User:AnomieBOT/source/tasks/SourceUploader.pm

From Wikipedia, the free encyclopedia
package tasks::SourceUploader;

=pod

=begin metadata

Bot:     AnomieBOT
Task:    SourceUploader
BRFA:    N/A
Status:  Begun 2008-08-15
Created: 2008-08-16

Updates the pages under [[User:AnomieBOT/source]] to reflect the current source
of the bot.

=end metadata

=cut

use utf8;
use strict;

use AnomieBOT::Task qw/:time/;
use vars qw/@ISA/;
@ISA=qw/AnomieBOT::Task/;

use Storable qw/dclone/;
use URI::Escape;
use tasks::SourceUploader::WikiPod;
use tasks::SourceUploader::Pod;
use Data::Dumper;
use Fcntl ':mode';

my %extensions=(
    'pl' => 'perl',
    'pm' => 'perl',
    'ini' => 'ini',
    'sh' => 'bash',
    'css' => 'css',
    'js' => 'js',
    'json' => 'json',
    'html' => 'html',
);

my @sizes=('', 'KiB', 'MiB', 'GiB', 'TiB', 'PiB', 'EiB', 'ZiB', 'YiB');


=pod

=for info
Per [[WP:BOT#Approval]], any bot or automated editing process that only
affects only the operators' user and talk pages (or subpages thereof),
and which are not otherwise disruptive, may be run without prior
approval.

=cut

sub approved {
    return 999;
}

sub new {
    my $class=shift;
    my $self=$class->SUPER::new();
    $self->{'loadexisting'}=1;
    $self->{'order'}=-1000;
    $self->{'fail'}=0;
    return $self;
}

sub build_sources {
    my $self = shift;

    my $basedir=$AnomieBOT::API::basedir;
    $basedir=~s{/$}{};
    if(!-d $basedir){
        $self->{'fail'}=1;
        AnomieBOT::API->warn("Cannot find source directory\n");
        return;
    }

    $self->{'summary'}='Updating published sources: ';
    if(!open(X, "<:utf8", 'ChangeLog')){
        $self->{'fail'}=1;
        AnomieBOT::API->warn("Cannot load changelog: $!\n");
        return;
    }
    local $_;
    my $intro=1;
    while(<X>){
        if(/^==.*==$/){
            last if !$intro;
            $intro=0;
            next;
        }
        $self->{'summary'}.=$_ if !$intro;
    }
    close(X);
    $self->{'summary'}=~s/\s+/ /g;
    $self->{'summary'}=~s/\s+$//;
    $self->{'summary'}=$self->{'summary'};

    my $mainbot="\x02BOT\x03";
    my %pages=();
    my %tasks=(''=>{
        '01 Current'           => [],
        '05 Awaiting approval' => [],
        '06 Past'              => [],
    });
    $tasks{'TaskList'}=dclone($tasks{''});
    my @dirs=($basedir);
    $self->{'shutoff_pages'}=[];
    $self->{'check_shutoff_notice'}=[];
    while(my $dir=shift @dirs){
        if(!opendir(D, $dir)){
            $self->{'fail'}=1;
            AnomieBOT::API->warn("Cannot open directory $dir: $!\n");
            return;
        }

        my @dirpage=();
        while(my $page=readdir(D)){
            next if substr($page,0,1) eq '.';

            my $p="$dir/$page";
            my $pp=substr($p,length($basedir));
            my @stat=stat($p);
            my $img='Gnome-fs-executable.svg';
            if(-d $p){
                next if ($stat[2]&(S_IROTH|S_IXOTH))!=(S_IROTH|S_IXOTH);
                push @dirs, $p;
                $img='Gnome-fs-directory-visiting.svg';
            } elsif(-f $p){
                next if ($stat[2]&(S_IROTH))!=(S_IROTH);
                if(!open(X, '<:utf8', $p)){
                    $self->{'fail'}=1;
                    AnomieBOT::API->warn("Cannot open file $p: $!\n");
                    return;
                }
                do {
                    local $/=undef;
                    $pages{$pp}=<X>;
                };
                close(X);
                my $top='';
                if($page eq 'ChangeLog'){
                    # Pull out most recent 10M of changelog entries
                    my @x=split(/\n/, $pages{$pp});
                    $pages{$pp}='';
                    while(defined(my $x=shift @x)){
                        last if(length($pages{$pp})>204800 && $x=~/^==.*==$/);
                        $pages{$pp}.=$x."\n";
                    }
                    $pages{$pp}.="\n__NOTOC__\n";
                    $pages{$pp}.="\x7b\x7bombox|type=notice|text=See the \x7b\x7bsubst:history|\x7b\x7bsubst:FULLPAGENAME\x7d\x7d|page history|subst=subst:\x7d\x7d for further ChangeLog entries\x7d\x7d\n" if @x;
                } elsif($page=~/\.([^.]+)$/ && exists($extensions{$1})){
                    if($extensions{$1} eq 'perl'){
                        # Try to construct POD documentation
                        my $parser = tasks::SourceUploader::WikiPod->new("User:$mainbot/source");
                        my $x='';
                        $parser->output_string(\$x);
                        $parser->parse_string_document($pages{$pp});
                        $pages{"$pp/doc"}=$x if($parser->content_seen);

                        # Handle embedded notices and metadata
                        $x='';
                        $parser=tasks::SourceUploader::Pod->new;
                        $parser->output_string(\$x);
                        $parser->parse_string_document($pages{$pp});
                        $top.=$x if($parser->content_seen);
                        my %metadata=$parser->metadata;
                        if(%metadata){
                            $x ="<noinclude>\n";
                            $x.="{| class=\"wikitable\"\n";
                            $x.="! Account !! Task !! Disable !! {{tlx|bots}} !! Status !! Description\n";
                            $x.="</noinclude>\n";
                            $x.="|- valign=\"top\"\n";
                            my $task=$metadata{'task'};
                            my $bot=$metadata{'bot'} // "\x02!!!\x03";
                            $x.="{{#if:{{{noacct|}}}||{{!}}align=\"center\"{{!}} [[User:$bot|$bot]]}}\n";
                            $x.="|align=\"center\"| [[User:$mainbot/source$pp|$task]]\n";
                            if(exists($metadata{'shutoff'}) && $metadata{'shutoff'} eq 'false'){
                                $x.="|align=\"center\"| No\n";
                            } else {
                                $x.="|align=\"center\"| <span class=\"plainlinks\">[{{fullurl:User:$bot/shutoff/$task|action=edit}} Here]</span>\n";
                                push @{$self->{'shutoff_pages'}}, "User:$bot/shutoff/$task";
                                push @{$self->{'check_shutoff_notice'}}, $bot;
                            }
                            if(exists($metadata{'exclusion'}) && $metadata{'exclusion'} eq 'false'){
                                $x.="|align=\"center\"| {{N}}\n";
                            } else {
                                $x.="|align=\"center\"| {{Y}}\n";
                            }
                            my $brfa=$metadata{'brfa'};
                            my $status=$metadata{'status'};
                            if($brfa eq 'N/A'){
                                $x.="| \x7b\x7bsort|$status|\x7d\x7d\x5b\x5bWikipedia:Bot policy#Approval|N/A\x5d\x5d, only edits bot's/owner's userspace. $status\n";
                            } elsif($brfa eq 'None'){
                                $x.="| $status\n";
                            } else {
                                $x.="| \x5b\x5b$brfa|$status\x5d\x5d\n";
                            }
                            if(exists($metadata{'+brfa'})){
                                $x.="<p style=\"margin:0;padding:0;font-size:smaller\">Supplemental:<br />\n";
                                for(my $i=0; $i<@{$metadata{'+brfa'}}; $i++){
                                    my $s=$metadata{'+status'}[$i];
                                    $s=$1 if $s=~/^Approved (\d{4}-\d{2}-\d{2})$/;
                                    $x.="+ \x5b\x5b".$metadata{'+brfa'}[$i].'|'.$s."\x5d\x5d<br />\n";
                                }
                                $x.="</p>\n";
                            }
                            $x.="|\n".$metadata{'*'}."\n";
                            $x.="<noinclude>\n";
                            $x.="|}\n";
                            $x.="</noinclude>";
                            $pages{"$pp/metadata"}=$x;
                            my $section=determine_task_section(".$pp", %metadata);
                            $tasks{"TaskList"}{$section}//=[];
                            push @{$tasks{"TaskList"}{$section}}, $metadata{'created'}." {{User:$mainbot/source$pp/metadata}}";
                            $tasks{"TaskList/$bot"}//=dclone($tasks{''});
                            $tasks{"TaskList/$bot"}{$section}//=[];
                            push @{$tasks{"TaskList/$bot"}{$section}}, $metadata{'created'}." {{User:$mainbot/source$pp/metadata|noacct=1}}";
                        }
                    }
                    $pages{$pp}="<syntaxhighlight lang=\"".$extensions{$1}."\">\n".$pages{$pp}."\n</syntax"."highlight>";
                } else {
                    $pages{$pp}="<pre>\n".$pages{$pp}."\n</pre>";
                }
                $top.="\x7b\x7bombox|text=See \x5b\x5b/doc\x5d\x5d for formatted documentation\x7d\x7d\n" if(exists($pages{"$pp/doc"}));
                $pages{$pp}=$top.$pages{$pp};
                $img='Gnome-fs-regular.svg';
            } else {
                AnomieBOT::API->warn("Unusual filetype on $p\n");
            }
            my $sz=$stat[7];
            my $i=0;
            while($sz>=1024 && $i<@sizes){ $sz/=1024; $i++; }
            ($sz="$sz")=~s/(\.\d\d)\d+/$1/;
            $sz.='&nbsp;'.$sizes[$i] if $i>0;
            my $dirpage="|-\n";
            $dirpage.="| \x5b\x5bImage:$img|32x32px\x5d\x5d";
            $dirpage.=" \x5b\x5b/$page|$page\x5d\x5d\n";
            $dirpage.="|align=\"right\"| \x7b\x7bsort|".sprintf("%020d",$stat[7])."|$sz\x7d\x7d\n";
            $dirpage.="|align=\"center\"| \x7b\x7bsort|".sprintf("%020d",$stat[9]).strftime("|%F %T (UTC)", gmtime($stat[9]))."\x7d\x7d\n";
            push @dirpage, { ts=>$stat[9], text=>$dirpage };
        }
        closedir(D);
        my $dirpage=substr($dir,length($basedir));
        $dirpage='/'.$dirpage if(substr($dirpage,0,1) ne '/');
        $dirpage ="==Index of $dirpage==\n";
        $dirpage.="{| class=\"wikitable sortable\" style=\"width:100%\"\n";
        $dirpage.="! Filename !! Size !! Modified\n";
        $dirpage.=join "", map $_->{'text'}, sort { $b->{'ts'}<=>$a->{'ts'} } @dirpage;
        $dirpage.="|}";
        $pages{substr($dir,length($basedir))}=$dirpage;
    }

    my %tasklists=();
    foreach my $pg (keys %tasks) {
        next if $pg eq '';
        my ($tl,$acct);
        if($pg=~/^TaskList\/(.*)$/){
            $tl="<noinclude>\nThis page lists basic information about every task coded for [[User:$1|$1]], including links to the bot approval and individual shutoff pages. For the full list of tasks for all associated bot accounts, see [[User:$mainbot/TaskList]].\n\n== Tasks ==\n</noinclude>";
            $acct='';
        } else {
            $tl="<noinclude>\nThis page lists basic information about every task coded for all $mainbot bots, including links to the bot approval and individual shutoff pages.\n\n== Tasks ==\n</noinclude>";
            $acct='! Account !';
        }
        foreach my $section (sort keys %{$tasks{$pg}}){
            my @links=sort @{$tasks{$pg}{$section}};
            $section=~s/^(\d+) //;
            my $n=$1;
            $tl.="\n=== $section ===\n";
            if(@links){
                $tl.="{| class=\"wikitable sortable\"\n";
                $tl.="$acct! Task !! Disable !! {{tlx|bots}} !! ".(($n==1 || $n==2)?'Approval':'Status')." !! Description\n";
                $tl.=join("\n", map { substr($_,11) } @links)."\n";
                $tl.="|}\n";
            } else {
                $tl.="None at this time.\n";
            }
        }
        $tasklists{$pg}=$tl;
    }

    $self->{'pages'}={%pages};
    $self->{'tasklists'}=\%tasklists;
}

sub run {
    my ($self, $api)=@_;

    return undef if $self->{'fail'};

    $self->build_sources() unless defined( $self->{'pages'} );
    return undef if $self->{'fail'};

    my @keys=keys(%{$self->{'pages'}});
    my @tlkeys=keys(%{$self->{'tasklists'}});
    my @shutoff=@{$self->{'shutoff_pages'}};
    my @checkshutoffnotice=@{$self->{'check_shutoff_notice'}};
    if(!@keys && !@tlkeys && !@shutoff && !@checkshutoffnotice){
        $api->debug(2, "Source uploaded, terminating");
        return undef;
    }

    $api->task('SourceUploader', 0, 10, qw/d::Redirects d::Talk/);

    my $src='User:'.$api->user.'/source';

    if($self->{'loadexisting'}){
        my $iter=$api->iterator(
            list        => 'allpages',
            apprefix    => $api->user.'/source',
            apnamespace => '2',
            aplimit     => 'max'
        );

        my ($k,$v,$k2,$v2,@x);
        while($_=$iter->next){
            if(!$_->{'_ok_'}){
                $api->warn("Failed to retrieve source tree: ".$_->{'error'}."\n");
                return 300;
            }
            $v2=substr($_->{'title'}, length($src));
            if(!exists($self->{'pages'}{$v2})){
                $self->{'pages'}{$v2}='';
                push @keys, $v2;
            }
        }

        $self->{'loadexisting'}=0;
    }

    while(@keys){
        my $page=shift @keys;
        my $text=$self->{'pages'}{$page};
        my $ret=$self->upload_page($api, $src.$page, $text);
        return $ret if $ret>60;
        next if $ret>0;
        $ret=redirect_talk($api, $src.$page);
        return $ret if $ret;
        delete($self->{'pages'}{$page});
    }

    while(@tlkeys){
        my $page=shift @tlkeys;
        my $text=$self->{'tasklists'}{$page};
        my $ret=$self->upload_page($api, 'User:'.$api->user."/$page", $text);
        return $ret if $ret>60;
        next if $ret>0;
        $ret=redirect_talk($api, 'User:'.$api->user."/$page");
        return $ret if $ret;
        delete($self->{'tasklists'}{$page});
    }

    while(@shutoff){
        my $page=shift @shutoff;
        my $tok=$api->edittoken($page);
        if($tok->{'code'} eq 'shutoff'){
            $api->warn("Task disabled: ".$tok->{'content'}."\n");
            return 300;
        }
        return 60 if($tok->{'code'} ne 'success');
        if(exists($tok->{'missing'})){
            my $r=$api->edit($tok, '{{subst:void}}', 'Creating empty shutoff page to avoid redlinks in summaries and to avoid confusing people', 1, 1);
            if($r->{'code'} ne 'success'){
                $api->warn("Write error for $page: ".$r->{'error'}."\n");
                return 60;
            } else {
                $api->log("Created $page");
            }
        }
        my $ret=redirect_talk($api, $page);
        return $ret if $ret;
        $self->{'shutoff_pages'}=[@shutoff];
    }

    my %x=map { $_=>1 } @checkshutoffnotice;
    @checkshutoffnotice=keys %x;
    $self->{'check_shutoff_notice'}=[@checkshutoffnotice];
    while(@checkshutoffnotice){
        my $bot=shift @checkshutoffnotice;

        my $res=$api->rawpage("Template:Editnotices/Group/User:$bot");
        if($res->{'code'} eq 'shutoff'){
            $api->warn("Task disabled: ".$res->{'content'}."\n");
            return 300;
        }
        $res->{'code'}='success' if($res->{'code'} eq 'httperror' && $res->{'httpcode'}==404);
        if($res->{'code'} ne 'success'){
            $api->warn("Could not load Template:Editnotices/Group/User:$bot: ".$res->{'error'}."\n");
            return 60;
        }
        $res=$res->{'content'}//'';
        unless($res=~/\{\{Editnotice subpages\|shutoff\|on base=no\}\}/){
            $api->whine("Please fix [[Template:Editnotices/Group/User:$bot]]", "To display the proper editnotice on shutoff pages for the AnomieBOT bot [[User:$bot]], please ensure that [[Template:Editnotices/Group/User:$bot]] contains <code><nowiki>{{Editnotice subpages|shutoff|on base=no}}</nowiki></code>. Thanks.");
        }

        $res=$api->rawpage("Template:Editnotices/Group/User:$bot/shutoff");
        if($res->{'code'} eq 'shutoff'){
            $api->warn("Task disabled: ".$res->{'content'}."\n");
            return 300;
        }
        $res->{'code'}='success' if($res->{'code'} eq 'httperror' && $res->{'httpcode'}==404);
        if($res->{'code'} ne 'success'){
            $api->warn("Could not load Template:Editnotices/Group/User:$bot/shutoff: ".$res->{'error'}."\n");
            return 60;
        }
        $res=$res->{'content'}//'';
        $res=~s/\s*$//;
        my $txt="{{#ifeq:{{PAGESIZE:{{FULLPAGENAME}}}}|0|\n{{ombox\n| type = delete\n| image = [[File:Shutdown button.svg|40px|link={{fullurl:Special:Block|wpTarget=".uri_escape($bot)."&wpExpiry=indefinite&wpHardBlock=1&wpAutoBlock=0&wpCreateAccount=0&wpReason=other&wpReason-other=Bot%20malfunctioning:%20}}|Emergency block button]]\n| text = To disable the task {{SUBPAGENAME}}, enter your reasoning in this page. If the task is currently running, the bot may make one more edit before noticing that it has been shut off, but rest assured it ''will'' notice.\n}}\n|{{ombox\n| text = To re-enable the task, blank the page.\n}}\n}}";
        unless($res=~/(?:^|\n)\Q$txt\E(?:$|\n)/){
            $api->whine("Please fix [[Template:Editnotices/Group/User:$bot/shutoff]]", "To display the proper editnotice on shutoff pages for the AnomieBOT bot [[User:$bot]], please ensure that [[Template:Editnotices/Group/User:$bot/shutoff]] contains the following text:\n<pre>$txt</pre>\nThanks.");
        }

        $self->{'check_shutoff_notice'}=[@checkshutoffnotice];
    }

    return 0;
}

sub upload_page {
    my $self=shift;
    my $api=shift;
    my $page=shift;
    my $text=shift;

    my $tok=$api->edittoken($page);
    if($tok->{'code'} eq 'shutoff'){
        $api->warn("Task disabled: ".$tok->{'content'}."\n");
        return 300;
    }
    return 60 if($tok->{'code'} ne 'success');
    my $bot=$api->user;
    $text=~s/\x02BOT\x03/$bot/go;
    $text=~s/\s+$//o;
    $tok->{'revisions'}[0]{'slots'}{'main'}{'*'}=~s/\s+$//o if !exists($tok->{'missing'});
    if(exists($tok->{'missing'}) ||
       $tok->{'revisions'}[0]{'slots'}{'main'}{'contentmodel'} ne 'wikitext' ||
       $tok->{'revisions'}[0]{'slots'}{'main'}{'*'} ne $text){
        my $r=$api->edit($tok, $text, $self->{'summary'}, 0, 1, contentmodel => 'wikitext' );
        if($r->{'code'} ne 'success'){
            $api->warn("Write error for $page: ".$r->{'error'}."\n");
            return 60;
        } else {
            $api->log("Updated $page");
        }
    } else {
        $api->debug(2, "No update needed for $page\n");
    }
    return 0;
}

sub redirect_talk {
    my $api=shift;
    my $page=shift;

    my $tpage=$page;
    $tpage=~s/^User:/User talk:/;
    my $txt=$tpage;
    $txt=~s!^(User talk:[^/]+)/.*!$1!;
    my $r={$api->resolve_redirects($txt)};
    if(exists($r->{''})){
        $api->warn("Error fetching redirects for $page: ".$r->{''}{'error'}."\n");
        return 60;
    }
    $txt=$r->{$txt};
    $txt="#REDIRECT [[$txt]]\n\nThis page is unwatched. Please direct comments to [[$txt]].";
    my $tok=$api->edittoken($tpage, EditRedir=>1, NoExclusion=>1);
    if($tok->{'code'} eq 'shutoff'){
        $api->warn("Task disabled: ".$tok->{'content'}."\n");
        return 300;
    }
    return 60 if($tok->{'code'} ne 'success');
    if(($tok->{'revisions'}[0]{'slots'}{'main'}{'*'} // '') ne $txt){
        my $r=$api->edit($tok, $txt, 'Redirect useless unwatched talk page to someplace useful.', 1, 1);
        if($r->{'code'} ne 'success'){
            $api->warn("Write error for $tpage: ".$r->{'error'}."\n");
            return 60;
        } else {
            $api->log("Created $tpage");
        }
    }
    return undef;
}

sub determine_task_section {
    my $file = shift;
    my %metadata = @_;

    my $t=$file; $t=~s{^.*/}{}; $t=~s/\.pm$//;
    return '99 Invalid metadata (no "Bot")' unless exists($metadata{'bot'});
    return '99 Invalid metadata (no/bad "Task")' unless(exists($metadata{'task'}) && $metadata{'task'} eq $t);
    return '99 Invalid metadata (no "BRFA")' unless exists($metadata{'brfa'});
    return '99 Invalid metadata (no "Status")' unless exists($metadata{'status'});
    return '99 Invalid metadata (no/bad "Created")' unless(exists($metadata{'created'}) && $metadata{'created'}=~/^\d{4}-\d\d-\d\d$/);
    if(exists($metadata{'+brfa'})){
        return '99 Invalid metadata (no "+Status")' unless exists($metadata{'+status'});
        return '99 Invalid metadata (mismatched +BRFA/+Status)' if scalar(@{$metadata{'+status'}}) != scalar(@{$metadata{'+brfa'}});
    }
    if(exists($metadata{'ondemand'})){
        return '99 Invalid metadata (bad "OnDemand")' unless($metadata{'ondemand'} eq 'true' || $metadata{'ondemand'} eq 'false');
    }
    if(exists($metadata{'shutoff'})){
        return '99 Invalid metadata (bad "Shutoff")' unless($metadata{'shutoff'} eq 'true' || $metadata{'shutoff'} eq 'false');
    }
    if(exists($metadata{'exclusion'})){
        return '99 Invalid metadata (bad "Exclusion")' unless($metadata{'exclusion'} eq 'true' || $metadata{'exclusion'} eq 'false');
    }
    my $status=$metadata{'status'};

    AnomieBOT::API::load($file);
    my $task='tasks::'.$metadata{'task'};
    return '99 Invalid metadata (missing "approved" method)' unless $task->can('approved');

    my $botnum=$task->approved;

    if($metadata{'brfa'} eq 'N/A'){
        return "01 Current" if($botnum>0 && $status=~/^Begun \d{4}-\d{2}-\d{2}$/);
        return "02 On demand" if(exists($metadata{'ondemand'}) && $metadata{'ondemand'} eq 'true');
    }
    if($status=~/^Approved \d{4}-\d{2}-\d{2}$/){
        return "01 Current" if $botnum>0;
        return "02 On demand" if(exists($metadata{'ondemand'}) && $metadata{'ondemand'} eq 'true');
        return '99 Invalid metadata ("Status: Approved" but not running or OnDemand)';
    }
    return "03 In trial" if($status=~/^In trial/ && $botnum>0);
    return '99 Invalid metadata (Running with a non-running Status)' if $botnum>0;
    return "04 In development" if($status eq 'Coding' || $status eq 'On hold');
    return "06 Past" if $status=~/^(?:Completed|Inactive) \d{4}-\d{2}-\d{2}$/;
    return '99 Invalid metadata (bad "BRFA")' if($metadata{'brfa'} eq 'N/A');
    return "05 Awaiting approval" if($status eq 'BRFA');
    return "07 Withdrawn" if $status eq 'Withdrawn';
    return "08 Rejected" if $status=~/^Rejected \d{4}-\d{2}-\d{2}$/;
    return '99 Invalid metadata (no match for Status)';
}

1;