Commit | Line | Data |
fb451ba6 |
1 | package DBM::Deep::Engine2; |
2 | |
3 | use base 'DBM::Deep::Engine'; |
4 | |
5 | use 5.6.0; |
6 | |
7 | use strict; |
8 | use warnings; |
9 | |
10 | our $VERSION = q(0.99_03); |
11 | |
12 | use Fcntl qw( :DEFAULT :flock ); |
13 | use Scalar::Util (); |
14 | |
15 | # File-wide notes: |
16 | # * Every method in here assumes that the _storage has been appropriately |
17 | # safeguarded. This can be anything from flock() to some sort of manual |
18 | # mutex. But, it's the caller's responsability to make sure that this has |
19 | # been done. |
20 | |
21 | # Setup file and tag signatures. These should never change. |
22 | sub SIG_FILE () { 'DPDB' } |
23 | sub SIG_HEADER () { 'h' } |
24 | sub SIG_INTERNAL () { 'i' } |
25 | sub SIG_HASH () { 'H' } |
26 | sub SIG_ARRAY () { 'A' } |
27 | sub SIG_NULL () { 'N' } |
28 | sub SIG_DATA () { 'D' } |
29 | sub SIG_INDEX () { 'I' } |
30 | sub SIG_BLIST () { 'B' } |
31 | sub SIG_FREE () { 'F' } |
32 | sub SIG_KEYS () { 'K' } |
33 | sub SIG_SIZE () { 1 } |
34 | |
35 | # This is the transaction ID for the HEAD |
36 | sub HEAD () { 0 } |
37 | |
38 | sub read_value { |
39 | my $self = shift; |
40 | my ($trans_id, $base_offset, $key) = @_; |
41 | |
f5677eab |
42 | # print "Trying to read $key from $base_offset ($trans_id)\n" if $key > 400; |
fb451ba6 |
43 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
44 | offset => $base_offset, |
45 | trans_id => $trans_id, |
46 | allow_head => 1, |
47 | }); |
48 | die "Attempt to use a deleted value" if $_is_del; |
49 | die "Internal error!" if !$_val_offset; |
50 | |
f5677eab |
51 | my ($key_tag) = $self->_find_key_offset({ |
fb451ba6 |
52 | offset => $_val_offset, |
53 | key_md5 => $self->_apply_digest( $key ), |
54 | }); |
f5677eab |
55 | return if !$key_tag; |
fb451ba6 |
56 | |
57 | my ($val_offset, $is_del) = $self->_find_value_offset({ |
f5677eab |
58 | offset => $key_tag->{start}, |
fb451ba6 |
59 | trans_id => $trans_id, |
60 | allow_head => 1, |
61 | }); |
62 | return if $is_del; |
63 | die "Internal error!" if !$val_offset; |
64 | |
65 | return $self->_read_value({ |
f5677eab |
66 | keyloc => $key_tag->{start}, |
fb451ba6 |
67 | offset => $val_offset, |
68 | }); |
69 | } |
70 | |
71 | sub key_exists { |
72 | my $self = shift; |
73 | my ($trans_id, $base_offset, $key) = @_; |
74 | |
75 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
76 | offset => $base_offset, |
77 | trans_id => $trans_id, |
78 | allow_head => 1, |
79 | }); |
80 | die "Attempt to use a deleted value" if $_is_del; |
81 | die "Internal error!" if !$_val_offset; |
82 | |
f5677eab |
83 | my ($key_tag) = $self->_find_key_offset({ |
fb451ba6 |
84 | offset => $_val_offset, |
85 | key_md5 => $self->_apply_digest( $key ), |
86 | }); |
f5677eab |
87 | return '' if !$key_tag->{start}; |
fb451ba6 |
88 | |
89 | my ($val_offset, $is_del) = $self->_find_value_offset({ |
f5677eab |
90 | offset => $key_tag->{start}, |
fb451ba6 |
91 | trans_id => $trans_id, |
92 | allow_head => 1, |
93 | }); |
f5677eab |
94 | die "Internal error!" if !$_val_offset; |
fb451ba6 |
95 | |
129ea236 |
96 | return '' if $is_del; |
fb451ba6 |
97 | |
129ea236 |
98 | return 1; |
fb451ba6 |
99 | } |
100 | |
101 | sub get_next_key { |
102 | my $self = shift; |
129ea236 |
103 | my ($trans_id, $base_offset) = @_; |
104 | |
105 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
106 | offset => $base_offset, |
107 | trans_id => $trans_id, |
108 | allow_head => 1, |
109 | }); |
110 | die "Attempt to use a deleted value" if $_is_del; |
111 | die "Internal error!" if !$_val_offset; |
fb451ba6 |
112 | |
113 | # If the previous key was not specifed, start at the top and |
114 | # return the first one found. |
115 | my $temp; |
129ea236 |
116 | if ( @_ > 2 ) { |
fb451ba6 |
117 | $temp = { |
129ea236 |
118 | prev_md5 => $self->_apply_digest($_[2]), |
fb451ba6 |
119 | return_next => 0, |
120 | }; |
121 | } |
122 | else { |
123 | $temp = { |
124 | prev_md5 => chr(0) x $self->{hash_size}, |
125 | return_next => 1, |
126 | }; |
127 | } |
128 | |
129ea236 |
129 | return $self->traverse_index( $temp, $_val_offset, 0 ); |
fb451ba6 |
130 | } |
131 | |
132 | sub delete_key { |
133 | my $self = shift; |
134 | my ($trans_id, $base_offset, $key) = @_; |
135 | |
136 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
137 | offset => $base_offset, |
138 | trans_id => $trans_id, |
139 | allow_head => 1, |
140 | }); |
141 | die "Attempt to use a deleted value" if $_is_del; |
142 | die "Internal error!" if !$_val_offset; |
143 | |
f5677eab |
144 | my ($key_tag, $bucket_tag) = $self->_find_key_offset({ |
fb451ba6 |
145 | offset => $_val_offset, |
146 | key_md5 => $self->_apply_digest( $key ), |
147 | }); |
f5677eab |
148 | return if !$key_tag->{start}; |
fb451ba6 |
149 | |
129ea236 |
150 | my $value = $self->read_value( $trans_id, $base_offset, $key ); |
fb451ba6 |
151 | if ( $trans_id ) { |
152 | $self->_mark_as_deleted({ |
153 | tag => $key_tag, |
154 | trans_id => $trans_id, |
155 | }); |
156 | } |
157 | else { |
fb451ba6 |
158 | if ( my @transactions = $self->_storage->current_transactions ) { |
159 | foreach my $other_trans_id ( @transactions ) { |
160 | next if $self->_has_keyloc_entry({ |
161 | tag => $key_tag, |
162 | trans_id => $other_trans_id, |
163 | }); |
164 | $self->write_value( $other_trans_id, $base_offset, $key, $value ); |
165 | } |
166 | } |
129ea236 |
167 | |
168 | $self->_mark_as_deleted({ |
169 | tag => $key_tag, |
170 | trans_id => $trans_id, |
171 | }); |
172 | # $self->_remove_key_offset({ |
173 | # offset => $_val_offset, |
174 | # key_md5 => $self->_apply_digest( $key ), |
175 | # }); |
fb451ba6 |
176 | } |
177 | |
129ea236 |
178 | return $value; |
fb451ba6 |
179 | } |
180 | |
181 | sub write_value { |
182 | my $self = shift; |
183 | my ($trans_id, $base_offset, $key, $value) = @_; |
184 | |
185 | # This verifies that only supported values will be stored. |
186 | { |
187 | my $r = Scalar::Util::reftype( $value ); |
188 | |
189 | last if !defined $r; |
190 | last if $r eq 'HASH'; |
191 | last if $r eq 'ARRAY'; |
192 | |
193 | $self->_throw_error( |
194 | "Storage of references of type '$r' is not supported." |
195 | ); |
196 | } |
197 | |
198 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
199 | offset => $base_offset, |
200 | trans_id => $trans_id, |
201 | allow_head => 1, |
202 | }); |
203 | die "Attempt to use a deleted value" if $_is_del; |
204 | die "Internal error!" if !$_val_offset; |
205 | |
f5677eab |
206 | my ($key_tag, $bucket_tag) = $self->_find_key_offset({ |
fb451ba6 |
207 | offset => $_val_offset, |
208 | key_md5 => $self->_apply_digest( $key ), |
209 | create => 1, |
210 | }); |
f5677eab |
211 | die "Cannot find/create new key offset!" if !$key_tag->{start}; |
fb451ba6 |
212 | |
213 | if ( $trans_id ) { |
f5677eab |
214 | if ( $key_tag->{is_new} ) { |
fb451ba6 |
215 | # Must mark the HEAD as deleted because it doesn't exist |
216 | $self->_mark_as_deleted({ |
217 | tag => $key_tag, |
218 | trans_id => HEAD, |
219 | }); |
220 | } |
221 | } |
222 | else { |
223 | # If the HEAD isn't new, then we must take other transactions |
224 | # into account. If it is, then there can be no other transactions. |
f5677eab |
225 | if ( !$key_tag->{is_new} ) { |
fb451ba6 |
226 | my $old_value = $self->read_value( $trans_id, $base_offset, $key ); |
227 | if ( my @transactions = $self->_storage->current_transactions ) { |
228 | foreach my $other_trans_id ( @transactions ) { |
229 | next if $self->_has_keyloc_entry({ |
230 | tag => $key_tag, |
231 | trans_id => $other_trans_id, |
232 | }); |
233 | $self->write_value( $other_trans_id, $base_offset, $key, $old_value ); |
234 | } |
235 | } |
236 | } |
237 | } |
238 | |
40956c06 |
239 | my $value_loc = $self->_storage->request_space( |
240 | $self->_length_needed( $value, $key ), |
241 | ); |
242 | |
243 | $self->_add_key_offset({ |
244 | tag => $key_tag, |
245 | trans_id => $trans_id, |
246 | loc => $value_loc, |
fb451ba6 |
247 | }); |
248 | |
f5677eab |
249 | $self->_write_value( $key_tag->{start}, $value_loc, $key, $value, $key ); |
40956c06 |
250 | |
fb451ba6 |
251 | return 1; |
252 | } |
253 | |
254 | sub _find_value_offset { |
255 | my $self = shift; |
256 | my ($args) = @_; |
257 | |
258 | my $key_tag = $self->load_tag( $args->{offset} ); |
259 | |
260 | my @head; |
261 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
262 | my ($loc, $trans_id, $is_deleted) = unpack( |
263 | "$self->{long_pack} C C", |
264 | substr( $key_tag->{content}, $i * $self->{key_size}, $self->{key_size} ), |
265 | ); |
266 | |
267 | if ( $trans_id == HEAD ) { |
268 | @head = ($loc, $is_deleted); |
269 | } |
270 | |
271 | next if $loc && $args->{trans_id} != $trans_id; |
272 | return( $loc, $is_deleted ); |
273 | } |
274 | |
275 | return @head if $args->{allow_head}; |
276 | return; |
277 | } |
278 | |
fb451ba6 |
279 | sub _find_key_offset { |
280 | my $self = shift; |
281 | my ($args) = @_; |
282 | |
283 | my $bucket_tag = $self->load_tag( $args->{offset} ) |
284 | or $self->_throw_error( "INTERNAL ERROR - Cannot find tag" ); |
285 | |
fb451ba6 |
286 | #XXX What happens when $ch >= $self->{hash_size} ?? |
287 | for (my $ch = 0; $bucket_tag->{signature} ne SIG_BLIST; $ch++) { |
288 | my $num = ord substr($args->{key_md5}, $ch, 1); |
289 | |
290 | my $ref_loc = $bucket_tag->{offset} + ($num * $self->{long_size}); |
291 | $bucket_tag = $self->index_lookup( $bucket_tag, $num ); |
292 | |
293 | if (!$bucket_tag) { |
294 | return if !$args->{create}; |
295 | |
296 | my $loc = $self->_storage->request_space( |
297 | $self->tag_size( $self->{bucket_list_size} ), |
298 | ); |
299 | |
300 | $self->_storage->print_at( $ref_loc, pack($self->{long_pack}, $loc) ); |
301 | |
302 | $bucket_tag = $self->write_tag( |
303 | $loc, SIG_BLIST, |
304 | chr(0)x$self->{bucket_list_size}, |
305 | ); |
306 | |
307 | $bucket_tag->{ref_loc} = $ref_loc; |
308 | $bucket_tag->{ch} = $ch; |
309 | $bucket_tag->{is_new} = 1; |
310 | |
311 | last; |
312 | } |
313 | |
314 | $bucket_tag->{ch} = $ch; |
315 | $bucket_tag->{ref_loc} = $ref_loc; |
316 | } |
317 | |
318 | # Need to create a new keytag, too |
319 | if ( $bucket_tag->{is_new} ) { |
f5677eab |
320 | # print "Creating new keytag\n"; |
fb451ba6 |
321 | my $keytag_loc = $self->_storage->request_space( |
322 | $self->tag_size( $self->{keyloc_size} ), |
323 | ); |
324 | |
325 | substr( $bucket_tag->{content}, 0, $self->{key_size} ) = |
326 | $args->{key_md5} . pack( "$self->{long_pack}", $keytag_loc ); |
327 | |
328 | $self->_storage->print_at( $bucket_tag->{offset}, $bucket_tag->{content} ); |
329 | |
f5677eab |
330 | my $key_tag = $self->write_tag( |
fb451ba6 |
331 | $keytag_loc, SIG_KEYS, |
332 | chr(0)x$self->{keyloc_size}, |
333 | ); |
334 | |
f5677eab |
335 | return( $key_tag, $bucket_tag ); |
fb451ba6 |
336 | } |
337 | else { |
40956c06 |
338 | my ($key, $subloc, $index); |
fb451ba6 |
339 | BUCKET: |
340 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
40956c06 |
341 | ($key, $subloc) = $self->_get_key_subloc( |
fb451ba6 |
342 | $bucket_tag->{content}, $i, |
343 | ); |
344 | |
345 | next BUCKET if $subloc && $key ne $args->{key_md5}; |
40956c06 |
346 | |
347 | # Keep track of where we are, in case we need to create a new |
348 | # entry. |
349 | $index = $i; |
350 | last; |
fb451ba6 |
351 | } |
40956c06 |
352 | |
f5677eab |
353 | # If we have a subloc to return or we don't want to create a new |
354 | # entry, we need to return now. |
355 | $args->{create} ||= 0; |
356 | # print "Found ($subloc) at $index ($args->{create})\n"; |
357 | return ($self->load_tag( $subloc ), $bucket_tag) if $subloc || !$args->{create}; |
40956c06 |
358 | |
359 | my $keytag_loc = $self->_storage->request_space( |
360 | $self->tag_size( $self->{keyloc_size} ), |
361 | ); |
362 | |
363 | # There's space left in this bucket |
364 | if ( defined $index ) { |
f5677eab |
365 | # print "There's space left in the bucket for $keytag_loc\n"; |
40956c06 |
366 | substr( $bucket_tag->{content}, $index * $self->{key_size}, $self->{key_size} ) = |
367 | $args->{key_md5} . pack( "$self->{long_pack}", $keytag_loc ); |
368 | |
369 | $self->_storage->print_at( $bucket_tag->{offset}, $bucket_tag->{content} ); |
370 | } |
371 | # We need to split the index |
372 | else { |
f5677eab |
373 | # print "Splitting the index for $keytag_loc\n"; |
40956c06 |
374 | $self->split_index( $bucket_tag, $args->{key_md5}, $keytag_loc ); |
375 | } |
376 | |
f5677eab |
377 | my $key_tag = $self->write_tag( |
40956c06 |
378 | $keytag_loc, SIG_KEYS, |
379 | chr(0)x$self->{keyloc_size}, |
380 | ); |
381 | |
f5677eab |
382 | return( $key_tag, $bucket_tag ); |
fb451ba6 |
383 | } |
384 | |
385 | return; |
386 | } |
387 | |
388 | sub _read_value { |
389 | my $self = shift; |
390 | my ($args) = @_; |
391 | |
129ea236 |
392 | return $self->read_from_loc( $args->{keyloc}, $args->{offset} ); |
fb451ba6 |
393 | } |
394 | |
395 | sub _mark_as_deleted { |
396 | my $self = shift; |
397 | my ($args) = @_; |
398 | |
399 | my $is_changed; |
400 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
401 | my ($loc, $trans_id, $is_deleted) = unpack( |
402 | "$self->{long_pack} C C", |
403 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ), |
404 | ); |
405 | |
129ea236 |
406 | last unless $loc || $is_deleted; |
fb451ba6 |
407 | |
408 | if ( $trans_id == $args->{trans_id} ) { |
409 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ) = pack( |
410 | "$self->{long_pack} C C", |
411 | $loc, $trans_id, 1, |
40956c06 |
412 | ); |
129ea236 |
413 | $is_changed = 1; |
414 | last; |
fb451ba6 |
415 | } |
416 | } |
417 | |
418 | if ( $is_changed ) { |
419 | $self->_storage->print_at( |
420 | $args->{tag}{offset}, $args->{tag}{content}, |
421 | ); |
422 | } |
423 | |
424 | return 1; |
425 | } |
426 | |
427 | sub _has_keyloc_entry { |
428 | my $self = shift; |
429 | my ($args) = @_; |
430 | |
431 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
432 | my ($loc, $trans_id, $is_deleted) = unpack( |
433 | "$self->{long_pack} C C", |
434 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ), |
435 | ); |
436 | |
437 | return 1 if $trans_id == $args->{trans_id}; |
438 | } |
439 | |
440 | return; |
441 | } |
442 | |
443 | sub _remove_key_offset { |
444 | my $self = shift; |
445 | my ($args) = @_; |
446 | |
447 | my $is_changed; |
448 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
449 | my ($loc, $trans_id, $is_deleted) = unpack( |
450 | "$self->{long_pack} C C", |
451 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ), |
452 | ); |
453 | |
454 | if ( $trans_id == $args->{trans_id} ) { |
455 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ) = ''; |
456 | $args->{tag}{content} .= chr(0) x $self->{key_size}; |
457 | $is_changed = 1; |
458 | redo; |
459 | } |
460 | } |
461 | |
462 | if ( $is_changed ) { |
463 | $self->_storage->print_at( |
464 | $args->{tag}{offset}, $args->{tag}{content}, |
465 | ); |
466 | } |
467 | |
468 | return 1; |
469 | } |
470 | |
40956c06 |
471 | sub _add_key_offset { |
fb451ba6 |
472 | my $self = shift; |
473 | my ($args) = @_; |
474 | |
40956c06 |
475 | my $is_changed; |
476 | for ( my $i = 0; $i < $self->{max_buckets}; $i++ ) { |
477 | my ($loc, $trans_id, $is_deleted) = unpack( |
478 | "$self->{long_pack} C C", |
479 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ), |
480 | ); |
fb451ba6 |
481 | |
40956c06 |
482 | if ( $trans_id == $args->{trans_id} || (!$loc && !$is_deleted) ) { |
483 | substr( $args->{tag}{content}, $i * $self->{key_size}, $self->{key_size} ) = pack( |
484 | "$self->{long_pack} C C", |
485 | $args->{loc}, $args->{trans_id}, 0, |
486 | ); |
487 | $is_changed = 1; |
488 | last; |
489 | } |
490 | } |
491 | |
492 | if ( $is_changed ) { |
493 | $self->_storage->print_at( |
494 | $args->{tag}{offset}, $args->{tag}{content}, |
495 | ); |
496 | } |
497 | else { |
498 | die "Why didn't _add_key_offset() change something?!\n"; |
499 | } |
500 | |
501 | return 1; |
fb451ba6 |
502 | } |
503 | |
504 | sub setup_fh { |
505 | my $self = shift; |
506 | my ($obj) = @_; |
507 | |
508 | # Need to remove use of $fh here |
509 | my $fh = $self->_storage->{fh}; |
510 | flock $fh, LOCK_EX; |
511 | |
512 | #XXX The duplication of calculate_sizes needs to go away |
513 | unless ( $obj->{base_offset} ) { |
514 | my $bytes_read = $self->read_file_header; |
515 | |
516 | $self->calculate_sizes; |
517 | |
518 | ## |
519 | # File is empty -- write header and master index |
520 | ## |
521 | if (!$bytes_read) { |
522 | $self->_storage->audit( "# Database created on" ); |
523 | |
524 | $self->write_file_header; |
525 | |
526 | $obj->{base_offset} = $self->_storage->request_space( |
527 | $self->tag_size( $self->{keyloc_size} ), |
528 | ); |
f5677eab |
529 | warn "INITIAL BASE OFFSET: $obj->{base_offset}\n"; |
fb451ba6 |
530 | |
531 | my $value_spot = $self->_storage->request_space( |
532 | $self->tag_size( $self->{index_size} ), |
533 | ); |
534 | |
535 | $self->write_tag( |
536 | $obj->{base_offset}, SIG_KEYS, |
129ea236 |
537 | pack( "$self->{long_pack} C C", $value_spot, HEAD, 0 ), |
538 | chr(0) x ($self->{index_size} - $self->{key_size}), |
fb451ba6 |
539 | ); |
540 | |
541 | $self->write_tag( |
542 | $value_spot, $obj->_type, |
543 | chr(0)x$self->{index_size}, |
544 | ); |
545 | |
546 | # Flush the filehandle |
547 | my $old_fh = select $fh; |
548 | my $old_af = $|; $| = 1; $| = $old_af; |
549 | select $old_fh; |
550 | } |
551 | else { |
552 | $obj->{base_offset} = $bytes_read; |
f5677eab |
553 | warn "REOPEN BASE OFFSET: $obj->{base_offset}\n"; |
fb451ba6 |
554 | |
129ea236 |
555 | my ($_val_offset, $_is_del) = $self->_find_value_offset({ |
556 | offset => $obj->{base_offset}, |
557 | trans_id => HEAD, |
558 | allow_head => 1, |
559 | }); |
560 | die "Attempt to use a deleted value" if $_is_del; |
561 | die "Internal error!" if !$_val_offset; |
562 | |
fb451ba6 |
563 | ## |
564 | # Get our type from master index header |
565 | ## |
129ea236 |
566 | my $tag = $self->load_tag($_val_offset); |
fb451ba6 |
567 | unless ( $tag ) { |
568 | flock $fh, LOCK_UN; |
569 | $self->_throw_error("Corrupted file, no master index record"); |
570 | } |
571 | |
572 | unless ($obj->_type eq $tag->{signature}) { |
573 | flock $fh, LOCK_UN; |
574 | $self->_throw_error("File type mismatch"); |
575 | } |
576 | } |
577 | } |
578 | else { |
579 | $self->calculate_sizes; |
580 | } |
581 | |
582 | #XXX We have to make sure we don't mess up when autoflush isn't turned on |
583 | $self->_storage->set_inode; |
584 | |
585 | flock $fh, LOCK_UN; |
586 | |
587 | return 1; |
588 | } |
589 | |
590 | 1; |
591 | __END__ |