Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
88.71% |
55 / 62 |
|
76.92% |
10 / 13 |
CRAP | |
0.00% |
0 / 1 |
| ActiveFormattingElements | |
88.71% |
55 / 62 |
|
76.92% |
10 / 13 |
44.54 | |
0.00% |
0 / 1 |
| push | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
5 | |||
| elementsMatchForNoahsArk | |
100.00% |
14 / 14 |
|
100.00% |
1 / 1 |
9 | |||
| pushMarker | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| clearToLastMarker | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| isEmpty | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| entries | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| contains | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| remove | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| replace | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
3 | |||
| indexOf | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
3.14 | |||
| insertAt | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| findLastBetweenMarkerAnd | |
100.00% |
7 / 7 |
|
100.00% |
1 / 1 |
5 | |||
| lastElement | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
20 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Phpdftk\Html\TreeConstruction; |
| 6 | |
| 7 | use Phpdftk\Html\Dom\Element; |
| 8 | |
| 9 | /** |
| 10 | * List of active formatting elements per WHATWG §13.2.4.3. |
| 11 | * |
| 12 | * Stores Element entries plus null sentinels ("markers") inserted at |
| 13 | * applet/object/marquee/template scope boundaries. The list is what makes |
| 14 | * the adoption agency algorithm work — it lets the tree builder remember |
| 15 | * which formatting elements (b, i, code, etc.) are "active" but possibly |
| 16 | * mis-nested with respect to the open-elements stack. |
| 17 | * |
| 18 | * Phase 1B.3 ships the data-structure operations; the full reconstruction |
| 19 | * loop (which materialises formatting elements onto the open-elements stack |
| 20 | * after they get popped by an unrelated end tag) is implemented as part of |
| 21 | * TreeBuilder so it has access to the insertion algorithms. |
| 22 | */ |
| 23 | final class ActiveFormattingElements |
| 24 | { |
| 25 | /** |
| 26 | * Entries are either Element (for an active formatting element) or null |
| 27 | * (for a marker). |
| 28 | * |
| 29 | * @var list<?Element> |
| 30 | */ |
| 31 | private array $entries = []; |
| 32 | |
| 33 | public function push(Element $element): void |
| 34 | { |
| 35 | // Noah's Ark clause per WHATWG §13.2.4.3: if three entries already |
| 36 | // exist between the last marker (or list start) and the end that |
| 37 | // have the same tag name, namespace, and attribute set, remove the |
| 38 | // earliest. This prevents O(N^2) blowups on pathological input like |
| 39 | // `<b><b><b>...<b>`. |
| 40 | $matches = []; |
| 41 | for ($i = count($this->entries) - 1; $i >= 0; $i--) { |
| 42 | $entry = $this->entries[$i]; |
| 43 | if ($entry === null) { |
| 44 | break; // hit a marker |
| 45 | } |
| 46 | if ($this->elementsMatchForNoahsArk($entry, $element)) { |
| 47 | $matches[] = $i; |
| 48 | if (count($matches) >= 3) { |
| 49 | // Remove the earliest of the three. |
| 50 | $earliest = $matches[count($matches) - 1]; |
| 51 | array_splice($this->entries, $earliest, 1); |
| 52 | break; |
| 53 | } |
| 54 | } |
| 55 | } |
| 56 | $this->entries[] = $element; |
| 57 | } |
| 58 | |
| 59 | /** |
| 60 | * Two elements match for the Noah's Ark clause iff: same local name, |
| 61 | * same namespace, and identical attribute sets (name → value). |
| 62 | */ |
| 63 | private function elementsMatchForNoahsArk(Element $a, Element $b): bool |
| 64 | { |
| 65 | if ($a->localName !== $b->localName || $a->namespaceURI !== $b->namespaceURI) { |
| 66 | return false; |
| 67 | } |
| 68 | $attrsA = []; |
| 69 | foreach ($a->attributes() as $attr) { |
| 70 | $attrsA[$attr->qualifiedName()] = $attr->value; |
| 71 | } |
| 72 | $attrsB = []; |
| 73 | foreach ($b->attributes() as $attr) { |
| 74 | $attrsB[$attr->qualifiedName()] = $attr->value; |
| 75 | } |
| 76 | if (count($attrsA) !== count($attrsB)) { |
| 77 | return false; |
| 78 | } |
| 79 | foreach ($attrsA as $name => $value) { |
| 80 | if (!array_key_exists($name, $attrsB) || $attrsB[$name] !== $value) { |
| 81 | return false; |
| 82 | } |
| 83 | } |
| 84 | return true; |
| 85 | } |
| 86 | |
| 87 | public function pushMarker(): void |
| 88 | { |
| 89 | $this->entries[] = null; |
| 90 | } |
| 91 | |
| 92 | public function clearToLastMarker(): void |
| 93 | { |
| 94 | while ($this->entries !== []) { |
| 95 | $popped = array_pop($this->entries); |
| 96 | if ($popped === null) { |
| 97 | return; |
| 98 | } |
| 99 | } |
| 100 | } |
| 101 | |
| 102 | public function isEmpty(): bool |
| 103 | { |
| 104 | return $this->entries === []; |
| 105 | } |
| 106 | |
| 107 | /** @return list<?Element> */ |
| 108 | public function entries(): array |
| 109 | { |
| 110 | return $this->entries; |
| 111 | } |
| 112 | |
| 113 | public function contains(Element $element): bool |
| 114 | { |
| 115 | foreach ($this->entries as $entry) { |
| 116 | if ($entry === $element) { |
| 117 | return true; |
| 118 | } |
| 119 | } |
| 120 | return false; |
| 121 | } |
| 122 | |
| 123 | public function remove(Element $element): void |
| 124 | { |
| 125 | foreach ($this->entries as $i => $entry) { |
| 126 | if ($entry === $element) { |
| 127 | array_splice($this->entries, $i, 1); |
| 128 | return; |
| 129 | } |
| 130 | } |
| 131 | } |
| 132 | |
| 133 | public function replace(Element $old, Element $new): void |
| 134 | { |
| 135 | foreach ($this->entries as $i => $entry) { |
| 136 | if ($entry === $old) { |
| 137 | $this->entries[$i] = $new; |
| 138 | return; |
| 139 | } |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | public function indexOf(Element $element): ?int |
| 144 | { |
| 145 | foreach ($this->entries as $i => $entry) { |
| 146 | if ($entry === $element) { |
| 147 | return $i; |
| 148 | } |
| 149 | } |
| 150 | return null; |
| 151 | } |
| 152 | |
| 153 | public function insertAt(int $index, Element $element): void |
| 154 | { |
| 155 | array_splice($this->entries, $index, 0, [$element]); |
| 156 | } |
| 157 | |
| 158 | /** |
| 159 | * Find the last element entry between the end of the list and the most |
| 160 | * recent marker (or the start) whose local name matches. Returns null if |
| 161 | * no match. |
| 162 | */ |
| 163 | public function findLastBetweenMarkerAnd(string $localName): ?Element |
| 164 | { |
| 165 | for ($i = array_key_last($this->entries); $i !== null && $i >= 0; $i--) { |
| 166 | $entry = $this->entries[$i]; |
| 167 | if ($entry === null) { |
| 168 | return null; // hit a marker |
| 169 | } |
| 170 | if ($entry->localName === $localName) { |
| 171 | return $entry; |
| 172 | } |
| 173 | } |
| 174 | return null; |
| 175 | } |
| 176 | |
| 177 | /** Last entry that is an Element (skip trailing markers). */ |
| 178 | public function lastElement(): ?Element |
| 179 | { |
| 180 | for ($i = array_key_last($this->entries); $i !== null && $i >= 0; $i--) { |
| 181 | $entry = $this->entries[$i]; |
| 182 | if ($entry !== null) { |
| 183 | return $entry; |
| 184 | } |
| 185 | } |
| 186 | return null; |
| 187 | } |
| 188 | } |