line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Search::Tools::HeatMap; |
2
|
17
|
|
|
17
|
|
61
|
use Moo; |
|
17
|
|
|
|
|
57
|
|
|
17
|
|
|
|
|
76
|
|
3
|
17
|
|
|
17
|
|
3115
|
use Carp; |
|
17
|
|
|
|
|
20
|
|
|
17
|
|
|
|
|
790
|
|
4
|
17
|
|
|
17
|
|
62
|
use Data::Dump qw( dump ); |
|
17
|
|
|
|
|
17
|
|
|
17
|
|
|
|
|
657
|
|
5
|
|
|
|
|
|
|
extends 'Search::Tools::Object'; |
6
|
|
|
|
|
|
|
|
7
|
17
|
|
|
17
|
|
58
|
use namespace::autoclean; |
|
17
|
|
|
|
|
19
|
|
|
17
|
|
|
|
|
93
|
|
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
our $VERSION = '1.004'; |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
# debugging only |
12
|
|
|
|
|
|
|
my $OPEN = '['; |
13
|
|
|
|
|
|
|
my $CLOSE = ']'; |
14
|
|
|
|
|
|
|
eval { require Term::ANSIColor; }; |
15
|
|
|
|
|
|
|
if ( !$@ ) { |
16
|
|
|
|
|
|
|
$OPEN .= Term::ANSIColor::color('bold red'); |
17
|
|
|
|
|
|
|
$CLOSE = Term::ANSIColor::color('reset') . $CLOSE; |
18
|
|
|
|
|
|
|
} |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
my @attrs = qw( window_size |
21
|
|
|
|
|
|
|
tokens |
22
|
|
|
|
|
|
|
spans |
23
|
|
|
|
|
|
|
as_sentences |
24
|
|
|
|
|
|
|
_treat_phrases_as_singles |
25
|
|
|
|
|
|
|
_qre |
26
|
|
|
|
|
|
|
_query |
27
|
|
|
|
|
|
|
_stemmer |
28
|
|
|
|
|
|
|
); |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
for my $attr (@attrs) { |
31
|
|
|
|
|
|
|
has $attr => ( is => 'rw' ); |
32
|
|
|
|
|
|
|
} |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
=head1 NAME |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
Search::Tools::HeatMap - locate the best matches in a snippet extract |
37
|
|
|
|
|
|
|
|
38
|
|
|
|
|
|
|
=head1 SYNOPSIS |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
use Search::Tools::Tokenizer; |
41
|
|
|
|
|
|
|
use Search::Tools::HeatMap; |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
my $tokens = $self->tokenizer->tokenize( $my_string, qr/^(interesting)$/ ); |
44
|
|
|
|
|
|
|
my $heatmap = Search::Tools::HeatMap->new( |
45
|
|
|
|
|
|
|
tokens => $tokens, |
46
|
|
|
|
|
|
|
window_size => 20, # default |
47
|
|
|
|
|
|
|
as_sentences => 0, # default |
48
|
|
|
|
|
|
|
); |
49
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
if ( $heatmap->has_spans ) { |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
my $tokens_arr = $tokens->as_array; |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
# stringify positions |
55
|
|
|
|
|
|
|
my @snips; |
56
|
|
|
|
|
|
|
for my $span ( @{ $heatmap->spans } ) { |
57
|
|
|
|
|
|
|
push( @snips, $span->{str} ); |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
my $occur_index = $self->occur - 1; |
60
|
|
|
|
|
|
|
if ( $#snips > $occur_index ) { |
61
|
|
|
|
|
|
|
@snips = @snips[ 0 .. $occur_index ]; |
62
|
|
|
|
|
|
|
} |
63
|
|
|
|
|
|
|
printf("%s\n", join( ' ... ', @snips )); |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
} |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
=head1 DESCRIPTION |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
Search::Tools::HeatMap implements a simple algorithm for locating |
70
|
|
|
|
|
|
|
the densest clusters of unique, hot terms in a TokenList. |
71
|
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
HeatMap is used internally by Snipper but documented here in case |
73
|
|
|
|
|
|
|
someone wants to abuse and/or improve it. |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
=head1 METHODS |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
=head2 new( tokens => I ) |
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
Create a new HeatMap. The I object may be either a |
80
|
|
|
|
|
|
|
Search::Tools::TokenList or Search::Tools::TokenListPP object. |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
=head2 BUILD |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
Builds the HeatMap object. Called internally by new(). |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
=cut |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
sub BUILD { |
89
|
30
|
|
|
30
|
1
|
485
|
my $self = shift; |
90
|
30
|
|
|
|
|
96
|
$self->_build; |
91
|
30
|
|
|
|
|
741
|
return $self; |
92
|
|
|
|
|
|
|
} |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
=head2 window_size |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
The max width of a span. Defaults to 20 tokens, including the |
97
|
|
|
|
|
|
|
matches. |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
Set this in new(). Access it later if you need to, but the spans |
100
|
|
|
|
|
|
|
will have already been created by new(). |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
=head2 as_sentences |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
Try to match clusters at sentence boundaries. Default is false. |
105
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
Set this in new(). |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
=head2 spans |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
Returns an array ref of matching clusters. Each span in the array |
111
|
|
|
|
|
|
|
is a hash ref with the following keys: |
112
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
=over |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
=item cluster |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
=item pos |
118
|
|
|
|
|
|
|
|
119
|
|
|
|
|
|
|
=item heat |
120
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
=item str |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
=item str_w_pos |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
This item is available only if debug() is true. |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
=item unique |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
=back |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=cut |
132
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
# TODO this is mostly integer math and might be much |
134
|
|
|
|
|
|
|
# faster if rewritten in XS once the algorithm is "final". |
135
|
|
|
|
|
|
|
sub _build { |
136
|
30
|
|
|
30
|
|
39
|
my $self = shift; |
137
|
30
|
50
|
|
|
|
544
|
my $tokens = $self->tokens or croak "tokens required"; |
138
|
30
|
|
50
|
|
|
126
|
my $window = $self->window_size || 20; |
139
|
30
|
|
100
|
|
|
152
|
my $as_sentences = $self->as_sentences || 0; |
140
|
30
|
100
|
|
|
|
131
|
return $as_sentences |
141
|
|
|
|
|
|
|
? $self->_as_sentences( $tokens, $window ) |
142
|
|
|
|
|
|
|
: $self->_no_sentences( $tokens, $window ); |
143
|
|
|
|
|
|
|
} |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
# currently _as_sentences() is mostly identical to _no_sentences() |
146
|
|
|
|
|
|
|
# with slightly fewer gymnastics. |
147
|
|
|
|
|
|
|
# Since we already know via sentence_starts where our boundaries are, |
148
|
|
|
|
|
|
|
# we do not have to call $tokens->get_window(). |
149
|
|
|
|
|
|
|
# Who knows how we might improve the sentence algorithm in future, |
150
|
|
|
|
|
|
|
# so already having it in its own method seems like a win. |
151
|
|
|
|
|
|
|
sub _as_sentences { |
152
|
13
|
|
|
13
|
|
24
|
my ( $self, $tokens, $window ) = @_; |
153
|
13
|
|
50
|
|
|
265
|
my $debug = $self->debug || 0; |
154
|
13
|
|
|
|
|
121
|
my $sentence_length = $window * 2; |
155
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
# build heatmap with sentence starts |
157
|
13
|
|
|
|
|
51
|
my $num_tokens = $tokens->len; |
158
|
13
|
|
|
|
|
41
|
my $tokens_arr = $tokens->as_array; |
159
|
13
|
|
|
|
|
26
|
my %heatmap = (); |
160
|
13
|
|
|
|
|
44
|
my $token_list_heat = $tokens->get_heat; |
161
|
13
|
|
|
|
|
43
|
my $heat_sentence_starts = $tokens->get_sentence_starts; |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
# this regex is a sanity check for phrases. we replace the \ with a |
164
|
|
|
|
|
|
|
# more promiscuous check because the single space is too naive |
165
|
|
|
|
|
|
|
# for real text (e.g. st. john's) |
166
|
13
|
|
|
|
|
24
|
my $qre = $self->{_qre}; |
167
|
13
|
|
|
|
|
20
|
my @phrases = @{ $self->{_query}->phrases }; |
|
13
|
|
|
|
|
62
|
|
168
|
13
|
|
|
|
|
53
|
my $n_terms = $self->{_query}->num_terms; |
169
|
13
|
|
|
|
|
80
|
my $query_has_phrase = $qre =~ s/(\\ )+/.+/g; |
170
|
|
|
|
|
|
|
|
171
|
13
|
50
|
|
|
|
41
|
if ($debug) { |
172
|
0
|
|
|
|
|
0
|
warn "heat_sentence_starts: " . dump($heat_sentence_starts); |
173
|
0
|
|
|
|
|
0
|
warn "token_list_heat: " . dump($token_list_heat); |
174
|
0
|
|
|
|
|
0
|
warn "n_terms: $n_terms"; |
175
|
0
|
|
|
|
|
0
|
warn "phrases: " . dump( \@phrases ); |
176
|
0
|
|
|
|
|
0
|
warn "query_has_phrase: $query_has_phrase"; |
177
|
|
|
|
|
|
|
} |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
# find the "sentence" that each hot token appears in. |
180
|
13
|
|
|
|
|
14
|
my @starts_ends; |
181
|
13
|
|
|
|
|
15
|
my $i = 0; |
182
|
13
|
|
|
|
|
23
|
my %heat_sentence_ends = (); # cache |
183
|
13
|
|
|
|
|
29
|
for (@$token_list_heat) { |
184
|
42
|
|
|
|
|
111
|
my $token = $tokens->get_token($_); |
185
|
42
|
|
|
|
|
100
|
my $token_pos = $token->pos; |
186
|
42
|
|
|
|
|
54
|
my $start = $heat_sentence_starts->[ $i++ ]; |
187
|
42
|
|
|
|
|
138
|
$heatmap{$token_pos} = $token->is_hot; |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
# a little optimization for when we've got |
190
|
|
|
|
|
|
|
# multiple hot tokens in the same sentence |
191
|
42
|
100
|
|
|
|
84
|
if ( exists $heat_sentence_ends{$start} ) { |
192
|
22
|
50
|
|
|
|
40
|
$debug |
193
|
|
|
|
|
|
|
and warn "found cached end $heat_sentence_ends{$start} " |
194
|
|
|
|
|
|
|
. "for start $start token $token_pos\n"; |
195
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
push( @starts_ends, |
197
|
22
|
|
|
|
|
47
|
[ $start, $token_pos, $heat_sentence_ends{$start} ] ); |
198
|
22
|
|
|
|
|
35
|
next; |
199
|
|
|
|
|
|
|
} |
200
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
# find the outermost limit of where this sentence might end |
202
|
20
|
|
|
|
|
19
|
my $max_end; |
203
|
|
|
|
|
|
|
|
204
|
|
|
|
|
|
|
# is there a "next" start? |
205
|
20
|
100
|
100
|
|
|
98
|
if ( defined $heat_sentence_starts->[$i] |
206
|
|
|
|
|
|
|
and $heat_sentence_starts->[$i] != $start ) |
207
|
|
|
|
|
|
|
{ |
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
# this token is unique in this non-final sentence |
210
|
3
|
|
|
|
|
8
|
$max_end = $heat_sentence_starts->[$i] - 1; |
211
|
|
|
|
|
|
|
} |
212
|
|
|
|
|
|
|
else { |
213
|
|
|
|
|
|
|
|
214
|
|
|
|
|
|
|
# this is the final sentence |
215
|
17
|
|
|
|
|
30
|
$max_end = $num_tokens - 1; |
216
|
|
|
|
|
|
|
} |
217
|
20
|
|
|
|
|
29
|
my $end = $start; |
218
|
|
|
|
|
|
|
|
219
|
|
|
|
|
|
|
# find the nearest sentence end to the start |
220
|
20
|
|
|
|
|
47
|
while ( $end < $max_end ) { |
221
|
1990
|
|
|
|
|
2299
|
my $tok = $tokens->get_token( $end++ ); |
222
|
1990
|
50
|
|
|
|
2868
|
if ( !$tok ) { |
223
|
0
|
0
|
|
|
|
0
|
$debug and warn "No token at end=$end"; |
224
|
0
|
|
|
|
|
0
|
last; |
225
|
|
|
|
|
|
|
} |
226
|
1990
|
100
|
|
|
|
4422
|
if ( $tok->is_sentence_end ) { |
227
|
10
|
|
|
|
|
17
|
$end--; # move back one position |
228
|
10
|
50
|
|
|
|
25
|
if ($debug) { |
229
|
0
|
|
|
|
|
0
|
warn "tok $_ is_sentence_end end=$end"; |
230
|
0
|
|
|
|
|
0
|
$tok->dump; |
231
|
|
|
|
|
|
|
} |
232
|
10
|
|
|
|
|
23
|
last; |
233
|
|
|
|
|
|
|
} |
234
|
|
|
|
|
|
|
} |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
# back up if we've exceeded the 0-based tokens array. |
237
|
20
|
50
|
|
|
|
50
|
$end = $num_tokens if $end > $num_tokens; |
238
|
|
|
|
|
|
|
|
239
|
20
|
50
|
|
|
|
45
|
$debug |
240
|
|
|
|
|
|
|
and warn "start=$start max_end=$max_end " |
241
|
|
|
|
|
|
|
. "sentence_length=$sentence_length end=$end " |
242
|
|
|
|
|
|
|
. "token_pos=$token_pos\n"; |
243
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
# if we didn't yet set the actual hot token, |
245
|
|
|
|
|
|
|
# include everything up to it. |
246
|
20
|
50
|
|
|
|
45
|
if ( $end < $token_pos ) { |
247
|
0
|
0
|
|
|
|
0
|
$debug |
248
|
|
|
|
|
|
|
and warn "resetting end=$token_pos\n"; |
249
|
|
|
|
|
|
|
|
250
|
0
|
|
|
|
|
0
|
$end = $token_pos; |
251
|
|
|
|
|
|
|
} |
252
|
20
|
|
|
|
|
57
|
push( @starts_ends, [ $start, $token_pos, $end ] ); |
253
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
# cache |
255
|
20
|
|
|
|
|
61
|
$heat_sentence_ends{$start} = $end; |
256
|
|
|
|
|
|
|
} |
257
|
|
|
|
|
|
|
|
258
|
13
|
50
|
|
|
|
36
|
$debug and warn "starts_ends: " . dump( \@starts_ends ); |
259
|
|
|
|
|
|
|
|
260
|
13
|
|
|
|
|
19
|
my @spans; |
261
|
|
|
|
|
|
|
my %seen_pos; |
262
|
|
|
|
|
|
|
START_END: |
263
|
13
|
|
|
|
|
44
|
for my $start_end (@starts_ends) { |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
# get full window, ignoring positions we've already seen. |
266
|
42
|
|
|
|
|
42
|
my $heat = 0; |
267
|
42
|
|
|
|
|
37
|
my %span; |
268
|
|
|
|
|
|
|
my @cluster_tokens; |
269
|
|
|
|
|
|
|
|
270
|
42
|
|
|
|
|
85
|
my ( $start, $hot_pos, $end ) = @$start_end; |
271
|
42
|
|
|
|
|
89
|
POS: for my $pos ( $start .. $end ) { |
272
|
5075
|
100
|
|
|
|
9889
|
next POS if $seen_pos{$pos}++; |
273
|
2000
|
100
|
|
|
|
2064
|
$heat += ( exists $heatmap{$pos} ? $heatmap{$pos} : 0 ); |
274
|
2000
|
|
|
|
|
2426
|
push( @cluster_tokens, $tokens->get_token($pos) ); |
275
|
|
|
|
|
|
|
} |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
# if we had already seen_pos all positions. |
278
|
42
|
100
|
|
|
|
109
|
next START_END unless @cluster_tokens; |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
# sanity: make sure we still have something hot |
281
|
20
|
|
|
|
|
29
|
my $has_hot = 0; |
282
|
20
|
|
|
|
|
24
|
my @cluster_pos; |
283
|
|
|
|
|
|
|
my @strings; |
284
|
20
|
|
|
|
|
34
|
TOK: for (@cluster_tokens) { |
285
|
2000
|
|
|
|
|
1855
|
my $pos = $_->pos; |
286
|
2000
|
100
|
|
|
|
2246
|
$has_hot++ if exists $heatmap{$pos}; |
287
|
2000
|
|
|
|
|
2235
|
push @strings, $_->str; |
288
|
2000
|
|
|
|
|
1799
|
push @cluster_pos, $pos; |
289
|
|
|
|
|
|
|
} |
290
|
20
|
50
|
|
|
|
43
|
next START_END unless $has_hot; |
291
|
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
# the final string is a sentence end, |
293
|
|
|
|
|
|
|
# but we only want the first char in it, |
294
|
|
|
|
|
|
|
# and not any whitespace, stray punctuation or other |
295
|
|
|
|
|
|
|
# non-word noise. |
296
|
20
|
|
|
|
|
122
|
$strings[$#strings] =~ s/^([\.\?\!]).*/$1/; |
297
|
|
|
|
|
|
|
|
298
|
20
|
|
|
|
|
48
|
$span{start_end} = $start_end; |
299
|
20
|
|
|
|
|
32
|
$span{heat} = $heat; |
300
|
20
|
|
|
|
|
33
|
$span{pos} = \@cluster_pos; |
301
|
20
|
|
|
|
|
30
|
$span{tokens} = \@cluster_tokens; |
302
|
20
|
|
|
|
|
130
|
$span{str} = join( '', @strings ); |
303
|
|
|
|
|
|
|
|
304
|
|
|
|
|
|
|
# spans with more *unique* hot tokens in a single span rank higher |
305
|
|
|
|
|
|
|
# spans with more *proximate* hot tokens in a single span rank higher |
306
|
20
|
|
|
|
|
40
|
my %uniq = (); |
307
|
20
|
|
|
|
|
19
|
my $i = 0; |
308
|
20
|
|
|
|
|
23
|
my $num_proximate = 1; # one for the single hot token |
309
|
20
|
|
|
|
|
36
|
for (@cluster_pos) { |
310
|
2000
|
100
|
|
|
|
2124
|
if ( exists $heatmap{$_} ) { |
311
|
42
|
|
|
|
|
107
|
$uniq{ lc $strings[$i] } += $heatmap{$_}; |
312
|
42
|
100
|
100
|
|
|
171
|
if ( $i && exists $heatmap{ $cluster_pos[ $i - 2 ] } ) { |
313
|
10
|
|
|
|
|
11
|
$num_proximate++; |
314
|
|
|
|
|
|
|
} |
315
|
|
|
|
|
|
|
} |
316
|
2000
|
|
|
|
|
1282
|
$i++; |
317
|
|
|
|
|
|
|
} |
318
|
20
|
|
|
|
|
45
|
$span{unique} = scalar keys %uniq; |
319
|
20
|
|
|
|
|
30
|
$span{proximate} = $num_proximate; |
320
|
|
|
|
|
|
|
|
321
|
|
|
|
|
|
|
# no false phrase matches if !_treat_phrases_as_singles |
322
|
|
|
|
|
|
|
# stemmer check because regex will likely fail |
323
|
|
|
|
|
|
|
# when stemmer is on |
324
|
20
|
100
|
66
|
|
|
80
|
if ( $query_has_phrase |
325
|
|
|
|
|
|
|
and !$self->{_treat_phrases_as_singles} ) |
326
|
|
|
|
|
|
|
{ |
327
|
7
|
100
|
|
|
|
16
|
if ( !$self->{_stemmer} ) { |
328
|
|
|
|
|
|
|
|
329
|
|
|
|
|
|
|
#warn "_treat_phrases_as_singles NOT true"; |
330
|
3
|
50
|
|
|
|
106
|
if ( $span{str} !~ m/$qre/ ) { |
331
|
0
|
0
|
|
|
|
0
|
$debug |
332
|
|
|
|
|
|
|
and warn |
333
|
|
|
|
|
|
|
"treat_phrases_as_singles=FALSE and '$span{str}' failed to match $qre\n"; |
334
|
0
|
|
|
|
|
0
|
next START_END; |
335
|
|
|
|
|
|
|
} |
336
|
|
|
|
|
|
|
} |
337
|
|
|
|
|
|
|
else { |
338
|
|
|
|
|
|
|
|
339
|
|
|
|
|
|
|
# if stemmer was on, we cannot rely on the regex, |
340
|
|
|
|
|
|
|
# but we assume that number of uniq terms must match query |
341
|
|
|
|
|
|
|
|
342
|
4
|
50
|
66
|
|
|
16
|
if ( $n_terms == $query_has_phrase |
343
|
|
|
|
|
|
|
&& $n_terms > $span{unique} ) |
344
|
|
|
|
|
|
|
{ |
345
|
|
|
|
|
|
|
|
346
|
0
|
0
|
|
|
|
0
|
$debug |
347
|
|
|
|
|
|
|
and warn |
348
|
|
|
|
|
|
|
"treat_phrases_as_singles=FALSE and '$span{str}' " |
349
|
|
|
|
|
|
|
. "expected $n_terms unique terms, got $span{unique}\n"; |
350
|
0
|
|
|
|
|
0
|
next START_END; |
351
|
|
|
|
|
|
|
} |
352
|
|
|
|
|
|
|
|
353
|
|
|
|
|
|
|
} |
354
|
|
|
|
|
|
|
} |
355
|
|
|
|
|
|
|
|
356
|
|
|
|
|
|
|
# just for debug |
357
|
20
|
50
|
|
|
|
53
|
if ($debug) { |
358
|
0
|
|
|
|
|
0
|
my $i = 0; |
359
|
|
|
|
|
|
|
$span{str_w_pos} = join( |
360
|
|
|
|
|
|
|
'', |
361
|
|
|
|
|
|
|
map { |
362
|
0
|
|
|
|
|
0
|
$strings[ $i++ ] |
363
|
|
|
|
|
|
|
. ( exists $heatmap{$_} ? $OPEN : '[' ) |
364
|
|
|
|
|
|
|
. $_ |
365
|
0
|
0
|
|
|
|
0
|
. ( exists $heatmap{$_} ? $CLOSE : ']' ) |
|
|
0
|
|
|
|
|
|
366
|
|
|
|
|
|
|
} @cluster_pos |
367
|
|
|
|
|
|
|
); |
368
|
|
|
|
|
|
|
} |
369
|
|
|
|
|
|
|
|
370
|
20
|
|
|
|
|
130
|
push @spans, \%span; |
371
|
|
|
|
|
|
|
|
372
|
|
|
|
|
|
|
} |
373
|
|
|
|
|
|
|
|
374
|
13
|
|
|
|
|
50
|
$self->{spans} = $self->_sort_spans( \@spans ); |
375
|
13
|
|
|
|
|
30
|
$self->{heatmap} = \%heatmap; |
376
|
|
|
|
|
|
|
|
377
|
13
|
|
|
|
|
195
|
return $self; |
378
|
|
|
|
|
|
|
} |
379
|
|
|
|
|
|
|
|
380
|
|
|
|
|
|
|
sub _sort_spans { |
381
|
|
|
|
|
|
|
return [ |
382
|
|
|
|
|
|
|
|
383
|
|
|
|
|
|
|
# sort by unique, |
384
|
|
|
|
|
|
|
# then by proximity |
385
|
|
|
|
|
|
|
# then by heat |
386
|
|
|
|
|
|
|
# then by pos |
387
|
|
|
|
|
|
|
|
388
|
|
|
|
|
|
|
sort { |
389
|
|
|
|
|
|
|
$b->{unique} <=> $a->{unique} |
390
|
|
|
|
|
|
|
|| $b->{proximate} <=> $a->{proximate} |
391
|
|
|
|
|
|
|
|| $b->{heat} <=> $a->{heat} |
392
|
51
|
50
|
66
|
|
|
267
|
|| $a->{pos}->[0] <=> $b->{pos}->[0] |
|
|
|
100
|
|
|
|
|
393
|
30
|
|
|
30
|
|
49
|
} @{ $_[1] } |
|
30
|
|
|
|
|
126
|
|
394
|
|
|
|
|
|
|
|
395
|
|
|
|
|
|
|
]; |
396
|
|
|
|
|
|
|
} |
397
|
|
|
|
|
|
|
|
398
|
|
|
|
|
|
|
sub _no_sentences { |
399
|
17
|
|
|
17
|
|
31
|
my ( $self, $tokens, $window ) = @_; |
400
|
17
|
|
|
|
|
52
|
my $lhs_window = int( $window / 2 ); |
401
|
17
|
|
50
|
|
|
299
|
my $debug = $self->debug || 0; |
402
|
|
|
|
|
|
|
|
403
|
17
|
|
|
|
|
154
|
my $num_tokens = $tokens->len; |
404
|
17
|
|
|
|
|
46
|
my $tokens_arr = $tokens->as_array; |
405
|
17
|
|
|
|
|
29
|
my %heatmap = (); |
406
|
17
|
|
|
|
|
50
|
my $token_list_heat = $tokens->get_heat; |
407
|
|
|
|
|
|
|
|
408
|
|
|
|
|
|
|
# this regex is a sanity check for phrases. we replace the \ with a |
409
|
|
|
|
|
|
|
# more promiscuous check because the single space is too naive |
410
|
|
|
|
|
|
|
# for real text (e.g. st. john's) |
411
|
17
|
|
|
|
|
30
|
my $qre = $self->{_qre}; |
412
|
17
|
|
|
|
|
21
|
my @phrases = @{ $self->{_query}->phrases }; |
|
17
|
|
|
|
|
65
|
|
413
|
17
|
|
|
|
|
64
|
my $n_terms = $self->{_query}->num_terms; |
414
|
17
|
|
|
|
|
70
|
my $query_has_phrase = $qre =~ s/(\\ )+/.+/g; |
415
|
|
|
|
|
|
|
|
416
|
17
|
50
|
|
|
|
43
|
if ($debug) { |
417
|
0
|
|
|
|
|
0
|
warn "token_list_heat: " . dump($token_list_heat); |
418
|
0
|
|
|
|
|
0
|
warn "n_terms: $n_terms"; |
419
|
0
|
|
|
|
|
0
|
warn "phrases: " . dump( \@phrases ); |
420
|
0
|
|
|
|
|
0
|
warn "query_has_phrase: $query_has_phrase"; |
421
|
|
|
|
|
|
|
} |
422
|
|
|
|
|
|
|
|
423
|
|
|
|
|
|
|
# build heatmap |
424
|
17
|
|
|
|
|
39
|
for (@$token_list_heat) { |
425
|
77
|
|
|
|
|
139
|
my $token = $tokens->get_token($_); |
426
|
77
|
|
|
|
|
247
|
$heatmap{ $token->pos } = $token->is_hot; |
427
|
|
|
|
|
|
|
} |
428
|
|
|
|
|
|
|
|
429
|
|
|
|
|
|
|
# make clusters |
430
|
|
|
|
|
|
|
|
431
|
|
|
|
|
|
|
# $proximity == (1/4 of $window)+1 is somewhat arbitrary, |
432
|
|
|
|
|
|
|
# but since we want to err in having too much context, |
433
|
|
|
|
|
|
|
# we aim high. Worst case scenario is where there are |
434
|
|
|
|
|
|
|
# multiple hot spots in a cluster and each is a full |
435
|
|
|
|
|
|
|
# $proximity length apart, which will grow the |
436
|
|
|
|
|
|
|
# eventual span far beyond $window size. We rely |
437
|
|
|
|
|
|
|
# on max_chars in Snipper to catch that worst case. |
438
|
17
|
|
|
|
|
45
|
my $proximity = int( $lhs_window / 2 ) + 1; |
439
|
17
|
|
|
|
|
76
|
my @positions = sort { $a <=> $b } keys %heatmap; |
|
151
|
|
|
|
|
142
|
|
440
|
17
|
|
|
|
|
38
|
my @clusters = ( [] ); |
441
|
17
|
|
|
|
|
21
|
my $i = 0; |
442
|
17
|
|
|
|
|
30
|
for my $pos (@positions) { |
443
|
|
|
|
|
|
|
|
444
|
|
|
|
|
|
|
# if we have advanced past the first position |
445
|
|
|
|
|
|
|
# and the previous position is not "close" to this one, |
446
|
|
|
|
|
|
|
# start a new cluster |
447
|
77
|
100
|
100
|
|
|
214
|
if ( $i && ( $pos - $positions[ $i - 1 ] ) > $proximity ) { |
448
|
33
|
|
|
|
|
46
|
push( @clusters, [$pos] ); |
449
|
|
|
|
|
|
|
} |
450
|
|
|
|
|
|
|
else { |
451
|
44
|
|
|
|
|
33
|
push( @{ $clusters[-1] }, $pos ); |
|
44
|
|
|
|
|
60
|
|
452
|
|
|
|
|
|
|
} |
453
|
77
|
|
|
|
|
72
|
$i++; |
454
|
|
|
|
|
|
|
} |
455
|
|
|
|
|
|
|
|
456
|
|
|
|
|
|
|
$debug |
457
|
17
|
50
|
|
|
|
36
|
and warn "proximity: $proximity clusters: " . dump \@clusters; |
458
|
|
|
|
|
|
|
|
459
|
|
|
|
|
|
|
# create spans from each cluster, each with a weight. |
460
|
|
|
|
|
|
|
# we do the initial sort so that clusters that overlap |
461
|
|
|
|
|
|
|
# other clusters via get_window() are weeded out via %seen_pos. |
462
|
17
|
|
|
|
|
26
|
my @spans; |
463
|
|
|
|
|
|
|
my %seen_pos; |
464
|
|
|
|
|
|
|
CLUSTER: |
465
|
17
|
|
|
|
|
32
|
for my $cluster ( |
466
|
|
|
|
|
|
|
sort { |
467
|
|
|
|
|
|
|
scalar(@$b) <=> scalar(@$a) |
468
|
61
|
50
|
66
|
|
|
206
|
|| $heatmap{ $b->[0] } <=> $heatmap{ $a->[0] } |
469
|
|
|
|
|
|
|
|| $a->[0] <=> $b->[0] |
470
|
|
|
|
|
|
|
} @clusters |
471
|
|
|
|
|
|
|
) |
472
|
|
|
|
|
|
|
{ |
473
|
|
|
|
|
|
|
|
474
|
|
|
|
|
|
|
# get full window, ignoring positions we've already seen. |
475
|
50
|
|
|
|
|
45
|
my $heat = 0; |
476
|
50
|
|
|
|
|
36
|
my %span; |
477
|
|
|
|
|
|
|
my @cluster_tokens; |
478
|
50
|
|
|
|
|
59
|
POS: for my $pos (@$cluster) { |
479
|
77
|
|
|
|
|
198
|
my ( $start, $end ) = $tokens->get_window( $pos, $window ); |
480
|
77
|
|
|
|
|
122
|
POS_TWO: for my $pos2 ( $start .. $end ) { |
481
|
3357
|
100
|
|
|
|
5728
|
next if $seen_pos{$pos2}++; |
482
|
1513
|
100
|
|
|
|
1431
|
$heat += ( exists $heatmap{$pos2} ? $heatmap{$pos2} : 0 ); |
483
|
1513
|
|
|
|
|
1851
|
push( @cluster_tokens, $tokens->get_token($pos2) ); |
484
|
|
|
|
|
|
|
} |
485
|
|
|
|
|
|
|
} |
486
|
|
|
|
|
|
|
|
487
|
|
|
|
|
|
|
# we may have skipped a $seen_pos from the $slice above |
488
|
|
|
|
|
|
|
# so make sure we still start/end on a match |
489
|
50
|
|
66
|
|
|
226
|
while ( @cluster_tokens && !$cluster_tokens[0]->is_match ) { |
490
|
11
|
|
|
|
|
49
|
shift @cluster_tokens; |
491
|
|
|
|
|
|
|
} |
492
|
50
|
|
66
|
|
|
2075
|
while ( @cluster_tokens && !$cluster_tokens[-1]->is_match ) { |
493
|
6
|
|
|
|
|
25
|
pop @cluster_tokens; |
494
|
|
|
|
|
|
|
} |
495
|
|
|
|
|
|
|
|
496
|
50
|
50
|
|
|
|
79
|
next CLUSTER unless @cluster_tokens; |
497
|
|
|
|
|
|
|
|
498
|
|
|
|
|
|
|
# sanity: make sure we still have something hot |
499
|
50
|
|
|
|
|
43
|
my $has_hot = 0; |
500
|
50
|
|
|
|
|
38
|
my @cluster_pos; |
501
|
|
|
|
|
|
|
my @strings; |
502
|
50
|
|
|
|
|
62
|
for (@cluster_tokens) { |
503
|
1496
|
|
|
|
|
1378
|
my $pos = $_->pos; |
504
|
1496
|
100
|
|
|
|
1711
|
$has_hot++ if exists $heatmap{$pos}; |
505
|
1496
|
|
|
|
|
1715
|
push @strings, $_->str; |
506
|
1496
|
|
|
|
|
1366
|
push @cluster_pos, $pos; |
507
|
|
|
|
|
|
|
} |
508
|
50
|
100
|
|
|
|
111
|
next CLUSTER unless $has_hot; |
509
|
|
|
|
|
|
|
|
510
|
39
|
|
|
|
|
52
|
$span{cluster} = $cluster; |
511
|
39
|
|
|
|
|
51
|
$span{heat} = $heat; |
512
|
39
|
|
|
|
|
44
|
$span{pos} = \@cluster_pos; |
513
|
39
|
|
|
|
|
48
|
$span{tokens} = \@cluster_tokens; |
514
|
39
|
|
|
|
|
125
|
$span{str} = join( '', @strings ); |
515
|
|
|
|
|
|
|
|
516
|
|
|
|
|
|
|
# spans with more *unique* hot tokens in a single span rank higher |
517
|
|
|
|
|
|
|
# spans with more *proximate* hot tokens in a single span rank higher |
518
|
39
|
|
|
|
|
53
|
my %uniq = (); |
519
|
39
|
|
|
|
|
34
|
my $i = 0; |
520
|
39
|
|
|
|
|
30
|
my $num_proximate = 1; # one for the single hot token |
521
|
39
|
|
|
|
|
46
|
for (@cluster_pos) { |
522
|
1341
|
100
|
|
|
|
1454
|
if ( exists $heatmap{$_} ) { |
523
|
77
|
|
|
|
|
154
|
$uniq{ lc $strings[$i] } += $heatmap{$_}; |
524
|
77
|
100
|
100
|
|
|
251
|
if ( $i && exists $heatmap{ $cluster_pos[ $i - 2 ] } ) { |
525
|
23
|
|
|
|
|
22
|
$num_proximate++; |
526
|
|
|
|
|
|
|
} |
527
|
|
|
|
|
|
|
} |
528
|
1341
|
|
|
|
|
849
|
$i++; |
529
|
|
|
|
|
|
|
} |
530
|
39
|
|
|
|
|
55
|
$span{unique} = scalar keys %uniq; |
531
|
39
|
|
|
|
|
45
|
$span{proximate} = $num_proximate; |
532
|
|
|
|
|
|
|
|
533
|
|
|
|
|
|
|
# no false phrase matches if !_treat_phrases_as_singles |
534
|
|
|
|
|
|
|
# stemmer check because regex will likely fail when stemmer is on |
535
|
39
|
100
|
66
|
|
|
98
|
if ( $query_has_phrase |
536
|
|
|
|
|
|
|
and !$self->{_treat_phrases_as_singles} ) |
537
|
|
|
|
|
|
|
{ |
538
|
3
|
100
|
|
|
|
11
|
if ( !$self->{_stemmer} ) { |
539
|
|
|
|
|
|
|
|
540
|
|
|
|
|
|
|
#warn "_treat_phrases_as_singles NOT true"; |
541
|
1
|
50
|
|
|
|
37
|
if ( $span{str} !~ m/$qre/ ) { |
542
|
1
|
50
|
|
|
|
3
|
$debug |
543
|
|
|
|
|
|
|
and warn |
544
|
|
|
|
|
|
|
"treat_phrases_as_singles=FALSE and '$span{str}' failed to match $qre\n"; |
545
|
1
|
|
|
|
|
6
|
next CLUSTER; |
546
|
|
|
|
|
|
|
} |
547
|
|
|
|
|
|
|
} |
548
|
|
|
|
|
|
|
else { |
549
|
|
|
|
|
|
|
|
550
|
|
|
|
|
|
|
# stemmer used, so check unique term count against n_terms |
551
|
2
|
50
|
66
|
|
|
13
|
if ( $n_terms == $query_has_phrase |
552
|
|
|
|
|
|
|
&& $n_terms > $span{unique} ) |
553
|
|
|
|
|
|
|
{ |
554
|
0
|
0
|
|
|
|
0
|
$debug |
555
|
|
|
|
|
|
|
and warn |
556
|
|
|
|
|
|
|
"treat_phrases_as_singles=FALSE and '$span{str}' " |
557
|
|
|
|
|
|
|
. "expected $n_terms but got $span{unique}\n"; |
558
|
0
|
|
|
|
|
0
|
next CLUSTER; |
559
|
|
|
|
|
|
|
} |
560
|
|
|
|
|
|
|
|
561
|
|
|
|
|
|
|
} |
562
|
|
|
|
|
|
|
} |
563
|
|
|
|
|
|
|
|
564
|
|
|
|
|
|
|
# just for debug |
565
|
38
|
50
|
|
|
|
61
|
if ($debug) { |
566
|
0
|
|
|
|
|
0
|
my $i = 0; |
567
|
|
|
|
|
|
|
$span{str_w_pos} = join( |
568
|
|
|
|
|
|
|
'', |
569
|
|
|
|
|
|
|
map { |
570
|
0
|
|
|
|
|
0
|
$strings[ $i++ ] |
571
|
|
|
|
|
|
|
. ( exists $heatmap{$_} ? $OPEN : '[' ) |
572
|
|
|
|
|
|
|
. $_ |
573
|
0
|
0
|
|
|
|
0
|
. ( exists $heatmap{$_} ? $CLOSE : ']' ) |
|
|
0
|
|
|
|
|
|
574
|
|
|
|
|
|
|
} @cluster_pos |
575
|
|
|
|
|
|
|
); |
576
|
|
|
|
|
|
|
} |
577
|
|
|
|
|
|
|
|
578
|
38
|
|
|
|
|
137
|
push @spans, \%span; |
579
|
|
|
|
|
|
|
|
580
|
|
|
|
|
|
|
} |
581
|
|
|
|
|
|
|
|
582
|
17
|
|
|
|
|
1288
|
$self->{spans} = $self->_sort_spans( \@spans ); |
583
|
17
|
|
|
|
|
33
|
$self->{heatmap} = \%heatmap; |
584
|
|
|
|
|
|
|
|
585
|
17
|
|
|
|
|
161
|
return $self; |
586
|
|
|
|
|
|
|
} |
587
|
|
|
|
|
|
|
|
588
|
|
|
|
|
|
|
=head2 has_spans |
589
|
|
|
|
|
|
|
|
590
|
|
|
|
|
|
|
Returns the number of spans found. |
591
|
|
|
|
|
|
|
|
592
|
|
|
|
|
|
|
=cut |
593
|
|
|
|
|
|
|
|
594
|
|
|
|
|
|
|
sub has_spans { |
595
|
30
|
|
|
30
|
1
|
43
|
return scalar @{ $_[0]->{spans} }; |
|
30
|
|
|
|
|
111
|
|
596
|
|
|
|
|
|
|
} |
597
|
|
|
|
|
|
|
|
598
|
|
|
|
|
|
|
1; |
599
|
|
|
|
|
|
|
|
600
|
|
|
|
|
|
|
__END__ |