Jump to content

User:AnomieBOT/source/tasks/TemplateReplacer14.pm

From Wikipedia, the free encyclopedia
package tasks::TemplateReplacer14;

=pod

=for warning
Due to breaking changes in AnomieBOT::API, this task will probably not run
anymore. If you really must run it, try getting a version from before
2018-08-12.

=begin metadata

Bot:     AnomieBOT
Task:    TemplateReplacer14
BRFA:    Wikipedia:Bots/Requests for approval/AnomieBOT 25
Status:  Completed 2009-03-26
Created: 2009-03-02

Replace the obsolete {{tlx|Infobox Television}} <code>imdb_id</code> and
<code>ttv_com_id</code> parameters with {{tl|imdb title}}, and {{tl|tv.com}} in
the External links section, respectively.

=end metadata

=cut

use utf8;
use strict;

use AnomieBOT::Task;
use Data::Dumper;
use vars qw/@ISA/;
@ISA=qw/AnomieBOT::Task/;

my $tv_extlink_templates_re=undef;
my $no_edit_just_to_remove_parameters=0;

sub new {
    my $class=shift;
    my $self=$class->SUPER::new();
    bless $self, $class;
    return $self;
}

=pod

=for info
Approved 2009-03-03, completed 2009-03-26<br />[[Wikipedia:Bots/Requests for approval/AnomieBOT 25]]

=cut

sub approved {
    return -1;
}

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

    $api->task('TemplateReplacer14', 0, 10, qw/d::Sections/);

    return 60 if(!defined($api->load_interlanguage_map($api)));

    if(!defined($tv_extlink_templates_re)){
        my @links=();
        my $res=$api->query([],
            list        => 'categorymembers',
            cmtitle     => 'Category:Television external link templates',
            cmnamespace => 10,
            cmlimit     => 'max',
        );
        if($res->{'code'} ne 'success'){
            $api->warn("Failed to retrieve tv external link template list: ".$res->{'error'}."\n");
            return 60;
        }
        #unshift @{$res->{'query'}{'categorymembers'}}, { title=>'Template:Official' };
        foreach (@{$res->{'query'}{'categorymembers'}}){
            my $t=$_->{'title'};
            next if $t=~m{/(?:doc|sandbox)$}i;
            $t="\Q".substr($t,9);
            $t=~s/^(.)/(?i:$1)/;
            $t=~s/ /[ _]/g;
            push @links, $t;
            my $res2=$api->query([],
                list          => 'backlinks',
                bltitle       => $_->{'title'},
                blfilterredir => 'redirects',
                bllimit       => 'max',
            );
            if($res2->{'code'} ne 'success'){
                $api->warn("Failed to retrieve redirects for ".$_->{'title'}.": ".$res2->{'error'}."\n");
                return 60;
            }
            foreach (@{$res2->{'query'}{'backlinks'}}){
                $_="\Q".substr($_->{'title'},9);
                s/^Template:(.)/(?i:$1)/;
                s/ /[ _]/g;
                push @links, $_;
            }
        }
        $tv_extlink_templates_re=join('|', @links);
        $tv_extlink_templates_re=qr/{{\s*(?:$tv_extlink_templates_re)\s*(?:\||}})/o;
    }

    my $req="[[User talk:Anomie#EL bot for film articles|request]]";

    # Spend a max of 5 minutes on this task before restarting
    my $endtime=time()+300;

    $self->_output_log($api);

    # Get a list of templates redirecting to our target
    my %templates=();
    $templates{"Template:Infobox Television"}=1;
    $res=$api->query([],
        list          => 'backlinks',
        bltitle       => "Template:Infobox Television",
        blfilterredir => 'redirects',
        bllimit       => 'max',
    );
    if($res->{'code'} ne 'success'){
        $api->warn("Failed to retrieve redirects for Template:Infobox Television: ".$res->{'error'}."\n");
        return 60;
    }
    $templates{$_->{'title'}}=1 foreach (@{$res->{'query'}{'backlinks'}});

    # Matching external links; "%X" is the text of the template parameter
    my @to_process=qw/imdb_id tv_com_id/;
    my %ext_links=(
        'imdb_id' => [
            'http://www.imdb.com/title/tt%X',
            'http://www.imdb.com/title/tt%X/',
            'http://imdb.com/title/tt%X',
            'http://imdb.com/title/tt%X/',
        ],
        'tv_com_id' => [
            'http://www.tv.com/show/%X/summary.html',
            'http://tv.com/show/%X/summary.html',
        ],
    );
    # External link generating templates; parameter 1 is the id, and optional
    # parameter 2 is the infobox's "show_name" parameter.
    my %ext_templates=(
        'imdb_id' => 'imdb title',
        'tv_com_id' => 'tv.com',
    );

    # Get the list of pages to check
    foreach my $cat ('Category:Television articles with an IMDb link in the infobox', 'Category:Television articles with a TV.com link in the infobox'){
        my %q=(
            list        => 'categorymembers',
            cmtitle     => $cat,
            cmnamespace => 0,
            cmlimit     => 'max',
        );
        do {
            $res=$api->query(%q);
            if($res->{'code'} ne 'success'){
                $api->warn("Failed to retrieve category member list for $cat: ".$res->{'error'}."\n");
                return 60;
            }
            if(exists($res->{'query-continue'})){
                $q{'cmcontinue'}=$res->{'query-continue'}{'categorymembers'}{'cmcontinue'};
            } else {
                delete $q{'cmcontinue'};
            }

            # Process found pages
            foreach (@{$res->{'query'}{'categorymembers'}}){
                my $pageid=$_->{'pageid'};
                next if exists($api->store->{$pageid});

                # Cleanup the log
                my $log={};
                $log=$api->store->{'log'} if exists($api->store->{'log'});
                delete $log->{$_}{$pageid} foreach (keys %$log);
                $api->store->{'log'}=$log;

                my $title=$_->{'title'};
                $api->log("Processing $title");

                # WTF?
                if(exists($_->{'missing'})){
                    $api->warn("$title is missing? WTF?\n");
                    next;
                }

                # Ok, check the page
                my $tok=$api->edittoken($title);
                if($tok->{'code'} eq 'shutoff'){
                    $api->warn("Task disabled: ".$tok->{'content'}."\n");
                    return 300;
                }
                if($tok->{'code'} ne 'success'){
                    $api->warn("Failed to get edit token for $title: ".$tok->{'error'}."\n");
                    next;
                }
                next if exists($tok->{'missing'});

                # Get page text
                my $intxt=$tok->{'revisions'}[0]{'*'};

                # Step 1: Find the parameters for the infobox. Also, strip the
                # parameters we are intending to process.
                my %infobox_params=();
                my $ct=0;
                my @process=();
                my $outtxt=$api->process_templates($intxt, sub {
                    my $name=shift;
                    my @params=@{shift()};
                    shift; # $wikitext
                    shift; # $data
                    my $oname=shift;

                    return undef unless exists($templates{"Template:$name"});
                    if($ct++>0){ # More than one infobox?
                        $self->_log($api, 'Multiple infoboxen', $pageid, $title, "$ct instances of the infobox detected.");
                        $api->store->{$pageid}=1;
                        return undef;
                    }
                    my @out=();
                    foreach ($api->process_paramlist(@params)){
                        $_->{'value'}=~s/^\s+|\s+$//g;
                        $infobox_params{$_->{'name'}}=$_->{'value'} unless $_->{'value'} eq '';
                        if(exists($ext_links{$_->{'name'}})){
                            push @process, $_->{'name'} unless $_->{'value'} eq '';
                        } else {
                            push @out, $_->{'text'};
                        }
                    }
                    return "{{$oname|".join("|", @out)."}}";
                });
                next if $ct>1;
                if($ct<1){
                    $self->_log($api, 'No infobox', $pageid, $title, "No instance of the infobox was found in the page.");
                    $api->store->{$pageid}=2;
                    next;
                }
                unless(@process){
                    # Nothing to do here.
                    $api->store->{$pageid}=1000000;
                    next;
                }

                # Step 2: Extract the external links section
                my $nowiki;
                ($outtxt,$nowiki)=$api->strip_nowiki($outtxt);
                my $comments=[];
                while(my ($k,$v)=each(%$nowiki)){
                    push @$comments, $k if $v=~/^<!--/;
                }
                $comments=join("|", @$comments);
                my @sections=();
                my $extlink_section=undef;
                my @split=("", "", split /((?:^|\n)==(=?)[^=\n](?:.*[^=\n])?\2==)(?=(?:\s*(?:$comments))*\s*(?:\n|$))/, $outtxt);
                for(my $i=0; $i<@split; $i+=3){
                    my $h=$api->replace_nowiki($split[$i+0], $nowiki);
                    $h=~s/^(\n?==)(=?)\s*External\s*(\2==)$/$1$2 External links $3/i;
                    $h=~s/^(\n?==)(=?)(.*)External link\(s\)(.*\2==)$/$1$2$3External links$4/i;
                    $h=~s/^(\n?==)(=?)(.*)External link((?!s).*\2==)$/$1$2$3External links$4/i;
                    $h=~s/^(\n?==)(=?)(.*)External references?(.*\2==)$/$1$2$3External links$4/i;
                    my $s=$h.$api->replace_nowiki($split[$i+2], $nowiki);
                    push @sections, \$s;
                    $extlink_section=\$s if $h=~/External links/i;
                }
                if(!defined($extlink_section)){
                    # Crap, we have to create an external links section.
                    $self->_log($api, 'Added "External links"', $pageid, $title, "No \"External links\" section was found in the page. Check if one was added in the right place.");
                    my $x=pop @sections;
                    my ($pre,$post)=$api->extract_end_content($$x);
                    return 60 if(!defined($pre));
                    $pre=~s/\s+$/\n/;
                    push @sections, \$pre;
                    my $dummy="\n== External links ==\n\n";
                    $extlink_section=\$dummy;
                    push @sections, $extlink_section;
                    push @sections, \$post;
                } elsif($extlink_section==$sections[-1]){
                    # Last section, strip off the post-content junk
                    my $x=pop @sections;
                    my ($pre,$post)=$api->extract_end_content($$x);
                    return 60 if(!defined($pre));
                    $extlink_section=\$pre;
                    push @sections, $extlink_section;
                    push @sections, \$post;
                }

                # Step 3: Process our parameters
                my $res=$api->query([],
                    action=>'parse',
                    text=>$$extlink_section,
                    prop=>'externallinks',
                );
                if($res->{'code'} ne 'success'){
                    $api->warn("Failed to parse external links section for $title: ".$res->{'error'}."\n");
                    return 60;
                }
                my @el=();
                @el=@{$res->{'parse'}{'externallinks'}} if exists($res->{'parse'}{'externallinks'});
                my $add='';
                my $has_website=0;
                PARAM:
                foreach my $param (@to_process){
                    next unless grep($_ eq $param, @process);
                    my $id=$infobox_params{$param};
                    foreach (@{$ext_links{$param}}){
                        my $link=$_;
                        $link=~s/%X/$id/g;
                        next PARAM if grep($_ eq $link, @el);
                    }
                    my $tmpl=$ext_templates{$param};
                    $add.="\n* {{$tmpl|$id";
                    $add.='|'.$infobox_params{'show_name'} if exists($infobox_params{'show_name'});
                    $add.="}}";
                    $has_website=1 if $param eq 'website';
                }

                # Step 4: Reassemble the page, if anything changed in step 3
                if($add ne ''){
                    if($has_website && $$extlink_section=~s/\n\*/$add\n*/){
                        # Move "website" to the top of the external links
                    } elsif($$extlink_section=~s/(\n\*\s*$tv_extlink_templates_re.*?)\n/$1$add\n/){
                        # Put it after any other existing tv external link
                        # templates
                    } elsif($$extlink_section=~s/(\s*\n===)/$add$1/){
                        # There is a subsection in there (e.g. "Reviews"), put
                        # the links before it.
                    } else {
                        # Just tack it on the end.
                        $$extlink_section=~s/(\s*)$/$add$1/;
                    }
                    $outtxt=join('', map { $$_ } @sections);
                } elsif($no_edit_just_to_remove_parameters){
                    $api->store->{$pageid}=1000001;
                    next;
                } else {
                    $outtxt=$api->replace_nowiki($outtxt, $nowiki);
                }

                # Step 5: Perform the edit.
                $process[-1]='and '.$process[-1] if @process>1;
                my $summary='Moving deprecated '.join((@process>2)?', ':' ', @process).' from {{Infobox Television}} to External links per '.$req;
                $api->log("$summary in $title");
                my $r=$api->edit($tok, $outtxt, $summary, 0, 1);
                if($r->{'code'} ne 'success'){
                    $api->warn("Write failed on $title: ".$r->{'error'}."\n");
                    next;
                }

                # Mark this page as done
                $api->store->{$pageid}=2000000;

                # If we've been at it long enough, let another task have a go.
                if(time()>=$endtime){
                    $self->_output_log($api);
                    return 0;
                }
            }
        } while(exists($q{'cmcontinue'}));
    }

    # No more pages to check, try again in 10 minutes or so in case of errors.
    $self->_output_log($api);
    return 600;
}

sub _log {
    my $self=shift;
    my $api=shift;
    my $section=shift;
    my $pageid=shift;
    my $title=shift;
    my $message=shift;

    my $log={};
    $log=$api->store->{'log'} if exists($api->store->{'log'});
    $log->{$section}={} unless exists($log->{$section});
    $log->{$section}{$pageid}=[$title, $message];
    $api->store->{'log'}=$log;
    $api->log("LOG: $title: $message");
}

sub _output_log {
    my $self=shift;
    my $api=shift;

    $api->log("Updating log");
    my $tok=$api->edittoken("User:AnomieBOT/TemplateReplacer14 log");
    if($tok->{'code'} eq 'shutoff'){
        $api->warn("Task disabled: ".$tok->{'content'}."\n");
        return 300;
    }
    if($tok->{'code'} ne 'success'){
        $api->warn("Failed to get edit token for log: ".$tok->{'error'}."\n");
        return;
    }
    my $header="This is a log of issues encountered during the processing of the task TemplateReplacer14. Do not edit this page, the bot will overwrite it.\n";
    my $intxt=exists($tok->{'missing'})?$header:$tok->{'revisions'}[0]{'*'};
    my $outtxt=$header;
    my $log={};
    $log=$api->store->{'log'} if exists($api->store->{'log'});
    foreach my $section (sort keys %$log){
        my @out=();
        foreach my $pageid (keys %{$log->{$section}}){
            next unless exists($api->store->{$pageid});
            my ($title,$message)=@{$log->{$section}{$pageid}};
            push @out, "* [[:$title]]: $message\n";
        }
        next unless @out;
        $outtxt.="\n== $section ==\n".join('', @out) if @out;
    }
    if($outtxt ne $intxt){
        my $r=$api->edit($tok, $outtxt, 'Updating log', 0, 0);
        if($r->{'code'} ne 'success'){
            $api->warn("Could not write log: ".$r->{'error'}."\n");
            return;
        }
    }
}

1;