line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Search::Indexer; |
2
|
|
|
|
|
|
|
|
3
|
2
|
|
|
2
|
|
28004
|
use strict; |
|
2
|
|
|
|
|
4
|
|
|
2
|
|
|
|
|
51
|
|
4
|
2
|
|
|
2
|
|
7
|
use warnings; |
|
2
|
|
|
|
|
3
|
|
|
2
|
|
|
|
|
49
|
|
5
|
2
|
|
|
2
|
|
6
|
no warnings 'uninitialized'; |
|
2
|
|
|
|
|
6
|
|
|
2
|
|
|
|
|
44
|
|
6
|
|
|
|
|
|
|
|
7
|
2
|
|
|
2
|
|
6
|
use Carp; |
|
2
|
|
|
|
|
6
|
|
|
2
|
|
|
|
|
136
|
|
8
|
2
|
|
|
2
|
|
1106
|
use BerkeleyDB; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
use locale; |
10
|
|
|
|
|
|
|
use Search::QueryParser; |
11
|
|
|
|
|
|
|
use List::MoreUtils qw/uniq/; |
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
# TODO : experiment with bit vectors (cf vec() and pack "b*" for combining |
14
|
|
|
|
|
|
|
# result sets |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
our $VERSION = "0.78"; |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
=head1 NAME |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
Search::Indexer - full-text indexer |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
=head1 SYNOPSIS |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
use Search::Indexer; |
25
|
|
|
|
|
|
|
my $ix = new Search::Indexer(dir => $dir, writeMode => 1); |
26
|
|
|
|
|
|
|
foreach my $docId (keys %docs) { |
27
|
|
|
|
|
|
|
$ix->add($docId, $docs{$docId}); |
28
|
|
|
|
|
|
|
} |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
my $result = $ix->search('+word -excludedWord +"exact phrase"'); |
31
|
|
|
|
|
|
|
my @docIds = keys %{$result->{scores}}; |
32
|
|
|
|
|
|
|
my $killedWords = join ", ", @{$result->{killedWords}}; |
33
|
|
|
|
|
|
|
print scalar(@docIds), " documents found\n", ; |
34
|
|
|
|
|
|
|
print "words $killedWords were ignored during the search\n" if $killedWords; |
35
|
|
|
|
|
|
|
foreach my $docId (@docIds) { |
36
|
|
|
|
|
|
|
my $score = $result->{scores}{$docId}; |
37
|
|
|
|
|
|
|
my $excerpts = join "\n", $ix->excerpts($docs{$docId}, $result->{regex}); |
38
|
|
|
|
|
|
|
print "DOCUMENT $docId, score $score:\n$excerpts\n\n"; |
39
|
|
|
|
|
|
|
} |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
my $result2 = $ix->search('word1 AND (word2 OR word3) AND NOT word4'); |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
$ix->remove($someDocId); |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
=head1 DESCRIPTION |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
This module provides support for indexing a collection of documents, |
48
|
|
|
|
|
|
|
for searching the collection, and displaying the sorted results, |
49
|
|
|
|
|
|
|
together with contextual excerpts of the original document. |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
=head2 Documents |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
As far as this module is concerned, a I is just a buffer of |
54
|
|
|
|
|
|
|
plain text, together with a unique identifying number. The caller is |
55
|
|
|
|
|
|
|
responsible for supplying unique numbers, and for converting the |
56
|
|
|
|
|
|
|
original source (HTML, PDF, whatever) into plain text. Documents could |
57
|
|
|
|
|
|
|
also contain more information (other fields like date, author, Dublin |
58
|
|
|
|
|
|
|
Core, etc.), but this must be handled externally, in a database or any |
59
|
|
|
|
|
|
|
other store. A candidate for storing metadata about documents |
60
|
|
|
|
|
|
|
could be L, which uses the same |
61
|
|
|
|
|
|
|
query parser. |
62
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
=head2 Search syntax |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
Searching requests may include plain terms, "exact phrases", |
66
|
|
|
|
|
|
|
'+' or '-' prefixes, boolean operators and parentheses. |
67
|
|
|
|
|
|
|
See L for details. |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
=head2 Index files |
70
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
The indexer uses three files in BerkeleyDB format : a) a mapping from |
72
|
|
|
|
|
|
|
words to wordIds; b) a mapping from wordIds to lists of documents ; c) |
73
|
|
|
|
|
|
|
a mapping from pairs (docId, wordId) to lists of positions within the |
74
|
|
|
|
|
|
|
document. This third file holds detailed information and therefore is |
75
|
|
|
|
|
|
|
quite big ; but it allows us to quickly retrieve "exact phrases" |
76
|
|
|
|
|
|
|
(sequences of adjacent words) in the document. |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
=head2 Indexing steps |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
Indexing of a document buffer goes through the following |
81
|
|
|
|
|
|
|
steps : |
82
|
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
=over |
84
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
=item * |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
terms are extracted, according to the I regular expression |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
=item * |
90
|
|
|
|
|
|
|
|
91
|
|
|
|
|
|
|
extracted terms are normalized or filtered out |
92
|
|
|
|
|
|
|
by the I callback function. This function can for example |
93
|
|
|
|
|
|
|
remove accented characters, perform lemmatization, suppress |
94
|
|
|
|
|
|
|
irrelevant terms (such as numbers), etc. |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
=item * |
97
|
|
|
|
|
|
|
|
98
|
|
|
|
|
|
|
normalized terms are eliminated if they belong to |
99
|
|
|
|
|
|
|
the I list (list of common words to exclude from the index). |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
=item * |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
remaining terms are stored, together with the positions where they |
104
|
|
|
|
|
|
|
occur in the document. |
105
|
|
|
|
|
|
|
|
106
|
|
|
|
|
|
|
=back |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
=head2 Limits |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
All ids are stored as unsigned 32-bit integers; therefore there is |
111
|
|
|
|
|
|
|
a limit of 4294967295 to the number of documents or to the number of |
112
|
|
|
|
|
|
|
different words. |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
=head2 Related modules |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
A short comparison with other CPAN indexing modules is |
117
|
|
|
|
|
|
|
given in the L"SEE ALSO"> section. |
118
|
|
|
|
|
|
|
|
119
|
|
|
|
|
|
|
This module depends on L for analyzing requests and |
120
|
|
|
|
|
|
|
on L for storing the indexes. |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
This module was designed together with L. |
123
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
=cut |
125
|
|
|
|
|
|
|
|
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
sub addToScore (\$$); |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
use constant { |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
# max size of various ids |
132
|
|
|
|
|
|
|
MAX_DOC_ID => 0xFFFFFFFF, # unsigned long (32 bits) |
133
|
|
|
|
|
|
|
MAX_POS_ID => 0xFFFFFFFF, # position_id |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# encodings for pack/unpack |
136
|
|
|
|
|
|
|
IXDPACK => 'wC', # docId : compressed int; freq : unsigned char |
137
|
|
|
|
|
|
|
IXDPACK_L => '(wC)*', # list of above |
138
|
|
|
|
|
|
|
IXPPACK => 'w*', # word positions : list of compressed ints |
139
|
|
|
|
|
|
|
IXPKEYPACK => 'ww', # key for ixp : (docId, wordId) |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
WRITECACHESIZE => (1 << 24), # arbitrary big value; seems good enough but need tuning |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
# default values for args to new() |
144
|
|
|
|
|
|
|
DEFAULT => { |
145
|
|
|
|
|
|
|
writeMode => 0, |
146
|
|
|
|
|
|
|
wregex => qr/\w+/, |
147
|
|
|
|
|
|
|
wfilter => sub { # default filter : lowercase and no accents |
148
|
|
|
|
|
|
|
my $word = lc($_[0]); |
149
|
|
|
|
|
|
|
$word =~ tr[çáàâäéèêëíìîïóòôöúùûüýÿ][caaaaeeeeiiiioooouuuuyy]; |
150
|
|
|
|
|
|
|
return $word; |
151
|
|
|
|
|
|
|
}, |
152
|
|
|
|
|
|
|
fieldname => '', |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
ctxtNumChars => 35, |
155
|
|
|
|
|
|
|
maxExcerpts => 5, |
156
|
|
|
|
|
|
|
preMatch => "", |
157
|
|
|
|
|
|
|
postMatch => "", |
158
|
|
|
|
|
|
|
positions => 1, |
159
|
|
|
|
|
|
|
} |
160
|
|
|
|
|
|
|
}; |
161
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
=head1 METHODS |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
=over |
165
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
=item C expr1, ...)> |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
Creates an indexer (either for a new index, or for |
169
|
|
|
|
|
|
|
accessing an existing index). Parameters are : |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
=over |
172
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
=item dir |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
Directory for index files. and possibly for the stopwords file. |
176
|
|
|
|
|
|
|
Default is current directory |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
=item writeMode |
179
|
|
|
|
|
|
|
|
180
|
|
|
|
|
|
|
Give a true value if you intend to write into the index. |
181
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
=item wregex |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
Regex for matching a word (C by default). |
185
|
|
|
|
|
|
|
Will affect both L and L method. |
186
|
|
|
|
|
|
|
This regex should not contain any capturing parentheses |
187
|
|
|
|
|
|
|
(use non-capturing parentheses C<< (?: ... ) >> instead). |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
=item wfilter |
190
|
|
|
|
|
|
|
|
191
|
|
|
|
|
|
|
Ref to a callback sub that may normalize or eliminate a word. Will |
192
|
|
|
|
|
|
|
affect both L and L method. The default wfilter |
193
|
|
|
|
|
|
|
translates words in lower case and translates latin1 (iso-8859-1) |
194
|
|
|
|
|
|
|
accented characters into plain characters. |
195
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
=item stopwords |
197
|
|
|
|
|
|
|
|
198
|
|
|
|
|
|
|
List of words that will be marked into the index as "words to exclude". |
199
|
|
|
|
|
|
|
This should usually occur when creating a new index ; but nothing prevents |
200
|
|
|
|
|
|
|
you to add other stopwords later. Since stopwords are stored in the |
201
|
|
|
|
|
|
|
index, they need not be specified when opening an index for searches or |
202
|
|
|
|
|
|
|
updates. |
203
|
|
|
|
|
|
|
|
204
|
|
|
|
|
|
|
The list may be supplied either as a ref to an array of scalars, or |
205
|
|
|
|
|
|
|
as a the name of a file containing the stopwords (full pathname |
206
|
|
|
|
|
|
|
or filename relative to I). |
207
|
|
|
|
|
|
|
|
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
=item fieldname |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
Will only affect the L method. |
212
|
|
|
|
|
|
|
Search queries are passed to a general parser |
213
|
|
|
|
|
|
|
(see L). |
214
|
|
|
|
|
|
|
Then, before being applied to the present indexer module, |
215
|
|
|
|
|
|
|
queries are pruned of irrelevant items. |
216
|
|
|
|
|
|
|
Query items are considered relevant if they have no |
217
|
|
|
|
|
|
|
associated field name, or if the associated field name is |
218
|
|
|
|
|
|
|
equal to this C. |
219
|
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
=back |
221
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
Below are some additional parameters that only affect the |
223
|
|
|
|
|
|
|
L method. |
224
|
|
|
|
|
|
|
|
225
|
|
|
|
|
|
|
=over |
226
|
|
|
|
|
|
|
|
227
|
|
|
|
|
|
|
=item ctxtNumChars |
228
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
Number of characters determining the size of contextual excerpts |
230
|
|
|
|
|
|
|
return by the L method. |
231
|
|
|
|
|
|
|
A I is a part of the document text, |
232
|
|
|
|
|
|
|
containg a matched word surrounded by I characters |
233
|
|
|
|
|
|
|
to the left and to the right. Default is 35. |
234
|
|
|
|
|
|
|
|
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
=item maxExcerpts |
237
|
|
|
|
|
|
|
|
238
|
|
|
|
|
|
|
Maximum number of contextual excerpts to retrieve per document. |
239
|
|
|
|
|
|
|
Default is 5. |
240
|
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
=item preMatch |
242
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
String to insert in contextual excerpts before a matched word. |
244
|
|
|
|
|
|
|
Default is C<"EbE">. |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
=item postMatch |
247
|
|
|
|
|
|
|
|
248
|
|
|
|
|
|
|
String to insert in contextual excerpts after a matched word. |
249
|
|
|
|
|
|
|
Default is C<"E/bE">. |
250
|
|
|
|
|
|
|
|
251
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
=item positions |
253
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
my $indexer = new Search::Indexer(dir => $dir, |
255
|
|
|
|
|
|
|
writeMode => 1, |
256
|
|
|
|
|
|
|
positions => 0); |
257
|
|
|
|
|
|
|
|
258
|
|
|
|
|
|
|
Truth value to tell whether or not, when creating a new index, |
259
|
|
|
|
|
|
|
word positions should be stored. The default is true. |
260
|
|
|
|
|
|
|
|
261
|
|
|
|
|
|
|
If you turn it off, index files will be much smaller, indexing |
262
|
|
|
|
|
|
|
will be faster, but results will be less precise, |
263
|
|
|
|
|
|
|
because the indexer can no longer find "exact phrases". |
264
|
|
|
|
|
|
|
So if you type C<"quick fox jumped">, the query will be |
265
|
|
|
|
|
|
|
translated into C, and therefore |
266
|
|
|
|
|
|
|
will retrieve documents in which those three words are present, but |
267
|
|
|
|
|
|
|
not necessarily in order. |
268
|
|
|
|
|
|
|
|
269
|
|
|
|
|
|
|
Another consequence of C<< positions => 0 >> is that |
270
|
|
|
|
|
|
|
there will be no automatic check of uniqueness of ids |
271
|
|
|
|
|
|
|
when adding documents into the index. |
272
|
|
|
|
|
|
|
|
273
|
|
|
|
|
|
|
=back |
274
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
=cut |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
sub new { |
278
|
|
|
|
|
|
|
my $class = shift; |
279
|
|
|
|
|
|
|
my $args = ref $_[0] eq 'HASH' ? $_[0] : {@_}; |
280
|
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
# parse options |
282
|
|
|
|
|
|
|
my $self = {}; |
283
|
|
|
|
|
|
|
$self->{$_} = exists $args->{$_} ? delete $args->{$_} : DEFAULT->{$_} |
284
|
|
|
|
|
|
|
foreach qw(writeMode wregex wfilter fieldname |
285
|
|
|
|
|
|
|
ctxtNumChars maxExcerpts preMatch postMatch positions); |
286
|
|
|
|
|
|
|
my $dir = delete $args->{dir} || "."; |
287
|
|
|
|
|
|
|
$dir =~ s{[/\\]$}{}; # remove trailing slash |
288
|
|
|
|
|
|
|
my $stopwords = delete $args->{stopwords}; |
289
|
|
|
|
|
|
|
|
290
|
|
|
|
|
|
|
# check if invalid options |
291
|
|
|
|
|
|
|
my @remaining = keys %$args; |
292
|
|
|
|
|
|
|
croak "unexpected option : $remaining[0]" if @remaining; |
293
|
|
|
|
|
|
|
croak "can't add 'positions' after index creation time" |
294
|
|
|
|
|
|
|
if $self->{writeMode} and $self->{positions} |
295
|
|
|
|
|
|
|
and -f "$dir/ixd.bdb" and not -f "$dir/ixp.bdb"; |
296
|
|
|
|
|
|
|
|
297
|
|
|
|
|
|
|
# BerkeleyDB environment should allow us to do proper locking for |
298
|
|
|
|
|
|
|
# concurrent access ; but seems to be incompatible with the |
299
|
|
|
|
|
|
|
# -Cachesize argument, so I commented it out ... need to learn more about |
300
|
|
|
|
|
|
|
# BerkeleyDB ... |
301
|
|
|
|
|
|
|
# my $dbEnv = new BerkeleyDB::Env |
302
|
|
|
|
|
|
|
# -Home => $dir, |
303
|
|
|
|
|
|
|
# -Flags => DB_INIT_CDB | DB_INIT_MPOOL | DB_CDB_ALLDB | |
304
|
|
|
|
|
|
|
# ($self->{writeMode} ? DB_CREATE : 0), |
305
|
|
|
|
|
|
|
# -Verbose => 1 |
306
|
|
|
|
|
|
|
# or croak "new BerkeleyDB::Env : $^E $BerkeleyDB::Error" ; |
307
|
|
|
|
|
|
|
|
308
|
|
|
|
|
|
|
|
309
|
|
|
|
|
|
|
my @bdb_args = (# -Env => $dbEnv, # commented out, see explanation above |
310
|
|
|
|
|
|
|
-Flags => ($self->{writeMode} ? DB_CREATE : DB_RDONLY), |
311
|
|
|
|
|
|
|
($self->{writeMode} ? (-Cachesize => WRITECACHESIZE) : ())); |
312
|
|
|
|
|
|
|
|
313
|
|
|
|
|
|
|
# 3 index files : |
314
|
|
|
|
|
|
|
# ixw : word => wordId (or -1 for stopwords) |
315
|
|
|
|
|
|
|
$self->{ixwDb} = tie %{$self->{ixw}}, |
316
|
|
|
|
|
|
|
'BerkeleyDB::Btree', |
317
|
|
|
|
|
|
|
-Filename => "$dir/ixw.bdb", @bdb_args |
318
|
|
|
|
|
|
|
or croak "open $dir/ixw.bdb : $^E $BerkeleyDB::Error"; |
319
|
|
|
|
|
|
|
|
320
|
|
|
|
|
|
|
# ixd : wordId => list of (docId, nOccur) |
321
|
|
|
|
|
|
|
$self->{ixdDb} = tie %{$self->{ixd}}, |
322
|
|
|
|
|
|
|
'BerkeleyDB::Hash', |
323
|
|
|
|
|
|
|
-Filename => "$dir/ixd.bdb", @bdb_args |
324
|
|
|
|
|
|
|
or croak "open $dir/ixd.bdb : $^E $BerkeleyDB::Error"; |
325
|
|
|
|
|
|
|
|
326
|
|
|
|
|
|
|
if (-f "$dir/ixp.bdb" || $self->{writeMode} && $self->{positions}) { |
327
|
|
|
|
|
|
|
# ixp : (docId, wordId) => list of positions of word in doc |
328
|
|
|
|
|
|
|
$self->{ixpDb} = tie %{$self->{ixp}}, |
329
|
|
|
|
|
|
|
'BerkeleyDB::Btree', |
330
|
|
|
|
|
|
|
-Filename => "$dir/ixp.bdb", @bdb_args |
331
|
|
|
|
|
|
|
or croak "open $dir/ixp.bdb : $^E $BerkeleyDB::Error"; |
332
|
|
|
|
|
|
|
} |
333
|
|
|
|
|
|
|
|
334
|
|
|
|
|
|
|
|
335
|
|
|
|
|
|
|
# optional list of stopwords may be given as a list or as a filename |
336
|
|
|
|
|
|
|
if ($stopwords) { |
337
|
|
|
|
|
|
|
$self->{writeMode} or croak "must be in writeMode to specify stopwords"; |
338
|
|
|
|
|
|
|
if (not ref $stopwords) { # if scalar, name of stopwords file |
339
|
|
|
|
|
|
|
open TMP, $stopwords or |
340
|
|
|
|
|
|
|
(open TMP, "$dir/$stopwords") or |
341
|
|
|
|
|
|
|
croak "open stopwords file $stopwords : $^E "; |
342
|
|
|
|
|
|
|
local $/ = undef; |
343
|
|
|
|
|
|
|
my $buf = ; |
344
|
|
|
|
|
|
|
$stopwords = [$buf =~ /$self->{wregex}/g]; |
345
|
|
|
|
|
|
|
close TMP; |
346
|
|
|
|
|
|
|
} |
347
|
|
|
|
|
|
|
foreach my $word (@$stopwords) { |
348
|
|
|
|
|
|
|
$self->{ixw}{$word} = -1; |
349
|
|
|
|
|
|
|
} |
350
|
|
|
|
|
|
|
} |
351
|
|
|
|
|
|
|
|
352
|
|
|
|
|
|
|
bless $self, $class; |
353
|
|
|
|
|
|
|
} |
354
|
|
|
|
|
|
|
|
355
|
|
|
|
|
|
|
|
356
|
|
|
|
|
|
|
|
357
|
|
|
|
|
|
|
|
358
|
|
|
|
|
|
|
|
359
|
|
|
|
|
|
|
=item C |
360
|
|
|
|
|
|
|
|
361
|
|
|
|
|
|
|
Add a new document to the index. |
362
|
|
|
|
|
|
|
I is the unique identifier for this doc |
363
|
|
|
|
|
|
|
(the caller is responsible for uniqueness). |
364
|
|
|
|
|
|
|
I is a scalar containing the text representation of this doc. |
365
|
|
|
|
|
|
|
|
366
|
|
|
|
|
|
|
=cut |
367
|
|
|
|
|
|
|
|
368
|
|
|
|
|
|
|
sub add { |
369
|
|
|
|
|
|
|
my $self = shift; |
370
|
|
|
|
|
|
|
my $docId = shift; |
371
|
|
|
|
|
|
|
# my $buf = shift; # using $_[0] instead for efficiency reasons |
372
|
|
|
|
|
|
|
|
373
|
|
|
|
|
|
|
croak "docId $docId is too large" if $docId > MAX_DOC_ID; |
374
|
|
|
|
|
|
|
|
375
|
|
|
|
|
|
|
# first check if this docId is already used |
376
|
|
|
|
|
|
|
if ($self->{ixp}) { # can only check if we have the positions index |
377
|
|
|
|
|
|
|
my $c = $self->{ixpDb}->db_cursor; |
378
|
|
|
|
|
|
|
my $k = pack IXPKEYPACK, $docId, 0; |
379
|
|
|
|
|
|
|
my $v; # not used, but needed by c_get() |
380
|
|
|
|
|
|
|
my $status = $c->c_get($k, $v, DB_SET_RANGE); |
381
|
|
|
|
|
|
|
if ($status == 0) { |
382
|
|
|
|
|
|
|
my ($check, $wordId) = unpack IXPKEYPACK, $k; |
383
|
|
|
|
|
|
|
croak "docId $docId is already used (wordId=$wordId)" |
384
|
|
|
|
|
|
|
if $docId == $check; |
385
|
|
|
|
|
|
|
} |
386
|
|
|
|
|
|
|
} |
387
|
|
|
|
|
|
|
|
388
|
|
|
|
|
|
|
# OK, let's extract words from the $_[0] buffer |
389
|
|
|
|
|
|
|
my %positions; |
390
|
|
|
|
|
|
|
for (my $nwords = 1; $_[0] =~ /$self->{wregex}/g; $nwords++) { |
391
|
|
|
|
|
|
|
|
392
|
|
|
|
|
|
|
my $word = $self->{wfilter}->($&) or next; |
393
|
|
|
|
|
|
|
my $wordId = $self->{ixw}{$word} || |
394
|
|
|
|
|
|
|
($self->{ixw}{$word} = ++$self->{ixw}{_NWORDS}); # create new wordId |
395
|
|
|
|
|
|
|
push @{$positions{$wordId}}, $nwords if $wordId > 0; |
396
|
|
|
|
|
|
|
} |
397
|
|
|
|
|
|
|
|
398
|
|
|
|
|
|
|
foreach my $wordId (keys %positions) { |
399
|
|
|
|
|
|
|
my $occurrences = @{$positions{$wordId}}; |
400
|
|
|
|
|
|
|
$occurrences = 255 if $occurrences > 255; |
401
|
|
|
|
|
|
|
|
402
|
|
|
|
|
|
|
$self->{ixd}{$wordId} .= pack(IXDPACK, $docId, $occurrences); |
403
|
|
|
|
|
|
|
if ($self->{ixp}) { |
404
|
|
|
|
|
|
|
my $ixpKey = pack IXPKEYPACK, $docId, $wordId; |
405
|
|
|
|
|
|
|
$self->{ixp}{$ixpKey} = pack(IXPPACK, @{$positions{$wordId}}); |
406
|
|
|
|
|
|
|
} |
407
|
|
|
|
|
|
|
} |
408
|
|
|
|
|
|
|
|
409
|
|
|
|
|
|
|
$self->{ixd}{NDOCS} = 0 if not defined $self->{ixd}{NDOCS}; |
410
|
|
|
|
|
|
|
$self->{ixd}{NDOCS} += 1; |
411
|
|
|
|
|
|
|
} |
412
|
|
|
|
|
|
|
|
413
|
|
|
|
|
|
|
|
414
|
|
|
|
|
|
|
=item C |
415
|
|
|
|
|
|
|
|
416
|
|
|
|
|
|
|
Removes a document from the index. |
417
|
|
|
|
|
|
|
If the index contains word positions (true by default), then |
418
|
|
|
|
|
|
|
only the C is needed; however, if the index was created |
419
|
|
|
|
|
|
|
without word positions, then the text representation |
420
|
|
|
|
|
|
|
of the document must be given as a scalar string in the second argument |
421
|
|
|
|
|
|
|
(of course this should be the same as the one that was supplied |
422
|
|
|
|
|
|
|
when calling the L method). |
423
|
|
|
|
|
|
|
|
424
|
|
|
|
|
|
|
|
425
|
|
|
|
|
|
|
=cut |
426
|
|
|
|
|
|
|
|
427
|
|
|
|
|
|
|
sub remove { |
428
|
|
|
|
|
|
|
my $self = shift; |
429
|
|
|
|
|
|
|
my $docId = shift; |
430
|
|
|
|
|
|
|
# my $buf = shift; # using $_[0] instead for efficiency reasons |
431
|
|
|
|
|
|
|
|
432
|
|
|
|
|
|
|
my $wordIds; |
433
|
|
|
|
|
|
|
|
434
|
|
|
|
|
|
|
if ($self->{ixp}) { # if using word positions |
435
|
|
|
|
|
|
|
not $_[0] or carp "remove() : unexpected 'buf' argument"; |
436
|
|
|
|
|
|
|
$wordIds= $self->wordIds($docId); |
437
|
|
|
|
|
|
|
} |
438
|
|
|
|
|
|
|
else { # otherwise : recompute word ids |
439
|
|
|
|
|
|
|
$wordIds = [grep {defined $_ and $_ > 0} |
440
|
|
|
|
|
|
|
map {$self->{ixw}{$_}} |
441
|
|
|
|
|
|
|
uniq map {$self->{wfilter}->($_)} |
442
|
|
|
|
|
|
|
($_[0] =~ /$self->{wregex}/g)]; |
443
|
|
|
|
|
|
|
} |
444
|
|
|
|
|
|
|
|
445
|
|
|
|
|
|
|
return if not @$wordIds; |
446
|
|
|
|
|
|
|
|
447
|
|
|
|
|
|
|
foreach my $wordId (@$wordIds) { |
448
|
|
|
|
|
|
|
my %docs = unpack IXDPACK_L, $self->{ixd}{$wordId}; |
449
|
|
|
|
|
|
|
delete $docs{$docId}; |
450
|
|
|
|
|
|
|
$self->{ixd}{$wordId} = pack IXDPACK_L, %docs; |
451
|
|
|
|
|
|
|
if ($self->{ixp}) { |
452
|
|
|
|
|
|
|
my $ixpKey = pack IXPKEYPACK, $docId, $wordId; |
453
|
|
|
|
|
|
|
delete $self->{ixp}{$ixpKey}; |
454
|
|
|
|
|
|
|
} |
455
|
|
|
|
|
|
|
} |
456
|
|
|
|
|
|
|
|
457
|
|
|
|
|
|
|
$self->{ixd}{NDOCS} -= 1; |
458
|
|
|
|
|
|
|
} |
459
|
|
|
|
|
|
|
|
460
|
|
|
|
|
|
|
=item C |
461
|
|
|
|
|
|
|
|
462
|
|
|
|
|
|
|
Returns a ref to an array of word Ids contained in the specified document |
463
|
|
|
|
|
|
|
(not available if the index was created with C<< positions => 0 >>) |
464
|
|
|
|
|
|
|
|
465
|
|
|
|
|
|
|
=cut |
466
|
|
|
|
|
|
|
|
467
|
|
|
|
|
|
|
sub wordIds { |
468
|
|
|
|
|
|
|
my $self = shift; |
469
|
|
|
|
|
|
|
my $docId_ini = shift; |
470
|
|
|
|
|
|
|
|
471
|
|
|
|
|
|
|
$self->{ixpDb} |
472
|
|
|
|
|
|
|
or croak "wordIds() not available (index was created with positions=>0)"; |
473
|
|
|
|
|
|
|
|
474
|
|
|
|
|
|
|
my @wordIds = (); |
475
|
|
|
|
|
|
|
my $c = $self->{ixpDb}->db_cursor; |
476
|
|
|
|
|
|
|
my ($k, $v); |
477
|
|
|
|
|
|
|
$k = pack IXPKEYPACK, $docId_ini, 0; |
478
|
|
|
|
|
|
|
my $status = $c->c_get($k, $v, DB_SET_RANGE); |
479
|
|
|
|
|
|
|
while ($status == 0) { |
480
|
|
|
|
|
|
|
my ($docId, $wordId) = unpack IXPKEYPACK, $k; |
481
|
|
|
|
|
|
|
last if $docId != $docId_ini; |
482
|
|
|
|
|
|
|
push @wordIds, $wordId; |
483
|
|
|
|
|
|
|
$status = $c->c_get($k, $v, DB_NEXT); |
484
|
|
|
|
|
|
|
} |
485
|
|
|
|
|
|
|
return \@wordIds; |
486
|
|
|
|
|
|
|
} |
487
|
|
|
|
|
|
|
|
488
|
|
|
|
|
|
|
|
489
|
|
|
|
|
|
|
=item C |
490
|
|
|
|
|
|
|
|
491
|
|
|
|
|
|
|
Returns a ref to an array of words found in the dictionary, |
492
|
|
|
|
|
|
|
starting with prefix (i.e. C<< $ix->words("foo") >> will |
493
|
|
|
|
|
|
|
return "foo", "food", "fool", "footage", etc.). |
494
|
|
|
|
|
|
|
|
495
|
|
|
|
|
|
|
=cut |
496
|
|
|
|
|
|
|
|
497
|
|
|
|
|
|
|
sub words { |
498
|
|
|
|
|
|
|
my $self = shift; |
499
|
|
|
|
|
|
|
my $prefix = shift; |
500
|
|
|
|
|
|
|
|
501
|
|
|
|
|
|
|
my $regex = qr/^$prefix/; |
502
|
|
|
|
|
|
|
my @words = (); |
503
|
|
|
|
|
|
|
my $c = $self->{ixwDb}->db_cursor; |
504
|
|
|
|
|
|
|
my ($k, $v); |
505
|
|
|
|
|
|
|
$k = $prefix; |
506
|
|
|
|
|
|
|
my $status = $c->c_get($k, $v, DB_SET_RANGE); |
507
|
|
|
|
|
|
|
while ($status == 0) { |
508
|
|
|
|
|
|
|
last if $k !~ $regex; |
509
|
|
|
|
|
|
|
push @words, $k; |
510
|
|
|
|
|
|
|
$status = $c->c_get($k, $v, DB_NEXT); |
511
|
|
|
|
|
|
|
} |
512
|
|
|
|
|
|
|
return \@words; |
513
|
|
|
|
|
|
|
} |
514
|
|
|
|
|
|
|
|
515
|
|
|
|
|
|
|
|
516
|
|
|
|
|
|
|
|
517
|
|
|
|
|
|
|
=item C |
518
|
|
|
|
|
|
|
|
519
|
|
|
|
|
|
|
Debugging function, prints indexed words with list of associated docs. |
520
|
|
|
|
|
|
|
|
521
|
|
|
|
|
|
|
=cut |
522
|
|
|
|
|
|
|
|
523
|
|
|
|
|
|
|
sub dump { |
524
|
|
|
|
|
|
|
my $self = shift; |
525
|
|
|
|
|
|
|
foreach my $word (sort keys %{$self->{ixw}}) { |
526
|
|
|
|
|
|
|
my $wordId = $self->{ixw}{$word}; |
527
|
|
|
|
|
|
|
if ($wordId == -1) { |
528
|
|
|
|
|
|
|
print "$word : STOPWORD\n"; |
529
|
|
|
|
|
|
|
} |
530
|
|
|
|
|
|
|
else { |
531
|
|
|
|
|
|
|
my %docs = unpack IXDPACK_L, $self->{ixd}{$wordId}; |
532
|
|
|
|
|
|
|
print "$word : ", join (" ", keys %docs), "\n"; |
533
|
|
|
|
|
|
|
} |
534
|
|
|
|
|
|
|
} |
535
|
|
|
|
|
|
|
} |
536
|
|
|
|
|
|
|
|
537
|
|
|
|
|
|
|
|
538
|
|
|
|
|
|
|
=item C |
539
|
|
|
|
|
|
|
|
540
|
|
|
|
|
|
|
Searches the index. See the L and L sections |
541
|
|
|
|
|
|
|
above for short descriptions of query strings, or |
542
|
|
|
|
|
|
|
L for details. The second argument is optional ; |
543
|
|
|
|
|
|
|
if true, all words without any prefix will implicitly take prefix '+' |
544
|
|
|
|
|
|
|
(mandatory words). |
545
|
|
|
|
|
|
|
|
546
|
|
|
|
|
|
|
The return value is a hash ref containing |
547
|
|
|
|
|
|
|
|
548
|
|
|
|
|
|
|
=over |
549
|
|
|
|
|
|
|
|
550
|
|
|
|
|
|
|
=item scores |
551
|
|
|
|
|
|
|
|
552
|
|
|
|
|
|
|
hash ref, where keys are docIds of matching documents, and values are |
553
|
|
|
|
|
|
|
the corresponding computed scores. |
554
|
|
|
|
|
|
|
|
555
|
|
|
|
|
|
|
=item killedWords |
556
|
|
|
|
|
|
|
|
557
|
|
|
|
|
|
|
ref to an array of terms from the query string which were ignored |
558
|
|
|
|
|
|
|
during the search (because they were filtered out or were stopwords) |
559
|
|
|
|
|
|
|
|
560
|
|
|
|
|
|
|
=item regex |
561
|
|
|
|
|
|
|
|
562
|
|
|
|
|
|
|
ref to a regular expression corresponding to all terms in the query |
563
|
|
|
|
|
|
|
string. This will be useful if you later want to get contextual |
564
|
|
|
|
|
|
|
excerpts from the found documents (see the L method). |
565
|
|
|
|
|
|
|
|
566
|
|
|
|
|
|
|
=back |
567
|
|
|
|
|
|
|
|
568
|
|
|
|
|
|
|
=cut |
569
|
|
|
|
|
|
|
|
570
|
|
|
|
|
|
|
|
571
|
|
|
|
|
|
|
sub search { |
572
|
|
|
|
|
|
|
my $self = shift; |
573
|
|
|
|
|
|
|
my $query_string = shift; |
574
|
|
|
|
|
|
|
my $implicitPlus = shift; |
575
|
|
|
|
|
|
|
|
576
|
|
|
|
|
|
|
$self->{qp} ||= new Search::QueryParser; |
577
|
|
|
|
|
|
|
|
578
|
|
|
|
|
|
|
my $q = $self->{qp}->parse($query_string, $implicitPlus); |
579
|
|
|
|
|
|
|
my $killedWords = {}; |
580
|
|
|
|
|
|
|
my $wordsRegexes = []; |
581
|
|
|
|
|
|
|
|
582
|
|
|
|
|
|
|
my $qt = $self->translateQuery($q, $killedWords, $wordsRegexes); |
583
|
|
|
|
|
|
|
|
584
|
|
|
|
|
|
|
my $tmp = {}; |
585
|
|
|
|
|
|
|
$tmp->{$_} = 1 foreach @$wordsRegexes; |
586
|
|
|
|
|
|
|
my $strRegex = "(?:" . join("|", keys %$tmp) . ")"; |
587
|
|
|
|
|
|
|
|
588
|
|
|
|
|
|
|
return {scores => $self->_search($qt), |
589
|
|
|
|
|
|
|
killedWords => [keys %$killedWords], |
590
|
|
|
|
|
|
|
regex => qr/$strRegex/i}; |
591
|
|
|
|
|
|
|
} |
592
|
|
|
|
|
|
|
|
593
|
|
|
|
|
|
|
|
594
|
|
|
|
|
|
|
sub _search { |
595
|
|
|
|
|
|
|
my ($self, $q) = @_; |
596
|
|
|
|
|
|
|
|
597
|
|
|
|
|
|
|
my $scores = undef; # hash {doc1 => score1, doc2 => score2 ...} |
598
|
|
|
|
|
|
|
|
599
|
|
|
|
|
|
|
# 1) deal with mandatory subqueries |
600
|
|
|
|
|
|
|
|
601
|
|
|
|
|
|
|
foreach my $subQ ( @{$q->{'+'}} ) { |
602
|
|
|
|
|
|
|
my $sc = $self->docsAndScores($subQ) or next; |
603
|
|
|
|
|
|
|
$scores = $sc and next if not $scores; # if first result set, just store |
604
|
|
|
|
|
|
|
|
605
|
|
|
|
|
|
|
# otherwise, intersect with previous result set |
606
|
|
|
|
|
|
|
foreach my $docId (keys %$scores) { |
607
|
|
|
|
|
|
|
delete $scores->{$docId} and next if not defined $sc->{$docId}; |
608
|
|
|
|
|
|
|
addToScore $scores->{$docId}, $sc->{$docId}; # otherwise |
609
|
|
|
|
|
|
|
} |
610
|
|
|
|
|
|
|
} |
611
|
|
|
|
|
|
|
|
612
|
|
|
|
|
|
|
my $noMandatorySubq = not $scores; |
613
|
|
|
|
|
|
|
|
614
|
|
|
|
|
|
|
# 2) deal with non-mandatory subqueries |
615
|
|
|
|
|
|
|
|
616
|
|
|
|
|
|
|
foreach my $subQ (@{$q->{''}}) { |
617
|
|
|
|
|
|
|
my $sc = $self->docsAndScores($subQ) or next; |
618
|
|
|
|
|
|
|
$scores = $sc and next if not $scores; # if first result set, just store |
619
|
|
|
|
|
|
|
|
620
|
|
|
|
|
|
|
# otherwise, combine with previous result set |
621
|
|
|
|
|
|
|
foreach my $docId (keys %$sc) { |
622
|
|
|
|
|
|
|
if (defined $scores->{$docId}) { # docId was already there, add new score |
623
|
|
|
|
|
|
|
addToScore $scores->{$docId}, $sc->{$docId}; |
624
|
|
|
|
|
|
|
} |
625
|
|
|
|
|
|
|
elsif ($noMandatorySubq){ # insert a new docId to the result set |
626
|
|
|
|
|
|
|
$scores->{$docId} = $sc->{$docId}; |
627
|
|
|
|
|
|
|
} |
628
|
|
|
|
|
|
|
# else do nothing (ignore this docId) |
629
|
|
|
|
|
|
|
} |
630
|
|
|
|
|
|
|
} |
631
|
|
|
|
|
|
|
|
632
|
|
|
|
|
|
|
return undef if not $scores or not %$scores; # no results |
633
|
|
|
|
|
|
|
|
634
|
|
|
|
|
|
|
# 3) deal with negative subqueries (remove corresponding docs from results) |
635
|
|
|
|
|
|
|
|
636
|
|
|
|
|
|
|
foreach my $subQ (@{$q->{'-'}}) { |
637
|
|
|
|
|
|
|
my $negScores = $self->docsAndScores($subQ) or next; |
638
|
|
|
|
|
|
|
delete $scores->{$_} foreach keys %$negScores; |
639
|
|
|
|
|
|
|
} |
640
|
|
|
|
|
|
|
|
641
|
|
|
|
|
|
|
return $scores; |
642
|
|
|
|
|
|
|
} |
643
|
|
|
|
|
|
|
|
644
|
|
|
|
|
|
|
|
645
|
|
|
|
|
|
|
|
646
|
|
|
|
|
|
|
sub docsAndScores { # returns a hash {docId => score} or undef (no info) |
647
|
|
|
|
|
|
|
my ($self, $subQ) = @_; |
648
|
|
|
|
|
|
|
|
649
|
|
|
|
|
|
|
# recursive call to _search if $subQ is a parenthesized query |
650
|
|
|
|
|
|
|
return $self->_search($subQ->{value}) if $subQ->{op} eq '()'; |
651
|
|
|
|
|
|
|
|
652
|
|
|
|
|
|
|
# otherwise, don't care about $subQ->{op} (assert $subQ->{op} eq ':') |
653
|
|
|
|
|
|
|
|
654
|
|
|
|
|
|
|
if (ref $subQ->{value}) { # several words, this is an "exact phrase" |
655
|
|
|
|
|
|
|
return $self->matchExactPhrase($subQ); |
656
|
|
|
|
|
|
|
} |
657
|
|
|
|
|
|
|
elsif ($subQ->{value} <= -1) {# this is a stopword |
658
|
|
|
|
|
|
|
return undef; |
659
|
|
|
|
|
|
|
} |
660
|
|
|
|
|
|
|
else { # scalar value, match single word |
661
|
|
|
|
|
|
|
my $scores = {unpack IXDPACK_L, ($self->{ixd}{$subQ->{value}} || "")}; |
662
|
|
|
|
|
|
|
my @k = keys %$scores; |
663
|
|
|
|
|
|
|
if (@k) { |
664
|
|
|
|
|
|
|
my $coeff = log(($self->{ixd}{NDOCS} + 1)/@k) * 100; |
665
|
|
|
|
|
|
|
$scores->{$_} = int($coeff * $scores->{$_}) foreach @k; |
666
|
|
|
|
|
|
|
} |
667
|
|
|
|
|
|
|
return $scores; |
668
|
|
|
|
|
|
|
} |
669
|
|
|
|
|
|
|
} |
670
|
|
|
|
|
|
|
|
671
|
|
|
|
|
|
|
|
672
|
|
|
|
|
|
|
sub matchExactPhrase { |
673
|
|
|
|
|
|
|
my ($self, $subQ) = @_; |
674
|
|
|
|
|
|
|
|
675
|
|
|
|
|
|
|
if (! $self->{ixp}) { # if not indexed with positions |
676
|
|
|
|
|
|
|
# translate into an AND query |
677
|
|
|
|
|
|
|
my $fake_query = {'+' => [map {{op => ':', |
678
|
|
|
|
|
|
|
value => $_ }} @{$subQ->{value}}]}; |
679
|
|
|
|
|
|
|
# and search for that one |
680
|
|
|
|
|
|
|
return $self->_search($fake_query); |
681
|
|
|
|
|
|
|
}; |
682
|
|
|
|
|
|
|
|
683
|
|
|
|
|
|
|
# otherwise, intersect word position sets |
684
|
|
|
|
|
|
|
my %pos; |
685
|
|
|
|
|
|
|
my $wordDelta = 0; |
686
|
|
|
|
|
|
|
my $scores = undef; |
687
|
|
|
|
|
|
|
foreach my $wordId (@{$subQ->{value}}) { |
688
|
|
|
|
|
|
|
my $sc = $self->docsAndScores({op=>':', value=>$wordId}); |
689
|
|
|
|
|
|
|
if (not $scores) { # no previous result set |
690
|
|
|
|
|
|
|
if ($sc) { |
691
|
|
|
|
|
|
|
$scores = $sc; |
692
|
|
|
|
|
|
|
foreach my $docId (keys %$scores) { |
693
|
|
|
|
|
|
|
my $ixpKey = pack IXPKEYPACK, $docId, $wordId; |
694
|
|
|
|
|
|
|
$pos{$docId} = [unpack IXPPACK, $self->{ixp}{$ixpKey}]; |
695
|
|
|
|
|
|
|
} |
696
|
|
|
|
|
|
|
} |
697
|
|
|
|
|
|
|
} |
698
|
|
|
|
|
|
|
else { # combine with previous result set |
699
|
|
|
|
|
|
|
$wordDelta++; |
700
|
|
|
|
|
|
|
foreach my $docId (keys %$scores) { |
701
|
|
|
|
|
|
|
if ($sc) { # if we have info about current word (is not a stopword) |
702
|
|
|
|
|
|
|
if (not defined $sc->{$docId}) { # current word not in current doc |
703
|
|
|
|
|
|
|
delete $scores->{$docId}; |
704
|
|
|
|
|
|
|
} else { # current word found in current doc, check if positions match |
705
|
|
|
|
|
|
|
my $ixpKey = pack IXPKEYPACK, $docId, $wordId; |
706
|
|
|
|
|
|
|
my @newPos = unpack IXPPACK, $self->{ixp}{$ixpKey}; |
707
|
|
|
|
|
|
|
$pos{$docId} = nearPositions($pos{$docId}, \@newPos, $wordDelta) |
708
|
|
|
|
|
|
|
and addToScore $scores->{$docId}, $sc->{$docId} |
709
|
|
|
|
|
|
|
or delete $scores->{$docId}; |
710
|
|
|
|
|
|
|
} |
711
|
|
|
|
|
|
|
} |
712
|
|
|
|
|
|
|
} # end foreach my $docId (keys %$scores) |
713
|
|
|
|
|
|
|
} |
714
|
|
|
|
|
|
|
} # end foreach my $wordId (@{$subQ->{value}}) |
715
|
|
|
|
|
|
|
|
716
|
|
|
|
|
|
|
return $scores; |
717
|
|
|
|
|
|
|
} |
718
|
|
|
|
|
|
|
|
719
|
|
|
|
|
|
|
|
720
|
|
|
|
|
|
|
|
721
|
|
|
|
|
|
|
|
722
|
|
|
|
|
|
|
|
723
|
|
|
|
|
|
|
|
724
|
|
|
|
|
|
|
|
725
|
|
|
|
|
|
|
|
726
|
|
|
|
|
|
|
|
727
|
|
|
|
|
|
|
|
728
|
|
|
|
|
|
|
|
729
|
|
|
|
|
|
|
|
730
|
|
|
|
|
|
|
|
731
|
|
|
|
|
|
|
|
732
|
|
|
|
|
|
|
|
733
|
|
|
|
|
|
|
|
734
|
|
|
|
|
|
|
|
735
|
|
|
|
|
|
|
sub nearPositions { |
736
|
|
|
|
|
|
|
my ($set1, $set2, $wordDelta) = @_; |
737
|
|
|
|
|
|
|
# returns the set of positions in $set2 which are "close enough" (<= $wordDelta) |
738
|
|
|
|
|
|
|
# to positions in $set1. Assumption : input sets are sorted. |
739
|
|
|
|
|
|
|
|
740
|
|
|
|
|
|
|
|
741
|
|
|
|
|
|
|
my @result; |
742
|
|
|
|
|
|
|
my ($i1, $i2) = (0, 0); # indices into sets |
743
|
|
|
|
|
|
|
|
744
|
|
|
|
|
|
|
while ($i1 < @$set1 and $i2 < @$set2) { |
745
|
|
|
|
|
|
|
my $delta = $set2->[$i2] - $set1->[$i1]; |
746
|
|
|
|
|
|
|
++$i1 and next if $delta > $wordDelta; |
747
|
|
|
|
|
|
|
push @result, $set2->[$i2] if $delta > 0; |
748
|
|
|
|
|
|
|
++$i2; |
749
|
|
|
|
|
|
|
} |
750
|
|
|
|
|
|
|
return @result ? \@result : undef; |
751
|
|
|
|
|
|
|
} |
752
|
|
|
|
|
|
|
|
753
|
|
|
|
|
|
|
|
754
|
|
|
|
|
|
|
|
755
|
|
|
|
|
|
|
sub addToScore (\$$) { # first score arg gets "incremented" by the second arg |
756
|
|
|
|
|
|
|
my ($ptScore1, $score2) = @_; |
757
|
|
|
|
|
|
|
$$ptScore1 = 0 if not defined $$ptScore1; |
758
|
|
|
|
|
|
|
$$ptScore1 += $score2 if $score2; # TODO : find better formula for score combination ! |
759
|
|
|
|
|
|
|
} |
760
|
|
|
|
|
|
|
|
761
|
|
|
|
|
|
|
|
762
|
|
|
|
|
|
|
sub translateQuery { # replace words by ids, remove irrelevant subqueries |
763
|
|
|
|
|
|
|
my ($self, $q, $killedWords, $wordsRegexes) = @_; |
764
|
|
|
|
|
|
|
|
765
|
|
|
|
|
|
|
my $r = {}; |
766
|
|
|
|
|
|
|
|
767
|
|
|
|
|
|
|
foreach my $k ('+', '-', '') { |
768
|
|
|
|
|
|
|
foreach my $subQ (@{$q->{$k}}) { |
769
|
|
|
|
|
|
|
|
770
|
|
|
|
|
|
|
# ignore items concerning other field names |
771
|
|
|
|
|
|
|
next if $subQ->{field} and $subQ->{field} ne $self->{fieldname}; |
772
|
|
|
|
|
|
|
|
773
|
|
|
|
|
|
|
my $val = $subQ->{value}; |
774
|
|
|
|
|
|
|
|
775
|
|
|
|
|
|
|
my $clone = undef; |
776
|
|
|
|
|
|
|
if ($subQ->{op} eq '()') { |
777
|
|
|
|
|
|
|
$clone = {op => '()', |
778
|
|
|
|
|
|
|
value => $self->translateQuery($val, $killedWords, $wordsRegexes)}; |
779
|
|
|
|
|
|
|
} |
780
|
|
|
|
|
|
|
elsif ($subQ->{op} eq ':') { |
781
|
|
|
|
|
|
|
# split query according to our notion of "term" |
782
|
|
|
|
|
|
|
my @words = ($val =~ /$self->{wregex}/g); |
783
|
|
|
|
|
|
|
|
784
|
|
|
|
|
|
|
# TODO : 1) accept '*' suffix; 2) find keys in $self->{ixw}; 3) rewrite into |
785
|
|
|
|
|
|
|
# an 'OR' query |
786
|
|
|
|
|
|
|
|
787
|
|
|
|
|
|
|
# my @words = ($str =~ /$self->{wregex}\*?/g); |
788
|
|
|
|
|
|
|
|
789
|
|
|
|
|
|
|
my $regex1 = join "\\W+", map quotemeta, @words; |
790
|
|
|
|
|
|
|
my $regex2 = join "\\W+", map quotemeta, |
791
|
|
|
|
|
|
|
map {$self->{wfilter}($_)} @words; |
792
|
|
|
|
|
|
|
foreach my $regex ($regex1, $regex2) { |
793
|
|
|
|
|
|
|
$regex = "\\b$regex" if $regex =~ /^\w/; |
794
|
|
|
|
|
|
|
$regex = "$regex\\b" if $regex =~ /\w$/; |
795
|
|
|
|
|
|
|
} |
796
|
|
|
|
|
|
|
push @$wordsRegexes, $regex1; |
797
|
|
|
|
|
|
|
push @$wordsRegexes, $regex2 unless $regex1 eq $regex2; |
798
|
|
|
|
|
|
|
|
799
|
|
|
|
|
|
|
# now translate into word ids |
800
|
|
|
|
|
|
|
foreach my $word (@words) { |
801
|
|
|
|
|
|
|
my $wf = $self->{wfilter}->($word); |
802
|
|
|
|
|
|
|
my $wordId = $wf ? ($self->{ixw}{$wf} || 0) : -1; |
803
|
|
|
|
|
|
|
$killedWords->{$word} = 1 if $wordId < 0; |
804
|
|
|
|
|
|
|
$word = $wordId; |
805
|
|
|
|
|
|
|
} |
806
|
|
|
|
|
|
|
|
807
|
|
|
|
|
|
|
$val = (@words>1) ? \@words : # several words : return an array |
808
|
|
|
|
|
|
|
(@words>0) ? $words[0] : # just one word : return its id |
809
|
|
|
|
|
|
|
0; # no word : return 0 (means "no info") |
810
|
|
|
|
|
|
|
|
811
|
|
|
|
|
|
|
$clone = {op => ':', value=> $val}; |
812
|
|
|
|
|
|
|
} |
813
|
|
|
|
|
|
|
push @{$r->{$k}}, $clone if $clone; |
814
|
|
|
|
|
|
|
} |
815
|
|
|
|
|
|
|
} |
816
|
|
|
|
|
|
|
|
817
|
|
|
|
|
|
|
return $r; |
818
|
|
|
|
|
|
|
} |
819
|
|
|
|
|
|
|
|
820
|
|
|
|
|
|
|
|
821
|
|
|
|
|
|
|
|
822
|
|
|
|
|
|
|
=item C |
823
|
|
|
|
|
|
|
|
824
|
|
|
|
|
|
|
Searches C for occurrences of C, |
825
|
|
|
|
|
|
|
extracts the occurences together with some context |
826
|
|
|
|
|
|
|
(a number of characters to the left and to the right), |
827
|
|
|
|
|
|
|
and highlights the occurences. See parameters C, |
828
|
|
|
|
|
|
|
C, C, C of the L method. |
829
|
|
|
|
|
|
|
|
830
|
|
|
|
|
|
|
=cut |
831
|
|
|
|
|
|
|
|
832
|
|
|
|
|
|
|
sub excerpts { |
833
|
|
|
|
|
|
|
my $self = shift; |
834
|
|
|
|
|
|
|
# $_[0] : text buffer ; no copy for efficiency reason |
835
|
|
|
|
|
|
|
my $regex = $_[1]; |
836
|
|
|
|
|
|
|
|
837
|
|
|
|
|
|
|
my $nc = $self->{ctxtNumChars}; |
838
|
|
|
|
|
|
|
|
839
|
|
|
|
|
|
|
# find start and end positions of matching fragments |
840
|
|
|
|
|
|
|
my $matches = []; # array of refs to [start, end, number_of_matches] |
841
|
|
|
|
|
|
|
while ($_[0] =~ /$regex/g) { |
842
|
|
|
|
|
|
|
my ($start, $end) = ($-[0], $+[0]); |
843
|
|
|
|
|
|
|
if (@$matches and $start <= $matches->[-1][1] + $nc) { |
844
|
|
|
|
|
|
|
# merge with the last fragment if close enough |
845
|
|
|
|
|
|
|
$matches->[-1][1] = $end; # extend the end position |
846
|
|
|
|
|
|
|
$matches->[-1][2] += 1; # increment the number of matches |
847
|
|
|
|
|
|
|
} |
848
|
|
|
|
|
|
|
else { |
849
|
|
|
|
|
|
|
push @$matches, [$start, $end, 1]; |
850
|
|
|
|
|
|
|
} |
851
|
|
|
|
|
|
|
} |
852
|
|
|
|
|
|
|
|
853
|
|
|
|
|
|
|
foreach (@$matches) { # extend start and end positions by $self->{ctxtNumChars} |
854
|
|
|
|
|
|
|
$_->[0] = ($_->[0] < $nc) ? 0 : $_->[0] - $nc; |
855
|
|
|
|
|
|
|
$_->[1] += $nc; |
856
|
|
|
|
|
|
|
} |
857
|
|
|
|
|
|
|
|
858
|
|
|
|
|
|
|
my $excerpts = []; |
859
|
|
|
|
|
|
|
foreach my $match (sort {$b->[2] <=> $a->[2]} @$matches) { |
860
|
|
|
|
|
|
|
last if @$excerpts >= $self->{maxExcerpts}; |
861
|
|
|
|
|
|
|
my $x = substr($_[0], $match->[0], $match->[1] - $match->[0]); # extract |
862
|
|
|
|
|
|
|
$x =~ s/$regex/$self->{preMatch}$&$self->{postMatch}/g ; # highlight |
863
|
|
|
|
|
|
|
push @$excerpts, "...$x..."; |
864
|
|
|
|
|
|
|
} |
865
|
|
|
|
|
|
|
return $excerpts; |
866
|
|
|
|
|
|
|
} |
867
|
|
|
|
|
|
|
|
868
|
|
|
|
|
|
|
=back |
869
|
|
|
|
|
|
|
|
870
|
|
|
|
|
|
|
=head1 TO DO |
871
|
|
|
|
|
|
|
|
872
|
|
|
|
|
|
|
=over |
873
|
|
|
|
|
|
|
|
874
|
|
|
|
|
|
|
=item * |
875
|
|
|
|
|
|
|
|
876
|
|
|
|
|
|
|
Find a proper formula for combining scores from several terms. |
877
|
|
|
|
|
|
|
Current implementation is ridiculously simple-minded (just an addition). |
878
|
|
|
|
|
|
|
Also study the literature to improve the scoring formula. |
879
|
|
|
|
|
|
|
|
880
|
|
|
|
|
|
|
=item * |
881
|
|
|
|
|
|
|
|
882
|
|
|
|
|
|
|
Handle concurrency through BerkeleyDB locks. |
883
|
|
|
|
|
|
|
|
884
|
|
|
|
|
|
|
=item * |
885
|
|
|
|
|
|
|
|
886
|
|
|
|
|
|
|
Maybe put all 3 index files as subDatabases in one single file. |
887
|
|
|
|
|
|
|
|
888
|
|
|
|
|
|
|
=item * |
889
|
|
|
|
|
|
|
|
890
|
|
|
|
|
|
|
Fine tuning of cachesize and other BerkeleyDB parameters. |
891
|
|
|
|
|
|
|
|
892
|
|
|
|
|
|
|
=item * |
893
|
|
|
|
|
|
|
|
894
|
|
|
|
|
|
|
Compare performances with other packages. |
895
|
|
|
|
|
|
|
|
896
|
|
|
|
|
|
|
=item * |
897
|
|
|
|
|
|
|
|
898
|
|
|
|
|
|
|
More functionalities : add NEAR operator and boost factors. |
899
|
|
|
|
|
|
|
|
900
|
|
|
|
|
|
|
=back |
901
|
|
|
|
|
|
|
|
902
|
|
|
|
|
|
|
|
903
|
|
|
|
|
|
|
|
904
|
|
|
|
|
|
|
=head1 SEE ALSO |
905
|
|
|
|
|
|
|
|
906
|
|
|
|
|
|
|
L is nice and compact, but |
907
|
|
|
|
|
|
|
limited in functionality (no +/- prefixes, no "exact phrase" search, |
908
|
|
|
|
|
|
|
no parentheses). |
909
|
|
|
|
|
|
|
|
910
|
|
|
|
|
|
|
L is a Perl port of the Java I search engine. |
911
|
|
|
|
|
|
|
Plucene has probably every feature you will ever need, but requires |
912
|
|
|
|
|
|
|
quite an investment to install and learn (more than 60 classes, |
913
|
|
|
|
|
|
|
dependencies on lots of external modules). |
914
|
|
|
|
|
|
|
I haven't done any benchmarks yet to compare performance. |
915
|
|
|
|
|
|
|
|
916
|
|
|
|
|
|
|
L is a more recent, more sophisticated search engine, |
917
|
|
|
|
|
|
|
which looks very powerful and should be probably faster and definitely |
918
|
|
|
|
|
|
|
more scalable than C; but also with a less compact |
919
|
|
|
|
|
|
|
API. I haven't performed any detailed comparison yet. |
920
|
|
|
|
|
|
|
|
921
|
|
|
|
|
|
|
|
922
|
|
|
|
|
|
|
=cut |
923
|
|
|
|
|
|
|
|
924
|
|
|
|
|
|
|
|
925
|
|
|
|
|
|
|
1; |