Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 41 |
|
0.00% |
0 / 13 |
CRAP | |
0.00% |
0 / 1 |
| DomSourceRange | |
0.00% |
0 / 41 |
|
0.00% |
0 / 13 |
420 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
2 | |||
| innerSubstr | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| innerStart | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| innerEnd | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| innerLength | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| openSubstr | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| closeSubstr | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| stripTags | |
0.00% |
0 / 9 |
|
0.00% |
0 / 1 |
2 | |||
| offset | |
0.00% |
0 / 8 |
|
0.00% |
0 / 1 |
2 | |||
| hasValidTagWidths | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
20 | |||
| fromTsr | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| fromArray | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| jsonSerialize | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
12 | |||
| 1 | <?php |
| 2 | declare( strict_types = 1 ); |
| 3 | |
| 4 | namespace Wikimedia\Parsoid\Core; |
| 5 | |
| 6 | use Wikimedia\Assert\Assert; |
| 7 | use Wikimedia\Parsoid\Tokens\SourceRange; |
| 8 | use Wikimedia\Parsoid\Utils\PHPUtils; |
| 9 | |
| 10 | /** |
| 11 | * Represents a DOM source range. That is, for a given DOM tree, gives |
| 12 | * the source offset range in the original wikitext for this DOM tree, |
| 13 | * as well as the opening and closing tag widths if appropriate. |
| 14 | */ |
| 15 | class DomSourceRange extends SourceRange { |
| 16 | /** |
| 17 | * Opening tag width. |
| 18 | * @var ?int |
| 19 | */ |
| 20 | public $openWidth; |
| 21 | |
| 22 | /** |
| 23 | * Closing tag width. |
| 24 | * @var ?int |
| 25 | */ |
| 26 | public $closeWidth; |
| 27 | |
| 28 | /** |
| 29 | * Width of trimmed whitespace between opening tag & first child. |
| 30 | * Defaults to zero since for most nodes, there is no ws trimming. |
| 31 | * -1 indicates that this information is invalid and should not be used. |
| 32 | * @var int |
| 33 | */ |
| 34 | public $leadingWS = 0; |
| 35 | |
| 36 | /** |
| 37 | * Width of trimmed whitespace between last child & closing tag. |
| 38 | * Defaults to zero since for most nodes, there is no ws trimming. |
| 39 | * -1 indicates that this information is invalid and should not be used. |
| 40 | * @var int |
| 41 | */ |
| 42 | public $trailingWS = 0; |
| 43 | |
| 44 | /** |
| 45 | * Create a new DOM source offset range (DSR). |
| 46 | * @param ?int $start The starting index (UTF-8 byte count, inclusive) |
| 47 | * @param ?int $end The ending index (UTF-8 byte count, exclusive) |
| 48 | * @param ?int $openWidth The width of the open container tag |
| 49 | * @param ?int $closeWidth The width of the close container tag |
| 50 | * @param int $leadingWS The width of WS chars between opening tag & first child |
| 51 | * @param int $trailingWS The width of WS chars between last child & closing tag |
| 52 | */ |
| 53 | public function __construct( |
| 54 | ?int $start, ?int $end, ?int $openWidth, ?int $closeWidth, |
| 55 | int $leadingWS = 0, |
| 56 | int $trailingWS = 0 |
| 57 | ) { |
| 58 | parent::__construct( $start, $end ); |
| 59 | $this->openWidth = $openWidth; |
| 60 | $this->closeWidth = $closeWidth; |
| 61 | $this->leadingWS = $leadingWS; |
| 62 | $this->trailingWS = $trailingWS; |
| 63 | } |
| 64 | |
| 65 | /** |
| 66 | * Return the substring of the given string corresponding to the |
| 67 | * inner portion of this range (that is, not including the opening |
| 68 | * and closing tag widths). |
| 69 | * @param string $str The source text string |
| 70 | * @return string |
| 71 | */ |
| 72 | public function innerSubstr( string $str ): string { |
| 73 | return PHPUtils::safeSubstr( $str, $this->innerStart(), $this->innerLength() ); |
| 74 | } |
| 75 | |
| 76 | /** |
| 77 | * Return the "inner start", that is, the start offset plus the open width. |
| 78 | * @return int |
| 79 | */ |
| 80 | public function innerStart(): int { |
| 81 | return $this->start + ( $this->openWidth ?? 0 ); |
| 82 | } |
| 83 | |
| 84 | /** |
| 85 | * Return the "inner end", that is, the end offset minus the close width. |
| 86 | * @return int |
| 87 | */ |
| 88 | public function innerEnd(): int { |
| 89 | return $this->end - ( $this->closeWidth ?? 0 ); |
| 90 | } |
| 91 | |
| 92 | /** |
| 93 | * Return the length of this source range, excluding the open and close |
| 94 | * tag widths. |
| 95 | * @return int |
| 96 | */ |
| 97 | public function innerLength(): int { |
| 98 | return $this->innerEnd() - $this->innerStart(); |
| 99 | } |
| 100 | |
| 101 | /** |
| 102 | * Return the substring of the given string corresponding to the |
| 103 | * open portion of this range. |
| 104 | * @param string $str The source text string |
| 105 | * @return string |
| 106 | */ |
| 107 | public function openSubstr( string $str ): string { |
| 108 | return PHPUtils::safeSubstr( $str, $this->start, $this->openWidth ); |
| 109 | } |
| 110 | |
| 111 | /** |
| 112 | * Return the substring of the given string corresponding to the |
| 113 | * close portion of this range. |
| 114 | * @param string $str The source text string |
| 115 | * @return string |
| 116 | */ |
| 117 | public function closeSubstr( string $str ): string { |
| 118 | return PHPUtils::safeSubstr( $str, $this->innerEnd(), $this->closeWidth ); |
| 119 | } |
| 120 | |
| 121 | /** |
| 122 | * Strip the tag open and close from the beginning and end of the |
| 123 | * provided string. This is similar to `DomSourceRange::innerSubstr()` |
| 124 | * but we assume that the string before `$this->start` and after |
| 125 | * `$this->end` has already been removed. (That is, that the input |
| 126 | * is `$this->substr( $originalWikitextSource )`.) |
| 127 | * |
| 128 | * @param string $src The source text string from `$this->start` |
| 129 | * (inclusive) to `$this->end` (exclusive). |
| 130 | * @return string |
| 131 | */ |
| 132 | public function stripTags( string $src ): string { |
| 133 | Assert::invariant( |
| 134 | strlen( $src ) === $this->length(), |
| 135 | "Input string not the expected length" |
| 136 | ); |
| 137 | return PHPUtils::safeSubstr( |
| 138 | $src, |
| 139 | $this->openWidth, |
| 140 | -$this->closeWidth |
| 141 | ); |
| 142 | } |
| 143 | |
| 144 | /** |
| 145 | * Return a new DOM source range shifted by $amount. |
| 146 | * @param int $amount The amount to shift by |
| 147 | * @return DomSourceRange |
| 148 | */ |
| 149 | public function offset( int $amount ): DomSourceRange { |
| 150 | return new DomSourceRange( |
| 151 | $this->start + $amount, |
| 152 | $this->end + $amount, |
| 153 | $this->openWidth, |
| 154 | $this->closeWidth, |
| 155 | $this->leadingWS, |
| 156 | $this->trailingWS |
| 157 | ); |
| 158 | } |
| 159 | |
| 160 | /** |
| 161 | * @return bool True if the tag widths are valid. |
| 162 | */ |
| 163 | public function hasValidTagWidths(): bool { |
| 164 | return $this->openWidth !== null && $this->closeWidth !== null && |
| 165 | $this->openWidth >= 0 && $this->closeWidth >= 0; |
| 166 | } |
| 167 | |
| 168 | /** |
| 169 | * Convert a TSR to a DSR with zero-width container open/close tags. |
| 170 | * @param SourceRange $tsr |
| 171 | * @return DomSourceRange |
| 172 | */ |
| 173 | public static function fromTsr( SourceRange $tsr ): DomSourceRange { |
| 174 | return new DomSourceRange( $tsr->start, $tsr->end, null, null ); |
| 175 | } |
| 176 | |
| 177 | /** |
| 178 | * Create a new DomSourceRange from an array of integers/null (such as |
| 179 | * created during JSON serialization). |
| 180 | * @param array<int|null> $dsr |
| 181 | * @return DomSourceRange |
| 182 | */ |
| 183 | public static function fromArray( array $dsr ): DomSourceRange { |
| 184 | $n = count( $dsr ); |
| 185 | Assert::invariant( $n === 2 || $n === 4 || $n === 6, 'Not enough elements in DSR array' ); |
| 186 | return new DomSourceRange( |
| 187 | $dsr[0], $dsr[1], $dsr[2] ?? null, $dsr[3] ?? null, $dsr[4] ?? 0, $dsr[5] ?? 0 |
| 188 | ); |
| 189 | } |
| 190 | |
| 191 | /** |
| 192 | * @inheritDoc |
| 193 | */ |
| 194 | public function jsonSerialize(): array { |
| 195 | $a = [ $this->start, $this->end, $this->openWidth, $this->closeWidth ]; |
| 196 | if ( $this->leadingWS !== 0 || $this->trailingWS !== 0 ) { |
| 197 | $a[] = $this->leadingWS; |
| 198 | $a[] = $this->trailingWS; |
| 199 | } |
| 200 | return $a; |
| 201 | } |
| 202 | } |