef;
return ();
- }
+ }
+ } elsif ( $http =~ m@^HTTP/[0-9.]+ ([4-9][0-9][0-9].*)$@ ) {
- my @urls;
- my %unique_urls;
-
- foreach (split(/ *</)) {
- if ( m/^meta /i ) {
-
- # Likewise, reject any web pages that have a KEYWORDS meta tag
- # that is too long.
- #
- if (m/name ?= ?\"?keywords\"?/i &&
- m/content ?= ?\"([^\"]+)\"/) {
- my $L = length($1);
- if ($L > 1000) {
- if ($verbose > 1) {
- print STDERR blurb() . "keywords of" .
- " length $L in $url: rejecting.\n";
- }
- $rejected_urls{$url} = $L;
- $body = undef;
- $_ = undef;
- return ();
- } elsif ( $verbose > 2 ) {
- print STDERR blurb() . "keywords of length $L" .
- " in $url (ok.)\n";
- }
- }
-
- } elsif ( m/^(img|a) .*(src|href) ?= ?\"? ?(.*?)[ >\"]/io ) {
-
- my $was_inline = ( "$1" eq "a" || "$1" eq "A" );
- my $link = $3;
- my ( $width ) = m/width ?=[ \"]*(\d+)/oi;
- my ( $height ) = m/height ?=[ \"]*(\d+)/oi;
- $_ = $link;
-
- if ( m@^/@o ) {
- my $site;
- ( $site = $base ) =~ s@^(http://[^/]*).*@$1@gio;
- $_ = "$site$link";
- } elsif ( ! m@^[^/:?]+:@ ) {
- $_ = "$base$link";
- s@/\./@/@g;
- while (s@/\.\./@/@g) {
- }
- }
-
- # skip non-http
- if ( ! m@^http://@io ) {
- next;
- }
-
- # skip non-image
- if ( ! m@[.](gif|jpg|jpeg|pjpg|pjpeg)$@io ) {
- next;
- }
-
- # skip really short or really narrow images
- if ( $width && $width < $min_width) {
- if ( $verbose > 2 ) {
- if (!$height) { $height = "?"; }
- print STDERR blurb() . "skip narrow image " .
- "$_ (${width}x$height)\n";
- }
- next;
- }
-
- if ( $height && $height < $min_height) {
- if ( $verbose > 2 ) {
- if (!$width) { $width = "?"; }
- print STDERR blurb() . "skip short image " .
- "$_ (${width}x$height)\n";
- }
- next;
- }
-
- # skip images with ratios that make them look like banners.
- if ( $min_ratio && $width && $height &&
- ($width * $min_ratio ) > $height ) {
- if ( $verbose > 2 ) {
- if (!$height) { $height = "?"; }
- print STDERR blurb() . "skip bad ratio " .
- "$_ (${width}x$height)\n";
- }
- next;
- }
-
- my $url = $_;
-
- if ( $unique_urls{$url} ) {
- if ( $verbose > 2 ) {
- print STDERR blurb() . "skip duplicate image $_\n";
- }
- next;
- }
-
- if ( $verbose > 2 ) {
- print STDERR blurb() . "got $url" .
- ($width && $height ? " (${width}x${height})" : "") .
- ($was_inline ? " (inline)" : "") . "\n";
- }
-
- $urls[++$#urls] = $url;
- $unique_urls{$url}++;
-
- # jpegs are preferable to gifs.
- $_ = $url;
- if ( ! m@[.]gif$@io ) {
- $urls[++$#urls] = $url;
- }
-
- # pointers to images are preferable to inlined images.
- if ( ! $was_inline ) {
- $urls[++$#urls] = $url;
- $urls[++$#urls] = $url;
- }
- }
- }
+ LOG (($verbose_net || $verbose_load), "failed: $1 ($url)");
- $_ = undef;
- $body = undef;
+ # http errors -- return nothing.
+ $body = undef;
+ return ();
- if ( $#urls == 0 ) {
- if ( $verbose > 2 ) {
- print STDERR blurb() . "no images on $base\n";
- }
- return ();
- }
+ } elsif (!$body) {
- return () if ( $#urls < 1 );
+ LOG (($verbose_net || $verbose_load), "document contains no data: $url");
+ return ();
- # pick a random element of the table
- my $i = ((rand() * 99999) % $#urls);
- $url = $urls[$i];
+ } else {
- if ( $verbose > 2 ) {
- print STDERR blurb() . "picked $url\n";
+ # ok!
+ return ( $url, $body );
}
- return $url;
+ } while (1);
}
-
-# Using the URL-randomizer, 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.
+# 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 pick_from_url_randomizer {
- my ( $timeout ) = @_;
+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");
+ }
+}
- if ( $verbose > 3 ) {
- print STDERR "\n\n$progname: picking from $random_redirector...\n\n";
- }
- my ( $base, $body ) = get_document ($random_redirector, undef, $timeout);
+############################################################################
+#
+# Extracting image URLs from HTML
+#
+############################################################################
- if (!$base || !$body) {
- $body = undef;
- return;
- }
- my $img = pick_image_from_body ($base, $body);
+# 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@^http://[^/]+/@io ) {
+ $base =~ s@[^/]+$@@go;
+ }
+
+ # if there are no slashes after the host at all, put one on the end.
+ if ( m@^http://[^/]+$@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 /i ) {
+
+ # Likewise, reject any web pages that have a KEYWORDS meta tag
+ # that is too long.
+ #
+ if (m/name ?= ?\"?keywords\"?/i &&
+ m/content ?= ?\"([^\"]+)\"/) {
+ my $L = length($1);
+ 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) .*(src|href) ?= ?\"? ?(.*?)[ >\"]/io ) {
+
+ my $was_inline = (! ( "$1" eq "a" || "$1" eq "A" ));
+ my $link = $3;
+ my ( $width ) = m/width ?=[ \"]*(\d+)/oi;
+ my ( $height ) = m/height ?=[ \"]*(\d+)/oi;
+ $_ = $link;
+
+ if ( m@^/@o ) {
+ my $site;
+ ( $site = $base ) =~ s@^(http://[^/]*).*@$1@gio;
+ $_ = "$site$link";
+ } elsif ( ! m@^[^/:?]+:@ ) {
+ $_ = "$base$link";
+ s@/\./@/@g;
+ 1 while (s@/[^/]+/\.\./@/@g);
+ }
+
+ # skip non-http
+ if ( ! m@^http://@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;
+ }
+
+
+ my $url = $_;
+
+ if ($unique_urls{$url}) {
+ LOG ($verbose_filter, " skip duplicate image $_");
+ next;
+ }
+
+ LOG ($verbose_filter,
+ " image $url" .
+ ($width && $height ? " (${width}x${height})" : "") .
+ ($was_inline ? " (inline)" : ""));
+
+ $urls[++$#urls] = $url;
+ $unique_urls{$url}++;
+
+ # JPEGs are preferable to GIFs and PNGs.
+ $_ = $url;
+ if ( ! m@[.](gif|png)$@io ) {
+ $urls[++$#urls] = $url;
+ }
+
+ # pointers to images are preferable to inlined images.
+ if ( ! $was_inline ) {
+ $urls[++$#urls] = $url;
+ $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 ();
+ }
- if ($img) {
- return ($base, $img, "yahoo");
- } else {
- 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;
}
+\f
+############################################################################
+#
+# 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");
+ 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 {
-
my $word = 0;
if (open (IN, "<$wordlist")) {
my $size = (stat(IN))[7];
$word = <IN>; # toss partial line
$word = <IN>; # keep next line
}
- if (!$word) {
+ if (!$word) {
seek( IN, 0, 0 );
$word = <IN>;
- }
+ }
close (IN);
}
$word =~ tr/A-Z/a-z/;
if ( $word =~ s/[ \t\n\r]/\+/g ) { # convert intra-word spaces to "+".
- $word = "\%22$word\%22"; # And put quotes (%22) around it.
+ $word = "\%22$word\%22"; # And put quotes (%22) around it.
}
return $word;
}
+sub random_words {
+ my ($or_p) = @_;
+ my $sep = ($or_p ? "%20OR%20" : "%20");
+ return (random_word . $sep .
+ random_word . $sep .
+ random_word . $sep .
+ random_word . $sep .
+ random_word);
+}
-# Using the image-randomizer, 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_from_image_randomizer {
- my ( $timeout, $which ) = @_;
-
- my $words = random_word;
- $words .= "%20" . random_word;
- $words .= "%20" . random_word;
- $words .= "%20" . random_word;
- $words .= "%20" . random_word;
-
- my $search_url = ($which == 0 ? $image_randomizer_1 :
- $which == 1 ? $image_randomizer_2 :
- $which == 2 ? $image_randomizer_3 :
- $image_randomizer_4) .
- $words;
-
- # Pick a random search-result page instead of always taking the first.
- # This assumes there are at least 10 pages...
- if ($which == 0) {
- $search_url .= "&pgno=" . (int(rand(9)) + 1);
- } elsif ($which == 2) {
- $search_url .= "&stq=" . (10 * (int(rand(9)) + 1));
- }
-
- if ( $verbose > 3 ) {
- $_ = $words; s/%20/ /g; print STDERR blurb() . "search words: $_\n";
- }
-
- if ( $verbose > 3 ) {
- print STDERR "\n\n$progname: picking from $search_url\n";
- }
+sub url_quote {
+ my ($s) = @_;
+ $s =~ s|([^-a-zA-Z0-9.\@/_\r\n])|sprintf("%%%02X", ord($1))|ge;
+ return $s;
+}
- my $start = time;
- my ( $base, $body ) = get_document ($search_url, undef, $timeout);
- if (defined ($timeout)) {
- $timeout -= (time - $start);
- if ($timeout <= 0) {
- $body = undef;
- return ();
- }
- }
+sub url_unquote {
+ my ($s) = @_;
+ $s =~ s/[+]/ /g;
+ $s =~ s/%([a-z0-9]{2})/chr(hex($1))/ige;
+ return $s;
+}
- return () if (! $body);
+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/</</gi; # far from exhaustive...
+ $s =~ s/>/</gi;
+ $s =~ s/"/\"/gi;
+ $s =~ s/&/&/gi;
+ return $s;
+}
- my @subpages;
- my $skipped = 0;
- my $search_count = "?";
- if ($which == 0 &&
- $body =~ m@found (approximately |about )?(<B>)?(\d+)(</B>)? image@) {
- $search_count = $3;
- } elsif ($which == 1 && $body =~ m@<NOBR>((\d{1,3})(,\d{3})*) @i) {
- $search_count = $1;
- } elsif ($which == 2 && $body =~ m@found ((\d{1,3})(,\d{3})*|\d+) Web p@) {
- $search_count = $1;
- }
- 1 while ($search_count =~ s/^(\d+)(\d{3})/$1,$2/);
+# 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") {
+# local *OUT;
+# my $file = "/tmp/wc.html";
+# open(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;
+
+ if ($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 ();
+ }
- my $length = length($body);
- my $href_count = 0;
+ LOG ($verbose_filter, "" . $#subpages+1 . " links on $search_url");
- $_ = $body;
+ return ($search_count, @subpages);
+}
-# s/Result [Pp]ages:.*$//s; # trim off page footer
-# s/^.*?IMAGE RESULTS//s; # trim off page header
- s/Have you tried these resources.*//s; # let's try it again
+sub depoison {
+ my (@urls) = @_;
+ my @urls2 = ();
+ foreach (@urls) {
+ my ($h) = m@^http://([^/: \t\r\n]+)@i;
- s/[\r\n\t ]+/ /g;
+ next unless defined($h);
- s/(<A )/\n$1/gi;
- foreach (split(/\n/)) {
- $href_count++;
- 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
+ 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;
+ }
- if ($which == 1) {
- # Kludge to decode HotBot pages
- next unless ($u =~ m@/director\.asp\?target=(http%3A[^&>]+)@);
- $u = url_decode($1);
- }
+ push @urls2, $_;
+ }
+ return @urls2;
+}
- next unless ($u =~ m@^http://@i); # skip non-http and relative urls.
-
- next if ($u =~ m@[/.]altavista\.com\b@i); # skip altavista builtins
- next if ($u =~ m@[/.]altavista\.[a-z]{2}\b@i); # altavista.fr, etc
- next if ($u =~ m@[/.]av\.com\b@i);
- next if ($u =~ m@[/.]virage\.com\b@i);
- next if ($u =~ m@[/.]photoloft\.com\b@i);
- next if ($u =~ m@[/.]shopping\.com\b@i);
- next if ($u =~ m@[/.]thetrip\.com\b@i);
- next if ($u =~ m@[/.]cmgi\.com\b@i);
- next if ($u =~ m@[/.]intelihealth\.com\b@i);
- next if ($u =~ m@[/.]wildweb\.com\b@i);
- next if ($u =~ m@[/.]digital\.com\b@i);
- next if ($u =~ m@[/.]doubleclick\.net\b@i);
- next if ($u =~ m@[/.]freeim\.org\b@i);
- next if ($u =~ m@[/.]clicktomarket\.com\b@i); # you cretins
- next if ($u =~ m@[/.]teragram\.com\b@i);
-
- # must lose this one for altavista, even though it loses images of
- # every single customer of akamai. Oh well, those people have lots
- # of money, and so their images are probably boring anyway.
- next if ($u =~ m@[/.]akamai\.net@i);
-
- if ($which == 0 && $u =~ m@[/.]corbis\.com@) {
- $skipped = 1;
- if ( $verbose > 3 ) {
- print STDERR blurb() . "skipping corbis URL: $u\n";
- }
- next;
-
- } elsif ($which == 3 &&
- ($u =~ m@^http://[^/]+$@ || # no slashes
- $u =~ m@/$@ || # ends in /
- ! ($u =~ m@dailynews\.yahoo\.com@))) { # not dailynews
-# $skipped = 1;
- if ( $verbose > 3 ) {
- print STDERR blurb() . "skipping non-AP URL: $u\n";
- }
- next;
-
- } elsif ( $rejected_urls{$u} ) {
- if ( $verbose > 3 ) {
- my $L = $rejected_urls{$u};
- print STDERR blurb() . "pre-rejecting sub-page: $u\n";
- }
- next;
-
- } elsif ( $verbose > 3 ) {
- print STDERR blurb() . "sub-page: $u\n";
- }
- $subpages[++$#subpages] = $u;
- }
+# 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;
+# and a debugging description string.
+#
+sub pick_image_from_pages {
+ my ($base, $total_hit_count, $unfiltered_link_count, $timeout, @pages) = @_;
- if ( $#subpages < 0 ) {
- if (!$skipped && $verbose > 1) {
- print STDERR blurb() . "found nothing on $base " .
- "($length bytes, $href_count links).\n";
- }
- $body = undef;
- $_ = undef;
- return ();
- }
+ $total_hit_count = "?" unless defined($total_hit_count);
- # pick a random element of the table
- my $i = ((rand() * 99999) % ($#subpages + 1));
- my $subpage = $subpages[$i];
+ @pages = depoison (@pages);
+ LOG ($verbose_load,
+ "" . ($#pages+1) . " candidates of $unfiltered_link_count links" .
+ " ($total_hit_count total)");
- if ( $verbose > 3 ) {
- print STDERR blurb() . "picked page $subpage\n";
- }
+ return () if ($#pages < 0);
+ my $i = int(rand($#pages+1));
+ my $page = $pages[$i];
- $body = undef;
- $_ = undef;
+ LOG ($verbose_load, "picked page $page");
- my ( $base2, $body2 ) = get_document ($subpage, $base, $timeout);
+ $suppress_audit = 1;
- if (!$base2 || !$body2) {
- $body2 = undef;
- return ();
- }
+ my ( $base2, $body2 ) = get_document ($page, $base, $timeout);
- my $img = pick_image_from_body ($base2, $body2);
+ if (!$base2 || !$body2) {
$body2 = undef;
+ return ();
+ }
- if ($img) {
- return ($base2, $img,
- ($which == 0 ? "imagevista" :
- $which == 1 ? "hotbot" :
- $which == 2 ? "altavista" :
- "ap") .
- "/$search_count");
- } else {
- return ();
- }
-}
+ my $img = pick_image_from_body ($base2, $body2);
+ $body2 = undef;
+ if ($img) {
+ return ($base2, $img);
+ } else {
+ return ();
+ }
+}
-# Using the photo site, generate a random URL that will hopefully point
-# to an image. Returns two URLs, both of which are the URL of the image.
-# Returns () if nothing found this time.
+\f
+############################################################################
#
-#sub pick_from_photo_randomizer {
-# my ( $timeout ) = @_;
-# my $n = ($photo_randomizer_lo +
-# int(rand() * ($photo_randomizer_hi - $photo_randomizer_lo)));
-# my $url = $photo_randomizer . $n;
-# return ( $url, $url, "photopoint" );
-#}
+# 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 image on a random page, and returns two URLs:
-# the page containing the image, and the image.
+
+# 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.
-# Uses the url-randomizer 1 time in 5, else the image randomizer.
-#
-my $total_0 = 0;
-my $total_1 = 0;
-my $total_2 = 0;
-my $total_3 = 0;
-my $total_4 = 0;
-my $count_0 = 0;
-my $count_1 = 0;
-my $count_2 = 0;
-my $count_3 = 0;
-my $count_4 = 0;
+#
+sub pick_from_yahoo_random_link {
+ my ( $timeout ) = @_;
-sub pick_image {
- my ( $timeout ) = @_;
- my $r = int(rand(100));
+ print STDERR "\n\n" if ($verbose_load);
+ LOG ($verbose_load, "URL: $yahoo_random_link");
- my ($base, $img, $source, $total, $count);
+ $last_search = $yahoo_random_link; # for warnings
- if ($r < 20) {
- ($base, $img, $source) = pick_from_url_randomizer ($timeout);
- $total = ++$total_0;
- $count = ++$count_0 if $img;
+ $suppress_audit = 1;
- } elsif ($r < 60) {
- ($base, $img, $source) = pick_from_image_randomizer ($timeout, 0);
- $total = ++$total_1;
- $count = ++$count_1 if $img;
+ my ( $base, $body ) = get_document ($yahoo_random_link, undef, $timeout);
+ if (!$base || !$body) {
+ $body = undef;
+ return;
+ }
- } elsif ($r < 65) {
- ($base, $img, $source) = pick_from_image_randomizer ($timeout, 3);
- $total = ++$total_4;
- $count = ++$count_4 if $img;
+ LOG ($verbose_load, "redirected to: $base");
-# } elsif ($r < 70) {
-# ($base, $img, $source) = pick_from_photo_randomizer ($timeout);
-# $total = ++$total_4;
-# $count = ++$count_4 if $img;
+ my $img = pick_image_from_body ($base, $body);
+ $body = undef;
-# } elsif ($r < 80) {
-# # HotBot sucks: 98% of the time, it says "no pages match your
-# # search", and then if I load the URL again by hand, it works.
-# # I don't understand what's going wrong here, but we're not getting
-# # any good data back from them, so forget it for now.
-#
-# ($base, $img, $source) = pick_from_image_randomizer ($timeout, 1);
-# $total = ++$total_2;
-# $count = ++$count_2 if $img;
+ if ($img) {
+ return ($base, $img);
+ } else {
+ return ();
+ }
+}
- } else {
- ($base, $img, $source) = pick_from_image_randomizer ($timeout, 2);
- $total = ++$total_3;
- $count = ++$count_3 if $img;
- }
+\f
+############################################################################
+#
+# Pick images from random pages returned by the Alta Vista Random Link
+#
+############################################################################
- if ($source && $total > 0) {
- $source .= " " . int(($count / $total) * 100) . "%";
- }
- return ($base, $img, $source);
-}
+# altavista
+my $alta_vista_random_link = "http://www.altavista.com/image/randomlink";
-# Does %-decoding.
+# 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 url_decode {
- ($_) = @_;
- tr/+/ /;
- s/%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))/eg;
- return $_;
-}
+sub pick_from_alta_vista_random_link {
+ my ( $timeout ) = @_;
+ print STDERR "\n\n" if ($verbose_load);
+ LOG ($verbose_load, "URL: $alta_vista_random_link");
-# 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 (($b<<8|$a), ($d<<8|$c));
-}
+ $last_search = $alta_vista_random_link; # for warnings
-# 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) {
- $ch = substr($body, $i, 1); $i++;
- }
- # markers can be padded with any number of 0xFF.
- while (ord($ch) == 0xFF) {
- $ch = substr($body, $i, 1); $i++;
- }
+ $suppress_audit = 1;
- # $ch contains the value of the marker.
- my $marker = ord($ch);
+ my ( $base, $body ) = get_document ($alta_vista_random_link,
+ undef, $timeout);
+ if (!$base || !$body) {
+ $body = undef;
+ return;
+ }
- if (($marker >= 0xC0) &&
- ($marker <= 0xCF) &&
- ($marker != 0xC4) &&
- ($marker != 0xCC)) { # it's a SOFn marker
- $i += 3;
- my $s = substr($body, $i, 4); $i += 4;
- my ($a,$b,$c,$d) = unpack("C"x4, $s);
- return (($c<<8|$d), ($a<<8|$b));
+ LOG ($verbose_load, "redirected to: $base");
- } else {
- # We must skip variables, since FFs in variable names aren't
- # valid JPEG markers.
- 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;
- }
- }
+ my $img = pick_image_from_body ($base, $body);
+ $body = undef;
+
+ if ($img) {
+ return ($base, $img);
+ } else {
return ();
+ }
}
-# Given the raw body of a GIF or JPEG document, returns the dimensions of
-# the image.
+\f
+############################################################################
#
-sub image_size {
- my ($body) = @_;
- my ($w, $h) = gif_size ($body);
- if ($w && $h) { return ($w, $h); }
- return jpeg_size ($body);
-}
+# Pick images by feeding random words into Alta Vista Image Search
+#
+############################################################################
-# returns the full path of the named program, or undef.
-#
-sub which {
- my ($prog) = @_;
- foreach (split (/:/, $ENV{PATH})) {
- if (-x "$_/$prog") {
- return $prog;
- }
- }
- return undef;
-}
+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 ) = @_;
-# Like rand(), but chooses numbers with a bell curve distribution.
-sub bellrand {
- ($_) = @_;
- $_ = 1.0 unless defined($_);
- $_ /= 3.0;
- return (rand($_) + rand($_) + rand($_));
-}
+ 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@^http://@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);
+}
+
+
+\f
+############################################################################
+#
+# Pick images by feeding random words into Google Image Search.
+# By Charles Gales <gales@us.ibm.com>
+#
+############################################################################
+
+
+my $google_images_url = "http://images.google.com/images" .
+ "?site=images" . # photos
+ "&btnG=Search" . # graphics
+ "&safe=off" . # no screening
+ "&imgsafe=off" .
+ "&q=";
+
+# googleimgs
+sub pick_from_google_images {
+ my ( $timeout ) = @_;
+
+ my $words = random_word; # only one word for Google
+ my $page = (int(rand(9)) + 1);
+ my $num = 20; # 20 images per page
+ my $search_url = $google_images_url . $words;
+
+ if ($page > 1) {
+ $search_url .= "&start=" . $page*$num; # page number
+ $search_url .= "&num=" . $num; #images per page
+ }
+
+ my ($search_hit_count, @subpages) =
+ pick_from_search_engine ($timeout, $search_url, $words);
+
+ my @candidates = ();
+ foreach my $u (@subpages) {
+ next unless ($u =~ m@imgres\?imgurl@i); # All pics start with this
+ next if ($u =~ m@[/.]google\.com\b@i); # skip google builtins
+
+ if ($u =~ m@^/imgres\?imgurl=(.*?)\&imgrefurl=(.*?)\&@) {
+ my $urlf = $2;
+ LOG ($verbose_filter, " candidate: $urlf");
+ push @candidates, $urlf;
+ }
+ }
+
+ return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
+ $timeout, @candidates);
+}
+
+
+\f
+############################################################################
+#
+# 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);
+
+ my $words = "$number";
+ my $page = (int(rand(40)) + 1);
+ my $num = 20; # 20 images per page
+ my $search_url = $google_images_url . $words;
+
+ if ($page > 1) {
+ $search_url .= "&start=" . $page*$num; # page number
+ $search_url .= "&num=" . $num; #images per page
+ }
+
+ my ($search_hit_count, @subpages) =
+ pick_from_search_engine ($timeout, $search_url, $words);
+
+ my @candidates = ();
+ my %referers;
+ foreach my $u (@subpages) {
+ next unless ($u =~ m@imgres\?imgurl@i); # All pics start with this
+ next if ($u =~ m@[/.]google\.com\b@i); # skip google builtins
+
+ if ($u =~ m@^/imgres\?imgurl=(.*?)\&imgrefurl=(.*?)\&@) {
+ my $ref = $2;
+ my $img = $1;
+ $img = "http://$img" unless ($img =~ m/^http:/i);
+
+ 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);
+}
+
+
+\f
+############################################################################
+#
+# 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(0);
+ 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@^http://@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);
+}
+
+
+\f
+############################################################################
+#
+# 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@^http://@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);
+}
+
+
+\f
+############################################################################
+#
+# 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@^http://click.lycos.com/director.asp
+# .*
+# \btarget=([^&]+)
+# .*
+# @x);
+# $u = url_decode($1);
+
+ next unless ($u =~ m@^http://@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);
+}
+
+
+\f
+############################################################################
+#
+# Pick images by feeding random words into news.yahoo.com
+#
+############################################################################
+
+my $yahoo_news_url = "http://search.news.yahoo.com/search/news" .
+ "?a=1" .
+ "&c=news_photos" .
+ "&s=-%24s%2C-date" .
+ "&n=100" .
+ "&o=o" .
+ "&2=" .
+ "&3=" .
+ "&p=";
+
+# yahoonews
+sub pick_from_yahoo_news_text {
+ my ( $timeout ) = @_;
+
+ $last_search = $yahoo_news_url; # for warnings
+
+ my $words = random_words(0);
+ 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) {
+ # only accept URLs on Yahoo's news site
+ next unless ($u =~ m@^http://dailynews\.yahoo\.com/@i ||
+ $u =~ m@^http://story\.news\.yahoo\.com/@i);
+
+ LOG ($verbose_filter, " candidate: $u");
+ push @candidates, $u;
+ }
+
+ return pick_image_from_pages ($search_url, $search_hit_count, $#subpages+1,
+ $timeout, @candidates);
+}
+
+
+\f
+############################################################################
+#
+# Pick images from LiveJournal's list of recently-posted images.
+#
+############################################################################
+
+my $livejournal_img_url = "http://www.livejournal.com/stats/latest-img.bml";
+
+# 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);
+ return () unless $body;
+
+ my @candidates = ();
+
+ $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 unless (m/\bURL=[\'\"]([^\'\"]+)[\'\"]/si);
+ my $page = html_unquote ($1);
+ my @pair = ($img, $page);
+ LOG ($verbose_filter, " candidate: $img");
+ push @candidates, \@pair;
+ }
+
+ return () if ($#candidates == -1);
+
+ my $i = int(rand($#candidates+1));
+ my ($img, $page) = @{$candidates[$i]};
+
+ LOG ($verbose_load, "picked image " .($i+1) . "/" . ($#candidates+1) .
+ ": $img");
+
+ return ($page, $img);
+}
+
+\f
+############################################################################
+#
+# 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/^http:/i);
+ next if ($u =~ m@^http://(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);
+}
+
+\f
+############################################################################
+#
+# Pick images by waiting for driftnet to populate a temp dir with files.
+# Requires driftnet version 0.1.5 or later.
+# (Driftnet is a program by Chris Lightfoot that sniffs your local ethernet
+# for images being downloaded by others.)
+# Driftnet/webcollage integration by jwz.
+#
+############################################################################
+
+# driftnet
+sub pick_from_driftnet {
+ my ( $timeout ) = @_;
+
+ my $id = $driftnet_magic;
+ my $dir = $driftnet_dir;
+ my $start = time;
+ my $now;
+
+ error ("\$driftnet_dir unset?") unless ($dir);
+ $dir =~ s@/+$@@;
+
+ error ("$dir unreadable") unless (-d "$dir/.");
+
+ $timeout = $http_timeout unless ($timeout);
+ $last_search = $id;
+
+ while ($now = time, $now < $start + $timeout) {
+ local *DIR;
+ opendir (DIR, $dir) || error ("$dir: $!");
+ while (my $file = readdir(DIR)) {
+ next if ($file =~ m/^\./);
+ $file = "$dir/$file";
+ closedir DIR;
+ LOG ($verbose_load, "picked file $file ($id)");
+ return ($id, $file);
+ }
+ closedir DIR;
+ }
+ LOG (($verbose_net || $verbose_load), "timed out for $id");
+ return ();
+}
+
+
+sub get_driftnet_file {
+ my ($file) = @_;
+
+ error ("\$driftnet_dir unset?") unless ($driftnet_dir);
+
+ my $id = $driftnet_magic;
+ my $re = qr/$driftnet_dir/;
+ error ("$id: $file not in $driftnet_dir?")
+ unless ($file =~ m@^$re@o);
+
+ local *IN;
+ open (IN, $file) || error ("$id: $file: $!");
+ my $body = '';
+ while (<IN>) { $body .= $_; }
+ 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));
+}
+
+\f
+############################################################################
+#
+# 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);
+}
+
+
+\f
+############################################################################
+#
+# 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;
+}
+
+
+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 = int($secs / $try);
+ print STDERR sprintf ("$blurb %-12s %4s (%d/%d);\t %2d 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@^http://([^ \t\n\r/:]+)@;
+ return unless defined ($site);
+
+ if ($base eq $driftnet_magic) {
+ $site = $driftnet_magic;
+ @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);
+}
+
+
+\f
+##############################################################################
+#
+# 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 (($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 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);
+}
+
+
+# returns the full path of the named program, or undef.
+#
+sub which {
+ my ($prog) = @_;
+ foreach (split (/:/, $ENV{PATH})) {
+ if (-x "$_/$prog") {
+ return $prog;
+ }
+ }
+ 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();
+ 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);
+ 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);
+ do {
+ my ($base, $img) = pick_image;
+ if ($img) {
+ $base =~ s/ /%20/g;
+ $img =~ s/ /%20/g;
+ print "$img $base\n";
+ }
+ } while (1);
}
##############################################################################
#
##############################################################################
+my $image_ppm = sprintf ("%s/webcollage-%08x",
+ ($ENV{TMPDIR} ? $ENV{TMPDIR} : "/tmp"),
+ rand(0xFFFFFFFF));
+my $image_tmp1 = sprintf ("%s/webcollage-1-%08x",
+ ($ENV{TMPDIR} ? $ENV{TMPDIR} : "/tmp"),
+ rand(0xFFFFFFFF));
+my $image_tmp2 = sprintf ("%s/webcollage-2-%08x",
+ ($ENV{TMPDIR} ? $ENV{TMPDIR} : "/tmp"),
+ rand(0xFFFFFFFF));
+
+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 {
- my ($sig) = @_;
- if ($verbose > 0) { print STDERR blurb() . "caught signal $sig.\n"; }
- unlink $image_ppm, $image_tmp1, $image_tmp2;
- exit 1;
+ unlink $image_ppm, $image_tmp1, $image_tmp2;
+ unlink $imagemap_html_tmp, $imagemap_jpg_tmp
+ if (defined ($imagemap_html_tmp));
}
# with whatever signal killed the sub-process, if any.
#
sub nontrapping_system {
- $! = 0;
-
- if ($verbose > 1) {
- $_ = join(" ", @_);
- s/\"[^\"]+\"/\"...\"/g;
- print STDERR blurb() . "executing \"$_\"\n";
- }
+ $! = 0;
- my $rc = system @_;
+ $_ = join(" ", @_);
+ s/\"[^\"]+\"/\"...\"/g;
- if ($rc == 0) {
- if ($verbose > 1) {
- print STDERR blurb() . "subproc exited normally.\n";
- }
- } elsif (($rc & 0xff) == 0) {
- $rc >>= 8;
- if ($verbose) {
- print blurb() . "subproc exited with status $rc.\n";
- }
- } else {
- if ($rc & 0x80) {
- if ($verbose) {
- print blurb() . "subproc dumped core.\n";
- }
- $rc &= ~0x80;
- }
- if ($verbose) {
- print blurb() . "subproc died with signal $rc.\n";
- }
- # die that way ourselves.
- kill $rc, $$;
+ 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;
+ return $rc;
}
-# Given the URL of a GIF or JPEG image, and the body of that image, writes a
-# PPM to the given output file. Returns the width/height of the image if
-# successful.
+# Given the URL of a GIF, JPEG, or PNG image, and the body of that image,
+# writes a PPM to the given output file. Returns the width/height of the
+# image if successful.
#
sub image_to_pnm {
- my ($url, $body, $output) = @_;
- my ($cmd, $cmd2, $w, $h);
-
- if ((@_ = gif_size ($body))) {
- ($w, $h) = @_;
- $cmd = "giftopnm";
- } elsif ((@_ = jpeg_size ($body))) {
- ($w, $h) = @_;
- $cmd = "djpeg";
- } else {
- return ();
- }
-
- $cmd2 = "exec $cmd"; # yes, this really is necessary. if we don't
- # do this, the process doesn't die properly.
- if ($verbose <= 1) {
- #
- # We get a "giftopnm: got a 'Application Extension' extension"
- # warning any time it's an animgif.
- #
- # Note that "giftopnm: EOF / read error on image data" is not
- # always a fatal error -- sometimes the image looks fine anyway.
- #
- $cmd2 .= " 2>/dev/null";
- }
+ my ($url, $body, $output) = @_;
+ my ($cmd, $cmd2, $w, $h);
+
+ if ((@_ = gif_size ($body))) {
+ ($w, $h) = @_;
+ $cmd = "giftopnm";
+ } elsif ((@_ = jpeg_size ($body))) {
+ ($w, $h) = @_;
+ $cmd = "djpeg";
+ } elsif ((@_ = png_size ($body))) {
+ ($w, $h) = @_;
+ $cmd = "pngtopnm";
+ } else {
+ LOG (($verbose_pbm || $verbose_load),
+ "not a GIF, JPG, or PNG" .
+ (($body =~ m@<(base|html|head|body|script|table|a href)>@i)
+ ? " (looks like HTML)" : "") .
+ ": $url");
+ $suppress_audit = 1;
+ return ();
+ }
- # There exist corrupted GIF and JPEG files that can make giftopnm and
- # djpeg lose their minds and go into a loop. So this gives those programs
- # a small timeout -- if they don't complete in time, kill them.
+ $cmd2 = "exec $cmd"; # yes, this really is necessary. if we don't
+ # do this, the process doesn't die properly.
+ if (!$verbose_pbm) {
#
- my $pid;
- @_ = eval {
- my $timed_out;
-
- local $SIG{ALRM} = sub {
- if ($verbose > 0) {
- print STDERR blurb() . "timed out ($cvt_timeout) for " .
- "$cmd on \"$url\" in pid $pid\n";
- }
- kill ('TERM', $pid) if ($pid);
- $timed_out = 1;
- $body = undef;
- };
-
- if (($pid = open(PIPE, "| $cmd2 > $output"))) {
- $timed_out = 0;
- alarm $cvt_timeout;
- print PIPE $body;
- $body = undef;
- close PIPE;
-
- if ($verbose > 3) { print STDERR blurb() . "awaiting $pid\n"; }
- waitpid ($pid, 0);
- if ($verbose > 3) { print STDERR blurb() . "$pid completed\n"; }
-
-
- my $size = (stat($output))[7];
- if ($size < 5) {
- if ($verbose) {
- print STDERR blurb() . "$cmd on ${w}x$h \"$url\" failed" .
- " ($size bytes)\n";
- }
- return ();
- }
-
- if ($verbose > 1) {
- print STDERR blurb() . "created ${w}x$h $output ($cmd)\n";
- }
- return ($w, $h);
- } else {
- print STDERR blurb() . "$cmd failed: $!\n";
- return ();
- }
+ # We get a "giftopnm: got a 'Application Extension' extension"
+ # warning any time it's an animgif.
+ #
+ # Note that "giftopnm: EOF / read error on image data" is not
+ # always a fatal error -- sometimes the image looks fine anyway.
+ #
+ $cmd2 .= " 2>/dev/null";
+ }
+
+ # There exist corrupted GIF and JPEG files that can make giftopnm and
+ # djpeg lose their minds and go into a loop. So this gives those programs
+ # a small timeout -- if they don't complete in time, kill them.
+ #
+ my $pid;
+ @_ = eval {
+ my $timed_out;
+
+ local $SIG{ALRM} = sub {
+ LOG ($verbose_pbm,
+ "timed out ($cvt_timeout) for $cmd on \"$url\" in pid $pid");
+ kill ('TERM', $pid) if ($pid);
+ $timed_out = 1;
+ $body = undef;
};
- die if ($@ && $@ ne "alarm\n"); # propagate errors
- if ($@) {
- # timed out
- $body = undef;
+
+ if (($pid = open(PIPE, "| $cmd2 > $output"))) {
+ $timed_out = 0;
+ alarm $cvt_timeout;
+ print PIPE $body;
+ $body = undef;
+ close PIPE;
+
+ LOG ($verbose_exec, "awaiting $pid");
+ waitpid ($pid, 0);
+ LOG ($verbose_exec, "$pid completed");
+
+ my $size = (stat($output))[7];
+ $size = -1 unless defined($size);
+ if ($size < 5) {
+ LOG ($verbose_pbm, "$cmd on ${w}x$h \"$url\" failed ($size bytes)");
return ();
+ }
+
+ LOG ($verbose_pbm, "created ${w}x$h $output ($cmd)");
+ return ($w, $h);
} else {
- # didn't
- alarm 0;
- $body = undef;
- return @_;
+ print STDERR blurb() . "$cmd failed: $!\n";
+ return ();
}
+ };
+ die if ($@ && $@ ne "alarm\n"); # propagate errors
+ if ($@) {
+ # timed out
+ $body = undef;
+ return ();
+ } else {
+ # didn't
+ alarm 0;
+ $body = undef;
+ return @_;
+ }
}
-sub x_output {
-
- my $win_cmd_1 = $ppm_to_root_window_cmd_1;
- my $win_cmd_2 = $ppm_to_root_window_cmd_2;
- my $win_cmd_3 = $ppm_to_root_window_cmd_3;
- $win_cmd_1 =~ s/^([^ \t\r\n]+).*$/$1/;
- $win_cmd_2 =~ s/^([^ \t\r\n]+).*$/$1/;
- $win_cmd_3 =~ s/^([^ \t\r\n]+).*$/$1/;
+sub pick_root_displayer {
+ my @names = ();
- # make sure the various programs we execute exist, right up front.
- foreach ("ppmmake", "giftopnm", "djpeg", "pnmpaste", "pnmscale",
- "pnmcut") {
- which ($_) || die blurb() . "$_ not found on \$PATH.\n";
+ 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");
}
+ }
- if (which($win_cmd_1)) {
- $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_1;
- } elsif (which($win_cmd_2)) {
- $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_2;
- } elsif (which($win_cmd_3)) {
- $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_3;
- } else {
- die blurb() . "didn't find $win_cmd_1, $win_cmd_2, or $win_cmd_3 on \$PATH.\n";
- }
+ $names[$#names] = "or " . $names[$#names];
+ error "none of: " . join (", ", @names) . " were found on \$PATH.";
+}
- $SIG{HUP} = \&x_cleanup;
- $SIG{INT} = \&x_cleanup;
- $SIG{QUIT} = \&x_cleanup;
- $SIG{ABRT} = \&x_cleanup;
- $SIG{KILL} = \&x_cleanup;
- $SIG{TERM} = \&x_cleanup;
-
- # Need this so that if giftopnm dies, we don't die.
- $SIG{PIPE} = 'IGNORE';
-
- if (!$img_width || !$img_height) {
- $_ = "xdpyinfo";
- which ($_) || die blurb() . "$_ not found on \$PATH.\n";
- $_ = `$_`;
- ($img_width, $img_height) = m/dimensions: *(\d+)x(\d+) /;
- if (!defined($img_height)) {
- die blurb() . "xdpyinfo failed.\n";
- }
- }
- 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) {
- print STDERR blurb() . "not a color or readable file: " .
- "$background\n";
- exit 1;
- } else {
- # default to assuming it's a color
- $bgcolor = $background;
- }
- }
+my $ppm_to_root_window_cmd = undef;
- # Create the sold-colored base image.
- #
- $_ = "ppmmake '$bgcolor' $img_width $img_height";
- if ($verbose > 1) {
- print STDERR blurb() . "creating base image: $_\n";
- }
- nontrapping_system "$_ > $image_ppm";
- # Paste the default background image in the middle of it.
- #
- if ($bgimage) {
- my ($iw, $ih);
-
- my $body = "";
- local *IMG;
- open(IMG, "<$bgimage") || die ("couldn't open $bgimage: $!\n");
- my $cmd;
- while (<IMG>) { $body .= $_; }
- close (IMG);
- if ((@_ = gif_size ($body))) {
- ($iw, $ih) = @_;
- $cmd = "giftopnm |";
- } elsif ((@_ = jpeg_size ($body))) {
- ($iw, $ih) = @_;
- $cmd = "djpeg |";
- } elsif ($body =~ m/^P\d\n(\d+) (\d+)\n/) {
- $iw = $1;
- $ih = $2;
- $cmd = "";
- } else {
- die blurb() . "$bgimage is not a GIF, JPEG, or PPM.\n";
- }
+sub x_or_pbm_output {
+ my ($window_id) = @_;
- my $x = int (($img_width - $iw) / 2);
- my $y = int (($img_height - $ih) / 2);
- if ($verbose > 1) {
- print STDERR blurb() . "pasting $bgimage (${iw}x$ih) into base ".
- "image at $x,$y\n";
- }
+ # Check for our helper program, to see whether we need to use PPM pipelines.
+ #
+ $_ = "webcollage-helper";
+ if (defined ($webcollage_helper) || which ($_)) {
+ $webcollage_helper = $_ unless (defined($webcollage_helper));
+ LOG ($verbose_pbm, "found \"$webcollage_helper\"");
+ $webcollage_helper .= " -v";
+ } else {
+ LOG (($verbose_pbm || $verbose_load), "no $_ program");
+ }
- $cmd .= "pnmpaste - $x $y $image_ppm > $image_tmp1";
- open (IMG, "| $cmd") || die ("running $cmd: $!\n");
- print IMG $body;
- $body = undef;
- close (IMG);
- if ($verbose > 1) {
- print STDERR blurb() . "subproc exited normally.\n";
- }
- rename ($image_tmp1, $image_ppm) ||
- die ("renaming $image_tmp1 to $image_ppm: $!\n");
- }
+ # make sure the various programs we execute exist, right up front.
+ #
+ my @progs = ("ppmmake"); # always need this one
- while (1) {
- my ($base, $img, $source) = pick_image();
- if ($img) {
- my ($headers, $body) = get_document ($img, $base);
- if ($body) {
- handle_image ($base, $img, $body, $source);
- $body = undef;
- }
- }
- unlink $image_tmp1, $image_tmp2;
- sleep $delay;
- }
-}
+ if (!defined($webcollage_helper)) {
+ # Only need these others if we don't have the helper.
+ @progs = (@progs,
+ "giftopnm", "pngtopnm", "djpeg",
+ "pnmpaste", "pnmscale", "pnmcut");
+ }
-sub handle_image {
- my ($base, $img, $body, $source) = @_;
+ foreach (@progs) {
+ which ($_) || error "$_ not found on \$PATH.";
+ }
- if ($verbose > 1) {
- print STDERR blurb() . "got $img (" . length($body) . ")\n";
- }
+ # find a root-window displayer program.
+ #
+ $ppm_to_root_window_cmd = pick_root_displayer();
- my ($iw, $ih) = image_to_pnm ($img, $body, $image_tmp1);
- $body = undef;
- return 0 unless ($iw && $ih);
+ if (defined ($window_id)) {
+ error ("-window-id only works if xscreensaver-getimage is installed")
+ unless ($ppm_to_root_window_cmd =~ m/^xscreensaver-getimage\b/);
- my $ow = $iw; # used only for error messages
- my $oh = $ih;
+ error ("unparsable window id: $window_id")
+ unless ($window_id =~ m/^\d+$|^0x[\da-f]+$/i);
+ $ppm_to_root_window_cmd =~ s/--?root\b/$window_id/ ||
+ error ("unable to munge displayer: $ppm_to_root_window_cmd");
+ }
- # 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) {
- if ($verbose > 1) {
- print STDERR blurb() . "running $filter_cmd\n";
- }
+ if (!$img_width || !$img_height) {
- my $rc = nontrapping_system "($filter_cmd) < $image_tmp1 >$image_tmp2";
- if ($rc != 0) {
- if ($verbose) {
- print STDERR blurb() . "failed command: \"$filter_cmd\"\n";
- print STDERR blurb() . "failed url: \"$img\" (${ow}x$oh)\n";
- }
- return;
- }
- rename ($image_tmp2, $image_tmp1);
-
- # re-get the width/height in case the filter resized it.
- local *IMG;
- open(IMG, "<$image_tmp1") || return 0;
- $_ = <IMG>;
- $_ = <IMG>;
- ($iw, $ih) = m/^(\d+) (\d+)$/;
- close (IMG);
- return 0 unless ($iw && $ih);
+ if (!defined ($window_id) &&
+ defined ($ENV{XSCREENSAVER_WINDOW})) {
+ $window_id = $ENV{XSCREENSAVER_WINDOW};
}
- my $target_w = $img_width;
- my $target_h = $img_height;
+ 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;
- my $cmd = "";
+ if (!defined($img_height)) {
+ error "xwininfo failed.";
+ }
+ }
+ }
+ my $bgcolor = "#000000";
+ my $bgimage = undef;
- # Usually scale the image to fit on the screen -- but sometimes scale it
- # to fit on half or a quarter of the screen. Note that we don't merely
- # scale it to fit, we instead cut it in half until it fits -- that should
- # give a wider distribution of sizes.
- #
- if (rand() < 0.3) { $target_w /= 2; $target_h /= 2; }
- 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);
- }
- if ($iw <= 10 || $ih <= 10) {
- if ($verbose > 1) {
- print STDERR blurb() . "scaling to ${iw}x$ih would " .
- "have been bogus.\n";
- }
- return 0;
- }
+ if ($background) {
+ if ($background =~ m/^\#[0-9a-f]+$/i) {
+ $bgcolor = $background;
- if ($verbose > 1) {
- print STDERR blurb() . "scaling to ${iw}x$ih\n";
- }
+ } elsif (-r $background) {
+ $bgimage = $background;
+
+ } elsif (! $background =~ m@^[-a-z0-9 ]+$@i) {
+ error "not a color or readable file: $background";
- $cmd .= " | pnmscale -xsize $iw -ysize $ih";
+ } else {
+ # default to assuming it's a color
+ $bgcolor = $background;
}
+ }
+ # Create the sold-colored base image.
+ #
+ $_ = "ppmmake '$bgcolor' $img_width $img_height";
+ LOG ($verbose_pbm, "creating base image: $_");
+ nontrapping_system "$_ > $image_ppm";
- my $src = $image_tmp1;
+ # Paste the default background image in the middle of it.
+ #
+ if ($bgimage) {
+ my ($iw, $ih);
- 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 $body = "";
+ local *IMG;
+ open(IMG, "<$bgimage") || error "couldn't open $bgimage: $!";
+ my $cmd;
+ while (<IMG>) { $body .= $_; }
+ close (IMG);
- # 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 ((@_ = gif_size ($body))) {
+ ($iw, $ih) = @_;
+ $cmd = "giftopnm |";
+
+ } elsif ((@_ = jpeg_size ($body))) {
+ ($iw, $ih) = @_;
+ $cmd = "djpeg |";
+
+ } elsif ((@_ = png_size ($body))) {
+ ($iw, $ih) = @_;
+ $cmd = "pngtopnm |";
+
+ } elsif ($body =~ m/^P\d\n(\d+) (\d+)\n/) {
+ $iw = $1;
+ $ih = $2;
+ $cmd = "";
+
+ } else {
+ error "$bgimage is not a GIF, JPEG, PNG, or PPM.";
}
- if ($verbose > 2 && $crop_chance > 0.1) {
- print STDERR blurb() . "crop chance: $crop_chance\n";
+ my $x = int (($img_width - $iw) / 2);
+ my $y = int (($img_height - $ih) / 2);
+ LOG ($verbose_pbm,
+ "pasting $bgimage (${iw}x$ih) into base image at $x,$y");
+
+ $cmd .= "pnmpaste - $x $y $image_ppm > $image_tmp1";
+ open (IMG, "| $cmd") || error "running $cmd: $!";
+ print IMG $body;
+ $body = undef;
+ close (IMG);
+ LOG ($verbose_exec, "subproc exited normally.");
+ rename ($image_tmp1, $image_ppm) ||
+ error "renaming $image_tmp1 to $image_ppm: $!";
+ }
+
+ 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";
- if (rand() < $crop_chance) {
+ unlink $image_tmp1, $image_tmp2;
+ sleep $delay;
+ }
+}
- my $ow = $crop_w;
- my $oh = $crop_h;
+sub paste_image {
+ my ($base, $img, $body, $source) = @_;
- 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));
- }
+ $current_state = "paste";
- if ($verbose > 1 &&
- ($crop_x != 0 || $crop_y != 0 ||
- $crop_w != $iw || $crop_h != $ih)) {
- print STDERR blurb() . "randomly cropping to " .
- "${crop_w}x$crop_h \@ $crop_x,$crop_y\n";
- }
- }
+ $suppress_audit = 0;
- # Where the image should logically land -- this might be negative.
- #
- my $x = int((rand() * ($img_width + $crop_w/2)) - $crop_w*3/4);
- my $y = int((rand() * ($img_height + $crop_h/2)) - $crop_h*3/4);
+ LOG ($verbose_pbm, "got $img (" . length($body) . ")");
- # 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) {
-
- if ($verbose > 1) {
- print STDERR blurb() . "cropping for effective paste of " .
- "${crop_w}x$crop_h \@ $x,$y\n";
- }
+ my ($iw, $ih);
- if ($x < 0) { $crop_x -= $x; $crop_w += $x; $x = 0; }
- if ($y < 0) { $crop_y -= $y; $crop_h += $y; $y = 0; }
+ # If we are using the webcollage-helper, then we do not need to convert this
+ # image to a PPM. But, if we're using a filter command, we still must, since
+ # that's what the filters expect (webcollage-helper can read PPMs, so that's
+ # fine.)
+ #
+ if (defined ($webcollage_helper) &&
+ !defined ($filter_cmd)) {
- if ($x + $crop_w >= $img_width) { $crop_w = $img_width - $x - 1; }
- if ($y + $crop_h >= $img_height) { $crop_h = $img_height - $y - 1; }
+ ($iw, $ih) = image_size ($body);
+ if (!$iw || !$ih) {
+ LOG (($verbose_pbm || $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 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;
- $cmd .= " | pnmcut $crop_x $crop_y $iw $ih";
- if ($verbose > 1) {
- print STDERR blurb() . "cropping to ${crop_w}x$crop_h \@ " .
- "$crop_x,$crop_y\n";
- }
- }
+ local *OUT;
+ open (OUT, ">$image_tmp1") || error ("writing $image_tmp1: $!");
+ print OUT $body || error ("writing $image_tmp1: $!");
+ close OUT || error ("writing $image_tmp1: $!");
- if ($verbose > 1) {
- print STDERR blurb() . "pasting ${iw}x$ih \@ $x,$y in $image_ppm\n";
+ } else {
+ ($iw, $ih) = image_to_pnm ($img, $body, $image_tmp1);
+ $body = undef;
+ if (!$iw || !$ih) {
+ LOG ($verbose_pbm, "unable to make PBM from $img");
+ return 0;
}
+ }
- $cmd .= " | pnmpaste - $x $y $image_ppm";
+ record_success ($load_method, $img, $base);
- $cmd =~ s@^ *\| *@@;
- my $rc = nontrapping_system "($cmd) < $image_tmp1 > $image_tmp2";
+ 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) {
+ LOG ($verbose_pbm, "running $filter_cmd");
+
+ my $rc = nontrapping_system "($filter_cmd) < $image_tmp1 >$image_tmp2";
if ($rc != 0) {
- if ($verbose) {
- print STDERR blurb() . "failed command: \"$cmd\"\n";
- print STDERR blurb() . "failed url: \"$img\" (${ow}x$oh)\n";
- }
- return;
+ LOG(($verbose_pbm || $verbose_load), "failed command: \"$filter_cmd\"");
+ LOG(($verbose_pbm || $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.
+ local *IMG;
+ open(IMG, "<$image_tmp1") || return 0;
+ $_ = <IMG>;
+ $_ = <IMG>;
+ ($iw, $ih) = m/^(\d+) (\d+)$/;
+ close (IMG);
+ return 0 unless ($iw && $ih);
+ }
+
+ my $target_w = $img_width; # max rectangle into which the image must fit
+ my $target_h = $img_height;
+
+ my $cmd = "";
+ my $scale = 1.0;
+
+
+ # 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_pbm, "scaling to ${iw}x$ih would have been bogus.");
+ return 0;
+ }
+
+ LOG ($verbose_pbm, "scaling to ${iw}x$ih ($scale)");
+
+ $cmd .= " | pnmscale -xsize $iw -ysize $ih";
+ }
+
+
+ my $src = $image_tmp1;
+
+ 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;
+
+ # 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_pbm, "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));
}
+ if ($crop_x != 0 || $crop_y != 0 ||
+ $crop_w != $iw || $crop_h != $ih) {
+ LOG ($verbose_pbm,
+ "randomly cropping to ${crop_w}x$crop_h \@ $crop_x,$crop_y");
+ }
+ }
+
+ # Where the image should logically land -- this might be negative.
+ #
+ my $x = int((rand() * ($img_width + $crop_w/2)) - $crop_w*3/4);
+ my $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_pbm,
+ "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;
+ $cmd .= " | pnmcut $crop_x $crop_y $iw $ih";
+ LOG ($verbose_pbm, "cropping to ${crop_w}x$crop_h \@ $crop_x,$crop_y");
+ }
+
+ LOG ($verbose_pbm, "pasting ${iw}x$ih \@ $x,$y in $image_ppm");
+
+ $cmd .= " | pnmpaste - $x $y $image_ppm";
+
+ $cmd =~ s@^ *\| *@@;
+
+ if (defined ($webcollage_helper)) {
+ $cmd = "$webcollage_helper $image_tmp1 $image_ppm " .
+ "$scale $opacity " .
+ "$crop_x $crop_y $x $y " .
+ "$iw $ih";
+ $_ = $cmd;
+
+ } else {
+ # use a PPM pipeline
+ $_ = "($cmd)";
+ $_ .= " < $image_tmp1 > $image_tmp2";
+ }
+
+ if ($verbose_pbm) {
+ $_ = "($_) 2>&1 | sed s'/^/" . blurb() . "/'";
+ } else {
+ $_ .= " 2> /dev/null";
+ }
+
+ my $rc = nontrapping_system ($_);
+
+ if (defined ($webcollage_helper) && -z $image_ppm) {
+ LOG (1, "failed command: \"$cmd\"");
+ print STDERR "\naudit log:\n\n\n";
+ print STDERR ("#" x 78) . "\n";
+ print STDERR blurb() . "$image_ppm has zero size\n";
+ showlog();
+ print STDERR "\n\n";
+ exit (1);
+ }
+
+ if ($rc != 0) {
+ LOG (($verbose_pbm || $verbose_load), "failed command: \"$cmd\"");
+ LOG (($verbose_pbm || $verbose_load), "failed URL: \"$img\" (${ow}x$oh)");
+ return;
+ }
+
+ if (!defined ($webcollage_helper)) {
rename ($image_tmp2, $image_ppm) || return;
+ }
- my $target = "$image_ppm";
+ my $target = "$image_ppm";
- # don't just tack this onto the end of the pipeline -- we don't want it
- # to end up in $image_ppm, because we don't want the results to be
- # cumulative.
- #
- if ($post_filter_cmd) {
- $target = $image_tmp1;
- $rc = nontrapping_system "($post_filter_cmd) < $image_ppm > $target";
- if ($rc != 0) {
- if ($verbose) {
- print STDERR blurb() . "filter failed: " .
- "\"$post_filter_cmd\"\n";
- }
- return;
- }
- }
+ # don't just tack this onto the end of the pipeline -- we don't want it
+ # to end up in $image_ppm, because we don't want the results to be
+ # cumulative.
+ #
+ if ($post_filter_cmd) {
- if (!$no_output_p) {
- my $tsize = (stat($target))[7];
- if ($tsize > 200) {
- $cmd = $ppm_to_root_window_cmd;
- $cmd =~ s/%%PPM%%/$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 .= "&";
-
- $rc = nontrapping_system ($cmd);
-
- if ($rc != 0) {
- if ($verbose) {
- print STDERR blurb() . "display failed: \"$cmd\"\n";
- }
- return;
- }
-
- } elsif ($verbose > 1) {
- print STDERR blurb() . "$target size is $tsize\n";
- }
+ my $cmd;
+
+ $target = $image_tmp1;
+ if (!defined ($webcollage_helper)) {
+ $cmd = "($post_filter_cmd) < $image_ppm > $target";
+ } else {
+ # Blah, my scripts need the JPEG data, but some other folks need
+ # the PPM data -- what to do? Ignore the problem, that's what!
+# $cmd = "djpeg < $image_ppm | ($post_filter_cmd) > $target";
+ $cmd = "($post_filter_cmd) < $image_ppm > $target";
}
- if ($verbose > 0) {
- print STDOUT "image: ${iw}x${ih} @ $x,$y $base $source\n";
+ $rc = nontrapping_system ($cmd);
+ if ($rc != 0) {
+ LOG ($verbose_pbm, "filter failed: \"$post_filter_cmd\"\n");
+ return;
+ }
+ }
+
+ if (!$no_output_p) {
+ my $tsize = (stat($target))[7];
+ if ($tsize > 200) {
+ $cmd = "$ppm_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 .= " &";
+
+ $rc = nontrapping_system ($cmd);
+
+ if ($rc != 0) {
+ LOG (($verbose_pbm || $verbose_load), "display failed: \"$cmd\"");
+ return;
+ }
+
+ } else {
+ LOG ($verbose_pbm, "$target size is $tsize");
}
+ }
+
+ $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_ppm, $img_width, $img_height);
+ }
- return 1;
+ clearlog();
+
+ return 1;
}
-sub main {
- $| = 1;
- srand(time ^ $$);
-
- my $root_p = 0;
-
- # historical suckage: the environment variable name is lower case.
- $http_proxy = $ENV{http_proxy} || $ENV{HTTP_PROXY};
-
- while ($_ = $ARGV[0]) {
- shift @ARGV;
- if ($_ eq "-display" ||
- $_ eq "-displ" ||
- $_ eq "-disp" ||
- $_ eq "-dis" ||
- $_ eq "-dpy" ||
- $_ eq "-d") {
- $ENV{DISPLAY} = shift @ARGV;
- } elsif ($_ eq "-root") {
- $root_p = 1;
- } elsif ($_ eq "-no-output") {
- $no_output_p = 1;
- } elsif ($_ eq "-urls-only") {
- $urls_only_p = 1;
- $no_output_p = 1;
- } elsif ($_ eq "-verbose") {
- $verbose++;
- } elsif (m/^-v+$/) {
- $verbose += length($_)-1;
- } elsif ($_ eq "-delay") {
- $delay = shift @ARGV;
- } elsif ($_ eq "-timeout") {
- $http_timeout = shift @ARGV;
- } elsif ($_ eq "-filter") {
- $filter_cmd = shift @ARGV;
- } elsif ($_ eq "-filter2") {
- $post_filter_cmd = shift @ARGV;
- } elsif ($_ eq "-background" || $_ eq "-bg") {
- $background = shift @ARGV;
- } elsif ($_ eq "-size") {
- $_ = shift @ARGV;
- if (m@^(\d+)x(\d+)$@) {
- $img_width = $1;
- $img_height = $2;
- } else {
- die blurb() . "argument to \"-size\" must be" .
- " of the form \"640x400\"\n";
- }
- } elsif ($_ eq "-proxy" || $_ eq "-http-proxy") {
- $http_proxy = shift @ARGV;
- } else {
- die "$copyright\nusage: $progname [-root]" .
- " [-display dpy] [-root] [-verbose] [-timeout secs]\n" .
- "\t\t [-delay secs] [-filter cmd] [-filter2 cmd]\n" .
- "\t\t [-http-proxy host[:port]]\n";
- }
- }
+sub update_imagemap {
+ my ($url, $x, $y, $w, $h, $image_ppm, $image_width, $image_height) = @_;
- if ($http_proxy && $http_proxy eq "") {
- $http_proxy = undef;
- }
- if ($http_proxy && $http_proxy =~ m@^http://([^/]*)/?$@ ) {
- # historical suckage: allow "http://host:port" as well as "host:port".
- $http_proxy = $1;
- }
+ $current_state = "imagemap";
+
+ my $max_areas = 200;
+
+ $url = html_quote ($url);
+ my $x2 = $x + $w;
+ my $y2 = $y + $h;
+ my $area = "<AREA SHAPE=RECT COORDS=\"$x,$y,$x2,$y2\" HREF=\"$url\">";
+ unshift @imagemap_areas, $area; # put one on the front
+ if ($#imagemap_areas >= $max_areas) {
+ pop @imagemap_areas; # take one off the back.
+ }
+
+ LOG ($verbose_pbm, "area: $x,$y,$x2,$y2 (${w}x$h)");
- if (!$root_p && !$no_output_p) {
- die "$copyright" .
- blurb() . "the -root argument is mandatory (for now.)\n";
+ my $map_name = $imagemap_base;
+ $map_name =~ s@^.*/@@;
+ $map_name = 'collage' if ($map_name eq '');
+
+ my $imagemap_html = $imagemap_base . ".html";
+ my $imagemap_jpg = $imagemap_base . ".jpg";
+
+ if (!defined ($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 $template_html = '';
+ {
+ local *IN;
+ if (open (IN, "<$imagemap_html")) {
+ while (<IN>) { $template_html .= $_; }
+ close IN;
+ LOG ($verbose_pbm, "read template $imagemap_html");
}
- if (!$no_output_p && !$ENV{DISPLAY}) {
- die blurb() . "\$DISPLAY is not set.\n";
+ if ($template_html =~ m/^\s*$/s) {
+ $template_html = ("<MAP NAME=\"$map_name\"></MAP>\n" .
+ "<IMG SRC=\"$imagemap_base.jpg\"" .
+ " USEMAP=\"$map_name\">\n");
+ LOG ($verbose_pbm, "created dummy template");
}
+ }
- if ($urls_only_p) {
- url_only_output;
+ # Write the jpg to a tmp file
+ #
+ {
+ my $cmd;
+ if (defined ($webcollage_helper)) {
+ $cmd = "cp -p $image_ppm $imagemap_jpg_tmp";
} else {
- x_output;
+ $cmd = "cjpeg < $image_ppm > $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 $areas = join ("\n\t", @imagemap_areas);
+ my $map = ("<MAP NAME=\"$map_name\">\n\t$areas\n</MAP>");
+ my $img = ("<IMG SRC=\"$imagemap_base.jpg\" " .
+ "BORDER=0 " .
+ "WIDTH=$image_width HEIGHT=$image_height " .
+ "USEMAP=\"#$map_name\">");
+ $body =~ s@(<MAP\s+NAME=\"[^\"]*\"\s*>).*?(</MAP>)@$map@is;
+ $body =~ s@<IMG\b[^<>]*\bUSEMAP\b[^<>]*>@$img@is;
+
+ # 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$2@si;
+ }
+
+ local *OUT;
+ open (OUT, ">$imagemap_html_tmp") || error ("$imagemap_html_tmp: $!");
+ print OUT $body || error ("$imagemap_html_tmp: $!");
+ close OUT || error ("$imagemap_html_tmp: $!");
+ LOG ($verbose_pbm, "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_pbm, "wrote $imagemap_html");
+ rename ($imagemap_jpg_tmp, $imagemap_jpg) ||
+ error "renaming $imagemap_jpg_tmp to $imagemap_jpg";
+ LOG ($verbose_pbm, "wrote $imagemap_jpg");
+}
+
+
+sub init_signals {
+
+ $SIG{HUP} = \&signal_cleanup;
+ $SIG{INT} = \&signal_cleanup;
+ $SIG{QUIT} = \&signal_cleanup;
+ $SIG{ABRT} = \&signal_cleanup;
+ $SIG{KILL} = \&signal_cleanup;
+ $SIG{TERM} = \&signal_cleanup;
+
+ # Need this so that if giftopnm dies, we don't die.
+ $SIG{PIPE} = 'IGNORE';
+}
+
+END { 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;
+
+ # historical suckage: the environment variable name is lower case.
+ $http_proxy = $ENV{http_proxy} || $ENV{HTTP_PROXY};
+
+ while ($_ = $ARGV[0]) {
+ shift @ARGV;
+ if ($_ eq "-display" ||
+ $_ eq "-displ" ||
+ $_ eq "-disp" ||
+ $_ eq "-dis" ||
+ $_ eq "-dpy" ||
+ $_ eq "-d") {
+ $ENV{DISPLAY} = shift @ARGV;
+ } elsif ($_ eq "-root") {
+ $root_p = 1;
+ } elsif ($_ eq "-window-id" || $_ eq "--window-id") {
+ $window_id = shift @ARGV;
+ $root_p = 1;
+ } elsif ($_ eq "-no-output") {
+ $no_output_p = 1;
+ } elsif ($_ eq "-urls-only") {
+ $urls_only_p = 1;
+ $no_output_p = 1;
+ } elsif ($_ eq "-imagemap") {
+ $imagemap_base = shift @ARGV;
+ $no_output_p = 1;
+ } elsif ($_ eq "-verbose") {
+ $verbose++;
+ } elsif (m/^-v+$/) {
+ $verbose += length($_)-1;
+ } elsif ($_ eq "-delay") {
+ $delay = shift @ARGV;
+ } elsif ($_ eq "-timeout") {
+ $http_timeout = shift @ARGV;
+ } elsif ($_ eq "-filter") {
+ $filter_cmd = shift @ARGV;
+ } elsif ($_ eq "-filter2") {
+ $post_filter_cmd = shift @ARGV;
+ } elsif ($_ eq "-background" || $_ eq "-bg") {
+ $background = shift @ARGV;
+ } elsif ($_ eq "-size") {
+ $_ = shift @ARGV;
+ if (m@^(\d+)x(\d+)$@) {
+ $img_width = $1;
+ $img_height = $2;
+ } else {
+ error "argument to \"-size\" must be of the form \"640x400\"";
+ }
+ } elsif ($_ eq "-proxy" || $_ eq "-http-proxy") {
+ $http_proxy = shift @ARGV;
+ } elsif ($_ eq "-dictionary" || $_ eq "-dict") {
+ $dict = shift @ARGV;
+ } elsif ($_ eq "-driftnet" || $_ eq "--driftnet") {
+ @search_methods = ( 100, "driftnet", \&pick_from_driftnet );
+ if (! ($ARGV[0] =~ m/^-/)) {
+ $driftnet_cmd = shift @ARGV;
+ } else {
+ $driftnet_cmd = $default_driftnet_cmd;
+ }
+ } elsif ($_ eq "-debug" || $_ eq "--debug") {
+ my $which = shift @ARGV;
+ my @rest = @search_methods;
+ 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\"");
+
+ } else {
+ print STDERR "$copyright\nusage: $progname " .
+ "[-root] [-display dpy] [-verbose] [-debug which]\n" .
+ "\t\t [-timeout secs] [-delay secs] [-size WxH]\n" .
+ "\t\t [-no-output] [-urls-only] [-imagemap filename]\n" .
+ "\t\t [-filter cmd] [-filter2 cmd] [-background color]\n" .
+ "\t\t [-dictionary dictionary-file] [-http-proxy host[:port]]\n" .
+ "\t\t [-driftnet [driftnet-program-and-args]]\n" .
+ "\n";
+ exit 1;
+ }
+ }
+
+ if ($http_proxy && $http_proxy eq "") {
+ $http_proxy = undef;
+ }
+ if ($http_proxy && $http_proxy =~ m@^http://([^/]*)/?$@ ) {
+ # historical suckage: allow "http://host:port" as well as "host:port".
+ $http_proxy = $1;
+ }
+
+ if (!$root_p && !$no_output_p) {
+ print STDERR $copyright;
+ error "the -root argument is mandatory (for now.)";
+ }
+
+ if (!$no_output_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;
+ $verbose_pbm = 1;
+
+ } elsif ($verbose == 6) {
+ $verbose_imgmap = 1;
+ $verbose_warnings = 1;
+ $verbose_load = 1;
+ $verbose_filter = 1;
+ $verbose_net = 1;
+ $verbose_pbm = 1;
+ $verbose_http = 1;
+
+ } elsif ($verbose >= 7) {
+ $verbose_imgmap = 1;
+ $verbose_warnings = 1;
+ $verbose_load = 1;
+ $verbose_filter = 1;
+ $verbose_net = 1;
+ $verbose_pbm = 1;
+ $verbose_http = 1;
+ $verbose_exec = 1;
+ }
+
+ if ($dict) {
+ error ("$dict does not exist") unless (-f $dict);
+ $wordlist = $dict;
+ } else {
+ pick_dictionary();
+ }
+
+ init_signals();
+
+ spawn_driftnet ($driftnet_cmd) if ($driftnet_cmd);
+
+ if ($urls_only_p) {
+ url_only_output ();
+ } else {
+ x_or_pbm_output ($window_id);
+ }
}
main;
projects / xscreensaver / blobdiff