#!/usr/bin/perl -w
-# Copyright © 2005-2008 Jamie Zawinski <jwz@jwz.org>
+# Copyright © 2005-2013 Jamie Zawinski <jwz@jwz.org>
#
# Permission to use, copy, modify, distribute, and sell this software and its
# documentation for any purpose is hereby granted without fee, provided that
#use diagnostics; # Fails on some MacOS 10.5 systems
use strict;
+# Some Linux systems don't install LWP by default!
+# Only error out if we're actually loading a URL instead of local data.
+BEGIN { eval 'use LWP::UserAgent;' }
+
use Socket;
use POSIX qw(strftime);
use Text::Wrap qw(wrap);
use bytes;
my $progname = $0; $progname =~ s@.*/@@g;
-my $version = q{ $Revision: 1.17 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
+my $version = q{ $Revision: 1.29 $ }; $version =~ s/^[^0-9]+([0-9.]+).*$/$1/;
my $verbose = 0;
my $http_proxy = undef;
my $text_literal = '';
my $text_file = '';
my $text_program = '';
-my $text_url = '';
+my $text_url = 'http://twitter.com/statuses/public_timeline.atom';
+# Default URL needs to be set and match what's in OSX/XScreenSaverView.m
my $wrap_columns = undef;
my $nyarlathotep_p = 0;
my ($text) = @_;
$text =~ s/(&(\#)?([[:alpha:]\d]+);?)/
{
- my $c;
- if ($2) {
- $c = chr($3); # the &#number is always decimal, right?
+ my $c = $3;
+ if (! defined($2)) {
+ $c = $entity_table{$c}; # for Á
} else {
- $c = $entity_table{$3};
+ if ($c =~ m@^x([\dA-F]+)$@si) { # for A
+ $c = chr(hex($1));
+ } elsif ($c =~ m@^\d+$@si) { # for A
+ $c = chr($c);
+ } else {
+ $c = undef;
+ }
}
-# print STDERR "$progname: warning: unknown HTML character entity \"$1\"\n"
-# unless $c;
- ($c ? $c : "[$3]");
+ ($c || "[$3]"); # for &unknown; => "[unknown]"
}
/gexi;
return $text;
}
if ($verbose > 1) {
- printf STDERR "$progname: mode: $text_mode\n";
- printf STDERR "$progname: literal: $text_literal\n";
- printf STDERR "$progname: file: $text_file\n";
- printf STDERR "$progname: program: $text_program\n";
- printf STDERR "$progname: url: $text_url\n";
+ print STDERR "$progname: mode: $text_mode\n";
+ print STDERR "$progname: literal: $text_literal\n";
+ print STDERR "$progname: file: $text_file\n";
+ print STDERR "$progname: program: $text_program\n";
+ print STDERR "$progname: url: $text_url\n";
}
$text_mode =~ tr/A-Z/a-z/;
$text_literal =~ s@\\n@\n@gs;
+ $text_literal =~ s@\\\n@\n@gs;
}
elsif ($text_mode eq '1') { $text_mode = 'literal'; }
elsif ($text_mode eq '2') { $text_mode = 'file'; }
elsif ($text_mode eq '3') { $text_mode = 'url'; }
+ elsif ($text_mode eq '4') { $text_mode = 'program'; }
$v = get_cocoa_pref_1 ($id, "textLiteral");
$text_literal = $v if defined ($v);
+ $text_literal =~ s@\\n@\n@gs;
+ $text_literal =~ s@\\\n@\n@gs;
$v = get_cocoa_pref_1 ($id, "textFile");
$text_file = $v if defined ($v);
} else { # $text_mode eq 'date'
- safe_system ("uname", "-n");
+ my $n = `uname -n`;
+ $n =~ s/\.local\n/\n/s;
+ print $n;
my $unamep = 1;
if (-f "/etc/redhat-release") { # "Fedora Core release 4 (Stentz)"
- system ("cat", "/etc/redhat-release");
+ safe_system ("cat", "/etc/redhat-release");
}
if (-f "/etc/release") { # "Solaris 10 3/05 s10_74L2a X86"
my $sp = # "iMac G5"
`/usr/sbin/system_profiler SPSoftwareDataType SPHardwareDataType`;
my ($v) = ($sp =~ m/^\s*System Version:\s*(.*)$/mi);
- my ($s) = ($sp =~ m/^\s*CPU Speed:\s*(.*)$/mi);
- my ($t) = ($sp =~ m/^\s*Machine Name:\s*(.*)$/mi);
+ my ($s) = ($sp =~ m/^\s*(?:CPU|Processor) Speed:\s*(.*)$/mi);
+ my ($t) = ($sp =~ m/^\s*(?:Machine|Model) Name:\s*(.*)$/mi);
print "$v\n" if ($v);
print "$s $t\n" if ($s && $t);
$unamep = !defined ($v);
}
-# Loads the given URL, returns: $http, $head, $body.
-#
-sub get_url_1($;$) {
- my ($url, $referer) = @_;
-
- if (! ($url =~ m@^(http|feed)://@i)) {
- error ("not an HTTP URL: $url");
- }
-
- my ($url_proto, $dummy, $serverstring, $path) = split(/\//, $url, 4);
- $path = "" unless $path;
-
- my ($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) {
- error ("unrecognised port in $url");
- }
-
- $iaddr = inet_aton($remote);
- error ("host not found: $remote") unless ($iaddr);
-
- $paddr = sockaddr_in($port2, $iaddr);
-
-
- my $head = "";
- my $body = "";
-
- $proto = getprotobyname('tcp');
- if (!socket(S, PF_INET, SOCK_STREAM, $proto)) {
- error ("socket: $!");
- }
- if (!connect(S, $paddr)) {
- error ("connect($serverstring): $!");
- }
-
- select(S); $| = 1; select(STDOUT);
-
- my $user_agent = "$progname/$version";
-
- my $hdrs = ("GET " . ($http_proxy ? $url : "/$path") . " HTTP/1.0\r\n" .
- "Host: $them\r\n" .
- "User-Agent: $user_agent\r\n");
- if ($referer) {
- $hdrs .= "Referer: $referer\r\n";
- }
- $hdrs .= "\r\n";
-
- if ($verbose > 3) {
- foreach (split('\r?\n', $hdrs)) {
- print STDERR " ==> $_\n";
- }
- }
- print S $hdrs;
- my $http = <S> || "";
-
- $_ = $http;
- s/[\r\n]+$//s;
- print STDERR " <== $_\n" if ($verbose > 3);
-
- while (<S>) {
- $head .= $_;
- s/[\r\n]+$//s;
- last if m@^$@;
- print STDERR " <== $_\n" if ($verbose > 3);
- }
-
- print STDERR " <== \n" if ($verbose > 4);
- my $lines = 0;
- while (<S>) {
- s/\r\n/\n/gs;
- print STDERR " <== $_" if ($verbose > 4);
- $body .= $_;
- $lines++;
- }
-
- print STDERR " <== [ body ]: $lines lines, " . length($body) . " bytes\n"
- if ($verbose == 4);
-
- close S;
-
- if (!$http) {
- error ("null response: $url");
- }
-
- return ( $http, $head, $body );
-}
-
-
-# Loads the given URL, processes redirects, returns (content-type, body).
-#
-sub get_url($;$) {
- my ($url, $referer) = @_;
-
- print STDERR "$progname: loading $url\n" if ($verbose > 2);
-
- my $orig_url = $url;
- my $loop_count = 0;
- my $max_loop_count = 10;
-
- do {
- my ( $http, $head, $body ) = get_url_1 ($url, $referer);
-
- $http =~ s/[\r\n]+$//s;
-
- if ( $http =~ m@^HTTP/[0-9.]+ 30[123]@ ) {
- $_ = $head;
-
- my ( $location ) = m@^location:[ \t]*(.*)$@im;
- if ( $location ) {
- $location =~ s/[\r\n]$//;
-
- print STDERR "$progname: redirect from $url to $location\n"
- if ($verbose > 3);
-
- $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 = $_ . $url;
- }
-
- } else {
- error ("no Location with \"$http\"");
- }
-
- if ($loop_count++ > $max_loop_count) {
- error ("too many redirects ($max_loop_count) from $orig_url");
- }
-
- } elsif ( $http =~ m@^HTTP/[0-9.]+ ([4-9][0-9][0-9].*)$@ ) {
- error ("failed: $1 ($url)");
-
- } else {
- my $ct = 'text/plain';
- $ct = $1 if ($head =~ m/^content-type:\s*([^\s]+)/mi);
- return ($ct, $body);
- }
- } while (1);
-}
-
-
# Make an educated guess as to what's in this document.
# We don't necessarily take the Content-Type header at face value.
# Returns 'html', 'rss', or 'text';
return 'text';
}
+
sub reformat_html($$) {
my ($body, $rss_p) = @_;
$_ = $body;
+ # In HTML, try to preserve newlines inside of PRE.
+ #
if (! $rss_p) {
- # In HTML, unfold lines (this breaks PRE. Sue me.)
+ s@(<PRE\b[^<>]*>\s*)(.*?)(</PRE)@{
+ my ($a, $b, $c) = ($1, $2, $3);
+ $b =~ s/[\r\n]/<BR>/gs;
+ $a . $b . $c;
+ }@gsexi;
+ }
+
+ if (! $rss_p) {
+ # In HTML, unfold lines.
# In RSS, assume \n means literal line break.
s@[\r\n]@ @gsi;
}
s@<[^<>]*>?@@gs; # lose all other HTML tags
$_ = de_entify ($_); # convert HTML entities
+ # For Wikipedia: delete anything inside {{ }} and unwrap [[tags]]
+ #
+ if ($rss_p eq 'wiki') {
+ s@/\*.*?\*/@@si; # /* ... */
+ 1 while (s/{{[^{}]*}}//gs); # {{ ... }}
+ s/\[\[([^:\[\]\|]+)\|([^\[\]]+)\]\]/$2/gs; # [[link|anchor]]
+ s/\[\[([^:\[\]\|]+)\]\]/$1/gs; # [[anchor]]
+ s/\[http:[^\[\]\s]+\s+([^\[\]]+)\]/$1/gs; # [url anchor]
+# s@\s*<ref>.*?</ref>@*@gs; # <ref>url<ref> -> "*"
+ s/<[^<>]*>//gs; # <tags> -- omit.
+ }
+
+
# elide any remaining non-Latin1 binary data...
s/([\177-\377]+(\s*[\177-\377]+)[^a-z\d]*)/«...» /g;
#s/([\177-\377]+(\s*[\177-\377]+)[^a-z\d]*)/«$1» /g;
if (!defined($wrap_columns) || $wrap_columns > 0) {
$Text::Wrap::columns = ($wrap_columns || 72);
+ $Text::Wrap::break = '[\s/]'; # wrap on slashes for URLs
$_ = wrap ("", " ", $_); # wrap the lines as a paragraph
s/[ \t]+$//gm; # lose whitespace at end of line again
}
sub reformat_rss($) {
my ($body) = @_;
+ my $wiki_p = ($body =~ m@<generator>[^<>]*Wiki@si);
+
$body =~ s/(<(ITEM|ENTRY)\b)/\001\001$1/gsi;
my @items = split (/\001\001/, $body);
$title = '' if ($body1 eq $title); # Identical in Twitter's atom feed.
- reformat_html ("$title<P>$body1", 1);
+ reformat_html ("$title<P>$body1", $wiki_p ? 'wiki' : 'rss');
print "\n";
}
}
if ($wrap_columns && $wrap_columns > 0) {
print STDERR "$progname: wrapping at $wrap_columns...\n" if ($verbose > 2);
$Text::Wrap::columns = $wrap_columns;
+ $Text::Wrap::break = '[\s/]'; # wrap on slashes for URLs
$body = wrap ("", "", $body);
$body =~ s/[ \t]+$//gm;
}
}
+# Figure out what the proxy server should be, either from environment
+# variables or by parsing the output of the (MacOS) program "scutil",
+# which tells us what the system-wide proxy settings are.
+#
+sub set_proxy($) {
+ my ($ua) = @_;
+
+ if (!defined($ENV{http_proxy}) && !defined($ENV{HTTP_PROXY})) {
+ my $proxy_data = `scutil --proxy 2>/dev/null`;
+ my ($server) = ($proxy_data =~ m/\bHTTPProxy\s*:\s*([^\s]+)/s);
+ my ($port) = ($proxy_data =~ m/\bHTTPPort\s*:\s*([^\s]+)/s);
+ if ($server) {
+ # Note: this ignores the "ExceptionsList".
+ $ENV{http_proxy} = "http://" . $server . ($port ? ":$port" : "") . "/";
+ print STDERR "$progname: MacOS proxy: $ENV{http_proxy}\n"
+ if ($verbose > 2)
+ }
+ }
+
+ $ua->env_proxy();
+}
+
+
sub get_url_text($) {
my ($url) = @_;
- # historical suckage: the environment variable name is lower case.
- $http_proxy = $ENV{http_proxy} || $ENV{HTTP_PROXY};
+ my $ua = eval 'LWP::UserAgent->new';
- if ($http_proxy && $http_proxy =~ m@^http://([^/]*)/?$@ ) {
- # historical suckage: allow "http://host:port" as well as "host:port".
- $http_proxy = $1;
+ if (! $ua) {
+ print STDOUT ("\n\tPerl is broken. Do this to repair it:\n" .
+ "\n\tsudo cpan LWP::UserAgent\n\n");
+ return;
}
- my ($ct, $body) = get_url ($url);
+ set_proxy ($ua);
+ $ua->agent ("$progname/$version");
+ my $res = $ua->get ($url);
+ my $body;
+ my $ct;
+
+ if ($res && $res->is_success) {
+ $body = $res->decoded_content || '';
+ $ct = $res->header ('Content-Type') || 'text/plain';
+
+ } else {
+ my $err = ($res ? $res->status_line : '') || '';
+ $err = 'unknown error' unless $err;
+ $err = "$url: $err";
+ # error ($err);
+ $body = "Error loading URL $err\n\n";
+ $ct = 'text/plain';
+ }
$ct = guess_content_type ($ct, $body);
if ($ct eq 'html') {
elsif (m/^--?date$/) { $text_mode = 'date';
$load_p = 0; }
elsif (m/^--?text$/) { $text_mode = 'literal';
- $text_literal = shift @ARGV;
+ $text_literal = shift @ARGV || '';
$load_p = 0; }
elsif (m/^--?file$/) { $text_mode = 'file';
- $text_file = shift @ARGV;
+ $text_file = shift @ARGV || '';
$load_p = 0; }
elsif (m/^--?program$/) { $text_mode = 'program';
- $text_program = shift @ARGV;
+ $text_program = shift @ARGV || '';
$load_p = 0; }
elsif (m/^--?url$/) { $text_mode = 'url';
- $text_url = shift @ARGV;
+ $text_url = shift @ARGV || '';
$load_p = 0; }
elsif (m/^--?col(umn)?s?$/) { $wrap_columns = 0 + shift @ARGV; }
elsif (m/^--?cocoa$/) { $cocoa_id = shift @ARGV; }
}
output();
+
+
+ if (defined ($cocoa_id)) {
+ #
+ # On MacOS, sleep for 10 seconds between when the last output is
+ # printed, and when this process exits. This is because MacOS
+ # 10.5.0 and later broke ptys in a new and exciting way: basically,
+ # once the process at the end of the pty exits, you have exactly
+ # 1 second to read all the queued data off the pipe before it is
+ # summarily flushed.
+ #
+ # Many of the screen savers were written to depend on being able
+ # to read a small number of bytes, and continue reading until they
+ # reached EOF. This is no longer possible.
+ #
+ # Note that the current MacOS behavior has all four of these
+ # awesome properties: 1) Inconvenient; 2) Has no sane workaround;
+ # 3) Different behavior than MacOS 10.1 through 10.4; and 4)
+ # Different behavior than every other Unix in the world.
+ #
+ # See http://jwz.livejournal.com/817438.html, and for those of
+ # you inside Apple, "Problem ID 5606018".
+ #
+ # One workaround would be to rewrite the savers to have an
+ # internal buffer, and always read as much data as possible as
+ # soon as a pipe has input available. However, that's a lot more
+ # work, so instead, let's just not exit right away, and hope that
+ # 10 seconds is enough.
+ #
+ # This will solve the problem for invocations of xscreensaver-text
+ # that produce little output (e.g., date-mode); and won't solve it
+ # in cases where a large amount of text is generated in a short
+ # amount of time (e.g., url-mode.)
+ #
+ sleep (10);
+ }
}
main();