1 |
# -*- Mode: Cperl -*- |
2 |
# Table.pm -- |
3 |
# ITIID : $ITI$ $Header $__Header$ |
4 |
# Author : Ulrich Pfeifer |
5 |
# Created On : Thu Aug 8 13:05:10 1996 |
6 |
# Last Modified By: Ulrich Pfeifer |
7 |
# Last Modified On: Wed Jan 23 14:15:15 2002 |
8 |
# Language : CPerl |
9 |
# Update Count : 152 |
10 |
# Status : Unknown, Use with caution! |
11 |
# |
12 |
# Copyright (c) 1996-1997, Ulrich Pfeifer |
13 |
# |
14 |
|
15 |
=head1 NAME |
16 |
|
17 |
WAIT::Table -- Module for maintaining Tables / Relations |
18 |
|
19 |
=head1 SYNOPSIS |
20 |
|
21 |
require WAIT::Table; |
22 |
|
23 |
=head1 DESCRIPTION |
24 |
|
25 |
=cut |
26 |
|
27 |
package WAIT::Table; |
28 |
our $VERSION = "2.000"; |
29 |
|
30 |
use WAIT::Table::Handle (); |
31 |
require WAIT::Parse::Base; |
32 |
|
33 |
use strict; |
34 |
use Carp qw(cluck croak confess); |
35 |
# use autouse Carp => qw( croak($) ); |
36 |
use BerkeleyDB; |
37 |
use Fcntl; |
38 |
|
39 |
=head2 Creating a Table. |
40 |
|
41 |
The constructor WAIT::Table-E<gt>new is normally called via the |
42 |
create_table method of a database handle. This is not enforced, but |
43 |
creating a table does not make any sense unless the table is |
44 |
registered by the database because the latter implements persistence |
45 |
of the meta data. Registering is done automatically by letting the |
46 |
database handle the creation of a table. |
47 |
|
48 |
my $db = WAIT::Database->create(name => 'sample'); |
49 |
my $tb = $db->create_table(name => 'test', |
50 |
access => $access, |
51 |
layout => $layout, |
52 |
attr => ['docid', 'headline'], |
53 |
); |
54 |
|
55 |
The constructor returns a handle for the table. This handle is hidden by the |
56 |
table module, to prevent direct access if called via Table. |
57 |
|
58 |
=over 10 |
59 |
|
60 |
=item C<access> => I<accessobj> |
61 |
|
62 |
A reference to an access object for the external parts (attributes) of |
63 |
tuples. As you may remember, the WAIT System does not enforce that |
64 |
objects are completely stored inside the system to avoid duplication. |
65 |
There is no (strong) point in storing all your HTML documents inside |
66 |
the system when indexing your WWW-Server. |
67 |
|
68 |
The access object is designed to work like as a tied hash. You pass |
69 |
the refernce to the object, not the tied hash though. An example |
70 |
implementation of an access class that works for manpages is |
71 |
WAIT::Document::Nroff. |
72 |
|
73 |
The implementation needs to take into account that WAIT will keep this |
74 |
object in a Data::Dumper or Storable database and re-use it when sman |
75 |
is run. So it is not good enough if we can produce the index with it |
76 |
now, when we create or actively access the table, WAIT also must be |
77 |
able to retrieve documents on its own, when we are in a different |
78 |
context. This happens specifically in a retrieval. To get this working |
79 |
seemlessly, the access-defining class must implement a close method. |
80 |
This method will be called before the Data::Dumper dump takes place. |
81 |
In that moment the access-defining class must get rid of all data |
82 |
structures that cannot be reconstructed via the Data::Dumper dump, |
83 |
such as database handles or C pointers. |
84 |
|
85 |
=item C<path> => I<dir> |
86 |
|
87 |
The path to database. Files for indexes will have I<path> |
88 |
as prefix. I<Mandatory>, but usually taken care of by the |
89 |
WAIT::Database handle when the constructor is called via |
90 |
WAIT::Database::create_table(). |
91 |
|
92 |
=item C<name> => I<name> |
93 |
|
94 |
The name of this table. I<Mandatory> |
95 |
|
96 |
=item C<attr> => [ I<attr> ... ] |
97 |
|
98 |
A reference to an array of attribute names. WAIT will keep the |
99 |
contents of these attributes in its table. I<Mandatory> |
100 |
|
101 |
=item C<djk> => [ I<attr> ... ] |
102 |
|
103 |
A reference to an array of attribute names which make up the |
104 |
I<disjointness key>. Don't think about it - it's of no use yet; |
105 |
|
106 |
=item C<layout> => I<layoutobj> |
107 |
|
108 |
A reference to an external parser object. Defaults to a new instance |
109 |
of C<WAIT::Parse::Base>. For an example implementation see |
110 |
WAIT::Parse::Nroff. A layout class can be implemented as a singleton |
111 |
class if you so like. |
112 |
|
113 |
=item C<keyset> => I<keyset> |
114 |
|
115 |
The set of attributes needed to identify a record. Defaults to all |
116 |
attributes. |
117 |
|
118 |
=item C<invindex> => I<inverted index> |
119 |
|
120 |
A reference to an anon array defining attributes of each record that |
121 |
need to be indexed. See the source of smakewhatis for how to set this |
122 |
up. |
123 |
|
124 |
=back |
125 |
|
126 |
=cut |
127 |
|
128 |
sub new { |
129 |
my $type = shift; |
130 |
my %parm = @_; |
131 |
my $self = {}; |
132 |
|
133 |
# Check for mandatory attrs early |
134 |
for my $x (qw(name attr env maindbfile tablename)) { |
135 |
$self->{$x} = $parm{$x} or confess "No $x specified"; |
136 |
} |
137 |
|
138 |
# Do that before we eventually add '_weight' to attributes. |
139 |
$self->{keyset} = $parm{keyset} || [[@{$parm{attr}}]]; |
140 |
|
141 |
$self->{mode} = O_CREAT | O_RDWR; |
142 |
|
143 |
# Determine and set up subclass |
144 |
$type = ref($type) || $type; |
145 |
if (defined $parm{djk}) { |
146 |
if (@{$parm{djk}} == @{$parm{attr}}) { |
147 |
# All attributes in DK (sloppy test here!) |
148 |
$type .= '::Independent'; |
149 |
require WAIT::Table::Independent; |
150 |
} else { |
151 |
$type .= '::Disjoint'; |
152 |
require WAIT::Table::Disjoint; |
153 |
} |
154 |
# Add '_weight' to attributes |
155 |
my %attr; |
156 |
@attr{@{$parm{attr}}} = (1) x @{$parm{attr}}; |
157 |
unshift @{$parm{attr}}, '_weight' unless $attr{'_weight'}; |
158 |
} |
159 |
|
160 |
$self->{path} = $parm{path} or confess "No path specified"; |
161 |
bless $self, $type; |
162 |
|
163 |
$self->{djk} = $parm{djk} if defined $parm{djk}; |
164 |
$self->{layout} = $parm{layout} || new WAIT::Parse::Base; |
165 |
$self->{access} = $parm{access} if defined $parm{access}; |
166 |
$self->{nextk} = 1; # next record to insert; first record unused |
167 |
$self->{deleted} = {}; # no deleted records yet |
168 |
$self->{indexes} = {}; |
169 |
|
170 |
# Checking for readers is not necessary, but let's go with the |
171 |
# generic method. |
172 |
|
173 |
# Call create_index() and create_index() for compatibility |
174 |
for (@{$self->{keyset}||[]}) { |
175 |
#carp "Specification of indexes at table create time is deprecated"; |
176 |
$self->create_index(@$_); |
177 |
} |
178 |
while (@{$parm{invindex}||[]}) { |
179 |
# carp "Specification of inverted indexes at table create time is deprecated"; |
180 |
my $att = shift @{$parm{invindex}}; |
181 |
my @spec = @{shift @{$parm{invindex}}}; |
182 |
my @opt = (); |
183 |
|
184 |
if (ref($spec[0])) { |
185 |
warn "Secondary pipelines are deprecated"; |
186 |
@opt = %{shift @spec}; |
187 |
} |
188 |
$self->create_inverted_index(attribute => $att, |
189 |
pipeline => \@spec, |
190 |
@opt); |
191 |
} |
192 |
|
193 |
$self; |
194 |
# end of backwarn compatibility stuff |
195 |
} |
196 |
|
197 |
for my $accessor (qw(maindbfile tablename)) { |
198 |
no strict 'refs'; |
199 |
*{$accessor} = sub { |
200 |
my($self) = @_; |
201 |
return $self->{$accessor} if $self->{$accessor}; |
202 |
require Carp; |
203 |
Carp::confess("accessor $accessor not there"); |
204 |
} |
205 |
} |
206 |
|
207 |
=head2 Creating an index |
208 |
|
209 |
$tb->create_index('docid'); |
210 |
|
211 |
C<create_index> |
212 |
must be called with a list of attributes. This must be a subset of the |
213 |
attributes specified when the table was created. Currently this |
214 |
method must be called before the first tuple is inserted in the |
215 |
table! |
216 |
|
217 |
=cut |
218 |
|
219 |
sub create_index { |
220 |
my $self= shift; |
221 |
|
222 |
confess "Cannot create index for table aready populated" |
223 |
if $self->{nextk} > 1; |
224 |
|
225 |
require WAIT::Index; |
226 |
|
227 |
my $name = join '-', @_; |
228 |
#### warn "WARNING: Suspect use of \$_ in method create_index. name[$name]_[$_]"; |
229 |
$self->{indexes}->{$name} = |
230 |
WAIT::Index->new( |
231 |
path => $self->path.'/'.$name, |
232 |
subname => $name, |
233 |
env => $self->{env}, |
234 |
maindbfile => $self->maindbfile, |
235 |
tablename => $self->tablename, |
236 |
attr => $_, |
237 |
); |
238 |
} |
239 |
|
240 |
=head2 Creating an inverted index |
241 |
|
242 |
$tb->create_inverted_index |
243 |
(attribute => 'au', |
244 |
pipeline => ['detex', 'isotr', 'isolc', 'split2', 'stop'], |
245 |
predicate => 'plain', |
246 |
); |
247 |
|
248 |
=over 5 |
249 |
|
250 |
=item C<attribute> |
251 |
|
252 |
The attribute to build the index on. This attribute may not be in the |
253 |
set attributes specified when the table was created. |
254 |
|
255 |
=item C<pipeline> |
256 |
|
257 |
A piplines specification is a reference to an array of method names |
258 |
(from package C<WAIT::Filter>) which are to be applied in sequence to |
259 |
the contents of the named attribute. The attribute name may not be in |
260 |
the attribute list. |
261 |
|
262 |
=item C<predicate> |
263 |
|
264 |
An indication which predicate the index implements. This may be |
265 |
e.g. 'plain', 'stemming' or 'soundex'. The indicator will be used for |
266 |
query processing. Currently there is no standard set of predicate |
267 |
names. The predicate defaults to the last member of the pipeline if |
268 |
omitted. |
269 |
|
270 |
=back |
271 |
|
272 |
Currently this method must be called before the first tuple is |
273 |
inserted in the table! |
274 |
|
275 |
=cut |
276 |
|
277 |
sub create_inverted_index { |
278 |
my $self = shift; |
279 |
my %parm = @_; |
280 |
|
281 |
confess "No attribute specified" unless $parm{attribute}; |
282 |
confess "No pipeline specified" unless $parm{pipeline}; |
283 |
|
284 |
$parm{predicate} ||= $parm{pipeline}->[-1]; |
285 |
|
286 |
confess "Cannot create index for table aready populated" |
287 |
if $self->{nextk} > 1; |
288 |
|
289 |
require WAIT::InvertedIndex; |
290 |
|
291 |
# backward compatibility stuff |
292 |
my %opt = %parm; |
293 |
for (qw(attribute pipeline predicate)) { |
294 |
delete $opt{$_}; |
295 |
} |
296 |
|
297 |
my $name = join '_', ($parm{attribute}, @{$parm{pipeline}}); |
298 |
my $idx = WAIT::InvertedIndex->new(path => $self->path.'/'.$name, |
299 |
subname=> $name, |
300 |
env => $self->{env}, |
301 |
maindbfile => $self->maindbfile, |
302 |
tablename => $self->tablename, |
303 |
filter => [@{$parm{pipeline}}], # clone |
304 |
name => $name, |
305 |
attr => $parm{attribute}, |
306 |
%opt, # backward compatibility stuff |
307 |
); |
308 |
# We will have to use $parm{predicate} here |
309 |
push @{$self->{inverted}->{$parm{attribute}}}, $idx; |
310 |
} |
311 |
|
312 |
sub dir { |
313 |
$_[0]->path; |
314 |
} |
315 |
|
316 |
=head2 C<$tb-E<gt>layout> |
317 |
|
318 |
Returns the reference to the associated parser object. |
319 |
|
320 |
=cut |
321 |
|
322 |
sub layout { $_[0]->{layout} } |
323 |
|
324 |
=head2 C<$tb-E<gt>fields> |
325 |
|
326 |
Returns the array of attribute names. |
327 |
|
328 |
=cut |
329 |
|
330 |
|
331 |
sub fields { keys %{$_[0]->{inverted}}} |
332 |
|
333 |
=head2 C<$tb-E<gt>drop> |
334 |
|
335 |
Must be called via C<WAIT::Database::drop_table> |
336 |
|
337 |
=cut |
338 |
|
339 |
sub drop { |
340 |
my $self = shift; |
341 |
|
342 |
if ((caller)[0] eq 'WAIT::Database') { # database knows about this |
343 |
$self->close; # just make sure |
344 |
|
345 |
# my $path = $self->path; |
346 |
|
347 |
for (values %{$self->{indexes}}) { |
348 |
$_->drop; |
349 |
} |
350 |
# unlink "$path/records"; |
351 |
# rmdir "$path/read" or warn "Could not rmdir '$path/read'"; |
352 |
|
353 |
} else { |
354 |
confess ref($self)."::drop called directly"; |
355 |
} |
356 |
} |
357 |
|
358 |
sub mrequire ($) { |
359 |
my $module = shift; |
360 |
|
361 |
$module =~ s{::}{/}g; |
362 |
$module .= '.pm'; |
363 |
require $module; |
364 |
} |
365 |
|
366 |
sub path { |
367 |
my($self) = @_; |
368 |
return $self->{path} if $self->{path}; |
369 |
require Data::Dumper; print STDERR "Line " . __LINE__ . ", File: " . __FILE__ . "\n" . Data::Dumper->new([$self],[qw(self)])->Indent(1)->Useqq(1)->Dump; # XXX |
370 |
require Carp; |
371 |
confess("NO path attr"); |
372 |
} |
373 |
|
374 |
sub open { |
375 |
my $self = shift; |
376 |
my $path = $self->path . '/records'; |
377 |
|
378 |
mrequire ref($self); # that's tricky eh? |
379 |
if (defined $self->{'layout'}) { |
380 |
mrequire ref($self->{'layout'}); |
381 |
} |
382 |
if (defined $self->{'access'}) { |
383 |
mrequire ref($self->{'access'}); |
384 |
} |
385 |
if (exists $self->{indexes}) { |
386 |
require WAIT::Index; |
387 |
for my $Ind (values %{$self->{indexes}}) { |
388 |
for my $x (qw(mode env maindbfile)) { |
389 |
$Ind->{$x} = $self->{$x}; |
390 |
} |
391 |
} |
392 |
} |
393 |
if (exists $self->{inverted}) { |
394 |
my ($att, $idx); |
395 |
for $att (keys %{$self->{inverted}}) { |
396 |
for $idx (@{$self->{inverted}->{$att}}) { |
397 |
for my $x (qw(mode env maindbfile)) { |
398 |
$idx->{$x} = $self->{$x}; |
399 |
} |
400 |
} |
401 |
} |
402 |
require WAIT::InvertedIndex; |
403 |
} |
404 |
|
405 |
# CONFUSION: WAIT knows two *modes*: read-only or read-write. |
406 |
# BerkeleyDB means file permissions when talking about Mode. |
407 |
# BerkeleyDB has the "Flags" attribute to specify |
408 |
# read/write/lock/etc subsystems. |
409 |
|
410 |
my $flags; |
411 |
if ($self->{mode} & O_RDWR) { |
412 |
$flags = DB_CREATE; # | DB_INIT_MPOOL | DB_PRIVATE | DB_INIT_CDB; |
413 |
#warn "DEBUG: Flags on table $path set to 'writing'"; |
414 |
} else { |
415 |
$flags = DB_RDONLY; |
416 |
#warn "DEBUG: Flags on table $path set to 'readonly'"; |
417 |
} |
418 |
unless (defined $self->{dbh}) { |
419 |
my $subname = $self->tablename . "/records"; |
420 |
$self->{dbh} = |
421 |
tie(%{$self->{db}}, 'BerkeleyDB::Btree', |
422 |
$self->{env} ? (Env => $self->{env}) : (), |
423 |
# Filename => $file, |
424 |
Filename => $self->maindbfile, |
425 |
Subname => $subname, |
426 |
Mode => 0664, |
427 |
Flags => $flags, |
428 |
$WAIT::Database::Cachesize?(Cachesize => $WAIT::Database::Cachesize):(), |
429 |
$WAIT::Database::Pagesize?(Pagesize => $WAIT::Database::Pagesize):(), |
430 |
) |
431 |
or confess "Cannot tie: $BerkeleyDB::Error\nDEBUG: Filename[$self->{maindbfile}]subname[$subname]Mode[0664]Flags[$flags]"; |
432 |
} |
433 |
$self; |
434 |
} |
435 |
|
436 |
sub fetch_extern { |
437 |
my $self = shift; |
438 |
|
439 |
# print "#@_", $self->{'access'}->{Mode}, "\n"; # DEBUGGING? |
440 |
if (exists $self->{'access'}) { |
441 |
mrequire ref($self->{'access'}); |
442 |
$self->{'access'}->FETCH(@_); |
443 |
} |
444 |
} |
445 |
|
446 |
sub fetch_extern_by_id { |
447 |
my $self = shift; |
448 |
|
449 |
$self->fetch_extern($self->fetch(@_)); |
450 |
} |
451 |
|
452 |
sub _find_index { |
453 |
my $self = shift; |
454 |
my (@att) = @_; |
455 |
my %att; |
456 |
my $name; |
457 |
|
458 |
@att{@att} = @att; |
459 |
|
460 |
KEY: for $name (keys %{$self->{indexes}}) { |
461 |
my @iat = split /-/, $name; |
462 |
for (@iat) { |
463 |
next KEY unless exists $att{$_}; |
464 |
} |
465 |
return $self->{indexes}->{$name}; |
466 |
} |
467 |
return undef; |
468 |
} |
469 |
|
470 |
sub have { |
471 |
my $self = shift; |
472 |
my %parm = @_; |
473 |
|
474 |
my $index = $self->_find_index(keys %parm) or return; # no index-no have |
475 |
|
476 |
defined $self->{db} or $self->open; |
477 |
return $index->have(@_); |
478 |
} |
479 |
|
480 |
sub insert { |
481 |
my $self = shift; |
482 |
my %parm = @_; |
483 |
|
484 |
defined $self->{db} or $self->open; |
485 |
|
486 |
# We should move all writing methods to a subclass to check only once |
487 |
$self->{mode} & O_RDWR or confess "Cannot insert into table opened in RD_ONLY mode"; |
488 |
|
489 |
my $tuple = join($;, map($parm{$_} || '', @{$self->{attr}})); |
490 |
my $key; |
491 |
my @deleted = keys %{$self->{deleted}}; |
492 |
my $gotkey = 0; |
493 |
|
494 |
if (@deleted) { |
495 |
$key = pop @deleted; |
496 |
delete $self->{deleted}->{$key}; |
497 |
# Sanity check |
498 |
if ($key && $key>0) { |
499 |
$gotkey=1; |
500 |
} else { |
501 |
warn(sprintf("WAIT database inconsistency during insert ". |
502 |
"key[%s]: Please rebuild index\n", |
503 |
$key |
504 |
)); |
505 |
} |
506 |
} |
507 |
unless ($gotkey) { |
508 |
$key = $self->{nextk}++; |
509 |
} |
510 |
$self->{db}->{$key} = $tuple; |
511 |
for (values %{$self->{indexes}}) { |
512 |
unless ($_->insert($key, %parm)) { |
513 |
# duplicate key, undo changes |
514 |
if ($key == $self->{nextk}-1) { |
515 |
$self->{nextk}--; |
516 |
} else { |
517 |
# warn "setting key[$key] deleted during insert"; |
518 |
$self->{deleted}->{$key}=1; |
519 |
} |
520 |
my $idx; |
521 |
for $idx (values %{$self->{indexes}}) { |
522 |
last if $idx eq $_; |
523 |
$idx->remove($key, %parm); |
524 |
} |
525 |
return undef; |
526 |
} |
527 |
} |
528 |
if (defined $self->{inverted}) { |
529 |
my $att; |
530 |
for $att (keys %{$self->{inverted}}) { |
531 |
if (defined $parm{$att}) { |
532 |
map $_->insert($key, $parm{$att}), @{$self->{inverted}->{$att}}; |
533 |
#map $_->sync, @{$self->{inverted}->{$att}} |
534 |
} |
535 |
} |
536 |
} |
537 |
$key |
538 |
} |
539 |
|
540 |
sub sync { |
541 |
my $self = shift; |
542 |
|
543 |
for (values %{$self->{indexes}}) { |
544 |
map $_->sync, $_; |
545 |
} |
546 |
if (defined $self->{inverted}) { |
547 |
my $att; |
548 |
for $att (keys %{$self->{inverted}}) { |
549 |
map $_->sync, @{$self->{inverted}->{$att}} |
550 |
} |
551 |
} |
552 |
} |
553 |
|
554 |
sub fetch { |
555 |
my $self = shift; |
556 |
my $key = shift; |
557 |
|
558 |
return () if exists $self->{deleted}->{$key}; |
559 |
|
560 |
defined $self->{db} or $self->open; |
561 |
$self->unpack($self->{db}->{$key}); |
562 |
} |
563 |
|
564 |
sub delete_by_key { |
565 |
my $self = shift; |
566 |
my $key = shift; |
567 |
|
568 |
unless ($key) { |
569 |
cluck "Warning: delete_by_key called without key. Looks like a bug in WAIT?"; |
570 |
return; |
571 |
} |
572 |
|
573 |
return $self->{deleted}->{$key} if defined $self->{deleted}->{$key}; |
574 |
my %tuple = $self->fetch($key); |
575 |
for (values %{$self->{indexes}}) { |
576 |
$_->delete($key, %tuple); |
577 |
} |
578 |
if (defined $self->{inverted}) { |
579 |
# User *must* provide the full record for this or the entries |
580 |
# in the inverted index will not be removed |
581 |
%tuple = (%tuple, @_); |
582 |
my $att; |
583 |
for $att (keys %{$self->{inverted}}) { |
584 |
if (defined $tuple{$att}) { |
585 |
map $_->delete($key, $tuple{$att}), @{$self->{inverted}->{$att}} |
586 |
} |
587 |
} |
588 |
} |
589 |
# warn "setting key[$key] deleted during delete_by_key"; |
590 |
++$self->{deleted}->{$key}; |
591 |
} |
592 |
|
593 |
sub delete { |
594 |
my $self = shift; |
595 |
my $tkey = $self->have(@_); |
596 |
# warn "tkey[$tkey]\@_[@_]"; |
597 |
defined $tkey && $self->delete_by_key($tkey, @_); |
598 |
} |
599 |
|
600 |
sub unpack { |
601 |
my($self, $tuple) = @_; |
602 |
|
603 |
unless (defined $tuple){ |
604 |
# require Carp; # unfortunately gives us "bizarre copy...." :-((((( |
605 |
warn("Debug: somebody called unpack without argument tuple!"); |
606 |
return; |
607 |
} |
608 |
|
609 |
my $att; |
610 |
my @result; |
611 |
my @tuple = split /$;/, $tuple; |
612 |
|
613 |
for $att (@{$self->{attr}}) { |
614 |
push @result, $att, shift @tuple; |
615 |
} |
616 |
@result; |
617 |
} |
618 |
|
619 |
sub set { |
620 |
my ($self, $iattr, $value) = @_; |
621 |
# in the rare case that they haven't written a single record yet, we |
622 |
# make sure, the inverted inherits our $self->{mode}: |
623 |
defined $self->{db} or $self->open; |
624 |
|
625 |
for my $att (keys %{$self->{inverted}}) { |
626 |
if ($] > 5.003) { # avoid bug in perl up to 5.003_05 |
627 |
my $idx; |
628 |
for $idx (@{$self->{inverted}->{$att}}) { |
629 |
$idx->set($iattr, $value); |
630 |
} |
631 |
} else { |
632 |
map $_->set($iattr, $value), @{$self->{inverted}->{$att}}; |
633 |
} |
634 |
} |
635 |
|
636 |
1; |
637 |
} |
638 |
|
639 |
sub close { |
640 |
my $self = shift; |
641 |
|
642 |
#cluck("DEBUG: Closing A Table"); |
643 |
|
644 |
if (exists $self->{'access'}) { |
645 |
eval {$self->{'access'}->close}; # dont bother if not opened |
646 |
} |
647 |
if ($WAIT::Index::VERSION) { |
648 |
for (values %{$self->{indexes}}) { |
649 |
$_->close(); |
650 |
} |
651 |
} |
652 |
if (defined $self->{inverted} && $WAIT::InvertedIndex::VERSION) { |
653 |
# require WAIT::InvertedIndex; Uli: we can avoid closing indexes: |
654 |
# if WAIT::InvertedIndex has not been loaded, they cannot have |
655 |
# been altered so far |
656 |
my $att; |
657 |
for $att (keys %{$self->{inverted}}) { |
658 |
if ($] > 5.003) { # avoid bug in perl up to 5.003_05 |
659 |
my $idx; |
660 |
for $idx (@{$self->{inverted}->{$att}}) { |
661 |
$idx->close; |
662 |
} |
663 |
} else { |
664 |
map $_->close(), @{$self->{inverted}->{$att}}; |
665 |
} |
666 |
} |
667 |
} |
668 |
if ($self->{dbh}) { |
669 |
delete $self->{dbh}; |
670 |
} |
671 |
untie %{$self->{db}}; |
672 |
for my $att (qw(env db path maindbfile)) { |
673 |
delete $self->{$att}; |
674 |
#cluck "DEBUG: Deleted att $att"; |
675 |
} |
676 |
|
677 |
1; |
678 |
} |
679 |
|
680 |
sub DESTROY { |
681 |
my $self = shift; |
682 |
|
683 |
delete $self->{env}; |
684 |
|
685 |
# require Data::Dumper; print STDERR "Line " . __LINE__ . ", File: " . __FILE__ . "\n" . Data::Dumper->new([$self],[qw(self)])->Indent(1)->Useqq(1)->Dump; # XXX |
686 |
|
687 |
} |
688 |
|
689 |
sub open_scan { |
690 |
my $self = shift; |
691 |
my $code = shift; |
692 |
|
693 |
$self->{dbh} or $self->open; |
694 |
require WAIT::Scan; |
695 |
new WAIT::Scan $self, $self->{nextk}-1, $code; |
696 |
} |
697 |
|
698 |
sub open_index_scan { |
699 |
my $self = shift; |
700 |
my $attr = shift; |
701 |
my $code = shift; |
702 |
my $name = join '-', @$attr; |
703 |
|
704 |
if (defined $self->{indexes}->{$name}) { |
705 |
$self->{indexes}->{$name}->open_scan($code); |
706 |
} else { |
707 |
confess "No such index '$name'"; |
708 |
} |
709 |
} |
710 |
|
711 |
eval {sub WAIT::Query::Raw::new} unless defined \&WAIT::Query::Raw::new; |
712 |
|
713 |
sub prefix { |
714 |
my ($self , $attr, $prefix) = @_; |
715 |
my %result; |
716 |
|
717 |
defined $self->{db} or $self->open; # require layout |
718 |
|
719 |
for (@{$self->{inverted}->{$attr}}) { |
720 |
my $result = $_->prefix($prefix); |
721 |
if (defined $result) { |
722 |
$result{$_->name} = $result; |
723 |
} |
724 |
} |
725 |
bless \%result, 'WAIT::Query::Raw'; |
726 |
} |
727 |
|
728 |
sub intervall { |
729 |
my ($self, $attr, $lb, $ub) = @_; |
730 |
my %result; |
731 |
|
732 |
defined $self->{db} or $self->open; # require layout |
733 |
|
734 |
for (@{$self->{inverted}->{$attr}}) { |
735 |
my $result = $_->intervall($lb, $ub); |
736 |
if (defined $result) { |
737 |
$result{$_->name} = $result; |
738 |
} |
739 |
} |
740 |
bless \%result, 'WAIT::Query::Raw'; |
741 |
} |
742 |
|
743 |
sub search_ref { |
744 |
my $self = shift; |
745 |
my ($query, $attr, $cont, $raw); |
746 |
if (ref $_[0]) { |
747 |
$query = shift; |
748 |
# require Data::Dumper; print STDERR "Line " . __LINE__ . ", File: " . __FILE__ . "\n" . Data::Dumper->new([$query],[qw()])->Indent(1)->Useqq(1)->Dump; # XXX |
749 |
|
750 |
$attr = $query->{attr}; |
751 |
$cont = $query->{cont}; |
752 |
$raw = $query->{raw}; |
753 |
} else { |
754 |
cluck("Using three argument search interface is deprecated, use hashref interface instead"); |
755 |
$attr = shift; |
756 |
$cont = shift; |
757 |
$raw = shift; |
758 |
$query = { |
759 |
attr => $attr, |
760 |
cont => $cont, |
761 |
raw => $raw, |
762 |
}; |
763 |
} |
764 |
|
765 |
my %result; |
766 |
|
767 |
defined $self->{db} or $self->open; # require layout |
768 |
|
769 |
if ($raw) { |
770 |
for (@{$self->{inverted}->{$attr}}) { |
771 |
my $name = $_->name; |
772 |
if (exists $raw->{$name} and @{$raw->{$name}}) { |
773 |
my $scale = 1/scalar(@{$raw->{$name}}); |
774 |
my %r = $_->search_raw($query, @{$raw->{$name}}); |
775 |
my ($key, $val); |
776 |
while (($key, $val) = each %r) { |
777 |
if (exists $result{$key}) { |
778 |
$result{$key} += $val*$scale; |
779 |
} else { |
780 |
$result{$key} = $val*$scale; |
781 |
} |
782 |
} |
783 |
} |
784 |
} |
785 |
} |
786 |
if (defined $cont and $cont ne '') { |
787 |
for (@{$self->{inverted}->{$attr}}) { |
788 |
my $r = $_->search_ref($query, $cont); |
789 |
my ($key, $val); |
790 |
while (($key, $val) = each %$r) { |
791 |
if (exists $result{$key}) { |
792 |
$result{$key} += $val; |
793 |
} else { |
794 |
$result{$key} = $val; |
795 |
} |
796 |
} |
797 |
} |
798 |
} |
799 |
# sanity check for deleted documents. |
800 |
# this should not be necessary !@#$ |
801 |
for (keys %result) { |
802 |
delete $result{$_} if $self->{deleted}->{$_} |
803 |
} |
804 |
\%result; |
805 |
} |
806 |
|
807 |
sub parse_query { |
808 |
my($self, $attr, $query) = @_; |
809 |
return unless defined $query && length $query; |
810 |
my %qt; |
811 |
for (@{$self->{inverted}->{$attr}}) { |
812 |
grep $qt{$_}++, $_->parse($query); |
813 |
} |
814 |
[keys %qt]; |
815 |
} |
816 |
|
817 |
sub hilight_positions { |
818 |
my ($self, $attr, $text, $query, $raw) = @_; |
819 |
my %pos; |
820 |
|
821 |
if (defined $raw) { |
822 |
for (@{$self->{inverted}->{$attr}}) { # objects of type |
823 |
# WAIT::InvertedIndex for |
824 |
# this index field $attr |
825 |
my $name = $_->name; |
826 |
if (exists $raw->{$name}) { |
827 |
my %qt; |
828 |
grep $qt{$_}++, @{$raw->{$name}}; |
829 |
for ($_->parse_pos($text)) { |
830 |
if (exists $qt{$_->[0]}) { |
831 |
$pos{$_->[1]} = max($pos{$_->[1]}, length($_->[0])); |
832 |
} |
833 |
} |
834 |
} |
835 |
} |
836 |
} |
837 |
if (defined $query) { |
838 |
for (@{$self->{inverted}->{$attr}}) { |
839 |
my %qt; |
840 |
|
841 |
grep $qt{$_}++, $_->parse($query); |
842 |
for ($_->parse_pos($text)) { |
843 |
if (exists $qt{$_->[0]}) { |
844 |
if (exists $pos{$_->[1]}) { # perl -w ;-) |
845 |
$pos{$_->[1]} = max($pos{$_->[1]}, length($_->[0])); |
846 |
} else { |
847 |
$pos{$_->[1]} = length($_->[0]); |
848 |
} |
849 |
} |
850 |
} |
851 |
} |
852 |
} |
853 |
|
854 |
\%pos; |
855 |
} |
856 |
|
857 |
sub hilight { |
858 |
my ($tb, $buf, $qplain, $qraw) = @_; |
859 |
my $layout = $tb->layout(); |
860 |
|
861 |
my @result; |
862 |
|
863 |
$qplain ||= {}; |
864 |
$qraw ||= {}; |
865 |
my @ttxt = $layout->tag($buf); |
866 |
while (@ttxt) { |
867 |
no strict 'refs'; |
868 |
my %tag = %{shift @ttxt}; |
869 |
my $txt = shift @ttxt; |
870 |
my $fld; |
871 |
|
872 |
my %hl; |
873 |
for $fld (grep defined $tag{$_}, keys %$qplain, keys %$qraw) { |
874 |
my $hp = $tb->hilight_positions($fld, $txt, |
875 |
$qplain->{$fld}, $qraw->{$fld}); |
876 |
for (keys %$hp) { |
877 |
if (exists $hl{$_}) { # -w ;-( |
878 |
$hl{$_} = max($hl{$_}, $hp->{$_}); |
879 |
} else { |
880 |
$hl{$_} = $hp->{$_}; |
881 |
} |
882 |
} |
883 |
} |
884 |
my $pos; |
885 |
my $qt = {_qt => 1, %tag}; |
886 |
my $pl = \%tag; |
887 |
my $last = length($txt); |
888 |
my @tmp; |
889 |
for $pos (sort {$b <=> $a} keys %hl) { |
890 |
unshift @tmp, $pl, substr($txt,$pos+$hl{$pos},$last-$pos-$hl{$pos}); |
891 |
unshift @tmp, $qt, substr($txt,$pos,$hl{$pos}); |
892 |
$last = $pos; |
893 |
} |
894 |
push @result, $pl, substr($txt,0,$last); |
895 |
push @result, @tmp; |
896 |
} |
897 |
@result; # no speed necessary |
898 |
} |
899 |
|
900 |
1; |