Commit | Line | Data |
9f98c4b2 |
1 | package # hide from the pauses |
2 | DBIx::Class::ResultSource::RowParser::Util; |
3 | |
4 | use strict; |
5 | use warnings; |
6 | |
ce556881 |
7 | use List::Util 'first'; |
01b25f12 |
8 | use DBIx::Class::_Util 'perlstring'; |
9f98c4b2 |
9 | |
cd784aab |
10 | use constant HAS_DOR => ( $] < 5.010 ? 0 : 1 ); |
11 | |
9f98c4b2 |
12 | use base 'Exporter'; |
13 | our @EXPORT_OK = qw( |
14 | assemble_simple_parser |
15 | assemble_collapsing_parser |
16 | ); |
17 | |
52864fbd |
18 | # working title - we are hoping to extract this eventually... |
19 | our $null_branch_class = 'DBIx::ResultParser::RelatedNullBranch'; |
20 | |
9f7d5590 |
21 | sub __wrap_in_strictured_scope { |
22 | " { use strict; use warnings; use warnings FATAL => 'uninitialized';\n$_[0]\n }" |
23 | } |
24 | |
9f98c4b2 |
25 | sub assemble_simple_parser { |
26 | #my ($args) = @_; |
27 | |
28 | # the non-collapsing assembler is easy |
29 | # FIXME SUBOPTIMAL there could be a yet faster way to do things here, but |
30 | # need to try an actual implementation and benchmark it: |
31 | # |
32 | # <timbunce_> First setup the nested data structure you want for each row |
33 | # Then call bind_col() to alias the row fields into the right place in |
34 | # the data structure, then to fetch the data do: |
35 | # push @rows, dclone($row_data_struct) while ($sth->fetchrow); |
36 | # |
37 | my $parser_src = sprintf('$_ = %s for @{$_[0]}', __visit_infmap_simple($_[0]) ); |
38 | |
39 | # change the quoted placeholders to unquoted alias-references |
40 | $parser_src =~ s/ \' \xFF__VALPOS__(\d+)__\xFF \' /"\$_->[$1]"/gex; |
41 | |
9f7d5590 |
42 | __wrap_in_strictured_scope($parser_src); |
9f98c4b2 |
43 | } |
44 | |
45 | # the simple non-collapsing nested structure recursor |
46 | sub __visit_infmap_simple { |
47 | my $args = shift; |
48 | |
49 | my $my_cols = {}; |
50 | my $rel_cols; |
51 | for (keys %{$args->{val_index}}) { |
52 | if ($_ =~ /^ ([^\.]+) \. (.+) /x) { |
53 | $rel_cols->{$1}{$2} = $args->{val_index}{$_}; |
54 | } |
55 | else { |
56 | $my_cols->{$_} = $args->{val_index}{$_}; |
57 | } |
58 | } |
59 | |
60 | my @relperl; |
61 | for my $rel (sort keys %$rel_cols) { |
62 | |
52864fbd |
63 | my $rel_struct = __visit_infmap_simple({ %$args, |
9f98c4b2 |
64 | val_index => $rel_cols->{$rel}, |
9f98c4b2 |
65 | }); |
66 | |
52864fbd |
67 | if (keys %$my_cols) { |
ce556881 |
68 | |
52864fbd |
69 | my $branch_null_checks = join ' && ', map |
ce556881 |
70 | { "( ! defined '\xFF__VALPOS__${_}__\xFF' )" } |
71 | sort { $a <=> $b } values %{$rel_cols->{$rel}} |
72 | ; |
73 | |
79adc44f |
74 | if ($args->{prune_null_branches}) { |
52864fbd |
75 | $rel_struct = sprintf ( '( (%s) ? undef : %s )', |
76 | $branch_null_checks, |
77 | $rel_struct, |
78 | ); |
79 | } |
80 | else { |
81 | $rel_struct = sprintf ( '( (%s) ? bless( (%s), %s ) : %s )', |
82 | $branch_null_checks, |
83 | $rel_struct, |
84 | perlstring($null_branch_class), |
85 | $rel_struct, |
86 | ); |
87 | } |
ce556881 |
88 | } |
52864fbd |
89 | |
90 | push @relperl, sprintf '( %s => %s )', |
91 | perlstring($rel), |
92 | $rel_struct, |
93 | ; |
94 | |
9f98c4b2 |
95 | } |
96 | |
ce556881 |
97 | my $me_struct; |
a8f62ee0 |
98 | $me_struct = __result_struct_to_source($my_cols) if keys %$my_cols; |
9f98c4b2 |
99 | |
ce556881 |
100 | if ($args->{hri_style}) { |
101 | $me_struct =~ s/^ \s* \{ | \} \s* $//gx |
102 | if $me_struct; |
103 | |
104 | return sprintf '{ %s }', join (', ', $me_struct||(), @relperl); |
105 | } |
106 | else { |
107 | return sprintf '[%s]', join (',', |
108 | $me_struct || 'undef', |
109 | @relperl ? sprintf ('{ %s }', join (',', @relperl)) : (), |
110 | ); |
111 | } |
9f98c4b2 |
112 | } |
113 | |
114 | sub assemble_collapsing_parser { |
115 | my $args = shift; |
116 | |
c863e102 |
117 | my ($top_node_key, $top_node_key_assembler, $variant_idcols); |
9f98c4b2 |
118 | |
119 | if (scalar @{$args->{collapse_map}{-identifying_columns}}) { |
120 | $top_node_key = join ('', map |
121 | { "{'\xFF__IDVALPOS__${_}__\xFF'}" } |
122 | @{$args->{collapse_map}{-identifying_columns}} |
123 | ); |
124 | } |
125 | elsif( my @variants = @{$args->{collapse_map}{-identifying_columns_variants}} ) { |
126 | |
127 | my @path_parts = map { sprintf |
128 | "( ( defined '\xFF__VALPOS__%d__\xFF' ) && (join qq(\xFF), '', %s, '') )", |
ce556881 |
129 | $_->[0], # checking just first is enough - one ID defined, all defined |
c863e102 |
130 | ( join ', ', map { ++$variant_idcols->{$_} and "'\xFF__IDVALPOS__${_}__\xFF'" } @$_ ), |
9f98c4b2 |
131 | } @variants; |
132 | |
133 | my $virtual_column_idx = (scalar keys %{$args->{val_index}} ) + 1; |
134 | |
7596ddca |
135 | $top_node_key = "{'\xFF__IDVALPOS__${virtual_column_idx}__\xFF'}"; |
9f98c4b2 |
136 | |
7596ddca |
137 | $top_node_key_assembler = sprintf "'\xFF__IDVALPOS__%d__\xFF' = (%s);", |
138 | $virtual_column_idx, |
139 | "\n" . join( "\n or\n", @path_parts, qq{"\0\$rows_pos\0"} ) |
140 | ; |
9f98c4b2 |
141 | |
142 | $args->{collapse_map} = { |
143 | %{$args->{collapse_map}}, |
144 | -custom_node_key => $top_node_key, |
145 | }; |
9f98c4b2 |
146 | } |
147 | else { |
148 | die('Unexpected collapse map contents'); |
149 | } |
150 | |
151 | my ($data_assemblers, $stats) = __visit_infmap_collapse ($args); |
152 | |
c863e102 |
153 | # variants do not necessarily overlap with true idcols |
154 | my @row_ids = sort { $a <=> $b } keys %{ { |
155 | %{ $variant_idcols || {} }, |
156 | %{ $stats->{idcols_seen} }, |
157 | } }; |
158 | |
159 | my $row_id_defs = sprintf "\@cur_row_ids{( %s )} = ( \n%s \n );", |
160 | join (', ', @row_ids ), |
161 | # in case we prune - we will never hit undefs/NULLs as pigeon-hole-criteria |
162 | ( $args->{prune_null_branches} |
163 | ? sprintf( '@{$cur_row_data}[( %s )]', join ', ', @row_ids ) |
164 | : join (",\n", map { |
165 | my $quoted_null_val = qq( "\0NULL\xFF\${rows_pos}\xFF${_}\0" ); |
166 | HAS_DOR |
167 | ? qq! ( \$cur_row_data->[$_] // $quoted_null_val ) ! |
168 | : qq! ( defined(\$cur_row_data->[$_]) ? \$cur_row_data->[$_] : $quoted_null_val ) ! |
169 | } @row_ids) |
170 | ) |
171 | ; |
172 | |
173 | my $parser_src = sprintf (<<'EOS', $row_id_defs, $top_node_key_assembler||'', $top_node_key, join( "\n", @{$data_assemblers||[]} ) ); |
9f98c4b2 |
174 | ### BEGIN LITERAL STRING EVAL |
aa1d8a87 |
175 | my $rows_pos = 0; |
c863e102 |
176 | my ($result_pos, @collapse_idx, $cur_row_data, %%cur_row_ids ); |
aa1d8a87 |
177 | |
9f98c4b2 |
178 | # this loop is a bit arcane - the rationale is that the passed in |
179 | # $_[0] will either have only one row (->next) or will have all |
180 | # rows already pulled in (->all and/or unordered). Given that the |
181 | # result can be rather large - we reuse the same already allocated |
182 | # array, since the collapsed prefetch is smaller by definition. |
183 | # At the end we cut the leftovers away and move on. |
3b4cd124 |
184 | while ($cur_row_data = ( |
185 | ( $rows_pos >= 0 and $_[0][$rows_pos++] ) |
186 | or |
187 | ( $_[1] and $rows_pos = -1 and $_[1]->() ) |
188 | ) ) { |
189 | |
c863e102 |
190 | # the undef checks may or may not be there |
5e6d06f4 |
191 | # depending on whether we prune or not |
7596ddca |
192 | # |
9f98c4b2 |
193 | # due to left joins some of the ids may be NULL/undef, and |
194 | # won't play well when used as hash lookups |
195 | # we also need to differentiate NULLs on per-row/per-col basis |
ce556881 |
196 | # (otherwise folding of optional 1:1s will be greatly confused |
c863e102 |
197 | %1$s |
9f98c4b2 |
198 | |
7596ddca |
199 | # in the case of an underdefined root - calculate the virtual id (otherwise no code at all) |
c863e102 |
200 | %2$s |
9f98c4b2 |
201 | |
aa1d8a87 |
202 | # if we were supplied a coderef - we are collapsing lazily (the set |
203 | # is ordered properly) |
204 | # as long as we have a result already and the next result is new we |
205 | # return the pre-read data and bail |
c863e102 |
206 | $_[1] and $result_pos and ! $collapse_idx[0]%3$s and (unshift @{$_[2]}, $cur_row_data) and last; |
9f98c4b2 |
207 | |
208 | # the rel assemblers |
c863e102 |
209 | %4$s |
9f98c4b2 |
210 | |
9f98c4b2 |
211 | } |
212 | |
aa1d8a87 |
213 | $#{$_[0]} = $result_pos - 1; # truncate the passed in array to where we filled it with results |
9f98c4b2 |
214 | ### END LITERAL STRING EVAL |
215 | EOS |
216 | |
217 | # !!! note - different var than the one above |
218 | # change the quoted placeholders to unquoted alias-references |
219 | $parser_src =~ s/ \' \xFF__VALPOS__(\d+)__\xFF \' /"\$cur_row_data->[$1]"/gex; |
06b3406d |
220 | $parser_src =~ s/ |
221 | \' \xFF__IDVALPOS__(\d+)__\xFF \' |
222 | / |
c863e102 |
223 | "\$cur_row_ids{$1}" |
06b3406d |
224 | /gex; |
9f98c4b2 |
225 | |
9f7d5590 |
226 | __wrap_in_strictured_scope($parser_src); |
9f98c4b2 |
227 | } |
228 | |
229 | |
230 | # the collapsing nested structure recursor |
231 | sub __visit_infmap_collapse { |
232 | my $args = {%{ shift() }}; |
233 | |
234 | my $cur_node_idx = ${ $args->{-node_idx_counter} ||= \do { my $x = 0} }++; |
235 | |
ce556881 |
236 | my ($my_cols, $rel_cols) = {}; |
9f98c4b2 |
237 | for ( keys %{$args->{val_index}} ) { |
238 | if ($_ =~ /^ ([^\.]+) \. (.+) /x) { |
239 | $rel_cols->{$1}{$2} = $args->{val_index}{$_}; |
240 | } |
241 | else { |
242 | $my_cols->{$_} = $args->{val_index}{$_}; |
243 | } |
244 | } |
245 | |
ce556881 |
246 | |
ce556881 |
247 | if ($args->{hri_style}) { |
248 | delete $my_cols->{$_} for grep { $rel_cols->{$_} } keys %$my_cols; |
249 | } |
9f98c4b2 |
250 | |
52864fbd |
251 | my $me_struct; |
a8f62ee0 |
252 | $me_struct = __result_struct_to_source($my_cols) if keys %$my_cols; |
ce556881 |
253 | |
52864fbd |
254 | $me_struct = sprintf( '[ %s ]', $me_struct||'' ) |
255 | unless $args->{hri_style}; |
256 | |
257 | |
258 | my $node_key = $args->{collapse_map}->{-custom_node_key} || join ('', map |
259 | { "{'\xFF__IDVALPOS__${_}__\xFF'}" } |
260 | @{$args->{collapse_map}->{-identifying_columns}} |
261 | ); |
ce556881 |
262 | my $node_idx_slot = sprintf '$collapse_idx[%d]%s', $cur_node_idx, $node_key; |
9f98c4b2 |
263 | |
52864fbd |
264 | |
9f98c4b2 |
265 | my @src; |
ce556881 |
266 | |
9f98c4b2 |
267 | if ($cur_node_idx == 0) { |
cd784aab |
268 | push @src, sprintf( '%s %s $_[0][$result_pos++] = %s;', |
9f98c4b2 |
269 | $node_idx_slot, |
cd784aab |
270 | (HAS_DOR ? '//=' : '||='), |
aa1d8a87 |
271 | $me_struct || '{}', |
272 | ); |
9f98c4b2 |
273 | } |
9f98c4b2 |
274 | else { |
ce556881 |
275 | my $parent_attach_slot = sprintf( '$collapse_idx[%d]%s%s{%s}', |
276 | @{$args}{qw/-parent_node_idx -parent_node_key/}, |
277 | $args->{hri_style} ? '' : '[1]', |
a5f5e470 |
278 | perlstring($args->{-node_rel_name}), |
9f98c4b2 |
279 | ); |
ce556881 |
280 | |
281 | if ($args->{collapse_map}->{-is_single}) { |
cd784aab |
282 | push @src, sprintf ( '%s %s %s%s;', |
ce556881 |
283 | $parent_attach_slot, |
cd784aab |
284 | (HAS_DOR ? '//=' : '||='), |
ce556881 |
285 | $node_idx_slot, |
aa1d8a87 |
286 | $me_struct ? " = $me_struct" : '', |
ce556881 |
287 | ); |
288 | } |
289 | else { |
290 | push @src, sprintf('(! %s) and push @{%s}, %s%s;', |
291 | $node_idx_slot, |
292 | $parent_attach_slot, |
293 | $node_idx_slot, |
294 | $me_struct ? " = $me_struct" : '', |
295 | ); |
296 | } |
9f98c4b2 |
297 | } |
298 | |
ce556881 |
299 | my $known_present_ids = { map { $_ => 1 } @{$args->{collapse_map}{-identifying_columns}} }; |
300 | my ($stats, $rel_src); |
301 | |
9f98c4b2 |
302 | for my $rel (sort keys %$rel_cols) { |
303 | |
ce556881 |
304 | my $relinfo = $args->{collapse_map}{$rel}; |
9f98c4b2 |
305 | |
ce556881 |
306 | ($rel_src, $stats->{$rel}) = __visit_infmap_collapse({ %$args, |
9f98c4b2 |
307 | val_index => $rel_cols->{$rel}, |
ce556881 |
308 | collapse_map => $relinfo, |
9f98c4b2 |
309 | -parent_node_idx => $cur_node_idx, |
310 | -parent_node_key => $node_key, |
a5f5e470 |
311 | -node_rel_name => $rel, |
9f98c4b2 |
312 | }); |
313 | |
ce556881 |
314 | my $rel_src_pos = $#src + 1; |
315 | push @src, @$rel_src; |
316 | |
317 | if ( |
ce556881 |
318 | $relinfo->{-is_optional} |
319 | and |
320 | defined ( my $first_distinct_child_idcol = first |
321 | { ! $known_present_ids->{$_} } |
322 | @{$relinfo->{-identifying_columns}} |
323 | ) |
324 | ) { |
325 | |
79adc44f |
326 | if ($args->{prune_null_branches}) { |
ce556881 |
327 | |
7596ddca |
328 | # start of wrap of the entire chain in a conditional |
79adc44f |
329 | splice @src, $rel_src_pos, 0, sprintf "( ! defined %s )\n ? %s%s{%s} = %s\n : do {", |
7596ddca |
330 | "'\xFF__VALPOS__${first_distinct_child_idcol}__\xFF'", |
52864fbd |
331 | $node_idx_slot, |
79adc44f |
332 | $args->{hri_style} ? '' : '[1]', |
52864fbd |
333 | perlstring($rel), |
79adc44f |
334 | ($args->{hri_style} && $relinfo->{-is_single}) ? 'undef' : '[]' |
7596ddca |
335 | ; |
336 | |
337 | # end of wrap |
338 | push @src, '};' |
52864fbd |
339 | } |
340 | else { |
341 | |
342 | splice @src, $rel_src_pos + 1, 0, sprintf ( '(defined %s) or bless (%s[1]{%s}, %s);', |
343 | "'\xFF__VALPOS__${first_distinct_child_idcol}__\xFF'", |
344 | $node_idx_slot, |
345 | perlstring($rel), |
346 | perlstring($null_branch_class), |
347 | ); |
348 | } |
ce556881 |
349 | } |
9f98c4b2 |
350 | } |
351 | |
352 | return ( |
ce556881 |
353 | \@src, |
9f98c4b2 |
354 | { |
355 | idcols_seen => { |
356 | ( map { %{ $_->{idcols_seen} } } values %$stats ), |
357 | ( map { $_ => 1 } @{$args->{collapse_map}->{-identifying_columns}} ), |
358 | } |
359 | } |
360 | ); |
361 | } |
362 | |
a8f62ee0 |
363 | sub __result_struct_to_source { |
364 | sprintf( '{ %s }', join (', ', map |
365 | { sprintf "%s => '\xFF__VALPOS__%d__\xFF'", perlstring($_), $_[0]{$_} } |
366 | sort keys %{$_[0]} |
367 | )); |
9f98c4b2 |
368 | } |
369 | |
370 | 1; |