line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Dancer::SearchApp; |
2
|
3
|
|
|
3
|
|
127049
|
use strict; |
|
3
|
|
|
|
|
13
|
|
|
3
|
|
|
|
|
147
|
|
3
|
3
|
|
|
3
|
|
12
|
use File::Basename 'basename'; |
|
3
|
|
|
|
|
6
|
|
|
3
|
|
|
|
|
236
|
|
4
|
3
|
|
|
3
|
|
2124
|
use Dancer; |
|
3
|
|
|
|
|
575509
|
|
|
3
|
|
|
|
|
16
|
|
5
|
3
|
|
|
3
|
|
230335
|
use Search::Elasticsearch::Async; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
use URI::Escape 'uri_unescape'; |
7
|
|
|
|
|
|
|
use URI::file; |
8
|
|
|
|
|
|
|
#use Search::Elasticsearch::TestServer; |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
use Dancer::SearchApp::Defaults 'get_defaults'; |
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
use Dancer::SearchApp::Entry; |
13
|
|
|
|
|
|
|
use Dancer::SearchApp::HTMLSnippet; |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
use vars qw($VERSION $es %indices); |
16
|
|
|
|
|
|
|
$VERSION = '0.06'; |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
=head1 NAME |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
Dancer::SearchApp - A simple local search engine |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
=head1 SYNOPSIS |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
=head1 QUICKSTART |
25
|
|
|
|
|
|
|
|
26
|
|
|
|
|
|
|
Also see L. |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
cpanm --look Dancer::SearchApp |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
# Install prerequisites |
31
|
|
|
|
|
|
|
cpanm --installdeps . |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
# Install Elasticsearch https://www.elastic.co/downloads/elasticsearch |
34
|
|
|
|
|
|
|
# Start Elasticsearch |
35
|
|
|
|
|
|
|
# Install Apache Tika from https://tika.apache.org/download.html into jar/ |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
# Launch the web frontend |
38
|
|
|
|
|
|
|
plackup --host 127.0.0.1 -p 8080 -Ilib -a bin\app.pl |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
# Edit filesystem configuration |
41
|
|
|
|
|
|
|
cat >>fs-import.yml |
42
|
|
|
|
|
|
|
fs: |
43
|
|
|
|
|
|
|
directories: |
44
|
|
|
|
|
|
|
- folder: "C:\\Users\\Corion\\Projekte\\App-StarTraders" |
45
|
|
|
|
|
|
|
recurse: true |
46
|
|
|
|
|
|
|
exclude: |
47
|
|
|
|
|
|
|
- ".git" |
48
|
|
|
|
|
|
|
- folder: "t\\documents" |
49
|
|
|
|
|
|
|
recurse: true |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
# Collect some content |
52
|
|
|
|
|
|
|
perl -Ilib -w bin/index-filesystem.pl -f |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
# Search in your browser |
55
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
=head1 CONFIGURATION |
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
Configuration happens through config.yml |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
elastic_search: |
61
|
|
|
|
|
|
|
home: "./elasticsearch-2.1.1/" |
62
|
|
|
|
|
|
|
index: "dancer-searchapp" |
63
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
The Elasticsearch instance to used can also be passed in C<%ENV> |
65
|
|
|
|
|
|
|
as C. |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
=cut |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
use Data::Dumper; |
70
|
|
|
|
|
|
|
$Data::Dumper::Sortkeys = 1; |
71
|
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
my $config = get_defaults( |
73
|
|
|
|
|
|
|
env => \%ENV, |
74
|
|
|
|
|
|
|
config => config(), |
75
|
|
|
|
|
|
|
#defaults => \% |
76
|
|
|
|
|
|
|
names => [ |
77
|
|
|
|
|
|
|
['elastic_search/index' => 'elastic_search/index' => 'SEARCHAPP_ES_INDEX', 'searchapp'], |
78
|
|
|
|
|
|
|
['elastic_search/nodes' => 'elastic_search/nodes' => 'SEARCHAPP_ES_NODES', 'localhost:9200'], |
79
|
|
|
|
|
|
|
], |
80
|
|
|
|
|
|
|
); |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
# A small helper subroutine that adds some API headers that result in |
83
|
|
|
|
|
|
|
# the API not being interpretable as pages to be displayed by a browser |
84
|
|
|
|
|
|
|
sub add_api_headers { |
85
|
|
|
|
|
|
|
header 'Content-Disposition' => 'attachment; filename="1.txt"'; |
86
|
|
|
|
|
|
|
header 'X-Content-Type-Options' => 'nosniff'; |
87
|
|
|
|
|
|
|
}; |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
sub search { |
90
|
|
|
|
|
|
|
if( ! $es ) { |
91
|
|
|
|
|
|
|
my $nodes = $config->{elastic_search}->{nodes}; |
92
|
|
|
|
|
|
|
$nodes = [split /,/, $nodes] # our config system doesn't provide for lists... |
93
|
|
|
|
|
|
|
unless ref $nodes; |
94
|
|
|
|
|
|
|
$es = Search::Elasticsearch->new( |
95
|
|
|
|
|
|
|
nodes => $nodes, |
96
|
|
|
|
|
|
|
); |
97
|
|
|
|
|
|
|
}; |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
$es |
100
|
|
|
|
|
|
|
}; |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
$Template::Stash::PRIVATE = $Template::Stash::PRIVATE = 1; |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
get '/' => sub { |
105
|
|
|
|
|
|
|
# Later, separate out the code paths between |
106
|
|
|
|
|
|
|
# search and index page only, to serve the index |
107
|
|
|
|
|
|
|
# page as a static file |
108
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
my $statistics; |
110
|
|
|
|
|
|
|
my $results; |
111
|
|
|
|
|
|
|
|
112
|
|
|
|
|
|
|
my $from = params->{'from'} || 0; |
113
|
|
|
|
|
|
|
$from =~ s!\D!!g; |
114
|
|
|
|
|
|
|
my $size = params->{'size'} || 25; |
115
|
|
|
|
|
|
|
$size =~ s!\D!!g; |
116
|
|
|
|
|
|
|
my $search_term = params->{'q'}; |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
if( defined $search_term) { |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
#warning "Reading ES indices\n"; |
121
|
|
|
|
|
|
|
%indices = %{ search->indices->get({index => ['*']}) }; |
122
|
|
|
|
|
|
|
#warning $_ for sort keys %indices; |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
my @restrict_type; |
125
|
|
|
|
|
|
|
my $type; |
126
|
|
|
|
|
|
|
if( $type = params->{'type'} and $type =~ m!([a-z0-9+-]+)/[a-z0-9+-]+!i) { |
127
|
|
|
|
|
|
|
#warn "Filtering for '$type'"; |
128
|
|
|
|
|
|
|
@restrict_type = (filter => { term => { mime_type => $type }}); |
129
|
|
|
|
|
|
|
}; |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
my $sanitized_search_term = $search_term; |
132
|
|
|
|
|
|
|
# Escape colons, as they're special in search queries... |
133
|
|
|
|
|
|
|
$sanitized_search_term =~ s!([:\\])!\\$1!g; |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# Move this to an async query, later |
136
|
|
|
|
|
|
|
my $index = $config->{elastic_search}->{index}; |
137
|
|
|
|
|
|
|
$results = search->search( |
138
|
|
|
|
|
|
|
# Wir suchen in allen Sprachindices |
139
|
|
|
|
|
|
|
index => [ grep { /^\Q$index\E/ } sort keys %indices ], |
140
|
|
|
|
|
|
|
body => { |
141
|
|
|
|
|
|
|
from => $from, |
142
|
|
|
|
|
|
|
size => $size, |
143
|
|
|
|
|
|
|
query => { |
144
|
|
|
|
|
|
|
# multi_match => { ... grep for the non-autocomplete stuff, and include the boosters |
145
|
|
|
|
|
|
|
bool => { |
146
|
|
|
|
|
|
|
must => { |
147
|
|
|
|
|
|
|
query_string => { |
148
|
|
|
|
|
|
|
query => $search_term, |
149
|
|
|
|
|
|
|
fields => ['title','folder','content', 'author'] #'creation_date'] |
150
|
|
|
|
|
|
|
}, |
151
|
|
|
|
|
|
|
}, |
152
|
|
|
|
|
|
|
@restrict_type, |
153
|
|
|
|
|
|
|
}, |
154
|
|
|
|
|
|
|
}, |
155
|
|
|
|
|
|
|
sort => { |
156
|
|
|
|
|
|
|
_score => { order => 'desc' }, |
157
|
|
|
|
|
|
|
}, |
158
|
|
|
|
|
|
|
"highlight" => { |
159
|
|
|
|
|
|
|
"pre_tags" => '', |
160
|
|
|
|
|
|
|
"post_tags" => '', |
161
|
|
|
|
|
|
|
"fields" => { |
162
|
|
|
|
|
|
|
# we want the whole content so we can strip it down |
163
|
|
|
|
|
|
|
# ourselves: |
164
|
|
|
|
|
|
|
"content" => {"number_of_fragments" => 0}, |
165
|
|
|
|
|
|
|
#"content" => {} |
166
|
|
|
|
|
|
|
} |
167
|
|
|
|
|
|
|
} |
168
|
|
|
|
|
|
|
} |
169
|
|
|
|
|
|
|
); |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
#warn Dumper $results->{hits}; |
172
|
|
|
|
|
|
|
} else { |
173
|
|
|
|
|
|
|
# Update the statistics |
174
|
|
|
|
|
|
|
#$statistics = search->search( |
175
|
|
|
|
|
|
|
# search_type => 'count', |
176
|
|
|
|
|
|
|
# index => config->{index}, |
177
|
|
|
|
|
|
|
# body => { |
178
|
|
|
|
|
|
|
# query => { |
179
|
|
|
|
|
|
|
# match_all => {} |
180
|
|
|
|
|
|
|
# } |
181
|
|
|
|
|
|
|
# } |
182
|
|
|
|
|
|
|
#); |
183
|
|
|
|
|
|
|
#warn Dumper $statistics; |
184
|
|
|
|
|
|
|
}; |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
if( $results ) { |
187
|
|
|
|
|
|
|
for( @{ $results->{ hits }->{hits} } ) { |
188
|
|
|
|
|
|
|
$_->{source} = Dancer::SearchApp::Entry->from_es( $_ ); |
189
|
|
|
|
|
|
|
for my $key ( qw( id index type )) { |
190
|
|
|
|
|
|
|
$_->{$key} = $_->{"_$key"}; # thanks, Template::Toolkit |
191
|
|
|
|
|
|
|
}; |
192
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
}; |
194
|
|
|
|
|
|
|
}; |
195
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
if( $results and exists params->{lucky}) { |
197
|
|
|
|
|
|
|
my $first = $results->{ hits }->{hits}->[0]; |
198
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
if( $first ) { |
200
|
|
|
|
|
|
|
my( $index, $type, $id ) = @{$first}{qw(index type id)}; |
201
|
|
|
|
|
|
|
warn "Redirecting/reproxying first document"; |
202
|
|
|
|
|
|
|
if( $type eq 'http' ) { |
203
|
|
|
|
|
|
|
return |
204
|
|
|
|
|
|
|
redirect $id |
205
|
|
|
|
|
|
|
} else { |
206
|
|
|
|
|
|
|
my $doc = $first->{source}; |
207
|
|
|
|
|
|
|
my $local = URI::file->new( $id )->file; |
208
|
|
|
|
|
|
|
return |
209
|
|
|
|
|
|
|
reproxy( $doc, $local, 'Inline', |
210
|
|
|
|
|
|
|
index => $index, |
211
|
|
|
|
|
|
|
type => $type, |
212
|
|
|
|
|
|
|
); |
213
|
|
|
|
|
|
|
} |
214
|
|
|
|
|
|
|
}; |
215
|
|
|
|
|
|
|
} else { |
216
|
|
|
|
|
|
|
|
217
|
|
|
|
|
|
|
if( $results and $results->{hits} and $results->{hits}->{hits} and $results->{hits}->{hits}->[0]->{highlight}) { |
218
|
|
|
|
|
|
|
# Rework the result snippets to show only the highlighted stuff, together |
219
|
|
|
|
|
|
|
# with the appropriate page number if available |
220
|
|
|
|
|
|
|
for my $document (@{ $results->{hits}->{hits} }) { |
221
|
|
|
|
|
|
|
my $html = $document->{highlight}->{content}->[0]; |
222
|
|
|
|
|
|
|
my @show = Dancer::SearchApp::HTMLSnippet->extract_highlights( |
223
|
|
|
|
|
|
|
html => $html, |
224
|
|
|
|
|
|
|
max_length => 300, |
225
|
|
|
|
|
|
|
); |
226
|
|
|
|
|
|
|
|
227
|
|
|
|
|
|
|
# Find the PDF page numbers from Tika |
228
|
|
|
|
|
|
|
for my $s (@show) { |
229
|
|
|
|
|
|
|
$s->{page} = () = (substr($html,0,$s->{start}) =~ /
|
230
|
|
|
|
|
|
|
}; |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
$document->{highlight}->{content} = |
233
|
|
|
|
|
|
|
[map { |
234
|
|
|
|
|
|
|
+{ snippet => substr( $html, $_->{start}, $_->{length} ), |
235
|
|
|
|
|
|
|
page => $_->{page}, |
236
|
|
|
|
|
|
|
} |
237
|
|
|
|
|
|
|
} @show |
238
|
|
|
|
|
|
|
]; |
239
|
|
|
|
|
|
|
}; |
240
|
|
|
|
|
|
|
}; |
241
|
|
|
|
|
|
|
|
242
|
|
|
|
|
|
|
template 'index', { |
243
|
|
|
|
|
|
|
results => ($results ? $results->{hits} : undef ), |
244
|
|
|
|
|
|
|
params => { |
245
|
|
|
|
|
|
|
q => $search_term, |
246
|
|
|
|
|
|
|
from => $from, |
247
|
|
|
|
|
|
|
size => $size, |
248
|
|
|
|
|
|
|
}, |
249
|
|
|
|
|
|
|
}; |
250
|
|
|
|
|
|
|
}; |
251
|
|
|
|
|
|
|
}; |
252
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
# Show (cached) elements |
254
|
|
|
|
|
|
|
get '/cache/:index/:type/:id' => sub { |
255
|
|
|
|
|
|
|
my $index = params->{index}; |
256
|
|
|
|
|
|
|
my $type = params->{type}; |
257
|
|
|
|
|
|
|
my $id = uri_unescape( params->{id} ); |
258
|
|
|
|
|
|
|
my $document = retrieve($index,$type,$id); |
259
|
|
|
|
|
|
|
#warn $document->basic_mime_type; |
260
|
|
|
|
|
|
|
|
261
|
|
|
|
|
|
|
$document->{type} = $type; |
262
|
|
|
|
|
|
|
$document->{index} = $index; |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
if( $document ) { |
265
|
|
|
|
|
|
|
return template 'view_document', { |
266
|
|
|
|
|
|
|
result => $document, |
267
|
|
|
|
|
|
|
backlink => scalar( request->referer ), |
268
|
|
|
|
|
|
|
# we should also save the page offset... |
269
|
|
|
|
|
|
|
} |
270
|
|
|
|
|
|
|
} else { |
271
|
|
|
|
|
|
|
status 404; |
272
|
|
|
|
|
|
|
return <
|
273
|
|
|
|
|
|
|
That file does (not) exist anymore in the index. |
274
|
|
|
|
|
|
|
SORRY |
275
|
|
|
|
|
|
|
# We could delete that item from the index here... |
276
|
|
|
|
|
|
|
# or schedule reindexing of the resource? |
277
|
|
|
|
|
|
|
} |
278
|
|
|
|
|
|
|
}; |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
# Reproxy elements from disk |
281
|
|
|
|
|
|
|
sub reproxy { |
282
|
|
|
|
|
|
|
my( $document, $local, $disposition, %options ) = @_; |
283
|
|
|
|
|
|
|
|
284
|
|
|
|
|
|
|
# Now, if the file exists both in the index and locally, let's reproxy the content |
285
|
|
|
|
|
|
|
if( $document and -f $local) { |
286
|
|
|
|
|
|
|
status 200; |
287
|
|
|
|
|
|
|
content_type( $document->mime_type ); |
288
|
|
|
|
|
|
|
header( "Content-Disposition" => sprintf '%s; filename="%s"', $disposition, basename $local); |
289
|
|
|
|
|
|
|
my $abs = File::Spec->rel2abs( $local, '.' ); |
290
|
|
|
|
|
|
|
open my $fh, '<', $local |
291
|
|
|
|
|
|
|
or die "Couldn't read local file '$local': $!"; |
292
|
|
|
|
|
|
|
binmode $fh; |
293
|
|
|
|
|
|
|
local $/; |
294
|
|
|
|
|
|
|
<$fh> |
295
|
|
|
|
|
|
|
|
296
|
|
|
|
|
|
|
} else { |
297
|
|
|
|
|
|
|
status 404; # sorry |
298
|
|
|
|
|
|
|
return <
|
299
|
|
|
|
|
|
|
That file does (not) exist anymore or is currently unreachable |
300
|
|
|
|
|
|
|
for this webserver. We'll need to implement |
301
|
|
|
|
|
|
|
cleaning up the index from dead items. |
302
|
|
|
|
|
|
|
SORRY |
303
|
|
|
|
|
|
|
# We could delete that item from the index here... |
304
|
|
|
|
|
|
|
# Or schedule reindexing of the resource? |
305
|
|
|
|
|
|
|
} |
306
|
|
|
|
|
|
|
}; |
307
|
|
|
|
|
|
|
|
308
|
|
|
|
|
|
|
sub retrieve { |
309
|
|
|
|
|
|
|
my( $index, $type, $id ) = @_; |
310
|
|
|
|
|
|
|
my $document; |
311
|
|
|
|
|
|
|
if( eval { |
312
|
|
|
|
|
|
|
$document = search->get(index => $index, type => $type, id => $id); |
313
|
|
|
|
|
|
|
1 |
314
|
|
|
|
|
|
|
}) { |
315
|
|
|
|
|
|
|
my $res = Dancer::SearchApp::Entry->from_es($document); |
316
|
|
|
|
|
|
|
return $res |
317
|
|
|
|
|
|
|
} else { |
318
|
|
|
|
|
|
|
warn "$@"; |
319
|
|
|
|
|
|
|
}; |
320
|
|
|
|
|
|
|
# Not found in the Elasticsearch index |
321
|
|
|
|
|
|
|
return undef |
322
|
|
|
|
|
|
|
} |
323
|
|
|
|
|
|
|
|
324
|
|
|
|
|
|
|
get '/open/:index/:type/:id' => sub { |
325
|
|
|
|
|
|
|
my $index = params->{index}; |
326
|
|
|
|
|
|
|
my $type = params->{type}; |
327
|
|
|
|
|
|
|
my $id = uri_unescape params->{id}; |
328
|
|
|
|
|
|
|
my $document = retrieve($index,$type,$id); |
329
|
|
|
|
|
|
|
if( $type eq 'http' ) { |
330
|
|
|
|
|
|
|
return |
331
|
|
|
|
|
|
|
redirect $id |
332
|
|
|
|
|
|
|
} else { |
333
|
|
|
|
|
|
|
my $local = URI::file->new( $id )->file; |
334
|
|
|
|
|
|
|
return |
335
|
|
|
|
|
|
|
reproxy( $document, $local, 'Attachment', |
336
|
|
|
|
|
|
|
index => $index, |
337
|
|
|
|
|
|
|
type => $type, |
338
|
|
|
|
|
|
|
); |
339
|
|
|
|
|
|
|
} |
340
|
|
|
|
|
|
|
}; |
341
|
|
|
|
|
|
|
|
342
|
|
|
|
|
|
|
get '/inline/:index/:type/:id' => sub { |
343
|
|
|
|
|
|
|
my $index = params->{index}; |
344
|
|
|
|
|
|
|
my $type = params->{type}; |
345
|
|
|
|
|
|
|
my $id = uri_unescape params->{id}; |
346
|
|
|
|
|
|
|
my $document = retrieve($index,$type,$id); |
347
|
|
|
|
|
|
|
|
348
|
|
|
|
|
|
|
my $local; |
349
|
|
|
|
|
|
|
if( 'http' eq $type ) { |
350
|
|
|
|
|
|
|
$document->content |
351
|
|
|
|
|
|
|
} else { |
352
|
|
|
|
|
|
|
$local = URI::file->new( $id )->file; |
353
|
|
|
|
|
|
|
}; |
354
|
|
|
|
|
|
|
|
355
|
|
|
|
|
|
|
reproxy( $document, $local, 'Inline', |
356
|
|
|
|
|
|
|
index => $index, |
357
|
|
|
|
|
|
|
type => $type, |
358
|
|
|
|
|
|
|
); |
359
|
|
|
|
|
|
|
|
360
|
|
|
|
|
|
|
}; |
361
|
|
|
|
|
|
|
|
362
|
|
|
|
|
|
|
# This is likely a really bad design choice which I will regret later. |
363
|
|
|
|
|
|
|
# Most likely, manually encoding to JSON would be the saner approach |
364
|
|
|
|
|
|
|
# instead of globally setting a serializer for all routes. |
365
|
|
|
|
|
|
|
set 'serializer' => 'JSON'; |
366
|
|
|
|
|
|
|
|
367
|
|
|
|
|
|
|
get '/suggest/:query.json' => sub { |
368
|
|
|
|
|
|
|
my( $q ) = params->{query}; |
369
|
|
|
|
|
|
|
#warn "Completing '$q'"; |
370
|
|
|
|
|
|
|
|
371
|
|
|
|
|
|
|
return [] unless $q and $q =~ /\S/; |
372
|
|
|
|
|
|
|
|
373
|
|
|
|
|
|
|
# Strip leading/trailing whitespace, Justin Case |
374
|
|
|
|
|
|
|
$q =~ s!^\s+!!; |
375
|
|
|
|
|
|
|
$q =~ s!\s+$!!; |
376
|
|
|
|
|
|
|
|
377
|
|
|
|
|
|
|
# Reinitialize indices |
378
|
|
|
|
|
|
|
# Some day, we could cache that/not refresh them all the time |
379
|
|
|
|
|
|
|
%indices = %{ search->indices->get({index => ['*']}) }; |
380
|
|
|
|
|
|
|
|
381
|
|
|
|
|
|
|
my @restrict_type; |
382
|
|
|
|
|
|
|
my $type; |
383
|
|
|
|
|
|
|
if( $type = params->{'type'} and $type =~ m!([a-z0-9+-]+)/[a-z0-9+-]+!i) { |
384
|
|
|
|
|
|
|
#warn "Filtering for '$type'"; |
385
|
|
|
|
|
|
|
@restrict_type = (filter => { term => { mime_type => $type }}); |
386
|
|
|
|
|
|
|
}; |
387
|
|
|
|
|
|
|
|
388
|
|
|
|
|
|
|
# This should be centralized |
389
|
|
|
|
|
|
|
# This is "did you mean X" |
390
|
|
|
|
|
|
|
#my @fields = ('title','content', 'author'); |
391
|
|
|
|
|
|
|
|
392
|
|
|
|
|
|
|
# Query all suggestive fields at once: |
393
|
|
|
|
|
|
|
#my %suggest_query = map {; |
394
|
|
|
|
|
|
|
# "my_suggestions_$_" => { |
395
|
|
|
|
|
|
|
# phrase => { |
396
|
|
|
|
|
|
|
# field => "$_.autocomplete", |
397
|
|
|
|
|
|
|
# #field => "$_", |
398
|
|
|
|
|
|
|
# #text => $q, |
399
|
|
|
|
|
|
|
# } |
400
|
|
|
|
|
|
|
# } |
401
|
|
|
|
|
|
|
#} @fields; |
402
|
|
|
|
|
|
|
|
403
|
|
|
|
|
|
|
my @fields = ('title_suggest'); |
404
|
|
|
|
|
|
|
|
405
|
|
|
|
|
|
|
# Query all suggestive fields at once: |
406
|
|
|
|
|
|
|
my %suggest_query = map {; |
407
|
|
|
|
|
|
|
"my_completions_$_" => { |
408
|
|
|
|
|
|
|
phrase => { |
409
|
|
|
|
|
|
|
field => "title_suggest", |
410
|
|
|
|
|
|
|
#field => "$_", |
411
|
|
|
|
|
|
|
#text => $q, |
412
|
|
|
|
|
|
|
} |
413
|
|
|
|
|
|
|
} |
414
|
|
|
|
|
|
|
} @fields; |
415
|
|
|
|
|
|
|
|
416
|
|
|
|
|
|
|
#warn Dumper \%suggest_query; |
417
|
|
|
|
|
|
|
|
418
|
|
|
|
|
|
|
# Move this to an async query, later |
419
|
|
|
|
|
|
|
my $index = $config->{elastic_search}->{index}; |
420
|
|
|
|
|
|
|
my $results = search->suggest( |
421
|
|
|
|
|
|
|
index => [ grep { /^\Q$index\E/ } sort keys %indices ], |
422
|
|
|
|
|
|
|
body => { |
423
|
|
|
|
|
|
|
foo => { |
424
|
|
|
|
|
|
|
text => $q, |
425
|
|
|
|
|
|
|
completion => { |
426
|
|
|
|
|
|
|
field => 'title_suggest', |
427
|
|
|
|
|
|
|
"fuzzy" => { "fuzziness" => 2 }, # edit distance of 2 |
428
|
|
|
|
|
|
|
} |
429
|
|
|
|
|
|
|
} |
430
|
|
|
|
|
|
|
#%suggest_query |
431
|
|
|
|
|
|
|
} |
432
|
|
|
|
|
|
|
); |
433
|
|
|
|
|
|
|
|
434
|
|
|
|
|
|
|
#warn Dumper $results; |
435
|
|
|
|
|
|
|
|
436
|
|
|
|
|
|
|
my %suggestions; |
437
|
|
|
|
|
|
|
my @res = map {; +{ |
438
|
|
|
|
|
|
|
tokens => [split //, $_->{text}], |
439
|
|
|
|
|
|
|
value => $_->{text}, |
440
|
|
|
|
|
|
|
url => $_->{_source}->{url}, |
441
|
|
|
|
|
|
|
} } |
442
|
|
|
|
|
|
|
sort { $b->{_score} <=> $a->{_score} || $b cmp $a } # sort by score+asciibetically descending |
443
|
|
|
|
|
|
|
map { $_->{options} ? @{ $_->{options} } : () } # unwrap again |
444
|
|
|
|
|
|
|
map { @$_ } # unwrap |
445
|
|
|
|
|
|
|
grep { ref $_ eq 'ARRAY' } values %$results |
446
|
|
|
|
|
|
|
; |
447
|
|
|
|
|
|
|
|
448
|
|
|
|
|
|
|
add_api_headers; |
449
|
|
|
|
|
|
|
return \@res; |
450
|
|
|
|
|
|
|
}; |
451
|
|
|
|
|
|
|
|
452
|
|
|
|
|
|
|
true; |
453
|
|
|
|
|
|
|
|
454
|
|
|
|
|
|
|
__END__ |