File: ImportIterator.pm

package info (click to toggle)
libur-perl 0.470%2Bds-2
  • links: PTS, VCS
  • area: main
  • in suites: bookworm
  • size: 7,184 kB
  • sloc: perl: 61,813; javascript: 255; xml: 108; sh: 13; makefile: 9
file content (819 lines) | stat: -rw-r--r-- 42,915 bytes parent folder | download | duplicates (3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
package UR::Context;

# Methods related to the import iterator (part of the loading process).
#
# They are broken out here for readability purposes.  The methods still live
# in the UR::Context namespace.

use strict;
use warnings;

our $VERSION = "0.47"; # UR $VERSION;


# A wrapper around the method of the same name in UR::DataSource::* to iterate over the
# possible data sources involved in a query.  The easy case (a query against a single data source)
# will return the $primary_template data structure.  If the query involves more than one data source,
# then this method also returns a list containing triples (@addl_loading_info) where each member is:
# 1) The secondary data source name
# 2) a listref of delegated properties joining the primary class to the secondary class
# 3) a rule template applicable against the secondary data source
sub _resolve_query_plan_for_ds_and_bxt {
    my($self,$primary_data_source,$rule_template) = @_;

    my $primary_query_plan = $primary_data_source->_resolve_query_plan($rule_template);

    unless ($primary_query_plan->{'joins_across_data_sources'}) {
        # Common, easy case
        return $primary_query_plan;
    }

    my @addl_loading_info;
    foreach my $secondary_data_source_id ( keys %{$primary_query_plan->{'joins_across_data_sources'}} ) {
        my $this_ds_delegations = $primary_query_plan->{'joins_across_data_sources'}->{$secondary_data_source_id};

        my %seen_properties;
        foreach my $delegated_property ( @$this_ds_delegations ) {
            my $delegated_property_name = $delegated_property->property_name;
            next if ($seen_properties{$delegated_property_name}++);

            my $operator = $rule_template->operator_for($delegated_property_name);
            $operator ||= '=';  # FIXME - shouldn't the template return this for us?
            my @secondary_params = ($delegated_property->to . ' ' . $operator);

            my $class_meta = UR::Object::Type->get($delegated_property->class_name);
            my $relation_property = $class_meta->property_meta_for_name($delegated_property->via);

            my $secondary_class = $relation_property->data_type;

            # we can also add in any properties in the property's joins that also appear in the rule
            my @property_pairs = $relation_property->get_property_name_pairs_for_join();
            foreach my $pair ( @property_pairs ) {
                my($primary_property, $secondary_property) = @$pair;
                next if ($seen_properties{$primary_property}++);
                next unless ($rule_template->specifies_value_for($primary_property));

                my $operator = $rule_template->operator_for($primary_property);
                $operator ||= '=';
                push @secondary_params, "$secondary_property $operator";
             }

            my $secondary_rule_template = UR::BoolExpr::Template->resolve($secondary_class, @secondary_params);

            # FIXME there should be a way to collect all the requests for the same datasource together...
            # FIXME - currently in the process of switching to object-based instead of class-based data sources
            # For now, data sources are still singleton objects, so this get() will work.  When we're fully on
            # regular-object-based data sources, then it'll probably change to UR::DataSource->get($secondary_data_source_id); 
            my $secondary_data_source = UR::DataSource->get($secondary_data_source_id) || $secondary_data_source_id->get();
            push @addl_loading_info,
                     $secondary_data_source,
                     [$delegated_property],
                     $secondary_rule_template;
        }
    }

    return ($primary_query_plan, @addl_loading_info);
}


# Used by _create_secondary_loading_comparators to convert a rule against the primary data source
# to a rule that can be used against a secondary data source
# FIXME this might be made simpler be leaning on infer_property_value_from_rule()?
sub _create_secondary_rule_from_primary {
    my($self,$primary_rule, $delegated_properties, $secondary_rule_template) = @_;

    my @secondary_values;
    my %seen_properties;  # FIXME - we've already been over this list in _resolve_query_plan_for_ds_and_bxt()...
    # FIXME - is there ever a case where @$delegated_properties will be more than one item?
    foreach my $property ( @$delegated_properties ) {
        my $value = $primary_rule->value_for($property->property_name);

        my $secondary_property_name = $property->to;
        my $pos = $secondary_rule_template->value_position_for_property_name($secondary_property_name);
        $secondary_values[$pos] = $value;
        $seen_properties{$property->property_name}++;

        my $class_meta = $property->class_meta;
        my $via_property = $class_meta->property_meta_for_name($property->via);
        my @pairs = $via_property->get_property_name_pairs_for_join();
        foreach my $pair ( @pairs ) {
            my($primary_property_name, $secondary_property_name) = @$pair;

            next if ($seen_properties{$primary_property_name}++);
            $value = $primary_rule->value_for($primary_property_name);
            next unless $value;

            $pos = $secondary_rule_template->value_position_for_property_name($secondary_property_name);
            $secondary_values[$pos] = $value;
        }
    }

    my $secondary_rule = $secondary_rule_template->get_rule_for_values(@secondary_values);

    return $secondary_rule;
}


# Since we'll be appending more "columns" of data to the listrefs returned by
# the primary datasource's query, we need to apply fixups to the column positions
# to all the secondary loading templates
# The column_position and object_num offsets needed for the next call of this method
# are returned
sub _fixup_secondary_loading_template_column_positions {
    my($self,$primary_loading_templates, $secondary_loading_templates, $column_position_offset, $object_num_offset) = @_;

    if (! defined($column_position_offset) or ! defined($object_num_offset)) {
        $column_position_offset = 0;
        foreach my $tmpl ( @{$primary_loading_templates} ) {
            $column_position_offset += scalar(@{$tmpl->{'column_positions'}});
        }
        $object_num_offset = scalar(@{$primary_loading_templates});
    }

    my $this_template_column_count;
    foreach my $tmpl ( @$secondary_loading_templates ) {
        foreach ( @{$tmpl->{'column_positions'}} ) {
            $_ += $column_position_offset;
        }
        foreach ( @{$tmpl->{'id_column_positions'}} ) {
            $_ += $column_position_offset;
        }
        $tmpl->{'object_num'} += $object_num_offset;

        $this_template_column_count += scalar(@{$tmpl->{'column_positions'}});
    }


    return ($column_position_offset + $this_template_column_count,
            $object_num_offset + scalar(@$secondary_loading_templates) );
}

# For queries that have to hit multiple data sources, this method creates two lists of
# closures.  The first is a list of object fabricators, where the loading templates
# have been given fixups to the column positions (see _fixup_secondary_loading_template_column_positions())
# The second is a list of closures for each data source (the @addl_loading_info stuff
# from _resolve_query_plan_for_ds_and_bxt) that's able to compare the row loaded from the
# primary data source and see if it joins to a row from this secondary datasource's database
sub _create_secondary_loading_closures {
    my($self, $primary_template, $rule, @addl_loading_info) = @_;

    my $loading_templates = $primary_template->{'loading_templates'};

    # Make a mapping of property name to column positions returned by the primary query
    my %primary_query_column_positions;
    foreach my $tmpl ( @$loading_templates ) {
        my $property_name_count = scalar(@{$tmpl->{'property_names'}});
        for (my $i = 0; $i < $property_name_count; $i++) {
            my $property_name = $tmpl->{'property_names'}->[$i];
            my $pos = $tmpl->{'column_positions'}->[$i];
            $primary_query_column_positions{$property_name} = $pos;
        }
    }

    my @secondary_object_importers;
    my @addl_join_comparators;

    # used to shift the apparent column position of the secondary loading template info
    my ($column_position_offset,$object_num_offset);

    while (@addl_loading_info) {
        my $secondary_data_source = shift @addl_loading_info;
        my $this_ds_delegations = shift @addl_loading_info;
        my $secondary_rule_template = shift @addl_loading_info;

        my $secondary_rule = $self->_create_secondary_rule_from_primary (
                                              $rule,
                                              $this_ds_delegations,
                                              $secondary_rule_template,
                                       );
        $secondary_data_source = $secondary_data_source->resolve_data_sources_for_rule($secondary_rule);
        my $secondary_template = $self->_resolve_query_plan_for_ds_and_bxt($secondary_data_source,$secondary_rule_template);

        # sets of triples where the first in the triple is the column index in the
        # $secondary_db_row (in the join_comparator closure below), the second is the
        # index in the $next_db_row.  And the last is a flag indicating if we should 
        # perform a numeric comparison.  This way we can preserve the order the comparisons
        # should be done in
        my @join_comparison_info;
        foreach my $property ( @$this_ds_delegations ) {
            # first, map column names in the joined class to column names in the primary class
            my %foreign_property_name_map;
            my @this_property_joins = $property->_resolve_join_chain();
            foreach my $join ( @this_property_joins ) {
                last if ($join->{foreign_class}->isa('UR::Value') and $join eq $this_property_joins[-1]);
                my @source_names = @{$join->{'source_property_names'}};
                my @foreign_names = @{$join->{'foreign_property_names'}};
                @foreign_property_name_map{@foreign_names} = @source_names;
            }

            # Now, find out which numbered column in the result query maps to those names
            my $secondary_loading_templates = $secondary_template->{'loading_templates'};
            foreach my $tmpl ( @$secondary_loading_templates ) {
                my $property_name_count = scalar(@{$tmpl->{'property_names'}});
                for (my $i = 0; $i < $property_name_count; $i++) {
                    my $property_name = $tmpl->{'property_names'}->[$i];
                    if ($foreign_property_name_map{$property_name}) {
                        # This is the one we're interested in...  Where does it come from in the primary query?
                        my $column_position = $tmpl->{'column_positions'}->[$i];
                        # What are the types involved?
                        my $primary_query_column_name = $foreign_property_name_map{$property_name};
                        my $primary_property_class_meta = $primary_template->{'class_name'}->__meta__;
                        my $primary_property_meta = $primary_property_class_meta->property_meta_for_name($primary_query_column_name);
                        unless ($primary_property_meta) {
                            Carp::croak("Can't resolve property metadata for property '$primary_query_column_name' of class ".$primary_template->{'class_name'});
                        }

                        my $secondary_class_meta = $secondary_template->{'class_name'}->__meta__;
                        my $secondary_property_meta = $secondary_class_meta->property_meta_for_name($property_name);
                        unless ($secondary_property_meta) {
                            Carp::croak("Can't resolve property metadata for property '$property_name' of class ".$secondary_template->{'class_name'});
                        }

                        my $comparison_type;
                        if ($primary_property_meta->is_numeric && $secondary_property_meta->is_numeric) {
                            $comparison_type = 1;
                        }

                        my $comparison_position;
                        if (exists $primary_query_column_positions{$primary_query_column_name} ) {
                            $comparison_position = $primary_query_column_positions{$primary_query_column_name};

                        } else {
                            # This isn't a real column we can get from the data source.  Maybe it's
                            # in the constant_property_names of the primary_loading_template?
                            unless (grep { $_ eq $primary_query_column_name}
                                    @{$loading_templates->[0]->{'constant_property_names'}}) {
                                die sprintf("Can't resolve datasource comparison to join %s::%s to %s:%s",
                                            $primary_template->{'class_name'}, $primary_query_column_name,
                                            $secondary_template->{'class_name'}, $property_name);
                            }
                            my $comparison_value = $rule->value_for($primary_query_column_name);
                            unless (defined $comparison_value) {
                                $comparison_value = $self->infer_property_value_from_rule($primary_query_column_name, $rule);
                            }
                            $comparison_position = \$comparison_value;
                        }
                        push @join_comparison_info, $column_position,
                                                    $comparison_position,
                                                    $comparison_type;


                    }
                }
            }
        }
        my $secondary_db_iterator = $secondary_data_source->create_iterator_closure_for_rule($secondary_rule);

        my $secondary_db_row;
        # For this closure, pass in the row we just loaded from the primary DB query.
        # This one will return the data from this secondary DB's row if the passed-in
        # row successfully joins to this secondary db iterator.  It returns an empty list
        # if there were no matches, and returns false if there is no more data from the query
        my $join_comparator = sub {
            my $next_db_row = shift;  # From the primary DB
            READ_DB_ROW:
            while(1) {
                return unless ($secondary_db_iterator);
                unless ($secondary_db_row) {
                    ($secondary_db_row) = $secondary_db_iterator->();
                    unless($secondary_db_row) {
                        # No more data to load 
                        $secondary_db_iterator = undef;
                        return;
                    }
                }

                for (my $i = 0; $i < @join_comparison_info; $i += 3) {
                    my $secondary_column = $join_comparison_info[$i];
                    my $primary_column = $join_comparison_info[$i+1];
                    my $is_numeric = $join_comparison_info[$i+2];

                    my $comparison;
                    if (ref $primary_column) {
                        # This was one of those constant value items
                        if ($is_numeric) {
                            $comparison = $secondary_db_row->[$secondary_column] <=> $$primary_column;
                        } else {
                            $comparison = $secondary_db_row->[$secondary_column] cmp $$primary_column;
                        }
                    } else {
                        if ($join_comparison_info[$i+2]) {
                            $comparison = $secondary_db_row->[$secondary_column] <=> $next_db_row->[$primary_column];
                        } else {
                            $comparison = $secondary_db_row->[$secondary_column] cmp $next_db_row->[$primary_column];
                        }
                    }
                    if ($comparison < 0) {
                        # less than, get the next row from the secondary DB
                        $secondary_db_row = undef;
                        redo READ_DB_ROW;
                    } elsif ($comparison == 0) {
                        # This one was the same, keep looking at the others
                    } else {
                        # greater-than, there's no match for this primary DB row
                        return 0;
                    }
                }
                # All the joined columns compared equal, return the data
                return $secondary_db_row;
            }
        };
        Sub::Name::subname('UR::Context::__join_comparator(closure)__', $join_comparator);
        push @addl_join_comparators, $join_comparator;


        # And for the object importer/fabricator, here's where we need to shift the column order numbers
        # over, because these closures will be called after all the db iterators' rows are concatenated
        # together.  We also need to make a copy of the loading_templates list so as to not mess up the
        # class' notion of where the columns are
        # FIXME - it seems wasteful that we need to re-created this each time.  Look into some way of using 
        # the original copy that lives in $primary_template->{'loading_templates'}?  Somewhere else?
        my @secondary_loading_templates;
        foreach my $tmpl ( @{$secondary_template->{'loading_templates'}} ) {
            my %copy;
            foreach my $key ( keys %$tmpl ) {
                my $value_to_copy = $tmpl->{$key};
                if (ref($value_to_copy) eq 'ARRAY') {
                    $copy{$key} = [ @$value_to_copy ];
                } elsif (ref($value_to_copy) eq 'HASH') {
                    $copy{$key} = { %$value_to_copy };
                } else {
                    $copy{$key} = $value_to_copy;
                }
            }
            push @secondary_loading_templates, \%copy;
        }
        ($column_position_offset,$object_num_offset) =
                $self->_fixup_secondary_loading_template_column_positions($primary_template->{'loading_templates'},
                                                                          \@secondary_loading_templates,
                                                                          $column_position_offset,$object_num_offset);

        #my($secondary_rule_template,@secondary_values) = $secondary_rule->get_template_and_values();
        my @secondary_values = $secondary_rule->values();
        foreach my $secondary_loading_template ( @secondary_loading_templates ) {
            my $secondary_object_importer = UR::Context::ObjectFabricator->create_for_loading_template(
                                                       $self,
                                                       $secondary_loading_template,
                                                       $secondary_template,
                                                       $secondary_rule,
                                                       $secondary_rule_template,
                                                       \@secondary_values,
                                                       $secondary_data_source
                                                );
            next unless $secondary_object_importer;
            push @secondary_object_importers, $secondary_object_importer;
        }


   }

    return (\@secondary_object_importers, \@addl_join_comparators);
}


# This returns an iterator that is used to bring objects in from an underlying
# context into this context.
sub _create_import_iterator_for_underlying_context {
    my ($self, $rule, $dsx, $this_get_serial) = @_;

    # TODO: instead of taking a data source, resolve this internally.
    # The underlying context itself should be responsible for its data sources.

    # Make an iterator for the primary data source.
    # Primary here meaning the one for the class we're explicitly requesting.
    # We may need to join to other data sources to complete the query.
    my ($db_iterator)
        = $dsx->create_iterator_closure_for_rule($rule);

    my ($rule_template, @values) = $rule->template_and_values();
    my ($query_plan,@addl_loading_info) = $self->_resolve_query_plan_for_ds_and_bxt($dsx,$rule_template);
    my $class_name = $query_plan->{class_name};

    my $group_by    = $rule_template->group_by;
    my $order_by    = $rule_template->order_by;
    my $aggregate   = $rule_template->aggregate;

    if (my $sub_typing_property) {
        # When the rule has a property specified which indicates a specific sub-type, catch this and re-call
        # this method recursively with the specific subclass name.
        my ($rule_template, @values) = $rule->template_and_values();
        my $rule_template_specifies_value_for_subtype   = $query_plan->{rule_template_specifies_value_for_subtype};
        my $class_table_name                            = $query_plan->{class_table_name};

        warn "Implement me carefully";

        if ($rule_template_specifies_value_for_subtype) {
            my $sub_classification_meta_class_name          = $query_plan->{sub_classification_meta_class_name};
            my $value = $rule->value_for($sub_typing_property);
            my $type_obj = $sub_classification_meta_class_name->get($value);
            if ($type_obj) {
                my $subclass_name = $type_obj->subclass_name($class_name);
                if ($subclass_name and $subclass_name ne $class_name) {
                    #$rule = $subclass_name->define_boolexpr($rule->params_list, $sub_typing_property => $value);
                    $rule = UR::BoolExpr->resolve_normalized($subclass_name, $rule->params_list, $sub_typing_property => $value);
                    return $self->_create_import_iterator_for_underlying_context($rule,$dsx,$this_get_serial);
                }
            }
            else {
                die "No $value for $class_name?\n";
            }
        }
        elsif (not $class_table_name) {
            die "No longer supported!";
            my $rule = UR::BoolExpr->resolve(
                           $class_name,
                           $rule_template->get_rule_for_values(@values)->params_list,
                        );
            return $self->_create_import_iterator_for_underlying_context($rule,$dsx,$this_get_serial)
        }
        else {
            # continue normally
            # the logic below will handle sub-classifying each returned entity
        }
    }


    my $loading_templates                           = $query_plan->{loading_templates};
    my $sub_typing_property                         = $query_plan->{sub_typing_property};
    my $next_db_row;
    my $rows = 0;                                   # number of rows the query returned

    my $recursion_desc                              = $query_plan->{recursion_desc};
    my($rule_template_without_recursion_desc, $rule_template_id_without_recursion);
    my($rule_without_recursion_desc, $rule_id_without_recursion);
    # These get set if you're doing a -recurse query, and the underlying data source doesn't support recursion
    my($by_hand_recursive_rule_template,$by_hand_recursive_source_property,@by_hand_recursive_source_values,$by_hand_recursing_iterator);
    if ($recursion_desc) {
        $rule_template_without_recursion_desc        = $query_plan->{rule_template_without_recursion_desc};
        $rule_template_id_without_recursion          = $rule_template_without_recursion_desc->id;
        $rule_without_recursion_desc                 = $rule_template_without_recursion_desc->get_rule_for_values(@values);
        $rule_id_without_recursion                   = $rule_without_recursion_desc->id;

        if ($query_plan->{'recurse_resolution_by_iteration'}) {
            # The data source does not support a recursive query.  Accomplish the same thing by
            # recursing back into _create_import_iterator_for_underlying_context for each level
            my $this;
            ($this,$by_hand_recursive_source_property) = @$recursion_desc;

            my @extra;
            $by_hand_recursive_rule_template = UR::BoolExpr::Template->resolve($class_name, "$this in");
            $by_hand_recursive_rule_template->recursion_desc($recursion_desc);
            if (!$by_hand_recursive_rule_template or @extra) {
                Carp::croak("Can't resolve recursive query: Class $class_name cannot filter by one or more properties: "
                            . join(', ', @extra));
            }
        }
    }

    my $rule_id = $rule->id;
    my $rule_template_id = $rule_template->id;

    my $needs_further_boolexpr_evaluation_after_loading = $query_plan->{'needs_further_boolexpr_evaluation_after_loading'};

    my %subordinate_iterator_for_class;

    # TODO: move the creation of the fabricators into the query plan object initializer.
    # instead of making just one import iterator, we make one per loading template
    # we then have our primary iterator use these to fabricate objects for each db row
    my @object_fabricators;
    if ($group_by) {
        # returning sets for each sub-group instead of instance objects...
        my $division_point = scalar(@$group_by)-1;
        my $subset_template = $rule_template->_template_for_grouped_subsets();
        my $set_class = $class_name . '::Set';
        my @aggregate_properties = ($aggregate ? @$aggregate : ());
        unshift(@aggregate_properties, 'count') unless (grep { $_ eq 'count' } @aggregate_properties);

        my $fab_subref = sub {
            my $row = $_[0];
            my @group_values = @$row[0..$division_point];
            my $ss_rule = $subset_template->get_rule_for_values(@values, @group_values);
            my $set = $set_class->get($ss_rule->id);
            unless ($set) {
                Carp::croak("Failed to fabricate $set_class for rule $ss_rule");
            }
            my $aggregates = $set->{__aggregates} ||= {};
            @$aggregates{@aggregate_properties} = @$row[$division_point+1..$#$row];
            return $set;
        };

        my $object_fabricator = UR::Context::ObjectFabricator->_create(
                                    fabricator => $fab_subref,
                                    context    => $self,
                                );
        unshift @object_fabricators, $object_fabricator;
    }
    else {
        # regular instances
        for my $loading_template (@$loading_templates) {
            my $object_fabricator =
                UR::Context::ObjectFabricator->create_for_loading_template(
                    $self,
                    $loading_template,
                    $query_plan,
                    $rule,
                    $rule_template,
                    \@values,
                    $dsx,
                );
            next unless $object_fabricator;
            unshift @object_fabricators, $object_fabricator;
        }
    }

    # For joins across data sources, we need to create importers/fabricators for those
    # classes, as well as callbacks used to perform the equivalent of an SQL join in
    # UR-space
    my @addl_join_comparators;
    if (@addl_loading_info) {
        if ($group_by) {
            Carp::croak("cross-datasource group-by is not supported yet");
        }
        my($addl_object_fabricators, $addl_join_comparators) =
                $self->_create_secondary_loading_closures( $query_plan,
                                                           $rule,
                                                           @addl_loading_info
                                                      );

        unshift @object_fabricators, @$addl_object_fabricators;
        push @addl_join_comparators, @$addl_join_comparators;
    }

    # To avoid calling the useless method 'fabricate' on a fabricator object for each object of each resultset row
    my @object_fabricator_closures = map { $_->fabricator } @object_fabricators;

    # Insert the key into all_objects_are_loaded to indicate that when we're done loading, we'll
    # have everything
    if ($query_plan->{'rule_matches_all'} and not $group_by) {
        $class_name->all_objects_are_loaded(undef);
    }

    #my $is_monitor_query = $self->monitor_query();

    # Make the iterator we'll return.
    my $next_object_to_return;
    my @object_ids_from_fabricators;
    my $underlying_context_iterator = sub {
        return undef unless $db_iterator;

        my $primary_object_for_next_db_row;

        LOAD_AN_OBJECT:
        until (defined $primary_object_for_next_db_row) { # note that we return directly when the db is out of data

            my ($next_db_row);
            ($next_db_row) = $db_iterator->() if ($db_iterator);

            if (! $next_db_row and $by_hand_recursive_rule_template and @by_hand_recursive_source_values) {
                # DB is out of results for this query, we need to handle recursion here in the context
                # and there are values to recurse on
                unless ($by_hand_recursing_iterator) {
                    # Do a new get() on the data source to recursively get more data
                    my $recurse_rule = $by_hand_recursive_rule_template->get_rule_for_values(\@by_hand_recursive_source_values);
                    $by_hand_recursing_iterator = $self->_create_import_iterator_for_underlying_context($recurse_rule,$dsx,$this_get_serial);
                }
                my $retval = $next_object_to_return;
                $next_object_to_return = $by_hand_recursing_iterator->();
                unless ($next_object_to_return) {
                    $by_hand_recursing_iterator = undef;
                    $by_hand_recursive_rule_template = undef;
                }
                return $retval;
            }

            unless ($next_db_row) {
                $db_iterator = undef;

                if ($rows == 0) {
                    # if we got no data at all from the sql then we give a status
                    # message about it and we update all_params_loaded to indicate
                    # that this set of parameters yielded 0 objects

                    my $rule_template_is_id_only = $query_plan->{rule_template_is_id_only};
                    if ($rule_template_is_id_only) {
                        my $id = $rule->value_for_id;
                        $UR::Context::all_objects_loaded->{$class_name}->{$id} = undef;
                    }
                    else {
                        $UR::Context::all_params_loaded->{$rule_template_id}->{$rule_id} = 0;
                    }
                }

                if ( $query_plan->{rule_matches_all} ) {
                    # No parameters.  We loaded the whole class.
                    # Doing a load w/o a specific ID w/o custom SQL loads the whole class.
                    # Set a flag so that certain optimizations can be made, such as 
                    # short-circuiting future loads of this class.        
                    #
                    # If the key still exists in the all_objects_are_loaded hash, then
                    # we can set it to true.  This is needed in the case where the user
                    # gets an iterator for all the objects of some class, but unloads
                    # one or more of the instances (be calling unload or through the 
                    # cache pruner) before the iterator completes.  If so, _abandon_object()
                    # will have removed the key from the hash
                    if (exists($UR::Context::all_objects_are_loaded->{$class_name})) {
                        $class_name->all_objects_are_loaded(1);
                    }
                }

                if ($recursion_desc) {
                    my @results = $class_name->is_loaded($rule_without_recursion_desc);
                    $UR::Context::all_params_loaded->{$rule_template_id_without_recursion}{$rule_id_without_recursion} = scalar(@results);
                    for my $object (@results) {
                        $object->{__load}->{$rule_template_id_without_recursion}->{$rule_id_without_recursion}++;
                    }
                }

                # Apply changes to all_params_loaded that each importer has collected
                foreach (@object_fabricators) {
                    $_->finalize if $_;
                }

                # If the SQL for the subclassed items was constructed properly, then each
                # of these iterators should be at the end, too.  Call them one more time
                # so they'll finalize their object fabricators.
                foreach my $class ( keys %subordinate_iterator_for_class ) {
                    my $obj = $subordinate_iterator_for_class{$class}->();
                    if ($obj) {
                        # The last time this happened, it was because a get() was done on an abstract
                        # base class with only 'id' as a param.  When the subclassified rule was
                        # turned into SQL in UR::DataSource::QueryPlan()
                        # it removed that one 'id' filter, since it assumed any class with more than
                        # one ID property (usually classes have a named whatever_id property, and an alias 'id'
                        # property) will have a rule that covered both ID properties
                        Carp::carp("Leftover objects in subordinate iterator for $class.  This shouldn't happen, but it's not fatal...");
                        while ($obj = $subordinate_iterator_for_class{$class}->()) {1;}
                    }
                }

                my $retval = $next_object_to_return;
                $next_object_to_return = undef;
                return $retval;
            }

            # we count rows processed mainly for more concise sanity checking
            $rows++;
            # For multi-datasource queries, does this row successfully join with all the other datasources?
            #
            # Normally, the policy is for the data source query to return (possibly) more than what you
            # asked for, and then we'd cache everything that may have been loaded.  In this case, we're
            # making the choice not to.  Reason being that a join across databases is likely to involve
            # a lot of objects, and we don't want to be stuffing our object cache with a lot of things
            # we're not interested in.  FIXME - in order for this to be true, then we could never query
            # these secondary data sources against, say, a calculated property because we're never turning
            # them into objects.  FIXME - fix this by setting the $needs_further_boolexpr_evaluation_after_loading
            # flag maybe?
            my @secondary_data;
            foreach my $callback (@addl_join_comparators) {
                # FIXME - (no, not another one...) There's no mechanism for duplicating SQL join's
                # behavior where if a row from a table joins to 2 rows in the secondary table, the 
                # first table's data will be in the result set twice.
                my $secondary_db_row = $callback->($next_db_row);
                unless (defined $secondary_db_row) {
                    # That data source has no more data, so there can be no more joins even if the
                    # primary data source has more data left to read
                    $db_iterator = undef;
                    $primary_object_for_next_db_row = undef;
                    last LOAD_AN_OBJECT;
                }
                unless ($secondary_db_row) {
                    # It returned 0
                    # didn't join (but there is still more data we can read later)... throw this row out.
                    $primary_object_for_next_db_row = undef;
                    redo LOAD_AN_OBJECT;
                }
                # $next_db_row is a read-only value from DBI, so we need to track our additional 
                # data seperately and smash them together before the object importer is called
                push(@secondary_data, @$secondary_db_row);
            }

            # get one or more objects from this row of results
            my $re_iterate = 0;
            my @imported;
            for (my $i = 0; $i < @object_fabricator_closures; $i++) {
                my $object_fabricator = $object_fabricator_closures[$i];

                # The usual case is that the query is just against one data source, and so the importer
                # callback is just given the row returned from the DB query.  For multiple data sources,
                # we need to smash together the primary and all the secondary lists
                my $imported_object;

                #my $object_creation_time;
                #if ($is_monitor_query) {
                #    $object_creation_time = Time::HiRes::time();
                #}

                if (@secondary_data) {
                    $imported_object = $object_fabricator->([@$next_db_row, @secondary_data]);
                } else {
                    $imported_object = $object_fabricator->($next_db_row);
                }

                #if ($is_monitor_query) {
                #    $self->_log_query_for_rule($class_name, $rule, sprintf("QUERY: object fabricator took %.4f s",Time::HiRes::time() - $object_creation_time));
                #}

                if ($imported_object and not ref($imported_object)) {
                    # object requires sub-classsification in a way which involves different db data.
                    $re_iterate = 1;
                }
                push @imported, $imported_object;

                # If the object ID for fabricator slot $i changes, then we can apply the 
                # all_params_loaded changes from iterators 0 .. $i-1 because we know we've
                # loaded all the hangoff data related to the previous object
                # remember that the last fabricator in the list is for the primary object
                if (defined $imported_object and ref($imported_object)) {
                    if (!defined $object_ids_from_fabricators[$i]) {
                        $object_ids_from_fabricators[$i] = $imported_object->id;
                    } elsif ($object_ids_from_fabricators[$i] ne $imported_object->id) {
                        for (my $j = 0; $j < $i; $j++) {
                            $object_fabricators[$j]->apply_all_params_loaded;
                        }
                        $object_ids_from_fabricators[$i] = $imported_object->id;
                    }
                }
            }

            $primary_object_for_next_db_row = $imported[-1];

            # The object importer will return undef for an object if no object
            # got created for that $next_db_row, and will return a string if the object
            # needs to be subclassed before being returned.  Don't put serial numbers on
            # these
            map { $_->{'__get_serial'} = $this_get_serial }
                grep { defined && ref }
                @imported;

            if ($re_iterate and defined($primary_object_for_next_db_row) and ! ref($primary_object_for_next_db_row)) {
                # It is possible that one or more objects go into subclasses which require more
                # data than is on the results row.  For each subclass (or set of subclasses),
                # we make a more specific, subordinate iterator to delegate-to.

                my $subclass_name = $primary_object_for_next_db_row;

                my $subclass_meta = UR::Object::Type->get(class_name => $subclass_name);
                my $table_subclass = $subclass_meta->most_specific_subclass_with_table();
                my $sub_iterator = $subordinate_iterator_for_class{$table_subclass};
                unless ($sub_iterator) {
                    #print "parallel iteration for loading $subclass_name under $class_name!\n";
                    my $sub_classified_rule_template = $rule_template->sub_classify($subclass_name);
                    my $sub_classified_rule = $sub_classified_rule_template->get_normalized_rule_for_values(@values);
                    $sub_iterator
                        = $subordinate_iterator_for_class{$table_subclass}
                            = $self->_create_import_iterator_for_underlying_context($sub_classified_rule,$dsx,$this_get_serial);
                }
                ($primary_object_for_next_db_row) = $sub_iterator->();
                if (! defined $primary_object_for_next_db_row) {
                    # the newly subclassed object 
                    redo LOAD_AN_OBJECT;
                }

            } # end of handling a possible subordinate iterator delegate

            unless (defined $primary_object_for_next_db_row) {
            #if (!$primary_object_for_next_db_row or $rule->evaluate($primary_object_for_next_db_row)) {
                redo LOAD_AN_OBJECT;
            }

            if ( !$group_by and (ref($primary_object_for_next_db_row) ne $class_name) and (not $primary_object_for_next_db_row->isa($class_name)) ) {
                $primary_object_for_next_db_row = undef;
                redo LOAD_AN_OBJECT;
            }

            if ($by_hand_recursive_source_property) {
                my @values = grep { defined } $primary_object_for_next_db_row->$by_hand_recursive_source_property;
                push @by_hand_recursive_source_values, @values;
            }

            if (! defined($next_object_to_return)
                or (Scalar::Util::refaddr($next_object_to_return) == Scalar::Util::refaddr($primary_object_for_next_db_row))
            ) {
                # The first time through the iterator, we need to buffer the object until
                # $primary_object_for_next_db_row is something different.
                $next_object_to_return = $primary_object_for_next_db_row;
                $primary_object_for_next_db_row = undef;
                redo LOAD_AN_OBJECT;
            }


        } # end of loop until we have a defined object to return

        #foreach my $object_fabricator ( @object_fabricators ) {
        #    # Don't apply all_params_loaded for primary fab until it's all done
        #    next if ($object_fabricator eq $object_fabricators[-1]);
        #    $object_fabricator->apply_all_params_loaded;
        #}

        my $retval = $next_object_to_return;
        $next_object_to_return = $primary_object_for_next_db_row;
        return $retval;
    };

    Sub::Name::subname('UR::Context::__underlying_context_iterator(closure)__', $underlying_context_iterator);
    return $underlying_context_iterator;
}



1;