X-Git-Url: http://git.hungrycats.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=hacks%2Fwebcollage;h=880dfb933f6125a3d9494783765932e19616a830;hb=0316d74da7982288abddd34e7a62698eb7f79965;hp=2a727006d90446729e39e1df73dfb05972e322b4;hpb=5832fe184606766fef23369159306c0a5799aeb0;p=xscreensaver

diff --git a/hacks/webcollage b/hacks/webcollage
index 2a727006..880dfb93 100755
--- a/hacks/webcollage
+++ b/hacks/webcollage
@@ -27,7 +27,7 @@ require POSIX;
 use Fcntl ':flock'; # import LOCK_* constants
 
 
-my $version = q{ $Revision: 1.36 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
+my $version = q{ $Revision: 1.44 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
 my $copyright = "WebCollage $version, Copyright (c) 1999" .
     " Jamie Zawinski <jwz\@jwz.org>\n" .
     "            http://www.jwz.org/xscreensaver/\n";
@@ -36,15 +36,15 @@ my $argv0 = $0;
 my $progname = $argv0; $progname =~ s@.*/@@g;
 
 my $random_redirector = "http://random.yahoo.com/bin/ryl";
-my $image_randomizer_1 = "http://image.altavista.com/cgi-bin/avncgi" .
-                         "?do=3" .
-                         "&verb=n" .
-                         "&oshape=n" .
-                         "&oorder=" .
-                         "&ophoto=1&oart=1&ocolor=1&obw=1" .
+my $image_randomizer_1 = "http://www.altavista.com/query" .
+                         "?mmdo=3" .
+                         "&nbq=12" .
                          "&stype=simage" .
-                         "&oprem=0" .
-                         "&query=";
+                         "&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=" .
@@ -56,9 +56,13 @@ my $image_randomizer_2 = "http://www.hotbot.com/?clickSrc=search" .
                          "&MT=";
 my $image_randomizer_3 = "http://www.altavista.com/cgi-bin/query?pg=q" .
                          "&text=yes&kl=XX&stype=stext&q=";
-my $photo_randomizer   = "http://albums.photopoint.com/j/View?u=1&a=1&p=";
-my $photo_randomizer_lo = 10000001;
-my $photo_randomizer_hi = 12400000;
+
+# 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";
@@ -71,10 +75,11 @@ my $http_proxy = undef;
 my $http_timeout = 30;
 my $cvt_timeout = 10;
 
-# if we have xli, use it to write to the root window.  else use xv.
-my $ppm_to_root_window_cmd_1 = "xli -quiet -onroot -center" .
+# 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_2 = "xv -root -rmode 5 -viewonly" .
+my $ppm_to_root_window_cmd_3 = "xv -root -rmode 5 -viewonly" .
                                " +noresetroot %%PPM%% -quit";
 
 my $ppm_to_root_window_cmd = undef;     # initialized by x_output()
@@ -158,6 +163,9 @@ sub get_document_1 {
     $paddr   = sockaddr_in($port2, $iaddr);
 
 
+    my $head = "";
+    my $body = "";
+
     @_ =
     eval {
         local $SIG{ALRM}  = sub {
@@ -187,16 +195,14 @@ sub get_document_1 {
             $cookie = "AV_ALL=1";
         }
 
-        print S ("GET " . ($http_proxy ? $url : "/$path") . " HTTP/1.0\n" .
-                 "Host: $them\n" .
-                 "User-Agent: $progname/$version\n" .
-                 ($referer ? "Referer: $referer\n" : "") .
-                 ($cookie ? "Cookie: $cookie\n" : "") .
-                 "\n");
+        print S ("GET " . ($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>;
 
-        my $head = "";
-        my $body = "";
         while (<S>) {
             $head .= $_;
             last if m@^[\r\n]@;
@@ -216,6 +222,8 @@ sub get_document_1 {
     die if ($@ && $@ ne "alarm\n");       # propagate errors
     if ($@) {
         # timed out
+        $head = undef;
+        $body = undef;
         return ();
     } else {
         # didn't
@@ -282,11 +290,13 @@ sub get_document {
 		    print STDERR "$progname: 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 {
@@ -369,6 +379,8 @@ sub pick_image_from_body {
                 " \"$url\": rejecting.\n";
         }
         $rejected_urls{$url} = -1;
+        $body = undef;
+        $_ = undef;
         return ();
     }
 
@@ -391,6 +403,8 @@ sub pick_image_from_body {
                             " length $L in $url: rejecting.\n";
                     }
                     $rejected_urls{$url} = $L;
+                    $body = undef;
+                    $_ = undef;
                     return ();
                 } elsif ( $verbose > 2 ) {
                     print STDERR "$progname: keywords of length $L" .
@@ -489,6 +503,9 @@ sub pick_image_from_body {
 	}
     }
 
+    $_ = undef;
+    $body = undef;
+
     if ( $#urls == 0 ) {
 	if ( $verbose > 2 ) {
 	    print STDERR "$progname: no images on $base\n";
@@ -523,8 +540,12 @@ sub pick_from_url_randomizer {
 
     my ( $base, $body ) = get_document ($random_redirector, undef, $timeout);
 
-    return if (!$base || !$body);
+    if (!$base || !$body) {
+        $body = undef;
+        return;
+    }
     my $img = pick_image_from_body ($base, $body);
+    $body = undef;
 
     if ($img) {
         return ($base, $img, "yahoo");
@@ -544,6 +565,10 @@ sub random_word {
             $word = <IN>;   # toss partial line
             $word = <IN>;   # keep next line
         }
+	if (!$word) {
+          seek( IN, 0, 0 );
+          $word = <IN>;
+	}
         close (IN);
     }
 
@@ -559,6 +584,10 @@ sub random_word {
     $word =~ s/izes$/ize/;
     $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.
+    }
+
     return $word;
 }
 
@@ -602,7 +631,10 @@ sub pick_from_image_randomizer {
     my ( $base, $body ) = get_document ($search_url, undef, $timeout);
     if (defined ($timeout)) {
         $timeout -= (time - $start);
-        return () if ($timeout <= 0);
+        if ($timeout <= 0) {
+            $body = undef;
+            return ();
+        }
     }
 
     return () if (! $body);
@@ -626,9 +658,11 @@ sub pick_from_image_randomizer {
     my $href_count = 0;
 
     $_ = $body;
-    s/[\r\n\t ]+/ /g;
 
-    s/Result Pages:.*$//;               # trim off page footer
+    s/Result [Pp]ages:.*$//s;            # trim off page footer
+    s/^.*?IMAGE RESULTS//s;              # trim off page header
+
+    s/[\r\n\t ]+/ /g;
 
     s/(<A )/\n$1/gi;
     foreach (split(/\n/)) {
@@ -647,6 +681,14 @@ sub pick_from_image_randomizer {
         next unless ($u =~ m@^http://@i);  # skip non-http and relative urls.
 
         next if ($u =~ m@[/.]altavista\.com@i);  # skip altavista builtins
+        next if ($u =~ m@[/.]av\.com@i);
+        next if ($u =~ m@[/.]virage\.com@i);
+        next if ($u =~ m@[/.]photoloft\.com@i);
+        next if ($u =~ m@[/.]shopping\.com@i);
+        next if ($u =~ m@[/.]thetrip\.com@i);
+        next if ($u =~ m@[/.]cmgi\.com@i);
+        next if ($u =~ m@[/.]intelihealth\.com@i);
+        next if ($u =~ m@[/.]wildweb\.com@i);
         next if ($u =~ m@[/.]digital\.com@i);
         next if ($u =~ m@[/.]doubleclick\.net@i);
 
@@ -676,6 +718,8 @@ sub pick_from_image_randomizer {
             print STDERR "$progname: found nothing on $base " .
                 "($length bytes, $href_count links).\n";
         }
+        $body = undef;
+        $_ = undef;
 	return ();
     }
 
@@ -688,12 +732,18 @@ sub pick_from_image_randomizer {
     }
 
 
+    $body = undef;
+    $_ = undef;
 
     my ( $base2, $body2 ) = get_document ($subpage, $base, $timeout);
 
-    return () if (!$base2 || !$body2);
+    if (!$base2 || !$body2) {
+        $body2 = undef;
+        return ();
+    }
 
     my $img = pick_image_from_body ($base2, $body2);
+    $body2 = undef;
 
     if ($img) {
         return ($base2, $img,
@@ -710,13 +760,13 @@ sub pick_from_image_randomizer {
 # to an image.  Returns two URLs, both of which are the URL of the image.
 # Returns () if nothing found this time.
 #
-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" );
-}
+#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" );
+#}
 
 
 # Picks a random image on a random page, and returns two URLs:
@@ -751,10 +801,10 @@ sub pick_image {
         $total = ++$total_1;
         $count = ++$count_1 if $img;
 
-    } elsif ($r < 70) {
-        ($base, $img, $source) = pick_from_photo_randomizer ($timeout);
-        $total = ++$total_4;
-        $count = ++$count_4 if $img;
+#    } elsif ($r < 70) {
+#        ($base, $img, $source) = pick_from_photo_randomizer ($timeout);
+#        $total = ++$total_4;
+#        $count = ++$count_4 if $img;
 
 #    } elsif ($r < 80) {
 #        # HotBot sucks: 98% of the time, it says "no pages match your
@@ -992,12 +1042,14 @@ sub image_to_pnm {
             }
             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 ($verbose > 3) { print STDERR "$progname: awaiting $pid\n"; }
@@ -1026,10 +1078,12 @@ sub image_to_pnm {
     die if ($@ && $@ ne "alarm\n");       # propagate errors
     if ($@) {
         # timed out
+        $body = undef;
         return ();
     } else {
         # didn't
         alarm 0;
+        $body = undef;
         return @_;
     }
 }
@@ -1038,8 +1092,10 @@ 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/;
 
     # make sure the various programs we execute exist, right up front.
     foreach ("ppmmake", "giftopnm", "djpeg", "pnmpaste", "pnmscale",
@@ -1051,8 +1107,10 @@ sub x_output {
         $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_1;
     } elsif (which($win_cmd_2)) {
         $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_2;
-    } else {
-        die "$progname: neither $win_cmd_1 nor $win_cmd_2 found on \$PATH.\n";
+    } elsif (which($win_cmd_3)) {
+        $ppm_to_root_window_cmd = $ppm_to_root_window_cmd_3;
+     } else {
+        die "$progname: didn't find $win_cmd_1, $win_cmd_2, or $win_cmd_3 on \$PATH.\n";
     }
 
     $SIG{HUP}  = \&x_cleanup;
@@ -1070,6 +1128,9 @@ sub x_output {
         which ($_) || die "$progname: $_ not found on \$PATH.\n";
         $_ = `$_`;
         ($img_width, $img_height) = m/dimensions: *(\d+)x(\d+) /;
+        if (!defined($img_height)) {
+            die "$progname: xdpyinfo failed.\n";
+        }
     }
 
     my $bgcolor = "#000000";
@@ -1134,6 +1195,7 @@ sub x_output {
         $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 "$progname: subproc exited normally.\n";
@@ -1148,6 +1210,7 @@ sub x_output {
             my ($headers, $body) = get_document ($img, $base);
             if ($body) {
                 handle_image ($base, $img, $body, $source);
+                $body = undef;
             }
         }
         unlink $image_tmp1, $image_tmp2;
@@ -1163,6 +1226,7 @@ sub handle_image {
     }
 
     my ($iw, $ih) = image_to_pnm ($img, $body, $image_tmp1);
+    $body = undef;
     return 0 unless ($iw && $ih);
 
     my $ow = $iw;  # used only for error messages
@@ -1471,7 +1535,7 @@ sub main {
 
     if (!$root_p && !$no_output_p) {
         die "$copyright" .
-            "$progname: the -root argument is manditory (for now.)\n";
+            "$progname: the -root argument is mandatory (for now.)\n";
     }
 
     if (!$no_output_p && !$ENV{DISPLAY}) {