| line | stmt | bran | cond | sub | pod | time | code | 
| 1 |  |  |  |  |  |  | # /=====================================================================\ # | 
| 2 |  |  |  |  |  |  | # |  NNexus Autolinker                                                  | # | 
| 3 |  |  |  |  |  |  | # | Job Request Module                                                  | # | 
| 4 |  |  |  |  |  |  | # |=====================================================================| # | 
| 5 |  |  |  |  |  |  | # | Part of the Planetary project: http://trac.mathweb.org/planetary    | # | 
| 6 |  |  |  |  |  |  | # |  Research software, produced as part of work done by:               | # | 
| 7 |  |  |  |  |  |  | # |  the KWARC group at Jacobs University                               | # | 
| 8 |  |  |  |  |  |  | # | Copyright (c) 2012                                                  | # | 
| 9 |  |  |  |  |  |  | # | Released under the MIT License (MIT)                                | # | 
| 10 |  |  |  |  |  |  | # |---------------------------------------------------------------------| # | 
| 11 |  |  |  |  |  |  | # | Adapted from the original NNexus code by                            | # | 
| 12 |  |  |  |  |  |  | # |                                  James Gardner and Aaron Krowne     | # | 
| 13 |  |  |  |  |  |  | # |---------------------------------------------------------------------| # | 
| 14 |  |  |  |  |  |  | # | Deyan Ginev                   #_#     | # | 
| 15 |  |  |  |  |  |  | # | http://kwarc.info/people/dginev                            (o o)    | # | 
| 16 |  |  |  |  |  |  | # \=========================================================ooo==U==ooo=/ # | 
| 17 |  |  |  |  |  |  | package NNexus::Job; | 
| 18 | 5 |  |  | 5 |  | 796424 | use strict; | 
|  | 5 |  |  |  |  | 11 |  | 
|  | 5 |  |  |  |  | 251 |  | 
| 19 | 5 |  |  | 5 |  | 29 | use warnings; | 
|  | 5 |  |  |  |  | 10 |  | 
|  | 5 |  |  |  |  | 210 |  | 
| 20 | 5 |  |  | 5 |  | 29 | use feature qw(say); | 
|  | 5 |  |  |  |  | 15 |  | 
|  | 5 |  |  |  |  | 488 |  | 
| 21 |  |  |  |  |  |  |  | 
| 22 | 5 |  |  | 5 |  | 2210 | use Mojo::DOM; | 
|  | 5 |  |  |  |  | 162770 |  | 
|  | 5 |  |  |  |  | 160 |  | 
| 23 | 5 |  |  | 5 |  | 2404 | use NNexus::Discover qw(mine_candidates); | 
|  | 5 |  |  |  |  | 14 |  | 
|  | 5 |  |  |  |  | 555 |  | 
| 24 | 5 |  |  | 5 |  | 2465 | use NNexus::Annotate qw(serialize_concepts); | 
|  | 5 |  |  |  |  | 9 |  | 
|  | 5 |  |  |  |  | 262 |  | 
| 25 | 5 |  |  | 5 |  | 2638 | use NNexus::Classification qw(disambiguate); | 
|  | 5 |  |  |  |  | 61 |  | 
|  | 5 |  |  |  |  | 816 |  | 
| 26 | 5 |  |  | 5 |  | 64 | use NNexus::Morphology qw(canonicalize_url); | 
|  | 5 |  |  |  |  | 8 |  | 
|  | 5 |  |  |  |  | 3817 |  | 
| 27 |  |  |  |  |  |  |  | 
| 28 |  |  |  |  |  |  | sub new { | 
| 29 | 14 |  |  | 14 | 1 | 450071 | my ($class,%opts) = @_; | 
| 30 | 14 |  | 100 |  |  | 111 | $opts{format} = lc($opts{format}||'html'); | 
| 31 | 14 |  |  |  |  | 46 | $opts{result} = {}; | 
| 32 | 14 | 100 |  |  |  | 72 | $opts{url} = canonicalize_url($opts{url}) if $opts{url}; | 
| 33 | 14 |  |  |  |  | 72 | bless \%opts, $class; } | 
| 34 |  |  |  |  |  |  |  | 
| 35 |  |  |  |  |  |  | sub execute { | 
| 36 | 13 |  |  | 13 | 1 | 66 | my ($self) = @_; | 
| 37 | 13 |  |  |  |  | 55 | my $function = $self->{function}; | 
| 38 | 13 | 100 |  |  |  | 55 | if ($function eq 'linkentry') {$self->_link_entry;} | 
|  | 8 | 50 |  |  |  | 29 |  | 
|  | 5 |  |  |  |  | 22 |  | 
| 39 | 0 |  |  |  |  | 0 | elsif ($function eq 'indexentry') {$self->_index_entry;} | 
| 40 |  |  |  |  |  |  | else {$self->_fail_with("Invalid action, aborting!"); } } | 
| 41 |  |  |  |  |  |  |  | 
| 42 |  |  |  |  |  |  | sub _fail_with { | 
| 43 | 0 |  |  | 0 |  | 0 | my ($self,$message)=@_; | 
| 44 |  |  |  |  |  |  | # TODO: Spec this out, maybe similar to LaTeXML? | 
| 45 | 0 |  |  |  |  | 0 | my $result = {payload=>q{},message=>$message,status=>'Failed!'}; | 
| 46 | 0 |  |  |  |  | 0 | $self->{result} = $result; } | 
| 47 |  |  |  |  |  |  |  | 
| 48 |  |  |  |  |  |  | sub _ok_with { | 
| 49 | 5 |  |  | 5 |  | 15 | my ($self,$payload,$message)=@_; | 
| 50 |  |  |  |  |  |  | # TODO: Spec this out, maybe similar to LaTeXML? | 
| 51 | 5 |  |  |  |  | 28 | my $result = {payload=>$payload,message=>$message,status=>'OK'}; | 
| 52 | 5 |  |  |  |  | 134 | $self->{result} = $result; } | 
| 53 |  |  |  |  |  |  |  | 
| 54 | 10 |  |  | 10 | 1 | 866 | sub response { $_[0]->{result};} | 
| 55 | 3 |  |  | 3 | 1 | 90 | sub result { $_[0]->{result}->{payload}; } | 
| 56 | 0 |  |  | 0 | 1 | 0 | sub message { $_[0]->{result}->{message}; } | 
| 57 | 0 |  |  | 0 | 1 | 0 | sub status { $_[0]->{result}->{status}; } | 
| 58 |  |  |  |  |  |  |  | 
| 59 |  |  |  |  |  |  | sub _link_entry { | 
| 60 | 8 |  |  | 8 |  | 11 | my ($self) = @_; | 
| 61 |  |  |  |  |  |  | # Process in 2 Steps: | 
| 62 |  |  |  |  |  |  | # I. Concept Discovery | 
| 63 | 8 |  |  |  |  | 67 | my ($concepts_mined,$text_length) = | 
| 64 |  |  |  |  |  |  | NNexus::Discover::mine_candidates( | 
| 65 |  |  |  |  |  |  | db=>$self->{db}, | 
| 66 |  |  |  |  |  |  | body=>$self->{body}, | 
| 67 |  |  |  |  |  |  | url=>$self->{url}, | 
| 68 |  |  |  |  |  |  | domain=>$self->{domain}, | 
| 69 |  |  |  |  |  |  | format=>$self->{format}, | 
| 70 |  |  |  |  |  |  | verbosity=>$self->{verbosity}); | 
| 71 |  |  |  |  |  |  | # II. Disambiguation | 
| 72 | 8 |  |  |  |  | 76 | my $concepts_refined = NNexus::Classification::disambiguate( | 
| 73 |  |  |  |  |  |  | $concepts_mined, | 
| 74 |  |  |  |  |  |  | text_length=>$text_length, | 
| 75 |  |  |  |  |  |  | include_all=>$self->{include_all}, | 
| 76 |  |  |  |  |  |  | verbosity=>$self->{verbosity}); | 
| 77 |  |  |  |  |  |  | # III. Annotation | 
| 78 | 8 |  | 100 |  |  | 40 | $self->{annotation} //= 'html'; | 
| 79 | 8 |  | 50 |  |  | 21 | $self->{embed} //= 1; | 
| 80 | 8 |  |  |  |  | 68 | my $serialized_result = | 
| 81 |  |  |  |  |  |  | NNexus::Annotate::serialize_concepts( | 
| 82 |  |  |  |  |  |  | body=>$self->{body}, | 
| 83 |  |  |  |  |  |  | concepts=>$concepts_refined, | 
| 84 |  |  |  |  |  |  | annotation=>$self->{annotation}, | 
| 85 |  |  |  |  |  |  | embed=>$self->{embed}, | 
| 86 |  |  |  |  |  |  | domain=>$self->{domain}, | 
| 87 |  |  |  |  |  |  | verbosity=>$self->{verbosity}); | 
| 88 | 8 |  |  |  |  | 48 | $self->{result}={payload=>$serialized_result,message=>'No obvious problems.', status=>'OK'}; | 
| 89 | 8 |  |  |  |  | 44 | $serialized_result; } | 
| 90 |  |  |  |  |  |  |  | 
| 91 |  |  |  |  |  |  | sub _index_entry { | 
| 92 | 5 |  |  | 5 |  | 10 | my ($self)=@_; | 
| 93 | 5 |  | 50 |  |  | 18 | my $domain = $self->{domain} || 'all'; | 
| 94 | 5 |  | 33 |  |  | 18 | my $url = $self->{url}||$self->{body}; | 
| 95 | 5 |  |  |  |  | 13 | my $dom = $self->{dom}; | 
| 96 | 5 | 100 | 66 |  |  | 38 | if ($dom && (! ref $dom)) { # Text: | 
| 97 | 2 |  |  |  |  | 20 | $dom = Mojo::DOM->new($dom); | 
| 98 |  |  |  |  |  |  | } | 
| 99 | 5 |  | 50 |  |  | 549 | my $should_update = $self->{should_update} // 1; | 
| 100 | 5 |  |  |  |  | 1704 | require NNexus::Index::Dispatcher; | 
| 101 | 5 |  |  |  |  | 67 | my $dispatcher = NNexus::Index::Dispatcher->new(db=>$self->{db},domain=>$domain, | 
| 102 |  |  |  |  |  |  | verbosity=>$self->{verbosity},should_update=>$should_update, | 
| 103 |  |  |  |  |  |  | start=>$url,dom=>$dom); | 
| 104 | 5 |  |  |  |  | 13 | my @invalidation_suggestions; | 
| 105 | 5 |  |  |  |  | 22 | while (my $payload = $dispatcher->index_step) { | 
| 106 | 5 |  |  |  |  | 7 | push @invalidation_suggestions, @{$payload}; } | 
|  | 5 |  |  |  |  | 22 |  | 
| 107 | 5 | 50 |  |  |  | 24 | my $report_url = ($url ne 'default') ? "http://$url" : 'the default domain root'; | 
| 108 | 5 |  |  |  |  | 39 | $self->_ok_with(\@invalidation_suggestions,"IndexConcepts succeeded in domain $domain, on $report_url"); } | 
| 109 |  |  |  |  |  |  |  | 
| 110 |  |  |  |  |  |  | 1; | 
| 111 |  |  |  |  |  |  |  | 
| 112 |  |  |  |  |  |  | __END__ |