| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* A node in parse tree. |
| 5 |
* |
| 6 |
* This class represents a node in the parse tree that is produced by WP_Parser. |
| 7 |
* A node corresponds to the related grammar rule that was matched by the parser. |
| 8 |
* Each node can contain children, consisting of other nodes and grammar tokens. |
| 9 |
* In this way, a parser node constitutes a recursive structure that represents |
| 10 |
* a parse (sub)tree at each level of the full grammar tree. |
| 11 |
*/ |
| 12 |
class WP_Parser_Node { |
| 13 |
/** |
| 14 |
* @TODO: Review and document these properties and their visibility. |
| 15 |
*/ |
| 16 |
public $rule_id; |
| 17 |
public $rule_name; |
| 18 |
private $children = array(); |
| 19 |
|
| 20 |
public function __construct( $rule_id, $rule_name ) { |
| 21 |
$this->rule_id = $rule_id; |
| 22 |
$this->rule_name = $rule_name; |
| 23 |
} |
| 24 |
|
| 25 |
public function append_child( $node ) { |
| 26 |
$this->children[] = $node; |
| 27 |
} |
| 28 |
|
| 29 |
/** |
| 30 |
* Flatten the matched rule fragments as if their children were direct |
| 31 |
* descendants of the current rule. |
| 32 |
* |
| 33 |
* What are rule fragments? |
| 34 |
* |
| 35 |
* When we initially parse the grammar file, it has compound rules such |
| 36 |
* as this one: |
| 37 |
* |
| 38 |
* query ::= EOF | ((simpleStatement | beginWork) ((SEMICOLON_SYMBOL EOF?) | EOF)) |
| 39 |
* |
| 40 |
* Building a parser that can understand such rules is way more complex than building |
| 41 |
* a parser that only follows simple rules, so we flatten those compound rules into |
| 42 |
* simpler ones. The above rule would be flattened to: |
| 43 |
* |
| 44 |
* query ::= EOF | %query0 |
| 45 |
* %query0 ::= %%query01 %%query02 |
| 46 |
* %%query01 ::= simpleStatement | beginWork |
| 47 |
* %%query02 ::= SEMICOLON_SYMBOL EOF_zero_or_one | EOF |
| 48 |
* EOF_zero_or_one ::= EOF | ε |
| 49 |
* |
| 50 |
* This factorization happens in "convert-grammar.php". |
| 51 |
* |
| 52 |
* "Fragments" are intermediate artifacts whose names are not in the original grammar. |
| 53 |
* They are extremely useful for the parser, but the API consumer should never have to |
| 54 |
* worry about them. Fragment names start with a percent sign ("%"). |
| 55 |
* |
| 56 |
* The code below inlines every fragment back in its parent rule. |
| 57 |
* |
| 58 |
* We could optimize this. The current $match may be discarded later on so any inlining |
| 59 |
* effort here would be wasted. However, inlining seems cheap and doing it bottom-up here |
| 60 |
* is **much** easier than reprocessing the parse tree top-down later on. |
| 61 |
* |
| 62 |
* The following parse tree: |
| 63 |
* |
| 64 |
* [ |
| 65 |
* 'query' => [ |
| 66 |
* [ |
| 67 |
* '%query01' => [ |
| 68 |
* [ |
| 69 |
* 'simpleStatement' => [ |
| 70 |
* MySQLToken(MySQLLexer::WITH_SYMBOL, 'WITH') |
| 71 |
* ], |
| 72 |
* '%query02' => [ |
| 73 |
* [ |
| 74 |
* 'simpleStatement' => [ |
| 75 |
* MySQLToken(MySQLLexer::WITH_SYMBOL, 'WITH') |
| 76 |
* ] |
| 77 |
* ], |
| 78 |
* ] |
| 79 |
* ] |
| 80 |
* ] |
| 81 |
* ] |
| 82 |
* ] |
| 83 |
* |
| 84 |
* Would be inlined as: |
| 85 |
* |
| 86 |
* [ |
| 87 |
* 'query' => [ |
| 88 |
* [ |
| 89 |
* 'simpleStatement' => [ |
| 90 |
* MySQLToken(MySQLLexer::WITH_SYMBOL, 'WITH') |
| 91 |
* ] |
| 92 |
* ], |
| 93 |
* [ |
| 94 |
* 'simpleStatement' => [ |
| 95 |
* MySQLToken(MySQLLexer::WITH_SYMBOL, 'WITH') |
| 96 |
* ] |
| 97 |
* ] |
| 98 |
* ] |
| 99 |
* ] |
| 100 |
*/ |
| 101 |
public function merge_fragment( $node ) { |
| 102 |
$this->children = array_merge( $this->children, $node->children ); |
| 103 |
} |
| 104 |
|
| 105 |
public function has_child(): bool { |
| 106 |
return count( $this->children ) > 0; |
| 107 |
} |
| 108 |
|
| 109 |
public function has_child_node( ?string $rule_name = null ): bool { |
| 110 |
foreach ( $this->children as $child ) { |
| 111 |
if ( |
| 112 |
$child instanceof WP_Parser_Node |
| 113 |
&& ( null === $rule_name || $child->rule_name === $rule_name ) |
| 114 |
) { |
| 115 |
return true; |
| 116 |
} |
| 117 |
} |
| 118 |
return false; |
| 119 |
} |
| 120 |
|
| 121 |
public function has_child_token( ?int $token_id = null ): bool { |
| 122 |
foreach ( $this->children as $child ) { |
| 123 |
if ( |
| 124 |
$child instanceof WP_Parser_Token |
| 125 |
&& ( null === $token_id || $child->id === $token_id ) |
| 126 |
) { |
| 127 |
return true; |
| 128 |
} |
| 129 |
} |
| 130 |
return false; |
| 131 |
} |
| 132 |
|
| 133 |
|
| 134 |
public function get_first_child() { |
| 135 |
return $this->children[0] ?? null; |
| 136 |
} |
| 137 |
|
| 138 |
public function get_first_child_node( ?string $rule_name = null ): ?WP_Parser_Node { |
| 139 |
foreach ( $this->children as $child ) { |
| 140 |
if ( |
| 141 |
$child instanceof WP_Parser_Node |
| 142 |
&& ( null === $rule_name || $child->rule_name === $rule_name ) |
| 143 |
) { |
| 144 |
return $child; |
| 145 |
} |
| 146 |
} |
| 147 |
return null; |
| 148 |
} |
| 149 |
|
| 150 |
public function get_first_child_token( ?int $token_id = null ): ?WP_Parser_Token { |
| 151 |
foreach ( $this->children as $child ) { |
| 152 |
if ( |
| 153 |
$child instanceof WP_Parser_Token |
| 154 |
&& ( null === $token_id || $child->id === $token_id ) |
| 155 |
) { |
| 156 |
return $child; |
| 157 |
} |
| 158 |
} |
| 159 |
return null; |
| 160 |
} |
| 161 |
|
| 162 |
public function get_first_descendant_node( ?string $rule_name = null ): ?WP_Parser_Node { |
| 163 |
$nodes = array( $this ); |
| 164 |
while ( count( $nodes ) ) { |
| 165 |
$node = array_shift( $nodes ); |
| 166 |
$child = $node->get_first_child_node( $rule_name ); |
| 167 |
if ( $child ) { |
| 168 |
return $child; |
| 169 |
} |
| 170 |
$children = $node->get_child_nodes(); |
| 171 |
if ( count( $children ) > 0 ) { |
| 172 |
array_push( $nodes, ...$children ); |
| 173 |
} |
| 174 |
} |
| 175 |
return null; |
| 176 |
} |
| 177 |
|
| 178 |
public function get_first_descendant_token( ?int $token_id = null ): ?WP_Parser_Token { |
| 179 |
$nodes = array( $this ); |
| 180 |
while ( count( $nodes ) ) { |
| 181 |
$node = array_shift( $nodes ); |
| 182 |
$child = $node->get_first_child_token( $token_id ); |
| 183 |
if ( $child ) { |
| 184 |
return $child; |
| 185 |
} |
| 186 |
$children = $node->get_child_nodes(); |
| 187 |
if ( count( $children ) > 0 ) { |
| 188 |
array_push( $nodes, ...$children ); |
| 189 |
} |
| 190 |
} |
| 191 |
return null; |
| 192 |
} |
| 193 |
|
| 194 |
public function get_children(): array { |
| 195 |
return $this->children; |
| 196 |
} |
| 197 |
|
| 198 |
public function get_child_nodes( ?string $rule_name = null ): array { |
| 199 |
$nodes = array(); |
| 200 |
foreach ( $this->children as $child ) { |
| 201 |
if ( |
| 202 |
$child instanceof WP_Parser_Node |
| 203 |
&& ( null === $rule_name || $child->rule_name === $rule_name ) |
| 204 |
) { |
| 205 |
$nodes[] = $child; |
| 206 |
} |
| 207 |
} |
| 208 |
return $nodes; |
| 209 |
} |
| 210 |
|
| 211 |
public function get_child_tokens( ?int $token_id = null ): array { |
| 212 |
$tokens = array(); |
| 213 |
foreach ( $this->children as $child ) { |
| 214 |
if ( |
| 215 |
$child instanceof WP_Parser_Token |
| 216 |
&& ( null === $token_id || $child->id === $token_id ) |
| 217 |
) { |
| 218 |
$tokens[] = $child; |
| 219 |
} |
| 220 |
} |
| 221 |
return $tokens; |
| 222 |
} |
| 223 |
|
| 224 |
public function get_descendants(): array { |
| 225 |
$nodes = array( $this ); |
| 226 |
$all_descendants = array(); |
| 227 |
while ( count( $nodes ) ) { |
| 228 |
$node = array_shift( $nodes ); |
| 229 |
$all_descendants = array_merge( $all_descendants, $node->get_children() ); |
| 230 |
$children = $node->get_child_nodes(); |
| 231 |
if ( count( $children ) > 0 ) { |
| 232 |
array_push( $nodes, ...$children ); |
| 233 |
} |
| 234 |
} |
| 235 |
return $all_descendants; |
| 236 |
} |
| 237 |
|
| 238 |
public function get_descendant_nodes( ?string $rule_name = null ): array { |
| 239 |
$nodes = array( $this ); |
| 240 |
$all_descendants = array(); |
| 241 |
while ( count( $nodes ) ) { |
| 242 |
$node = array_shift( $nodes ); |
| 243 |
$all_descendants = array_merge( $all_descendants, $node->get_child_nodes( $rule_name ) ); |
| 244 |
$children = $node->get_child_nodes(); |
| 245 |
if ( count( $children ) > 0 ) { |
| 246 |
array_push( $nodes, ...$children ); |
| 247 |
} |
| 248 |
} |
| 249 |
return $all_descendants; |
| 250 |
} |
| 251 |
|
| 252 |
public function get_descendant_tokens( ?int $token_id = null ): array { |
| 253 |
$nodes = array( $this ); |
| 254 |
$all_descendants = array(); |
| 255 |
while ( count( $nodes ) ) { |
| 256 |
$node = array_shift( $nodes ); |
| 257 |
$all_descendants = array_merge( $all_descendants, $node->get_child_tokens( $token_id ) ); |
| 258 |
$children = $node->get_child_nodes(); |
| 259 |
if ( count( $children ) > 0 ) { |
| 260 |
array_push( $nodes, ...$children ); |
| 261 |
} |
| 262 |
} |
| 263 |
return $all_descendants; |
| 264 |
} |
| 265 |
|
| 266 |
/** |
| 267 |
* Get the byte offset in the input SQL string where this node begins. |
| 268 |
* |
| 269 |
* @return int |
| 270 |
*/ |
| 271 |
public function get_start(): int { |
| 272 |
return $this->get_first_descendant_token()->start; |
| 273 |
} |
| 274 |
|
| 275 |
/** |
| 276 |
* Get the byte length of this node in the input SQL string. |
| 277 |
* |
| 278 |
* @return int |
| 279 |
*/ |
| 280 |
public function get_length(): int { |
| 281 |
$tokens = $this->get_descendant_tokens(); |
| 282 |
$last_token = end( $tokens ); |
| 283 |
$start = $this->get_start(); |
| 284 |
return $last_token->start + $last_token->length - $start; |
| 285 |
} |
| 286 |
|
| 287 |
/* |
| 288 |
* @TODO: Let's implement a more powerful AST-querying API. |
| 289 |
* See: https://github.com/WordPress/sqlite-database-integration/pull/164#discussion_r1855230501 |
| 290 |
*/ |
| 291 |
} |
| 292 |
|