User:ImplSearchBot/Code
<lang perl>#!/usr/bin/perl -w use strict; use Getopt::Long; use MediaWiki::Bot; use JSON qw/to_json from_json/;
my $usage = "Usage: $0 --username=(username) --password=(password) [--posttosite=yes]";
my %options;
- I don't care to pollute my global scope.
{ my $wiki = 'rosettacode.org'; my $username; my $password; my $verbosity = 2; # verbosity level. 0 is silent. 1 is error only. 2 is updates. 3 is process, more is noisy. my $post; # Is this an actual run? my $cachepath = "cache/"; my $cacheonly; # Don't query the wiki for data. Just pull from cache. my $nosvn; my $result = GetOptions( "wiki=s" => \$wiki, "username=s" => \$username, "password=s" => \$password, "verbosity=s" => \$verbosity, "post" => \$post, "cacheonly" => \$cacheonly, "nosvn" => \$nosvn, "cachepath=s" => \$cachepath); $options{'wiki'} = $wiki;
$options{'nosvn'} = $nosvn if defined $nosvn;
$options{'username'} = $username if defined $username;
$options{'password'} = $password if defined $password;
$options{'post'} = $post if defined $post;
$options{'verbosity'} = $verbosity;
$cachepath .= '/' if('/' ne substr($cachepath, -1, 1));
$options{'cachepath'} = $cachepath;
$options{'cacheonly'} = $cacheonly if defined $cacheonly;
# This could be reversed as a qualified statement, but I don't know
# if that it'd be any less ugly.
unless( exists $options{'cacheonly'})
{
if( exists $options{'post'} )
{
unless(exists $options{'username'} and exists $options{'password'})
{
use Data::Dumper;
print Dumper(%options);
die $usage;
}
}
}
}
- Statistic tracking.
my $starttime = time; my $pagesedited = 0;
- Tracking for svn checkin at end.
- We *should* be the only ones writing to the cache path for now.
- Eventually, we'll keep track of the vision we last ran at, and update to that.
- In the mean time, since the structure of this is a bit unstable for the moment,
- we'll do an update to HEAD just to catch anything silly I might have done in the
- mean time.
&svn('update', $options{'cachepath'}); opendir(CACHEDIR, $options{'cachepath'})
or die "Unable to open cache directory";
my @initialcache = readdir(CACHEDIR); closedir(CACHEDIR);
&out(scalar @initialcache . " categories initially cached\n", 4);
- Get our editor
my $editor = &geteditor();
- Get a complete listing of the tasks.
&out("Getting tasks\n", 3); my $alltasks = &getcategory('Category:Programming Tasks');
- Get a complete listing of the languages.
&out("Getting the languages.\n", 3); my $alllanguages = &getcategory('Category:Programming Languages');
- We want the language name, not the fully-qualified wiki name.
$_ =~ s/^Category:// foreach (@$alllanguages);
- Get a list of the languages for which we've already provided bodies for the related omit categories.
- Store it as a hash, so the lookup will be faster.
my $omitcatcontents = &getcategory('Category:Maintenance/OmitCategoriesCreated'); my %createdomitcategories = map {$_, 1} @$omitcatcontents;
&out("Identifying work to do\n", 3);
my %impldiff;
foreach my $lang (@$alllanguages) { my $val = &getlangwork($lang); next unless defined $val; $impldiff{$lang} = $val; }
foreach my $language (keys %impldiff) { my $pagename = "Tasks not implemented in $language"; &out("Preparing data for:$pagename\n", 4); my $hashref = $impldiff{$language}->{'impl'}; my %implemented = %$hashref; $hashref = $impldiff{$language}->{'omit'}; my %omitted = %$hashref; my $omitcount = scalar keys %omitted;
# Language metadata
my $taskcount = scalar @$alltasks;
my $unimpcount = $taskcount - scalar keys %implemented;
my $targetcount = ($taskcount - $omitcount);
# Language-specific page data. my $unimplisting = ""; my $omitlisting = ""; my $pagedata; # Not assembled until the end.
foreach my $taskname (@$alltasks) { # We want the task name, not the fully-qualified wiki name. my $baretaskname = $taskname; $baretaskname =~ s/^Category://;
# Add the task to the unimplemented list, if it's unimplemented. $unimplisting .= "* $baretaskname\n" unless(exists $implemented{$taskname});
# Add the task to the omission list, if it's omitted. $omitlisting .= "* $baretaskname\n" if(exists $omitted{$taskname}) }
# Prepare template fields my $langfield = "|$language"; my $unimpfield = "|$unimpcount"; my $tcfield = "|$targetcount";
my $impperccalc = 0; $impperccalc = (($targetcount - $unimpcount) / $targetcount) * 100 unless ($targetcount == 0);
my $imppercfield = sprintf "|%u", $impperccalc;
my $unimpltemplatename = "unimp_body_$language"; my $omittemplatename = "unimp_omit_body_$language";
# Prepare the listing page format. $pagedata = 'Template:Unimpl header' . $langfield . $unimpfield . $tcfield . $imppercfield . ''; $pagedata .= "Template:$unimpltemplatename" . "$unimpfield"; $pagedata .= "Template:Omit header" . "$langfield"; $pagedata .= "Template:$omittemplatename"; $pagedata .= "Template:Unimpl footer$langfield";
# Post the template containing the listing of unimplemented tasks. &postpage("Template:$unimpltemplatename", "$unimplisting", "ImplSearchBot:Updating list body of unimplemented tasks.", 1);
# Post the template containing the listing of omitted tasks. &postpage("Template:$omittemplatename", "$omitlisting", "ImplSearchBot:Updating list body of unimplemented tasks.", 1);
# I'll uncomment this if the layout has to change significantly again. In the mean time, that's over a hundred pages we don't have to edit... # # Update the layout of the listing page, because it's changed. # &postpage($pagename, $pagedata, "ImplSearchBot:Updating layout of listing page.",1);
my $omitcategoryname = "Category:$language/Omit"; my $omitcatpagedata = "Tasks listed here have been marked as "un-implementable" in $language. Solutions may be impossible to do, too complex to be of any valuable instruction in $language, or prohibited by the task definition.
But hey; if you think you can prove us wrong, go for it. :-) "; &postpage($omitcategoryname, $omitcatpagedata, "ImplSearchBot:Updating Omit category body", 0) unless ( exists $createdomitcategories{$omitcategoryname} );
}
- To add here: Post stats on activities:
- 1. Pages edited
- 2. Time last run started
- 3. Time to completion from start to just prior to stat post point.
my $runtime = time() - $starttime; my $statsdata = "Pages edited last run: $pagesedited\nTime to post all per-language updates: $runtime seconds\n"; &out("Updating stats page. Runtime ($runtime), Pages edited ($pagesedited)\n", 2);
&postpage("User:ImplSearchBot/Stats", $statsdata, 0);
&out("Updating bot code page\n", 4);
open my $sourcefile, '<', $0 or die "Finished without updating bot source page";
my $botsource; $botsource .= $_ while <$sourcefile>;
close $sourcefile;
my $tag = "lang";
&postpage("User:ImplSearchBot/Code", "<$tag perl>$botsource</$tag>", 0);
&out("Updating cache\n", 4); &commitcache();
&out("Done\n", 3);
exit(0);
sub geteditor { # If we're not posting, and we're only drawing from cache # We don't actually need to pull from the wiki. if( exists $options{'cacheonly'} ) { return undef unless exists $options{'post'}; }
# Handles interaction with the wiki. # Note that I had to modify HTTP::Message to make it work # HTTP::Message silently failed when presented by MW # with an encoding type of "application/json" or some such. &out("Creating editor\n", 3); my $editor = MediaWiki::Bot->new('ImpleSearchBot'); $editor->{debug} = 1;
# Tell the editor to edit Rosetta Code. I'm sure Wikipedia didn't like # my initial attempts from before I added this line. &out("Trying to set wiki.\n",3); $editor->set_wiki('rosettacode.org','w');
# If we're not posting, we don't need to log in if we're pulling from cache. # Otherwise, attempt to log in. &out("Trying to log in.\n", 3); my $loginres = $editor->login($options{'username'}, $options{'password'});
die "Unable to login: " . $loginres unless $loginres == "Success";
return $editor; }
sub out { my $string = shift; my $loglevel = shift;
chomp $string;
if($options{'verbosity'} >= $loglevel) { if($loglevel == 1) { warn $string; # use stderr. } else { print "$string\n"; } } }
sub sanitizenamefs { my $pagename = shift; $pagename =~ tr/:\//__/; return $pagename; }
sub diffcat_simple { my $first = shift; my $second = shift; my %firsthash = map { $_, 1 } @$first;
my @new = ();
foreach my $secondelement (@$second) { push @new, $secondelement unless exists $firsthash{$secondelement}; }
return @new; }
sub diffcat { my $first = shift; my $second = shift; my @newinfirst = &diffcat_simple($second, $first); my @newinsecond = &diffcat_simple($first, $second); return (\@newinfirst, \@newinsecond); }
sub postpage { my $pagename = shift; my $pagedata = shift; my $remark = shift; my $minoredit = shift;
++$pagesedited;
unless( exists $options{'post'} ) { # save it to disk, and out of the way. $pagename = "test/" . &sanitizenamefs($pagename); $pagename .= ".wikitxt";
&out("Saving: $pagename\n", 3);
open my $outfile, '>', $pagename or &log("Failed to open $pagename: $!", 1);
return unless defined $outfile;
print $outfile $pagedata; close $outfile; } else { &out("Posting $pagename\n", 3);
$editor->edit($pagename, $pagedata, "ImplSearchBot:$remark", $minoredit) or &out("Failed to post page: " . $editor->{'errstr'}, 1); } }
- This allows us to compare site state between now and when we
- last ran, so that we can learn to avoid doing unnecessary work.
- (Saves on server resources and bloated edit statistics...)
- Also, I plan on publishing the cache files in a version control
- system like SVN or Git, to allow multiple bots
- to share the data and to provide history. We're using JSON
- Instead of Perl's native Data::Dumper as JSON has broader
- cross-language support, making it easier for others to use the data.
- SVN has the advantage that I already know how to use it.
- Git has the advantage in that I can use GitHub and not tax
- my Slice with Git traffic, and I don't have to punch a hole
- in the firewall to access more services.
sub cachedata { my $dataname = shift; my $data = shift; my $filename = $options{'cachepath'} . &sanitizenamefs($dataname . ".json"); unless( exists $options{'post'} ) { # This is just a test run. Best not to overwrite our cache. &out("Skipping caching of $dataname to $filename...", 4); return; } my $outfile; unless(open $outfile, '>', $filename) { &out("Failed to open $filename for caching: $!\n", 1); return; }
print $outfile to_json($data); close $outfile; &out(scalar $data . " members cached to $filename.\n", 5); }
sub getcacheddata { my $dataname = shift; my $filename = $options{'cachepath'} . &sanitizenamefs($dataname . ".json"); &out("Getting cached data for $dataname from $filename...", 4); my $infile; unless (open $infile, '<', $filename) { &out("Failed to load cached data $filename: $!\n", 1); return []; }
my $jsondata; $jsondata .= $_ while <$infile>; close $infile;
my $cacheddata = from_json($jsondata); &out(scalar @$cacheddata . " cache members retrieved from $filename\n", 5);
return $cacheddata; }
sub getcategory { my $categoryname = shift;
# Return the cache data if we're not supposed to query the database. return &getcacheddata($categoryname) if( exists $options{'cacheonly'} );
&out("Getting category contents for $categoryname...", 4); my @categorycontents = $editor->get_pages_in_category($categoryname); &out(scalar @categorycontents . " members retrieved for $categoryname\n", 5);
my ($added, $removed) = &diffcat(&getcacheddata($categoryname), \@categorycontents);
&out(scalar @$removed . " removed, " . @$added . " added to $categoryname\n", &getloglevelfromdiff(2, $removed, $added));
&cachedata("$categoryname", \@categorycontents);
return \@categorycontents; }
sub getwork { my $categoryname = shift; &out("Getting work for $categoryname.\n",4); my $cacheddata = &getcacheddata($categoryname);
# If we're on a cache-only basis, we'll just say we have no old data, # and that our cached data is our new data. my $olddata; my $newdata; if (exists $options{'cacheonly'}) { $olddata = []; $newdata = $cacheddata; } else { $olddata = $cacheddata; $newdata = &getcategory($categoryname); }
my ($removed, $added) = &diffcat($olddata, $newdata); my $work = []; $work = $newdata if((scalar @$removed + scalar @$added) > 0);
&out(scalar @$work . " items to process for $categoryname.\n", 3);
return @$work; }
sub getloglevelfromdiff { my $base = shift; my $first = shift; my $second = shift; return $base if((scalar @$first + scalar @$second) > 0); return $base + 1; }
sub getlangwork { my $language = shift; &out("Getting lang work for $language\n", 4); my %impl; foreach my $workitem (&getwork("Category:$language")) { $impl{$workitem} = 1; } my %omit; foreach my $workitem (&getwork("Category:$language/Omit")) { $omit{$workitem} = 1; }
# No work to do? return undef if( ( 0 == scalar keys %omit) && (0 == scalar keys %impl));
# work to do. return {'impl' => \%impl, 'omit' => \%omit }; }
sub commitcache { # First, find out if we've added any files. my $cachepath = $options{'cachepath'}; opendir(CACHEDIR, $cachepath);
my @current = readdir(CACHEDIR); close(CACHEDIR);
# We need to run svn adds if we've created any new files. # Maybe we'll use SVN::Client some day. Not right now.
my ($added, $removed) = &diffcat(\@current, \@initialcache); &out("Detected " . scalar @$added . " new cache files and " . scalar @$removed . " removed\n",&getloglevelfromdiff(2, $added, $removed));
if ((scalar @$added + scalar @$removed ) > 0) { foreach my $cachefile (@$added) { &svn('add', $cachepath . $cachefile);; }
foreach my $cachefile (@$removed) { &svn('remove', $cachepath . $cachefile);; } }
&svn('ci', '--message="ImplSearchBot run"', $cachepath); &svn('update', $cachepath); }
sub svn { return if(exists $options{'nosvn'}); my @args = @_;
my $string = "system 'svn'"; $string .= ", '$_'" foreach (@args);
$string .= "\n";
&out($string, 3);
system 'svn', @args if(exists $options{'post'}); } </lang>