createTestStories

#!/usr/bin/perl -w
# This code is a part of Slash, and is released under the GPL.
# Copyright 1997-2005 by Open Source Technology Group. See README
# and COPYING for more information, or see http://slashcode.com/.
# $Id$
# Based on Cliff's createTestComments, bugs are probably his fault :)

BEGIN {
	{
		require Silly::Werder;
		Silly::Werder->import;
	}
	die "Installation of Silly::Werder is required for this util, sorry!\n"
		if $@;
}

use strict;
use File::Basename;
use Getopt::Std;
use Data::Dumper;
use Date::Manip qw( UnixDate ParseDateString );
use Slash;
use Slash::Utility;
use Benchmark;
use Slash::Relocate;

use vars qw( $slashdb $werder $constants );

(my $VERSION) = ' $Revision$ ' =~ /\$Revision:\s+([^\s]+)/;
my $PROGNAME = basename($0);

my (%opts, %family_tree);
# Remember to doublecheck these match usage()!
usage('Options used incorrectly') unless getopts('hvu:t:', \%opts);
usage() if $opts{'h'};
version() if $opts{'v'};
$opts{'u'} ||= 'slash';
$opts{'t'} ||= 0;
$opts{num_stories} = $ARGV[0] || 10;
usage('Invalid number of stories') 
	if $opts{num_stories} !~ /^\d+$/ || $opts{num_stories} < 0;

createEnvironment($opts{u});
setCurrentSkin(determineCurrentSkin());
$slashdb = getCurrentDB();
$werder = new Silly::Werder;
$werder->set_syllables_num(2, 6);
$constants = getCurrentStatic();

# main program logic (in braces to offset nicely)
{
 
	# Determine user list size (for -R)
	my $max_uid = $slashdb->countUsers({ max => 1});
	my $min_uid = $slashdb->sqlSelect("MIN(uid)", "users");

	my $rootdir = getCurrentSkin('rootdir');
 
	my $t0 = new Benchmark;
	my $count = $opts{num_stories};
	while ($count) {
		# Let's have a little fun with this...
		my $title = make_werds(1 + rand 8);
		$title =~ s/\W+$// if rand(1) < 0.9;
		$title =~ s/<[^>]+>//g;
		$title =~ s/\b(\w)/\U$1/g if rand(1) < 0.4;
		my $introtext = make_werds(20 + rand 40 + rand 50, 0.3);
		$introtext .= typical_remark() if rand(1) < 0.05;
		$introtext = slashizeLinks($introtext);
		$introtext = balanceTags($introtext);
		my $bodytext = '';
		if (rand(1) < 0.1) {
			$bodytext = make_werds(20 + (rand 30) * (rand 50), 0.1);
			$bodytext =~ s{([.?!] )}{
				$1 . ((rand(1) < 0.2) ? "<P>" : "")
			}xge;
		}
		$bodytext = slashizeLinks($bodytext);
		$bodytext = balanceTags($bodytext);
		my $dept = lc make_werds(2+rand 6);
		$dept =~ s/<[^>]+>//g;
		$dept =~ s/\s+/-/g;
		$dept =~ s/\W+$// if rand(1) < 0.9;
		my $author = getRandomAuthorID();
		my $submitter = getRandomUserID($min_uid, $max_uid);
		my $time = UnixDate(ParseDateString(
			"epoch ".get_time($opts{t})),
			"%Y-%m-%d %H-%M-%S");

		my @topics = getRandomTopicList();

		my $topics = { };
		%$topics = map { $_ => 20 } @topics;

		# If this story will display in multiple nexuses, we might
		# want to manually remove it from one, just for fun.
		my $nexuses = $slashdb->getNexuslistFromChosen($topics);
		if (@$nexuses >= 2 && rand(1) < 0.2) {
			my $n = $nexuses->[ rand(@$nexuses) ];
			$topics->{$n} = 0;
		}

		if (rand(1) < 0.80) {
			promote_to_mainpage($topics);
		}

		my $commentstatus = rand(1) < .95 ? "enabled" : "disabled";
		
		my $story = {
			subid		=> '',
			uid		=> $author,
			dept		=> $dept,
			'time'		=> $time,
			title		=> $title,
			topics_chosen	=> $topics,
			submitter	=> $submitter,
			introtext	=> $introtext,
			bodytext	=> $bodytext,
			rendered	=> undef,
			commentstatus   => $commentstatus
		};

		# Small chance of ND'ing this story.
		$story->{neverdisplay} = 1 if rand(1) < 0.1;

		my $sid = $slashdb->createStory($story);
		my $reloDB = getObject("Slash::Relocate");
		if ($reloDB && $sid) {
			$slashdb->setStory($sid, { 
				introtext => $reloDB->href2SlashTag($introtext, $sid),
				bodytext => $reloDB->href2SlashTag($bodytext, $sid),
			});
		}
		if ($sid) {
			print "($count) Story created: $sid $time $title\n";
		} else {
			warn "($count) Story could not be created";
		}
		$count--;
	}
	my $t1 = new Benchmark;
	print "$opts{num_stories} stories created in: ",
		timestr(timediff($t1, $t0), 'noc'),"\n";
}

# subroutines

sub promote_to_mainpage {
	my($topics_hr) = @_;
	my @topics = keys %$topics_hr;
	# Pick several topics to be mainpage-worthy.
	my %m_tid = map { $_ => rand(1) < 0.5 ? (rand(1) < 0.3 ? 40 : 30) : 10 }
		@topics;
	# Need at least one.
	my $rand_key = $topics[rand(@topics)];
	$m_tid{$rand_key} = 30;
	# elevate those to the mainpage
	foreach (@topics) {
		$topics_hr->{$_} = $m_tid{$_};
	}
	# Did it work?  If not, add mainpage manually.
	my $mainpage_nexus = $constants->{mainpage_nexus_tid};
	my $rendered_hr = $slashdb->renderTopics($topics_hr);
	if (!$rendered_hr->{$mainpage_nexus}) {
		$topics_hr->{$mainpage_nexus} = 30;
	}
}


sub get_time {
	my($direction) = @_;
	my($min, $max);
	my $now = time;
	if (!$direction) {
		$min = $now - 86400*1;
		$max = $now + 86400*5;
	} elsif ($direction < 0) {
		$min = $now + 86400*$direction;
		$max = $now;
	} else {
		$min = $now;
		$max = $now + 86400*$direction;
	}
	my $t = int(rand($max-$min)) + $min;
	return int($t/60) * 60;
}

sub getRandomAuthorID {
	# Yes, inefficient.  I do not care.
	my $rand_uid = 1;
	my $n_tries = 100;
	my $uids = $slashdb->sqlSelectAll("uid", "users", "seclev >= 100");
	my @uids = map { $_->[0] } @$uids;
	my $ok = 0;
	while (!$ok) {
		die "can't get random author! '@uids'" if --$n_tries<0;
		$rand_uid = $uids[ rand @uids ];
		$ok = 1;
		$ok = 0 if $rand_uid == $constants->{anonymous_coward_uid};
	}
	return $rand_uid;
}

sub getRandomTopicID {
	# Yes, inefficient.  I do not care.
	my $tids = $slashdb->sqlSelectAll("tid", "topics", "tid < 10000");
	my @tids = map { $_->[0] } @$tids;
	my $rand_tid = @tids[ rand @tids ];
	return $rand_tid;
}

sub getRandomTopicList {
	my @topics = ( );
	my $r = rand(1);
	my $min_num = 1;
	   if ($r < 0.1) {	$min_num = 3	}
	elsif ($r < 0.5) {	$min_num = 2	}
	for (1..$min_num) { push @topics, getRandomTopicID(); }

	# First try to get into a section with weight 10.
	my $in_skin = topiclistInSkin(10, @topics);
	my $n_tries = 0;
	while (!$in_skin && $n_tries < 10) {
		++$n_tries;
		$topics[ rand(@topics) ] = getRandomTopicID();
		$in_skin = topiclistInSkin(10, @topics);
	}

	# No?  OK, try adding more topics.
	$n_tries = 0;
	while (!$in_skin && $n_tries < 10) {
		++$n_tries;
		my $new_tid = getRandomTopicID();
		if (@topics <= 5 && rand(1) < 0.5) {
			push @topics, $new_tid;
		} else {
			$topics[ rand(@topics) ] = $new_tid;
		}
		$in_skin = topiclistInSkin(10, @topics);
	}

	my %topics = ( map { ($_, 1) } @topics );
	@topics = keys %topics;
	return @topics;
}

sub topiclistInSkin {
	my($weight, @topics) = @_;
	my $chosen_hr = { map { ( $_, $weight ) } @topics };
	my $rendered_hr = $slashdb->renderTopics($chosen_hr);
	my $skid = $slashdb->getPrimarySkidFromRendered($rendered_hr);
	return $skid ? 1 : 0;
}

sub getRandomUserID {
	my($min_uid, $max_uid) = @_;
	my $rand_uid = $slashdb->sqlSelect("uid", "users",
		"uid BETWEEN $min_uid AND $max_uid", "ORDER BY RAND() LIMIT 1");
	return $rand_uid;
}

sub make_werds {
	my($werds, $link_prob) = @_;
	$werds = int($werds);
	$link_prob ||= 0;
	if ($werds > 40) {
		$werder->set_werds_num(5,20);
	} elsif ($werds > 20) {
		$werder->set_werds_num(3,10);
	} else {
		$werder->set_werds_num($werds, $werds);
	}
	my @links = qw(
		http://slashdot.org/		http://www.brunching.com/
		http://www.newsforge.com	http://ostg.com
		http://whitehouse.gov		http://www.whitehouse.gov/
		http://us.imdb.com/		http://www.kernel.org
		http://ic.ac.uk			http://cr.yp.to/
		http://www.verizonreallyreallyreallysucks.com/
		ftp://ftp.kernel.org/		ftp://ftp.gnu.org/gnu/Licenses/COPYING-2.0
		news:news.admin.net-abuse.email gopher://ccat.sas.upenn.edu:3333/11/Fiction/
		http://List-Etiquette.com/about/
		http://slashdot.org/features/99/03/31/0137221.shtml
		http://www.nap.edu/readingroom/books/newpath/chap2.html
		http://www.forthnet.gr/forthnet/isoc/short.history.of.internet
		http://www.freenix.fr/unix/linux/HOWTO/French-HOWTO.html
		http://www.linux.org.mx/	http://www.linux.net.mx/linux.php
		http://www.linux.org.uk/diary/	http://hepwww.ph.qmw.ac.uk/HEPpc/
		http://www.w3.org		http://peipa.essex.ac.uk/tp-linux/
		http://www.crosswinds.net/~tvquotes/futurama/2.html
		http://www.google.com/search?q=cache:h58W7o4qU-w:www.space.com/sciencefiction/tv/futurama_gore_000516_wg.html+Futurama+Fry&amp;hl=en
		http://dmoz.org/Regional/Europe/United_Kingdom/Wales/Isle_of_Anglesey/Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch/
		www.google.com			bad_link b.a.d l.i.n.k . .. ... ....
		mailto:jamie@slashdot.org
	);
	push @links, 'http://directory.google.com/Top/Arts/Literature/Authors/C/Carroll,_Lewis/Works/Hunting_of_the_Snark,_The/';
	my $comment;
	my $cur_werds = 0;
	my $tag = "";
	my $quote = q{};
	while ($cur_werds < $werds) {
		my $new_werds = '';
		if (int(rand(6))==0 or $werds < $cur_werds * 1.1) {
			if (int(rand(2))) {
				$new_werds = $werder->question;
			} else {
				$new_werds = $werder->exclaimation; # Dave, you spelled this word wrong
			}
		} else {
			$new_werds = $werder->sentence;
		}
		if ($link_prob && rand(1) < $link_prob) {
			my @new_werds = split / /, $new_werds;
			if ($#new_werds > 2) {
				my $a_href = q{<a href="} . $links[rand @links] . q{">} . (rand(1)<0.1?" ":"");
				my $close_a = (rand(1)<0.1?" ":"") . "</a>";
				my $insert_start = int(rand($#new_werds-2));
				my $insert_end = $insert_start + int(rand(4))+1;
				$insert_end = $#new_werds - 1 if $insert_end > $#new_werds - 1;
				$new_werds = join(" ",
					@new_werds[0..$insert_start-1],
					$a_href . $new_werds[$insert_start],
					@new_werds[$insert_start+1..$insert_end-1],
					$new_werds[$insert_end] . $close_a,
					@new_werds[$insert_end+1..$#new_werds]
				);
			}
		}
		if ($cur_werds and !$tag and rand > $cur_werds/$werds+0.3) {
			if (rand(1) < 0.95) { $tag = "i" }
			else { $tag = "b" }
			if (rand(1) < 0.50) { $quote = int(rand(3)) ? q{'} : q{"} };
			$new_werds = "<$tag>$quote$new_werds";
		}
		if ($tag and rand(1) < 0.5) {
			$new_werds = "$new_werds$quote</$tag>";
			$tag = ""; $quote = q{};
		}
		$comment .= " $new_werds";
		$comment =~ s{\s+}{ }g;
		$cur_werds = $comment =~ tr/ / /;
	}
	$comment .= "</$tag>" if $tag;
	$comment =~ s{</(\w+)>\s+<\1>}{ }g;
	$comment =~ s/^\s*(.+)\s*$/$1/;
	$comment =~ s/^\s+//; $comment =~ s/\s+$//;
	$comment;
}

sub typical_remark {
	my @t = (
		"What do you think?",
		"This totally sucks.",
		"That's too bad.",
		"This sucks. Hope they change their mind.",
		"<b>Sweet!</b>",
		"Ahahahaha!",
		"What a relief.",
		"I'm so sick of people submitting this.",
		"OK, now we've posted it, you can stop submitting!!",
		"This is just dumb.",
		"This is so lame.",
		"Click for more.",
		"Some people just don't get it.",
		"Why don't they just go open-source?",
		"That's free as in beer.",
		"That's free as in speech.",
		"Dude, just apt-get.",
		"I wouldn't know, I use Debian.",
		"Someone make a .deb of this.",
		"<b>Update at 14:59 by <a href=\"me\@privacy.net\">J</a></b>: Sorry, I got this completely wrong. My bad.",
		"<b>Update at 18:57 by <a href=\"me\@privacy.net\">H</a></b>: Yes, we posted this last month - sorry.",
	);
	return " " . $t[rand @t];
}

sub usage {
	print "*** $_[0]\n" if $_[0];
	# Remember to doublecheck these match getopts()!
	print <<EOT;

Usage: $PROGNAME [OPTIONS] [#stories]

This utility creates test stories for a given Slash site. This program is for
testing purposes, only, particularly for those ambitious Slash users out there
who want to try their hand at modifying the comment or moderation systems.

Main options:
	-h	Help (this message)
	-v	Version
	-u	Virtual user (default is "slash")

EOT
	exit;
}


sub version {
	print <<EOT;

$PROGNAME $VERSION

This code is a part of Slash, and is released under the GPL.
Copyright 1997-2005 by Open Source Technology Group. See README
and COPYING for more information, or see http://slashcode.com/.

EOT
	exit;
}

__END__