rename DBIC::Storage::PP and get ready to re-release
[scpubgit/Q-Branch.git] / lib / SQL / Abstract / Tree.pm
CommitLineData
01dd4e4f 1package SQL::Abstract::Tree;
2
3use strict;
4use warnings;
5use Carp;
6
a97eb57c 7use List::Util;
1536de15 8
1536de15 9use base 'Class::Accessor::Grouped';
10
11__PACKAGE__->mk_group_accessors( simple => $_ ) for qw(
12 newline indent_string indent_amount colormap indentmap
13);
14
01dd4e4f 15# Parser states for _recurse_parse()
16use constant PARSE_TOP_LEVEL => 0;
17use constant PARSE_IN_EXPR => 1;
18use constant PARSE_IN_PARENS => 2;
19use constant PARSE_RHS => 3;
20
21# These SQL keywords always signal end of the current expression (except inside
22# of a parenthesized subexpression).
23# Format: A list of strings that will be compiled to extended syntax (ie.
24# /.../x) regexes, without capturing parentheses. They will be automatically
25# anchored to word boundaries to match the whole token).
26my @expression_terminator_sql_keywords = (
27 'SELECT',
7853a177 28 'UPDATE',
29 'INSERT \s+ INTO',
30 'DELETE \s+ FROM',
3d910890 31 'FROM',
7853a177 32 'SET',
01dd4e4f 33 '(?:
34 (?:
35 (?: \b (?: LEFT | RIGHT | FULL ) \s+ )?
36 (?: \b (?: CROSS | INNER | OUTER ) \s+ )?
37 )?
38 JOIN
39 )',
40 'ON',
41 'WHERE',
7853a177 42 'VALUES',
01dd4e4f 43 'EXISTS',
44 'GROUP \s+ BY',
45 'HAVING',
46 'ORDER \s+ BY',
47 'LIMIT',
48 'OFFSET',
49 'FOR',
50 'UNION',
51 'INTERSECT',
52 'EXCEPT',
53 'RETURNING',
8d0dd7dc 54 'ROW_NUMBER \s* \( \s* \) \s+ OVER',
01dd4e4f 55);
56
57# These are binary operator keywords always a single LHS and RHS
58# * AND/OR are handled separately as they are N-ary
59# * so is NOT as being unary
60# * BETWEEN without paranthesis around the ANDed arguments (which
61# makes it a non-binary op) is detected and accomodated in
62# _recurse_parse()
63my $stuff_around_mathops = qr/[\w\s\`\'\"\)]/;
64my @binary_op_keywords = (
65 ( map
66 {
67 ' ^ ' . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ",
68 " (?<= $stuff_around_mathops)" . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ",
69 }
70 (qw/< > != <> = <= >=/)
71 ),
72 ( map
73 { '\b (?: NOT \s+)?' . $_ . '\b' }
74 (qw/IN BETWEEN LIKE/)
75 ),
76);
77
78my $tokenizer_re_str = join("\n\t|\n",
79 ( map { '\b' . $_ . '\b' } @expression_terminator_sql_keywords, 'AND', 'OR', 'NOT'),
80 @binary_op_keywords,
81);
82
83my $tokenizer_re = qr/ \s* ( $tokenizer_re_str | \( | \) | \? ) \s* /xi;
84
85sub _binary_op_keywords { @binary_op_keywords }
86
7e5600e9 87my %indents = (
7853a177 88 select => 0,
89 update => 0,
90 'insert into' => 0,
91 'delete from' => 0,
3d910890 92 from => 1,
91916220 93 where => 0,
7853a177 94 join => 1,
95 'left join' => 1,
96 on => 2,
91916220 97 'group by' => 0,
98 'order by' => 0,
7853a177 99 set => 1,
100 into => 1,
91916220 101 values => 1,
7e5600e9 102);
103
75c3a063 104my %profiles = (
105 console => {
1536de15 106 indent_string => ' ',
75c3a063 107 indent_amount => 2,
1536de15 108 newline => "\n",
3be357b0 109 colormap => {},
7e5600e9 110 indentmap => { %indents },
3be357b0 111 },
112 console_monochrome => {
113 indent_string => ' ',
114 indent_amount => 2,
115 newline => "\n",
116 colormap => {},
7e5600e9 117 indentmap => { %indents },
118 },
119 html => {
120 indent_string => '&nbsp;',
121 indent_amount => 2,
122 newline => "<br />\n",
123 colormap => {
7853a177 124 select => ['<span class="select">' , '</span>'],
125 'insert into' => ['<span class="insert-into">' , '</span>'],
126 update => ['<span class="select">' , '</span>'],
127 'delete from' => ['<span class="delete-from">' , '</span>'],
128 where => ['<span class="where">' , '</span>'],
129 from => ['<span class="from">' , '</span>'],
130 join => ['<span class="join">' , '</span>'],
131 on => ['<span class="on">' , '</span>'],
132 'group by' => ['<span class="group-by">', '</span>'],
133 'order by' => ['<span class="order-by">', '</span>'],
134 set => ['<span class="set">', '</span>'],
135 into => ['<span class="into">', '</span>'],
136 values => ['<span class="values">', '</span>'],
1536de15 137 },
7e5600e9 138 indentmap => { %indents },
75c3a063 139 },
140 none => {
1536de15 141 colormap => {},
142 indentmap => {},
75c3a063 143 },
144);
145
3be357b0 146eval {
147 require Term::ANSIColor;
148 $profiles{console}->{colormap} = {
7853a177 149 select => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')],
150 'insert into' => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')],
151 update => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')],
152 'delete from' => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')],
153
154 set => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')],
c1b89c4f 155 from => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')],
7853a177 156
157 where => [Term::ANSIColor::color('green'), Term::ANSIColor::color('reset')],
158 values => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')],
159
160 join => [Term::ANSIColor::color('magenta'), Term::ANSIColor::color('reset')],
161 'left join' => [Term::ANSIColor::color('magenta'), Term::ANSIColor::color('reset')],
162 on => [Term::ANSIColor::color('blue'), Term::ANSIColor::color('reset')],
163
164 'group by' => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')],
165 'order by' => [Term::ANSIColor::color('yellow'), Term::ANSIColor::color('reset')],
3be357b0 166 };
167};
168
75c3a063 169sub new {
170 my ($class, $args) = @_;
171
172 my $profile = delete $args->{profile} || 'none';
173 my $data = {%{$profiles{$profile}}, %{$args||{}}};
174
175 bless $data, $class
176}
d695b0ad 177
01dd4e4f 178sub parse {
d695b0ad 179 my ($self, $s) = @_;
01dd4e4f 180
181 # tokenize string, and remove all optional whitespace
182 my $tokens = [];
183 foreach my $token (split $tokenizer_re, $s) {
184 push @$tokens, $token if (length $token) && ($token =~ /\S/);
185 }
186
d695b0ad 187 my $tree = $self->_recurse_parse($tokens, PARSE_TOP_LEVEL);
01dd4e4f 188 return $tree;
189}
190
191sub _recurse_parse {
d695b0ad 192 my ($self, $tokens, $state) = @_;
01dd4e4f 193
194 my $left;
195 while (1) { # left-associative parsing
196
197 my $lookahead = $tokens->[0];
198 if ( not defined($lookahead)
199 or
200 ($state == PARSE_IN_PARENS && $lookahead eq ')')
201 or
202 ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords ) )
203 or
204 ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords, @binary_op_keywords, 'AND', 'OR', 'NOT' ) )
205 ) {
206 return $left;
207 }
208
209 my $token = shift @$tokens;
210
211 # nested expression in ()
212 if ($token eq '(' ) {
d695b0ad 213 my $right = $self->_recurse_parse($tokens, PARSE_IN_PARENS);
214 $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse($right);
215 $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse($right);
01dd4e4f 216
217 $left = $left ? [@$left, [PAREN => [$right] ]]
218 : [PAREN => [$right] ];
219 }
220 # AND/OR
221 elsif ($token =~ /^ (?: OR | AND ) $/xi ) {
222 my $op = uc $token;
d695b0ad 223 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
01dd4e4f 224
225 # Merge chunks if logic matches
226 if (ref $right and $op eq $right->[0]) {
227 $left = [ (shift @$right ), [$left, map { @$_ } @$right] ];
228 }
229 else {
230 $left = [$op => [$left, $right]];
231 }
232 }
233 # binary operator keywords
234 elsif (grep { $token =~ /^ $_ $/xi } @binary_op_keywords ) {
235 my $op = uc $token;
d695b0ad 236 my $right = $self->_recurse_parse($tokens, PARSE_RHS);
01dd4e4f 237
238 # A between with a simple LITERAL for a 1st RHS argument needs a
239 # rerun of the search to (hopefully) find the proper AND construct
240 if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') {
241 unshift @$tokens, $right->[1][0];
d695b0ad 242 $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
01dd4e4f 243 }
244
245 $left = [$op => [$left, $right] ];
246 }
247 # expression terminator keywords (as they start a new expression)
248 elsif (grep { $token =~ /^ $_ $/xi } @expression_terminator_sql_keywords ) {
249 my $op = uc $token;
d695b0ad 250 my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR);
01dd4e4f 251 $left = $left ? [ $left, [$op => [$right] ]]
252 : [ $op => [$right] ];
253 }
254 # NOT (last as to allow all other NOT X pieces first)
255 elsif ( $token =~ /^ not $/ix ) {
256 my $op = uc $token;
d695b0ad 257 my $right = $self->_recurse_parse ($tokens, PARSE_RHS);
01dd4e4f 258 $left = $left ? [ @$left, [$op => [$right] ]]
259 : [ $op => [$right] ];
260
261 }
262 # literal (eat everything on the right until RHS termination)
263 else {
d695b0ad 264 my $right = $self->_recurse_parse ($tokens, PARSE_RHS);
265 $left = $left ? [ $left, [LITERAL => [join ' ', $token, $self->unparse($right)||()] ] ]
266 : [ LITERAL => [join ' ', $token, $self->unparse($right)||()] ];
01dd4e4f 267 }
268 }
269}
270
d695b0ad 271sub format_keyword {
272 my ($self, $keyword) = @_;
273
1536de15 274 if (my $around = $self->colormap->{lc $keyword}) {
d695b0ad 275 $keyword = "$around->[0]$keyword$around->[1]";
276 }
277
278 return $keyword
279}
280
a24cc3a0 281sub whitespace {
282 my ($self, $keyword, $depth) = @_;
e171c446 283
284 my $before = '';
1536de15 285 if (defined $self->indentmap->{lc $keyword}) {
286 $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword});
a24cc3a0 287 }
1a67e3a0 288 $before = '' if $depth == 0 and lc $keyword eq 'select';
b4e0e260 289 return [$before, ' '];
a24cc3a0 290}
291
1536de15 292sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) }
a24cc3a0 293
a97eb57c 294sub _is_key {
295 my ($self, $tree) = @_;
0569a14f 296 $tree = $tree->[0] while ref $tree;
297
a97eb57c 298 defined $tree && defined $self->indentmap->{lc $tree};
0569a14f 299}
300
01dd4e4f 301sub unparse {
a24cc3a0 302 my ($self, $tree, $depth) = @_;
303
e171c446 304 $depth ||= 0;
01dd4e4f 305
306 if (not $tree ) {
307 return '';
308 }
a24cc3a0 309
310 my $car = $tree->[0];
311 my $cdr = $tree->[1];
312
313 if (ref $car) {
e171c446 314 return join ('', map $self->unparse($_, $depth), @$tree);
01dd4e4f 315 }
a24cc3a0 316 elsif ($car eq 'LITERAL') {
317 return $cdr->[0];
01dd4e4f 318 }
a24cc3a0 319 elsif ($car eq 'PAREN') {
e171c446 320 return '(' .
a24cc3a0 321 join(' ',
0569a14f 322 map $self->unparse($_, $depth + 2), @{$cdr}) .
a97eb57c 323 ($self->_is_key($cdr)?( $self->newline||'' ).$self->indent($depth + 1):'') . ') ';
01dd4e4f 324 }
a24cc3a0 325 elsif ($car eq 'OR' or $car eq 'AND' or (grep { $car =~ /^ $_ $/xi } @binary_op_keywords ) ) {
e171c446 326 return join (" $car ", map $self->unparse($_, $depth), @{$cdr});
01dd4e4f 327 }
328 else {
a24cc3a0 329 my ($l, $r) = @{$self->whitespace($car, $depth)};
e171c446 330 return sprintf "$l%s %s$r", $self->format_keyword($car), $self->unparse($cdr, $depth);
01dd4e4f 331 }
332}
333
d695b0ad 334sub format { my $self = shift; $self->unparse($self->parse(@_)) }
01dd4e4f 335
3361;
337
3be357b0 338=pod
339
340=head1 SYNOPSIS
341
342 my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' });
343
344 print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2');
345
346 # SELECT *
347 # FROM foo
348 # WHERE foo.a > 2
349
6b1bf9f8 350=head1 METHODS
351
352=head2 new
353
354 my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' });
355
356=head2 format
357
358 $sqlat->format('SELECT * FROM bar')
359
360Returns a formatting string based on wthe string passed in