Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
100.00% |
2 / 2 |
|
100.00% |
2 / 2 |
CRAP | |
100.00% |
1 / 1 |
| Box | |
100.00% |
2 / 2 |
|
100.00% |
2 / 2 |
2 | |
100.00% |
1 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| addChild | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| 1 | <?php |
| 2 | |
| 3 | declare(strict_types=1); |
| 4 | |
| 5 | namespace Phpdftk\HtmlToPdf\Box; |
| 6 | |
| 7 | use Phpdftk\Css\Cascade\CascadedValues; |
| 8 | use Phpdftk\Html\Dom\Element; |
| 9 | use Phpdftk\HtmlToPdf\Layout\BoxGeometry; |
| 10 | use Phpdftk\HtmlToPdf\Layout\LineBox; |
| 11 | use Phpdftk\HtmlToPdf\Layout\MultiColumnLayout; |
| 12 | |
| 13 | /** |
| 14 | * Base of the box tree produced by {@see BoxGenerator}. |
| 15 | * |
| 16 | * Each non-anonymous box carries its originating DOM element + the |
| 17 | * cascade's resolved property bag for that element. Anonymous boxes |
| 18 | * (wrappers synthesised by CSS Display 3 box-generation rules) have a |
| 19 | * null `$element` but inherit a style derived from their parent. |
| 20 | * |
| 21 | * `$children` is in document / logical order; layout (Phase 1F) walks the |
| 22 | * tree to compute positions and dimensions without re-ordering. |
| 23 | * |
| 24 | * Phase 1E.1 ships the structural box tree only — every box's geometry |
| 25 | * (`x`, `y`, `width`, `height`) is layout's responsibility, not box |
| 26 | * generation's. Keep that boundary sharp; this class doesn't track |
| 27 | * positions or sizes. |
| 28 | */ |
| 29 | abstract class Box |
| 30 | { |
| 31 | /** @var list<Box> */ |
| 32 | public array $children = []; |
| 33 | |
| 34 | /** Resolved geometry — populated by layout, blank until {@see \Phpdftk\HtmlToPdf\Layout\BlockLayout} runs. */ |
| 35 | public BoxGeometry $geometry; |
| 36 | |
| 37 | /** |
| 38 | * Line boxes produced by {@see \Phpdftk\HtmlToPdf\Layout\InlineLayout} |
| 39 | * when this box's children form an inline formatting context. Empty |
| 40 | * for block-context parents and for boxes without inline content. |
| 41 | * |
| 42 | * @var list<LineBox> |
| 43 | */ |
| 44 | public array $lineBoxes = []; |
| 45 | |
| 46 | /** |
| 47 | * Set by {@see \Phpdftk\HtmlToPdf\Layout\BlockLayout} when this box's |
| 48 | * cascade declares a multi-column container (CSS Multi-column 1). |
| 49 | * Null on every other box. The painter reads this to stroke |
| 50 | * `column-rule` between adjacent columns. |
| 51 | */ |
| 52 | public ?MultiColumnLayout $multiColumn = null; |
| 53 | |
| 54 | /** |
| 55 | * Set by {@see \Phpdftk\HtmlToPdf\Layout\BlockLayout} on every box |
| 56 | * that falls entirely after a `line-clamp` container's clamp point. |
| 57 | * CSS Overflow 4 §6 discards that content: the boxes still exist in |
| 58 | * the tree (their geometry is what located the clamp point in the |
| 59 | * first place) but nothing about them is painted — not their |
| 60 | * background, not their borders, not their out-of-flow descendants. |
| 61 | */ |
| 62 | public bool $hiddenByLineClamp = false; |
| 63 | |
| 64 | /** |
| 65 | * True when this box is an out-of-flow (abs-pos / fixed) box whose |
| 66 | * computed `display` was inline-level BEFORE the CSS Display §2.7 |
| 67 | * blockification that rewrites out-of-flow displays to `block` (set by |
| 68 | * {@see \Phpdftk\HtmlToPdf\Box\BoxGenerator}). It preserves the |
| 69 | * discriminator the cascade otherwise loses, so the static-position |
| 70 | * recovery in {@see \Phpdftk\HtmlToPdf\Layout\BlockLayout} can apply the |
| 71 | * CSS 2.1 §10.6.4 inline-continuation static position (end/top of the |
| 72 | * previous sibling's last line box) ONLY to boxes that would have laid |
| 73 | * out inline in normal flow. A box whose static display was block-level |
| 74 | * (this flag false) uses the ordinary block-flow static position |
| 75 | * (container inline-start / next block position) instead. |
| 76 | */ |
| 77 | public bool $wasInlineLevel = false; |
| 78 | |
| 79 | /** |
| 80 | * The HTML "ordinal value" this box's `::marker` counts with, when the |
| 81 | * box is a `display: list-item` one. Null on every other box. |
| 82 | * |
| 83 | * Assigned by {@see BoxGenerator} during the generation walk rather |
| 84 | * than derived from the DOM on demand, because the answer depends on |
| 85 | * facts only box generation knows: HTML's "list owner" is the nearest |
| 86 | * ancestor `ol` / `ul` / `menu` THAT GENERATES BOXES, and an item that |
| 87 | * generates no boxes consumes no ordinal. A DOM-only walk can see |
| 88 | * neither, and cannot see that a `display: list-item` `<span>` is an |
| 89 | * item at all. |
| 90 | * |
| 91 | * Both marker paths read it, which is what keeps an `inside` marker |
| 92 | * (materialised as inline content by BoxGenerator) and an `outside` |
| 93 | * one (painted beside the box by |
| 94 | * {@see \Phpdftk\HtmlToPdf\Painter\Painter}) numbering the same list |
| 95 | * identically. |
| 96 | */ |
| 97 | public ?int $listItemOrdinal = null; |
| 98 | |
| 99 | /** |
| 100 | * True when this box must not draw a `::marker`, even though its |
| 101 | * `display` is `list-item`. |
| 102 | * |
| 103 | * HTML §15.3.9 says a `<fieldset>` "is expected to not generate a |
| 104 | * `::marker` pseudo-element" — its contents render through an |
| 105 | * anonymous content box that owns no marker position. The suppression |
| 106 | * is unconditional, so it cannot be spelled as a `list-style-type: |
| 107 | * none` UA rule an author could override with `list-style-type: |
| 108 | * decimal`. |
| 109 | * |
| 110 | * The counter still advances: `display: list-item` implies |
| 111 | * `counter-increment: list-item 1` whether or not anything is drawn, |
| 112 | * so a marker-less fieldset still consumes an ordinal from the list |
| 113 | * around it. |
| 114 | */ |
| 115 | public bool $suppressesListMarker = false; |
| 116 | |
| 117 | /** |
| 118 | * Set by {@see \Phpdftk\HtmlToPdf\Layout\BlockLayout} on an |
| 119 | * out-of-flow box that CSS Anchor Positioning 1 §10 |
| 120 | * (`position-visibility`) makes *strongly hidden*. A strongly hidden |
| 121 | * box is skipped by the painter along with its whole subtree — |
| 122 | * including descendants that set `visibility: visible` and |
| 123 | * out-of-flow descendants that escape it — which is what separates |
| 124 | * it from ordinary `visibility: hidden`. |
| 125 | */ |
| 126 | public bool $hiddenByPositionVisibility = false; |
| 127 | |
| 128 | /** |
| 129 | * `'before'` / `'after'` when this box was generated for a |
| 130 | * pseudo-element of {@see $element}, null when it is the element's |
| 131 | * own box. A pseudo-element shares its originating element's |
| 132 | * `Element` instance, so this is the only thing that tells the two |
| 133 | * apart — which CSS Anchor Positioning 1 §3.2 needs, because a |
| 134 | * pseudo-element's IMPLICIT anchor is its originating element. |
| 135 | */ |
| 136 | public ?string $pseudoElement = null; |
| 137 | |
| 138 | /** |
| 139 | * True when this box is an `<iframe>` / `<frame>` whose children are |
| 140 | * the box tree of the document it EMBEDS (HTML §7.3, a child |
| 141 | * navigable), rather than fallback content. |
| 142 | * |
| 143 | * The distinction matters because a frame is a replaced element, and |
| 144 | * layout otherwise skips a replaced box's children on the (correct, |
| 145 | * for `<img>` / `<video>` / `<object>`) grounds that they are fallback |
| 146 | * content the resource replaced. A frame is the one replaced element |
| 147 | * whose "resource" IS a box tree, so its children are exactly what |
| 148 | * has to be laid out and painted. |
| 149 | */ |
| 150 | public bool $hostsChildNavigable = false; |
| 151 | |
| 152 | public function __construct( |
| 153 | public readonly ?Element $element, |
| 154 | public readonly CascadedValues $style, |
| 155 | ) { |
| 156 | $this->geometry = new BoxGeometry(); |
| 157 | } |
| 158 | |
| 159 | public function addChild(self $child): void |
| 160 | { |
| 161 | $this->children[] = $child; |
| 162 | } |
| 163 | } |