Commit | Line | Data |
c443438f |
1 | package #hide from PAUSE |
2 | DBIx::Class::Storage::DBIHacks; |
d28bb90d |
3 | |
4 | # |
5 | # This module contains code that should never have seen the light of day, |
6 | # does not belong in the Storage, or is otherwise unfit for public |
6a6394f1 |
7 | # display. The arrival of SQLA2 should immediately obsolete 90% of this |
d28bb90d |
8 | # |
9 | |
10 | use strict; |
11 | use warnings; |
12 | |
13 | use base 'DBIx::Class::Storage'; |
14 | use mro 'c3'; |
15 | |
6298a324 |
16 | use List::Util 'first'; |
17 | use Scalar::Util 'blessed'; |
ea5c7509 |
18 | use Sub::Name 'subname'; |
6298a324 |
19 | use namespace::clean; |
d28bb90d |
20 | |
21 | # |
052e8431 |
22 | # This code will remove non-selecting/non-restricting joins from |
4b1b5ea3 |
23 | # {from} specs, aiding the RDBMS query optimizer |
052e8431 |
24 | # |
25 | sub _prune_unused_joins { |
ea95892e |
26 | my $self = shift; |
4b1b5ea3 |
27 | my ($from, $select, $where, $attrs) = @_; |
052e8431 |
28 | |
ea95892e |
29 | return $from unless $self->_use_join_optimizer; |
30 | |
052e8431 |
31 | if (ref $from ne 'ARRAY' || ref $from->[0] ne 'HASH' || ref $from->[1] ne 'ARRAY') { |
32 | return $from; # only standard {from} specs are supported |
33 | } |
34 | |
4b1b5ea3 |
35 | my $aliastypes = $self->_resolve_aliastypes_from_select_args(@_); |
36 | |
37 | # a grouped set will not be affected by amount of rows. Thus any |
38 | # {multiplying} joins can go |
39 | delete $aliastypes->{multiplying} if $attrs->{group_by}; |
40 | |
052e8431 |
41 | my @newfrom = $from->[0]; # FROM head is always present |
42 | |
a4812caa |
43 | my %need_joins; |
44 | for (values %$aliastypes) { |
45 | # add all requested aliases |
46 | $need_joins{$_} = 1 for keys %$_; |
47 | |
48 | # add all their parents (as per joinpath which is an AoH { table => alias }) |
49 | $need_joins{$_} = 1 for map { values %$_ } map { @$_ } values %$_; |
50 | } |
052e8431 |
51 | for my $j (@{$from}[1..$#$from]) { |
539ffe87 |
52 | push @newfrom, $j if ( |
4b1b5ea3 |
53 | (! $j->[0]{-alias}) # legacy crap |
539ffe87 |
54 | || |
55 | $need_joins{$j->[0]{-alias}} |
56 | ); |
052e8431 |
57 | } |
58 | |
59 | return \@newfrom; |
60 | } |
61 | |
052e8431 |
62 | # |
d28bb90d |
63 | # This is the code producing joined subqueries like: |
8273e845 |
64 | # SELECT me.*, other.* FROM ( SELECT me.* FROM ... ) JOIN other ON ... |
d28bb90d |
65 | # |
66 | sub _adjust_select_args_for_complex_prefetch { |
67 | my ($self, $from, $select, $where, $attrs) = @_; |
68 | |
69 | $self->throw_exception ('Nothing to prefetch... how did we get here?!') |
36fd7f07 |
70 | if not @{$attrs->{_prefetch_selector_range}}; |
d28bb90d |
71 | |
72 | $self->throw_exception ('Complex prefetches are not supported on resultsets with a custom from attribute') |
73 | if (ref $from ne 'ARRAY' || ref $from->[0] ne 'HASH' || ref $from->[1] ne 'ARRAY'); |
74 | |
d28bb90d |
75 | # generate inner/outer attribute lists, remove stuff that doesn't apply |
76 | my $outer_attrs = { %$attrs }; |
77 | delete $outer_attrs->{$_} for qw/where bind rows offset group_by having/; |
78 | |
186ba34c |
79 | my $inner_attrs = { %$attrs }; |
c9733800 |
80 | delete $inner_attrs->{$_} for qw/for collapse _prefetch_selector_range _collapse_order_by select as/; |
81 | |
82 | # bring over all non-collapse-induced order_by into the inner query (if any) |
83 | # the outer one will have to keep them all |
84 | delete $inner_attrs->{order_by}; |
85 | if (my $ord_cnt = @{$outer_attrs->{order_by}} - @{$outer_attrs->{_collapse_order_by}} ) { |
86 | $inner_attrs->{order_by} = [ |
87 | @{$outer_attrs->{order_by}}[ 0 .. $ord_cnt - 1] |
88 | ]; |
89 | } |
946f6260 |
90 | |
d28bb90d |
91 | # generate the inner/outer select lists |
92 | # for inside we consider only stuff *not* brought in by the prefetch |
93 | # on the outside we substitute any function for its alias |
94 | my $outer_select = [ @$select ]; |
95 | my $inner_select = []; |
36fd7f07 |
96 | |
27e0370d |
97 | my ($root_source, $root_source_offset); |
98 | |
99 | for my $i (0 .. $#$from) { |
100 | my $node = $from->[$i]; |
101 | my $h = (ref $node eq 'HASH') ? $node |
102 | : (ref $node eq 'ARRAY' and ref $node->[0] eq 'HASH') ? $node->[0] |
103 | : next |
104 | ; |
105 | |
106 | if ( ($h->{-alias}||'') eq $attrs->{alias} and $root_source = $h->{-rsrc} ) { |
107 | $root_source_offset = $i; |
108 | last; |
109 | } |
110 | } |
111 | |
112 | $self->throw_exception ('Complex prefetches are not supported on resultsets with a custom from attribute') |
113 | unless $root_source; |
114 | |
115 | # use the heavy duty resolver to take care of aliased/nonaliased naming |
116 | my $colinfo = $self->_resolve_column_info($from); |
117 | my $selected_root_columns; |
118 | |
36fd7f07 |
119 | my ($p_start, $p_end) = @{$outer_attrs->{_prefetch_selector_range}}; |
120 | for my $i (0 .. $p_start - 1, $p_end + 1 .. $#$outer_select) { |
d28bb90d |
121 | my $sel = $outer_select->[$i]; |
122 | |
123 | if (ref $sel eq 'HASH' ) { |
124 | $sel->{-as} ||= $attrs->{as}[$i]; |
125 | $outer_select->[$i] = join ('.', $attrs->{alias}, ($sel->{-as} || "inner_column_$i") ); |
126 | } |
27e0370d |
127 | elsif (! ref $sel and my $ci = $colinfo->{$sel}) { |
128 | $selected_root_columns->{$ci->{-colname}} = 1; |
129 | } |
d28bb90d |
130 | |
131 | push @$inner_select, $sel; |
bb9bffea |
132 | |
133 | push @{$inner_attrs->{as}}, $attrs->{as}[$i]; |
d28bb90d |
134 | } |
135 | |
27e0370d |
136 | # We will need to fetch all native columns in the inner subquery, which may be a part |
137 | # of an *outer* join condition. We can not just fetch everything because a potential |
138 | # has_many restricting join collapse *will not work* on heavy data types. |
139 | # Time for more horrible SQL parsing, aughhhh |
140 | |
141 | # MASSIVE FIXME - in fact when we are fully transitioned to DQ and the support is |
142 | # is sane - we will need to trim the select list to *only* fetch stuff that is |
143 | # necessary to build joins. In the current implementation if I am selecting a blob |
144 | # and the group_by kicks in - we are fucked, and all the user can do is not select |
145 | # that column. This is silly! |
146 | |
147 | my $retardo_sqla_cache = {}; |
148 | for my $cond ( map { $_->[1] } @{$from}[$root_source_offset + 1 .. $#$from] ) { |
149 | for my $col (@{$self->_extract_condition_columns($cond, $retardo_sqla_cache)}) { |
150 | my $ci = $colinfo->{$col}; |
151 | if ( |
152 | $ci |
153 | and |
154 | $ci->{-source_alias} eq $attrs->{alias} |
155 | and |
156 | ! $selected_root_columns->{$ci->{-colname}}++ |
157 | ) { |
158 | # adding it to both to keep limits not supporting dark selectors happy |
159 | push @$inner_select, $ci->{-fq_colname}; |
160 | push @{$inner_attrs->{as}}, $ci->{-fq_colname}; |
161 | } |
162 | } |
163 | } |
164 | |
ea95892e |
165 | # construct the inner $from and lock it in a subquery |
48580715 |
166 | # we need to prune first, because this will determine if we need a group_by below |
53c29913 |
167 | # the fake group_by is so that the pruner throws away all non-selecting, non-restricting |
168 | # multijoins (since we def. do not care about those inside the subquery) |
ea95892e |
169 | |
6395604e |
170 | my $inner_subq = do { |
ea95892e |
171 | |
172 | # must use it here regardless of user requests |
173 | local $self->{_use_join_optimizer} = 1; |
174 | |
175 | my $inner_from = $self->_prune_unused_joins ($from, $inner_select, $where, { |
176 | group_by => ['dummy'], %$inner_attrs, |
177 | }); |
178 | |
887a0aef |
179 | my $inner_aliastypes = |
180 | $self->_resolve_aliastypes_from_select_args( $inner_from, $inner_select, $where, $inner_attrs ); |
181 | |
a4812caa |
182 | # we need to simulate collapse in the subq if a multiplying join is pulled |
183 | # by being a non-selecting restrictor |
0a3441ee |
184 | if ( |
185 | ! $inner_attrs->{group_by} |
186 | and |
887a0aef |
187 | first { |
188 | $inner_aliastypes->{restricting}{$_} |
189 | and |
190 | ! $inner_aliastypes->{selecting}{$_} |
191 | } ( keys %{$inner_aliastypes->{multiplying}||{}} ) |
0a3441ee |
192 | ) { |
14e26c5f |
193 | my $unprocessed_order_chunks; |
194 | ($inner_attrs->{group_by}, $unprocessed_order_chunks) = $self->_group_over_selection ( |
0a3441ee |
195 | $inner_from, $inner_select, $inner_attrs->{order_by} |
196 | ); |
14e26c5f |
197 | |
198 | $self->throw_exception ( |
199 | 'A required group_by clause could not be constructed automatically due to a complex ' |
200 | . 'order_by criteria. Either order_by columns only (no functions) or construct a suitable ' |
201 | . 'group_by by hand' |
202 | ) if $unprocessed_order_chunks; |
0a3441ee |
203 | } |
d28bb90d |
204 | |
ea95892e |
205 | # we already optimized $inner_from above |
206 | local $self->{_use_join_optimizer} = 0; |
d28bb90d |
207 | |
ea95892e |
208 | # generate the subquery |
6395604e |
209 | $self->_select_args_to_query ( |
ea95892e |
210 | $inner_from, |
211 | $inner_select, |
212 | $where, |
213 | $inner_attrs, |
214 | ); |
d28bb90d |
215 | }; |
216 | |
217 | # Generate the outer from - this is relatively easy (really just replace |
218 | # the join slot with the subquery), with a major caveat - we can not |
219 | # join anything that is non-selecting (not part of the prefetch), but at |
220 | # the same time is a multi-type relationship, as it will explode the result. |
221 | # |
222 | # There are two possibilities here |
223 | # - either the join is non-restricting, in which case we simply throw it away |
224 | # - it is part of the restrictions, in which case we need to collapse the outer |
225 | # result by tackling yet another group_by to the outside of the query |
226 | |
27e0370d |
227 | # work on a shallow copy |
052e8431 |
228 | $from = [ @$from ]; |
052e8431 |
229 | |
d28bb90d |
230 | my @outer_from; |
53c29913 |
231 | |
27e0370d |
232 | # we may not be the head |
233 | if ($root_source_offset) { |
234 | # first generate the outer_from, up to the substitution point |
235 | @outer_from = splice @$from, 0, $root_source_offset; |
6395604e |
236 | |
27e0370d |
237 | my $root_node = shift @$from; |
238 | |
239 | push @outer_from, [ |
240 | { |
241 | -alias => $attrs->{alias}, |
242 | -rsrc => $root_node->[0]{-rsrc}, |
243 | $attrs->{alias} => $inner_subq, |
244 | }, |
245 | @{$root_node}[1 .. $#$root_node], |
246 | ]; |
247 | } |
248 | else { |
249 | my $root_node = shift @$from; |
250 | |
251 | @outer_from = { |
252 | -alias => $attrs->{alias}, |
253 | -rsrc => $root_node->{-rsrc}, |
254 | $attrs->{alias} => $inner_subq, |
255 | }; |
d28bb90d |
256 | } |
257 | |
ea95892e |
258 | # scan the *remaining* from spec against different attributes, and see which joins are needed |
052e8431 |
259 | # in what role |
260 | my $outer_aliastypes = |
539ffe87 |
261 | $self->_resolve_aliastypes_from_select_args( $from, $outer_select, $where, $outer_attrs ); |
052e8431 |
262 | |
a4812caa |
263 | # unroll parents |
264 | my ($outer_select_chain, $outer_restrict_chain) = map { +{ |
265 | map { $_ => 1 } map { values %$_} map { @$_ } values %{ $outer_aliastypes->{$_} || {} } |
266 | } } qw/selecting restricting/; |
267 | |
d28bb90d |
268 | # see what's left - throw away if not selecting/restricting |
a4812caa |
269 | # also throw in a group_by if a non-selecting multiplier, |
270 | # to guard against cross-join explosions |
36fd7f07 |
271 | my $need_outer_group_by; |
d28bb90d |
272 | while (my $j = shift @$from) { |
273 | my $alias = $j->[0]{-alias}; |
274 | |
a4812caa |
275 | if ( |
276 | $outer_select_chain->{$alias} |
277 | ) { |
278 | push @outer_from, $j |
d28bb90d |
279 | } |
a4812caa |
280 | elsif ($outer_restrict_chain->{$alias}) { |
d28bb90d |
281 | push @outer_from, $j; |
a4812caa |
282 | $need_outer_group_by ||= $outer_aliastypes->{multiplying}{$alias} ? 1 : 0; |
d28bb90d |
283 | } |
284 | } |
285 | |
36fd7f07 |
286 | if ($need_outer_group_by and ! $outer_attrs->{group_by}) { |
287 | |
288 | my $unprocessed_order_chunks; |
289 | ($outer_attrs->{group_by}, $unprocessed_order_chunks) = $self->_group_over_selection ( |
290 | \@outer_from, $outer_select, $outer_attrs->{order_by} |
291 | ); |
292 | |
293 | $self->throw_exception ( |
294 | 'A required group_by clause could not be constructed automatically due to a complex ' |
295 | . 'order_by criteria. Either order_by columns only (no functions) or construct a suitable ' |
296 | . 'group_by by hand' |
297 | ) if $unprocessed_order_chunks; |
298 | |
299 | } |
300 | |
d28bb90d |
301 | # This is totally horrific - the $where ends up in both the inner and outer query |
302 | # Unfortunately not much can be done until SQLA2 introspection arrives, and even |
303 | # then if where conditions apply to the *right* side of the prefetch, you may have |
304 | # to both filter the inner select (e.g. to apply a limit) and then have to re-filter |
305 | # the outer select to exclude joins you didin't want in the first place |
306 | # |
307 | # OTOH it can be seen as a plus: <ash> (notes that this query would make a DBA cry ;) |
308 | return (\@outer_from, $outer_select, $where, $outer_attrs); |
309 | } |
310 | |
1a736efb |
311 | # |
312 | # I KNOW THIS SUCKS! GET SQLA2 OUT THE DOOR SO THIS CAN DIE! |
313 | # |
ad630f4b |
314 | # Due to a lack of SQLA2 we fall back to crude scans of all the |
315 | # select/where/order/group attributes, in order to determine what |
316 | # aliases are neded to fulfill the query. This information is used |
317 | # throughout the code to prune unnecessary JOINs from the queries |
318 | # in an attempt to reduce the execution time. |
319 | # Although the method is pretty horrific, the worst thing that can |
1a736efb |
320 | # happen is for it to fail due to some scalar SQL, which in turn will |
321 | # result in a vocal exception. |
539ffe87 |
322 | sub _resolve_aliastypes_from_select_args { |
052e8431 |
323 | my ( $self, $from, $select, $where, $attrs ) = @_; |
546f1cd9 |
324 | |
ad630f4b |
325 | $self->throw_exception ('Unable to analyze custom {from}') |
326 | if ref $from ne 'ARRAY'; |
546f1cd9 |
327 | |
ad630f4b |
328 | # what we will return |
964a3c71 |
329 | my $aliases_by_type; |
546f1cd9 |
330 | |
ad630f4b |
331 | # see what aliases are there to work with |
332 | my $alias_list; |
539ffe87 |
333 | for (@$from) { |
334 | my $j = $_; |
ad630f4b |
335 | $j = $j->[0] if ref $j eq 'ARRAY'; |
539ffe87 |
336 | my $al = $j->{-alias} |
337 | or next; |
338 | |
339 | $alias_list->{$al} = $j; |
a4812caa |
340 | $aliases_by_type->{multiplying}{$al} ||= $j->{-join_path}||[] if ( |
341 | # not array == {from} head == can't be multiplying |
342 | ( ref($_) eq 'ARRAY' and ! $j->{-is_single} ) |
343 | or |
344 | # a parent of ours is already a multiplier |
345 | ( grep { $aliases_by_type->{multiplying}{$_} } @{ $j->{-join_path}||[] } ) |
346 | ); |
546f1cd9 |
347 | } |
546f1cd9 |
348 | |
1a736efb |
349 | # get a column to source/alias map (including unqualified ones) |
350 | my $colinfo = $self->_resolve_column_info ($from); |
351 | |
ad630f4b |
352 | # set up a botched SQLA |
353 | my $sql_maker = $self->sql_maker; |
07f31d19 |
354 | |
4c2b30d6 |
355 | # these are throw away results, do not pollute the bind stack |
4c2b30d6 |
356 | local $sql_maker->{select_bind}; |
0542ec57 |
357 | local $sql_maker->{where_bind}; |
358 | local $sql_maker->{group_bind}; |
359 | local $sql_maker->{having_bind}; |
3f5b99fe |
360 | |
361 | # we can't scan properly without any quoting (\b doesn't cut it |
362 | # everywhere), so unless there is proper quoting set - use our |
363 | # own weird impossible character. |
364 | # Also in the case of no quoting, we need to explicitly disable |
365 | # name_sep, otherwise sorry nasty legacy syntax like |
366 | # { 'count(foo.id)' => { '>' => 3 } } will stop working >:( |
367 | local $sql_maker->{quote_char} = $sql_maker->{quote_char}; |
368 | local $sql_maker->{name_sep} = $sql_maker->{name_sep}; |
369 | |
370 | unless (defined $sql_maker->{quote_char} and length $sql_maker->{quote_char}) { |
e493ecb2 |
371 | $sql_maker->{quote_char} = ["\x00", "\xFF"]; |
372 | # if we don't unset it we screw up retarded but unfortunately working |
373 | # 'MAX(foo.bar)' => { '>', 3 } |
3f5b99fe |
374 | $sql_maker->{name_sep} = ''; |
375 | } |
376 | |
377 | my ($lquote, $rquote, $sep) = map { quotemeta $_ } ($sql_maker->_quote_chars, $sql_maker->name_sep); |
07f31d19 |
378 | |
1a736efb |
379 | # generate sql chunks |
380 | my $to_scan = { |
381 | restricting => [ |
382 | $sql_maker->_recurse_where ($where), |
a7e643b1 |
383 | $sql_maker->_parse_rs_attrs ({ |
1a736efb |
384 | map { $_ => $attrs->{$_} } (qw/group_by having/) |
385 | }), |
386 | ], |
387 | selecting => [ |
1a736efb |
388 | $sql_maker->_recurse_fields ($select), |
bac358c9 |
389 | ( map { $_->[0] } $self->_extract_order_criteria ($attrs->{order_by}, $sql_maker) ), |
1a736efb |
390 | ], |
391 | }; |
07f31d19 |
392 | |
1a736efb |
393 | # throw away empty chunks |
394 | $_ = [ map { $_ || () } @$_ ] for values %$to_scan; |
07f31d19 |
395 | |
1a736efb |
396 | # first loop through all fully qualified columns and get the corresponding |
397 | # alias (should work even if they are in scalarrefs) |
ad630f4b |
398 | for my $alias (keys %$alias_list) { |
1a736efb |
399 | my $al_re = qr/ |
3f5b99fe |
400 | $lquote $alias $rquote $sep |
1a736efb |
401 | | |
3f5b99fe |
402 | \b $alias \. |
1a736efb |
403 | /x; |
404 | |
1a736efb |
405 | for my $type (keys %$to_scan) { |
406 | for my $piece (@{$to_scan->{$type}}) { |
a4812caa |
407 | $aliases_by_type->{$type}{$alias} ||= $alias_list->{$alias}{-join_path}||[] |
408 | if ($piece =~ $al_re); |
1a736efb |
409 | } |
ad630f4b |
410 | } |
1a736efb |
411 | } |
412 | |
413 | # now loop through unqualified column names, and try to locate them within |
414 | # the chunks |
415 | for my $col (keys %$colinfo) { |
3f5b99fe |
416 | next if $col =~ / \. /x; # if column is qualified it was caught by the above |
1a736efb |
417 | |
3f5b99fe |
418 | my $col_re = qr/ $lquote $col $rquote /x; |
07f31d19 |
419 | |
1a736efb |
420 | for my $type (keys %$to_scan) { |
421 | for my $piece (@{$to_scan->{$type}}) { |
a4812caa |
422 | if ($piece =~ $col_re) { |
423 | my $alias = $colinfo->{$col}{-source_alias}; |
424 | $aliases_by_type->{$type}{$alias} ||= $alias_list->{$alias}{-join_path}||[]; |
425 | } |
1a736efb |
426 | } |
07f31d19 |
427 | } |
428 | } |
429 | |
430 | # Add any non-left joins to the restriction list (such joins are indeed restrictions) |
ad630f4b |
431 | for my $j (values %$alias_list) { |
07f31d19 |
432 | my $alias = $j->{-alias} or next; |
a4812caa |
433 | $aliases_by_type->{restricting}{$alias} ||= $j->{-join_path}||[] if ( |
07f31d19 |
434 | (not $j->{-join_type}) |
435 | or |
436 | ($j->{-join_type} !~ /^left (?: \s+ outer)? $/xi) |
437 | ); |
438 | } |
439 | |
964a3c71 |
440 | return $aliases_by_type; |
07f31d19 |
441 | } |
442 | |
bac358c9 |
443 | # This is the engine behind { distinct => 1 } |
0a3441ee |
444 | sub _group_over_selection { |
445 | my ($self, $from, $select, $order_by) = @_; |
446 | |
447 | my $rs_column_list = $self->_resolve_column_info ($from); |
448 | |
449 | my (@group_by, %group_index); |
450 | |
36fd7f07 |
451 | # the logic is: if it is a { func => val } we assume an aggregate, |
452 | # otherwise if \'...' or \[...] we assume the user knows what is |
453 | # going on thus group over it |
0a3441ee |
454 | for (@$select) { |
455 | if (! ref($_) or ref ($_) ne 'HASH' ) { |
456 | push @group_by, $_; |
457 | $group_index{$_}++; |
458 | if ($rs_column_list->{$_} and $_ !~ /\./ ) { |
459 | # add a fully qualified version as well |
460 | $group_index{"$rs_column_list->{$_}{-source_alias}.$_"}++; |
461 | } |
07f31d19 |
462 | } |
463 | } |
ad630f4b |
464 | |
0a3441ee |
465 | # add any order_by parts that are not already present in the group_by |
466 | # we need to be careful not to add any named functions/aggregates |
bac358c9 |
467 | # i.e. order_by => [ ... { count => 'foo' } ... ] |
14e26c5f |
468 | my @leftovers; |
bac358c9 |
469 | for ($self->_extract_order_criteria($order_by)) { |
0a3441ee |
470 | # only consider real columns (for functions the user got to do an explicit group_by) |
14e26c5f |
471 | if (@$_ != 1) { |
472 | push @leftovers, $_; |
473 | next; |
474 | } |
bac358c9 |
475 | my $chunk = $_->[0]; |
14e26c5f |
476 | my $colinfo = $rs_column_list->{$chunk} or do { |
477 | push @leftovers, $_; |
478 | next; |
479 | }; |
0a3441ee |
480 | |
481 | $chunk = "$colinfo->{-source_alias}.$chunk" if $chunk !~ /\./; |
482 | push @group_by, $chunk unless $group_index{$chunk}++; |
483 | } |
484 | |
14e26c5f |
485 | return wantarray |
486 | ? (\@group_by, (@leftovers ? \@leftovers : undef) ) |
487 | : \@group_by |
488 | ; |
07f31d19 |
489 | } |
490 | |
d28bb90d |
491 | sub _resolve_ident_sources { |
492 | my ($self, $ident) = @_; |
493 | |
494 | my $alias2source = {}; |
495 | my $rs_alias; |
496 | |
497 | # the reason this is so contrived is that $ident may be a {from} |
498 | # structure, specifying multiple tables to join |
6298a324 |
499 | if ( blessed $ident && $ident->isa("DBIx::Class::ResultSource") ) { |
d28bb90d |
500 | # this is compat mode for insert/update/delete which do not deal with aliases |
501 | $alias2source->{me} = $ident; |
502 | $rs_alias = 'me'; |
503 | } |
504 | elsif (ref $ident eq 'ARRAY') { |
505 | |
506 | for (@$ident) { |
507 | my $tabinfo; |
508 | if (ref $_ eq 'HASH') { |
509 | $tabinfo = $_; |
510 | $rs_alias = $tabinfo->{-alias}; |
511 | } |
512 | if (ref $_ eq 'ARRAY' and ref $_->[0] eq 'HASH') { |
513 | $tabinfo = $_->[0]; |
514 | } |
515 | |
4376a157 |
516 | $alias2source->{$tabinfo->{-alias}} = $tabinfo->{-rsrc} |
517 | if ($tabinfo->{-rsrc}); |
d28bb90d |
518 | } |
519 | } |
520 | |
521 | return ($alias2source, $rs_alias); |
522 | } |
523 | |
524 | # Takes $ident, \@column_names |
525 | # |
526 | # returns { $column_name => \%column_info, ... } |
527 | # also note: this adds -result_source => $rsrc to the column info |
528 | # |
09e14fdc |
529 | # If no columns_names are supplied returns info about *all* columns |
530 | # for all sources |
d28bb90d |
531 | sub _resolve_column_info { |
532 | my ($self, $ident, $colnames) = @_; |
533 | my ($alias2src, $root_alias) = $self->_resolve_ident_sources($ident); |
534 | |
52416317 |
535 | my (%seen_cols, @auto_colnames); |
d28bb90d |
536 | |
537 | # compile a global list of column names, to be able to properly |
538 | # disambiguate unqualified column names (if at all possible) |
539 | for my $alias (keys %$alias2src) { |
540 | my $rsrc = $alias2src->{$alias}; |
541 | for my $colname ($rsrc->columns) { |
542 | push @{$seen_cols{$colname}}, $alias; |
3f5b99fe |
543 | push @auto_colnames, "$alias.$colname" unless $colnames; |
d28bb90d |
544 | } |
545 | } |
546 | |
09e14fdc |
547 | $colnames ||= [ |
548 | @auto_colnames, |
549 | grep { @{$seen_cols{$_}} == 1 } (keys %seen_cols), |
550 | ]; |
551 | |
52416317 |
552 | my (%return, $colinfos); |
d28bb90d |
553 | foreach my $col (@$colnames) { |
52416317 |
554 | my ($source_alias, $colname) = $col =~ m/^ (?: ([^\.]+) \. )? (.+) $/x; |
d28bb90d |
555 | |
52416317 |
556 | # if the column was seen exactly once - we know which rsrc it came from |
557 | $source_alias ||= $seen_cols{$colname}[0] |
558 | if ($seen_cols{$colname} and @{$seen_cols{$colname}} == 1); |
d28bb90d |
559 | |
52416317 |
560 | next unless $source_alias; |
561 | |
562 | my $rsrc = $alias2src->{$source_alias} |
563 | or next; |
564 | |
565 | $return{$col} = { |
6395604e |
566 | %{ |
567 | ( $colinfos->{$source_alias} ||= $rsrc->columns_info )->{$colname} |
568 | || |
569 | $self->throw_exception( |
570 | "No such column '$colname' on source " . $rsrc->source_name |
571 | ); |
572 | }, |
d28bb90d |
573 | -result_source => $rsrc, |
52416317 |
574 | -source_alias => $source_alias, |
81bf295c |
575 | -fq_colname => $col eq $colname ? "$source_alias.$col" : $col, |
576 | -colname => $colname, |
d28bb90d |
577 | }; |
81bf295c |
578 | |
579 | $return{"$source_alias.$colname"} = $return{$col} if $col eq $colname; |
d28bb90d |
580 | } |
581 | |
582 | return \%return; |
583 | } |
584 | |
289ac713 |
585 | # The DBIC relationship chaining implementation is pretty simple - every |
586 | # new related_relationship is pushed onto the {from} stack, and the {select} |
587 | # window simply slides further in. This means that when we count somewhere |
588 | # in the middle, we got to make sure that everything in the join chain is an |
589 | # actual inner join, otherwise the count will come back with unpredictable |
590 | # results (a resultset may be generated with _some_ rows regardless of if |
591 | # the relation which the $rs currently selects has rows or not). E.g. |
592 | # $artist_rs->cds->count - normally generates: |
593 | # SELECT COUNT( * ) FROM artist me LEFT JOIN cd cds ON cds.artist = me.artistid |
594 | # which actually returns the number of artists * (number of cds || 1) |
595 | # |
596 | # So what we do here is crawl {from}, determine if the current alias is at |
597 | # the top of the stack, and if not - make sure the chain is inner-joined down |
598 | # to the root. |
599 | # |
31a8aaaf |
600 | sub _inner_join_to_node { |
289ac713 |
601 | my ($self, $from, $alias) = @_; |
602 | |
603 | # subqueries and other oddness are naturally not supported |
604 | return $from if ( |
605 | ref $from ne 'ARRAY' |
606 | || |
607 | @$from <= 1 |
608 | || |
609 | ref $from->[0] ne 'HASH' |
610 | || |
611 | ! $from->[0]{-alias} |
612 | || |
7eb76996 |
613 | $from->[0]{-alias} eq $alias # this last bit means $alias is the head of $from - nothing to do |
289ac713 |
614 | ); |
615 | |
616 | # find the current $alias in the $from structure |
617 | my $switch_branch; |
618 | JOINSCAN: |
619 | for my $j (@{$from}[1 .. $#$from]) { |
620 | if ($j->[0]{-alias} eq $alias) { |
621 | $switch_branch = $j->[0]{-join_path}; |
622 | last JOINSCAN; |
623 | } |
624 | } |
625 | |
7eb76996 |
626 | # something else went quite wrong |
289ac713 |
627 | return $from unless $switch_branch; |
628 | |
629 | # So it looks like we will have to switch some stuff around. |
630 | # local() is useless here as we will be leaving the scope |
631 | # anyway, and deep cloning is just too fucking expensive |
8273e845 |
632 | # So replace the first hashref in the node arrayref manually |
289ac713 |
633 | my @new_from = ($from->[0]); |
faeb2407 |
634 | my $sw_idx = { map { (values %$_), 1 } @$switch_branch }; #there's one k/v per join-path |
289ac713 |
635 | |
636 | for my $j (@{$from}[1 .. $#$from]) { |
637 | my $jalias = $j->[0]{-alias}; |
638 | |
639 | if ($sw_idx->{$jalias}) { |
640 | my %attrs = %{$j->[0]}; |
641 | delete $attrs{-join_type}; |
642 | push @new_from, [ |
643 | \%attrs, |
644 | @{$j}[ 1 .. $#$j ], |
645 | ]; |
646 | } |
647 | else { |
648 | push @new_from, $j; |
649 | } |
650 | } |
651 | |
652 | return \@new_from; |
653 | } |
654 | |
ea5c7509 |
655 | # yet another atrocity: attempt to extract all columns from a |
656 | # where condition by hooking _quote |
657 | sub _extract_condition_columns { |
27e0370d |
658 | my ($self, $cond, $sql_maker_cache) = @_; |
ea5c7509 |
659 | |
660 | return [] unless $cond; |
661 | |
27e0370d |
662 | my $sm = $sql_maker_cache->{condparser} ||= $self->{_sql_ident_capturer} ||= do { |
ea5c7509 |
663 | # FIXME - replace with a Moo trait |
664 | my $orig_sm_class = ref $self->sql_maker; |
665 | my $smic_class = "${orig_sm_class}::_IdentCapture_"; |
666 | |
667 | unless ($smic_class->isa('SQL::Abstract')) { |
668 | |
669 | no strict 'refs'; |
670 | *{"${smic_class}::_quote"} = subname "${smic_class}::_quote" => sub { |
671 | my ($self, $ident) = @_; |
672 | if (ref $ident eq 'SCALAR') { |
673 | $ident = $$ident; |
674 | my $storage_quotes = $self->sql_quote_char || '"'; |
675 | my ($ql, $qr) = map |
676 | { quotemeta $_ } |
677 | (ref $storage_quotes eq 'ARRAY' ? @$storage_quotes : ($storage_quotes) x 2 ) |
678 | ; |
679 | |
680 | while ($ident =~ / |
681 | $ql (\w+) $qr |
682 | | |
683 | ([\w\.]+) |
684 | /xg) { |
685 | $self->{_captured_idents}{$1||$2}++; |
686 | } |
bac6c4fb |
687 | } |
037e8dca |
688 | else { |
ea5c7509 |
689 | $self->{_captured_idents}{$ident}++; |
037e8dca |
690 | } |
ea5c7509 |
691 | return $ident; |
692 | }; |
693 | |
694 | *{"${smic_class}::_get_captured_idents"} = subname "${smic_class}::_get_captures" => sub { |
695 | (delete shift->{_captured_idents}) || {}; |
696 | }; |
697 | |
698 | $self->inject_base ($smic_class, $orig_sm_class); |
699 | |
bac6c4fb |
700 | } |
ea5c7509 |
701 | |
702 | $smic_class->new(); |
703 | }; |
704 | |
27e0370d |
705 | $sm->_recurse_where($cond); |
ea5c7509 |
706 | |
27e0370d |
707 | return [ sort keys %{$sm->_get_captured_idents} ]; |
ea5c7509 |
708 | } |
709 | |
bac358c9 |
710 | sub _extract_order_criteria { |
1a736efb |
711 | my ($self, $order_by, $sql_maker) = @_; |
c0748280 |
712 | |
1a736efb |
713 | my $parser = sub { |
714 | my ($sql_maker, $order_by) = @_; |
c0748280 |
715 | |
1a736efb |
716 | return scalar $sql_maker->_order_by_chunks ($order_by) |
717 | unless wantarray; |
c0748280 |
718 | |
1a736efb |
719 | my @chunks; |
bac358c9 |
720 | for ($sql_maker->_order_by_chunks ($order_by) ) { |
721 | my $chunk = ref $_ ? $_ : [ $_ ]; |
722 | $chunk->[0] =~ s/\s+ (?: ASC|DESC ) \s* $//ix; |
1a736efb |
723 | push @chunks, $chunk; |
bac6c4fb |
724 | } |
1a736efb |
725 | |
726 | return @chunks; |
727 | }; |
728 | |
729 | if ($sql_maker) { |
730 | return $parser->($sql_maker, $order_by); |
bac6c4fb |
731 | } |
732 | else { |
1a736efb |
733 | $sql_maker = $self->sql_maker; |
734 | local $sql_maker->{quote_char}; |
735 | return $parser->($sql_maker, $order_by); |
bac6c4fb |
736 | } |
bac6c4fb |
737 | } |
738 | |
7cec4356 |
739 | sub _order_by_is_stable { |
5f11e54f |
740 | my ($self, $ident, $order_by, $where) = @_; |
c0748280 |
741 | |
5f11e54f |
742 | my $colinfo = $self->_resolve_column_info($ident, [ |
743 | (map { $_->[0] } $self->_extract_order_criteria($order_by)), |
744 | $where ? @{$self->_extract_fixed_condition_columns($where)} :(), |
745 | ]); |
c0748280 |
746 | |
7cec4356 |
747 | return undef unless keys %$colinfo; |
748 | |
749 | my $cols_per_src; |
750 | $cols_per_src->{$_->{-source_alias}}{$_->{-colname}} = $_ for values %$colinfo; |
751 | |
752 | for (values %$cols_per_src) { |
753 | my $src = (values %$_)[0]->{-result_source}; |
754 | return 1 if $src->_identifying_column_set($_); |
c0748280 |
755 | } |
756 | |
7cec4356 |
757 | return undef; |
758 | } |
759 | |
5f11e54f |
760 | # returns an arrayref of column names which *definitely* have som |
761 | # sort of non-nullable equality requested in the given condition |
762 | # specification. This is used to figure out if a resultset is |
763 | # constrained to a column which is part of a unique constraint, |
764 | # which in turn allows us to better predict how ordering will behave |
765 | # etc. |
766 | # |
767 | # this is a rudimentary, incomplete, and error-prone extractor |
768 | # however this is OK - it is conservative, and if we can not find |
769 | # something that is in fact there - the stack will recover gracefully |
770 | # Also - DQ and the mst it rode in on will save us all RSN!!! |
771 | sub _extract_fixed_condition_columns { |
772 | my ($self, $where, $nested) = @_; |
773 | |
774 | return unless ref $where eq 'HASH'; |
775 | |
776 | my @cols; |
777 | for my $lhs (keys %$where) { |
778 | if ($lhs =~ /^\-and$/i) { |
779 | push @cols, ref $where->{$lhs} eq 'ARRAY' |
780 | ? ( map { $self->_extract_fixed_condition_columns($_, 1) } @{$where->{$lhs}} ) |
781 | : $self->_extract_fixed_condition_columns($where->{$lhs}, 1) |
782 | ; |
783 | } |
784 | elsif ($lhs !~ /^\-/) { |
785 | my $val = $where->{$lhs}; |
786 | |
787 | push @cols, $lhs if (defined $val and ( |
788 | ! ref $val |
789 | or |
790 | (ref $val eq 'HASH' and keys %$val == 1 and defined $val->{'='}) |
791 | )); |
792 | } |
793 | } |
794 | return $nested ? @cols : \@cols; |
c0748280 |
795 | } |
bac6c4fb |
796 | |
d28bb90d |
797 | 1; |