Commit | Line | Data |
01dd4e4f |
1 | package SQL::Abstract::Tree; |
2 | |
3 | use strict; |
4 | use warnings; |
5 | use Carp; |
6 | |
1536de15 |
7 | |
1536de15 |
8 | use base 'Class::Accessor::Grouped'; |
9 | |
10 | __PACKAGE__->mk_group_accessors( simple => $_ ) for qw( |
11 | newline indent_string indent_amount colormap indentmap |
12 | ); |
13 | |
01dd4e4f |
14 | # Parser states for _recurse_parse() |
15 | use constant PARSE_TOP_LEVEL => 0; |
16 | use constant PARSE_IN_EXPR => 1; |
17 | use constant PARSE_IN_PARENS => 2; |
18 | use constant PARSE_RHS => 3; |
19 | |
20 | # These SQL keywords always signal end of the current expression (except inside |
21 | # of a parenthesized subexpression). |
22 | # Format: A list of strings that will be compiled to extended syntax (ie. |
23 | # /.../x) regexes, without capturing parentheses. They will be automatically |
24 | # anchored to word boundaries to match the whole token). |
25 | my @expression_terminator_sql_keywords = ( |
26 | 'SELECT', |
27 | 'FROM', |
28 | '(?: |
29 | (?: |
30 | (?: \b (?: LEFT | RIGHT | FULL ) \s+ )? |
31 | (?: \b (?: CROSS | INNER | OUTER ) \s+ )? |
32 | )? |
33 | JOIN |
34 | )', |
35 | 'ON', |
36 | 'WHERE', |
37 | 'EXISTS', |
38 | 'GROUP \s+ BY', |
39 | 'HAVING', |
40 | 'ORDER \s+ BY', |
41 | 'LIMIT', |
42 | 'OFFSET', |
43 | 'FOR', |
44 | 'UNION', |
45 | 'INTERSECT', |
46 | 'EXCEPT', |
47 | 'RETURNING', |
48 | ); |
49 | |
50 | # These are binary operator keywords always a single LHS and RHS |
51 | # * AND/OR are handled separately as they are N-ary |
52 | # * so is NOT as being unary |
53 | # * BETWEEN without paranthesis around the ANDed arguments (which |
54 | # makes it a non-binary op) is detected and accomodated in |
55 | # _recurse_parse() |
56 | my $stuff_around_mathops = qr/[\w\s\`\'\"\)]/; |
57 | my @binary_op_keywords = ( |
58 | ( map |
59 | { |
60 | ' ^ ' . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ", |
61 | " (?<= $stuff_around_mathops)" . quotemeta ($_) . "(?= \$ | $stuff_around_mathops ) ", |
62 | } |
63 | (qw/< > != <> = <= >=/) |
64 | ), |
65 | ( map |
66 | { '\b (?: NOT \s+)?' . $_ . '\b' } |
67 | (qw/IN BETWEEN LIKE/) |
68 | ), |
69 | ); |
70 | |
71 | my $tokenizer_re_str = join("\n\t|\n", |
72 | ( map { '\b' . $_ . '\b' } @expression_terminator_sql_keywords, 'AND', 'OR', 'NOT'), |
73 | @binary_op_keywords, |
74 | ); |
75 | |
76 | my $tokenizer_re = qr/ \s* ( $tokenizer_re_str | \( | \) | \? ) \s* /xi; |
77 | |
78 | sub _binary_op_keywords { @binary_op_keywords } |
79 | |
7e5600e9 |
80 | my %indents = ( |
1bb3956e |
81 | select => 0, |
82 | where => 1, |
83 | from => 1, |
84 | join => 1, |
85 | on => 2, |
86 | 'group by' => 1, |
7e5600e9 |
87 | ); |
88 | |
75c3a063 |
89 | my %profiles = ( |
90 | console => { |
1536de15 |
91 | indent_string => ' ', |
75c3a063 |
92 | indent_amount => 2, |
1536de15 |
93 | newline => "\n", |
3be357b0 |
94 | colormap => {}, |
7e5600e9 |
95 | indentmap => { %indents }, |
3be357b0 |
96 | }, |
97 | console_monochrome => { |
98 | indent_string => ' ', |
99 | indent_amount => 2, |
100 | newline => "\n", |
101 | colormap => {}, |
7e5600e9 |
102 | indentmap => { %indents }, |
103 | }, |
104 | html => { |
105 | indent_string => ' ', |
106 | indent_amount => 2, |
107 | newline => "<br />\n", |
108 | colormap => { |
1bb3956e |
109 | select => ['<span class="select">' , '</span>'], |
110 | where => ['<span class="where">' , '</span>'], |
111 | from => ['<span class="from">' , '</span>'], |
112 | join => ['<span class="join">' , '</span>'], |
113 | on => ['<span class="on">' , '</span>'], |
114 | 'group by' => ['<span class="group-by">', '</span>'], |
1536de15 |
115 | }, |
7e5600e9 |
116 | indentmap => { %indents }, |
75c3a063 |
117 | }, |
118 | none => { |
1536de15 |
119 | colormap => {}, |
120 | indentmap => {}, |
75c3a063 |
121 | }, |
122 | ); |
123 | |
3be357b0 |
124 | eval { |
125 | require Term::ANSIColor; |
126 | $profiles{console}->{colormap} = { |
127 | select => [Term::ANSIColor::color('red'), Term::ANSIColor::color('reset')], |
128 | where => [Term::ANSIColor::color('green'), Term::ANSIColor::color('reset')], |
129 | from => [Term::ANSIColor::color('cyan'), Term::ANSIColor::color('reset')], |
130 | }; |
131 | }; |
132 | |
75c3a063 |
133 | sub new { |
134 | my ($class, $args) = @_; |
135 | |
136 | my $profile = delete $args->{profile} || 'none'; |
137 | my $data = {%{$profiles{$profile}}, %{$args||{}}}; |
138 | |
139 | bless $data, $class |
140 | } |
d695b0ad |
141 | |
01dd4e4f |
142 | sub parse { |
d695b0ad |
143 | my ($self, $s) = @_; |
01dd4e4f |
144 | |
145 | # tokenize string, and remove all optional whitespace |
146 | my $tokens = []; |
147 | foreach my $token (split $tokenizer_re, $s) { |
148 | push @$tokens, $token if (length $token) && ($token =~ /\S/); |
149 | } |
150 | |
d695b0ad |
151 | my $tree = $self->_recurse_parse($tokens, PARSE_TOP_LEVEL); |
01dd4e4f |
152 | return $tree; |
153 | } |
154 | |
155 | sub _recurse_parse { |
d695b0ad |
156 | my ($self, $tokens, $state) = @_; |
01dd4e4f |
157 | |
158 | my $left; |
159 | while (1) { # left-associative parsing |
160 | |
161 | my $lookahead = $tokens->[0]; |
162 | if ( not defined($lookahead) |
163 | or |
164 | ($state == PARSE_IN_PARENS && $lookahead eq ')') |
165 | or |
166 | ($state == PARSE_IN_EXPR && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords ) ) |
167 | or |
168 | ($state == PARSE_RHS && grep { $lookahead =~ /^ $_ $/xi } ('\)', @expression_terminator_sql_keywords, @binary_op_keywords, 'AND', 'OR', 'NOT' ) ) |
169 | ) { |
170 | return $left; |
171 | } |
172 | |
173 | my $token = shift @$tokens; |
174 | |
175 | # nested expression in () |
176 | if ($token eq '(' ) { |
d695b0ad |
177 | my $right = $self->_recurse_parse($tokens, PARSE_IN_PARENS); |
178 | $token = shift @$tokens or croak "missing closing ')' around block " . $self->unparse($right); |
179 | $token eq ')' or croak "unexpected token '$token' terminating block " . $self->unparse($right); |
01dd4e4f |
180 | |
181 | $left = $left ? [@$left, [PAREN => [$right] ]] |
182 | : [PAREN => [$right] ]; |
183 | } |
184 | # AND/OR |
185 | elsif ($token =~ /^ (?: OR | AND ) $/xi ) { |
186 | my $op = uc $token; |
d695b0ad |
187 | my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
188 | |
189 | # Merge chunks if logic matches |
190 | if (ref $right and $op eq $right->[0]) { |
191 | $left = [ (shift @$right ), [$left, map { @$_ } @$right] ]; |
192 | } |
193 | else { |
194 | $left = [$op => [$left, $right]]; |
195 | } |
196 | } |
197 | # binary operator keywords |
198 | elsif (grep { $token =~ /^ $_ $/xi } @binary_op_keywords ) { |
199 | my $op = uc $token; |
d695b0ad |
200 | my $right = $self->_recurse_parse($tokens, PARSE_RHS); |
01dd4e4f |
201 | |
202 | # A between with a simple LITERAL for a 1st RHS argument needs a |
203 | # rerun of the search to (hopefully) find the proper AND construct |
204 | if ($op eq 'BETWEEN' and $right->[0] eq 'LITERAL') { |
205 | unshift @$tokens, $right->[1][0]; |
d695b0ad |
206 | $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
207 | } |
208 | |
209 | $left = [$op => [$left, $right] ]; |
210 | } |
211 | # expression terminator keywords (as they start a new expression) |
212 | elsif (grep { $token =~ /^ $_ $/xi } @expression_terminator_sql_keywords ) { |
213 | my $op = uc $token; |
d695b0ad |
214 | my $right = $self->_recurse_parse($tokens, PARSE_IN_EXPR); |
01dd4e4f |
215 | $left = $left ? [ $left, [$op => [$right] ]] |
216 | : [ $op => [$right] ]; |
217 | } |
218 | # NOT (last as to allow all other NOT X pieces first) |
219 | elsif ( $token =~ /^ not $/ix ) { |
220 | my $op = uc $token; |
d695b0ad |
221 | my $right = $self->_recurse_parse ($tokens, PARSE_RHS); |
01dd4e4f |
222 | $left = $left ? [ @$left, [$op => [$right] ]] |
223 | : [ $op => [$right] ]; |
224 | |
225 | } |
226 | # literal (eat everything on the right until RHS termination) |
227 | else { |
d695b0ad |
228 | my $right = $self->_recurse_parse ($tokens, PARSE_RHS); |
229 | $left = $left ? [ $left, [LITERAL => [join ' ', $token, $self->unparse($right)||()] ] ] |
230 | : [ LITERAL => [join ' ', $token, $self->unparse($right)||()] ]; |
01dd4e4f |
231 | } |
232 | } |
233 | } |
234 | |
d695b0ad |
235 | sub format_keyword { |
236 | my ($self, $keyword) = @_; |
237 | |
1536de15 |
238 | if (my $around = $self->colormap->{lc $keyword}) { |
d695b0ad |
239 | $keyword = "$around->[0]$keyword$around->[1]"; |
240 | } |
241 | |
242 | return $keyword |
243 | } |
244 | |
a24cc3a0 |
245 | sub whitespace { |
246 | my ($self, $keyword, $depth) = @_; |
e171c446 |
247 | |
248 | my $before = ''; |
1536de15 |
249 | my $after = ' '; |
250 | if (defined $self->indentmap->{lc $keyword}) { |
251 | $before = $self->newline . $self->indent($depth + $self->indentmap->{lc $keyword}); |
a24cc3a0 |
252 | } |
1a67e3a0 |
253 | $before = '' if $depth == 0 and lc $keyword eq 'select'; |
e171c446 |
254 | return [$before, $after]; |
a24cc3a0 |
255 | } |
256 | |
1536de15 |
257 | sub indent { ($_[0]->indent_string||'') x ( ( $_[0]->indent_amount || 0 ) * $_[1] ) } |
a24cc3a0 |
258 | |
0569a14f |
259 | sub _is_select { |
260 | my $tree = shift; |
261 | $tree = $tree->[0] while ref $tree; |
262 | |
ca8aec12 |
263 | defined $tree && lc $tree eq 'select'; |
0569a14f |
264 | } |
265 | |
01dd4e4f |
266 | sub unparse { |
a24cc3a0 |
267 | my ($self, $tree, $depth) = @_; |
268 | |
e171c446 |
269 | $depth ||= 0; |
01dd4e4f |
270 | |
271 | if (not $tree ) { |
272 | return ''; |
273 | } |
a24cc3a0 |
274 | |
275 | my $car = $tree->[0]; |
276 | my $cdr = $tree->[1]; |
277 | |
278 | if (ref $car) { |
e171c446 |
279 | return join ('', map $self->unparse($_, $depth), @$tree); |
01dd4e4f |
280 | } |
a24cc3a0 |
281 | elsif ($car eq 'LITERAL') { |
282 | return $cdr->[0]; |
01dd4e4f |
283 | } |
a24cc3a0 |
284 | elsif ($car eq 'PAREN') { |
e171c446 |
285 | return '(' . |
a24cc3a0 |
286 | join(' ', |
0569a14f |
287 | map $self->unparse($_, $depth + 2), @{$cdr}) . |
1536de15 |
288 | (_is_select($cdr)?( $self->newline||'' ).$self->indent($depth + 1):'') . ')'; |
01dd4e4f |
289 | } |
a24cc3a0 |
290 | elsif ($car eq 'OR' or $car eq 'AND' or (grep { $car =~ /^ $_ $/xi } @binary_op_keywords ) ) { |
e171c446 |
291 | return join (" $car ", map $self->unparse($_, $depth), @{$cdr}); |
01dd4e4f |
292 | } |
293 | else { |
a24cc3a0 |
294 | my ($l, $r) = @{$self->whitespace($car, $depth)}; |
e171c446 |
295 | return sprintf "$l%s %s$r", $self->format_keyword($car), $self->unparse($cdr, $depth); |
01dd4e4f |
296 | } |
297 | } |
298 | |
d695b0ad |
299 | sub format { my $self = shift; $self->unparse($self->parse(@_)) } |
01dd4e4f |
300 | |
301 | 1; |
302 | |
3be357b0 |
303 | =pod |
304 | |
305 | =head1 SYNOPSIS |
306 | |
307 | my $sqla_tree = SQL::Abstract::Tree->new({ profile => 'console' }); |
308 | |
309 | print $sqla_tree->format('SELECT * FROM foo WHERE foo.a > 2'); |
310 | |
311 | # SELECT * |
312 | # FROM foo |
313 | # WHERE foo.a > 2 |
314 | |