| 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
|
|
375318
|
use strict; |
|
|
5
|
|
|
|
|
9
|
|
|
|
5
|
|
|
|
|
149
|
|
|
19
|
5
|
|
|
5
|
|
17
|
use warnings; |
|
|
5
|
|
|
|
|
6
|
|
|
|
5
|
|
|
|
|
142
|
|
|
20
|
5
|
|
|
5
|
|
19
|
use feature qw(say); |
|
|
5
|
|
|
|
|
11
|
|
|
|
5
|
|
|
|
|
333
|
|
|
21
|
|
|
|
|
|
|
|
|
22
|
5
|
|
|
5
|
|
1426
|
use Mojo::DOM; |
|
|
5
|
|
|
|
|
131983
|
|
|
|
5
|
|
|
|
|
131
|
|
|
23
|
5
|
|
|
5
|
|
1955
|
use NNexus::Discover qw(mine_candidates); |
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
use NNexus::Annotate qw(serialize_concepts); |
|
25
|
|
|
|
|
|
|
use NNexus::Classification qw(disambiguate); |
|
26
|
|
|
|
|
|
|
use NNexus::Morphology qw(canonicalize_url); |
|
27
|
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
sub new { |
|
29
|
|
|
|
|
|
|
my ($class,%opts) = @_; |
|
30
|
|
|
|
|
|
|
$opts{format} = lc($opts{format}||'html'); |
|
31
|
|
|
|
|
|
|
$opts{result} = {}; |
|
32
|
|
|
|
|
|
|
$opts{url} = canonicalize_url($opts{url}) if $opts{url}; |
|
33
|
|
|
|
|
|
|
bless \%opts, $class; } |
|
34
|
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
sub execute { |
|
36
|
|
|
|
|
|
|
my ($self) = @_; |
|
37
|
|
|
|
|
|
|
my $function = $self->{function}; |
|
38
|
|
|
|
|
|
|
if ($function eq 'linkentry') {$self->_link_entry;} |
|
39
|
|
|
|
|
|
|
elsif ($function eq 'indexentry') {$self->_index_entry;} |
|
40
|
|
|
|
|
|
|
else {$self->_fail_with("Invalid action, aborting!"); } } |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
sub _fail_with { |
|
43
|
|
|
|
|
|
|
my ($self,$message)=@_; |
|
44
|
|
|
|
|
|
|
# TODO: Spec this out, maybe similar to LaTeXML? |
|
45
|
|
|
|
|
|
|
my $result = {payload=>q{},message=>$message,status=>'Failed!'}; |
|
46
|
|
|
|
|
|
|
$self->{result} = $result; } |
|
47
|
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
sub _ok_with { |
|
49
|
|
|
|
|
|
|
my ($self,$payload,$message)=@_; |
|
50
|
|
|
|
|
|
|
# TODO: Spec this out, maybe similar to LaTeXML? |
|
51
|
|
|
|
|
|
|
my $result = {payload=>$payload,message=>$message,status=>'OK'}; |
|
52
|
|
|
|
|
|
|
$self->{result} = $result; } |
|
53
|
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
sub response { $_[0]->{result};} |
|
55
|
|
|
|
|
|
|
sub result { $_[0]->{result}->{payload}; } |
|
56
|
|
|
|
|
|
|
sub message { $_[0]->{result}->{message}; } |
|
57
|
|
|
|
|
|
|
sub status { $_[0]->{result}->{status}; } |
|
58
|
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
sub _link_entry { |
|
60
|
|
|
|
|
|
|
my ($self) = @_; |
|
61
|
|
|
|
|
|
|
# Process in 2 Steps: |
|
62
|
|
|
|
|
|
|
# I. Concept Discovery |
|
63
|
|
|
|
|
|
|
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
|
|
|
|
|
|
|
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
|
|
|
|
|
|
|
$self->{annotation} //= 'html'; |
|
79
|
|
|
|
|
|
|
$self->{embed} //= 1; |
|
80
|
|
|
|
|
|
|
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
|
|
|
|
|
|
|
$self->{result}={payload=>$serialized_result,message=>'No obvious problems.', status=>'OK'}; |
|
89
|
|
|
|
|
|
|
$serialized_result; } |
|
90
|
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
sub _index_entry { |
|
92
|
|
|
|
|
|
|
my ($self)=@_; |
|
93
|
|
|
|
|
|
|
my $domain = $self->{domain} || 'all'; |
|
94
|
|
|
|
|
|
|
my $url = $self->{url}||$self->{body}; |
|
95
|
|
|
|
|
|
|
my $dom = $self->{dom}; |
|
96
|
|
|
|
|
|
|
if ($dom && (! ref $dom)) { # Text: |
|
97
|
|
|
|
|
|
|
$dom = Mojo::DOM->new($dom); |
|
98
|
|
|
|
|
|
|
} |
|
99
|
|
|
|
|
|
|
my $should_update = $self->{should_update} // 1; |
|
100
|
|
|
|
|
|
|
require NNexus::Index::Dispatcher; |
|
101
|
|
|
|
|
|
|
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
|
|
|
|
|
|
|
my @invalidation_suggestions; |
|
105
|
|
|
|
|
|
|
while (my $payload = $dispatcher->index_step) { |
|
106
|
|
|
|
|
|
|
push @invalidation_suggestions, @{$payload}; } |
|
107
|
|
|
|
|
|
|
my $report_url = ($url ne 'default') ? "http://$url" : 'the default domain root'; |
|
108
|
|
|
|
|
|
|
$self->_ok_with(\@invalidation_suggestions,"IndexConcepts succeeded in domain $domain, on $report_url"); } |
|
109
|
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
1; |
|
111
|
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
__END__ |