mail-boxes-etc / lib / dompdf / src / Frame / FrameTree.php

FrameTree.php in MBE eShip trunk, at lib/dompdf/src/Frame/FrameTree.php

325 lines 8.8 KB

Raw Download Zip

1	<?php
2	/**
3	* @package dompdf
4	* @link https://github.com/dompdf/dompdf
5	* @license http://www.gnu.org/copyleft/lesser.html GNU Lesser General Public License
6	*/
7	namespace Dompdf\Frame;
8
9	use DOMDocument;
10	use DOMNode;
11	use DOMElement;
12	use DOMXPath;
13
14	use Dompdf\Exception;
15	use Dompdf\Frame;
16	use IteratorAggregate;
17
18	/**
19	* Represents an entire document as a tree of frames
20	*
21	* The FrameTree consists of {@link Frame} objects each tied to specific
22	* DOMNode objects in a specific DomDocument. The FrameTree has the same
23	* structure as the DomDocument, but adds additional capabilities for
24	* styling and layout.
25	*
26	* @package dompdf
27	*/
28	class FrameTree implements IteratorAggregate
29	{
30	/**
31	* Tags to ignore while parsing the tree
32	*
33	* @var array
34	*/
35	protected static $HIDDEN_TAGS = [
36	"area",
37	"base",
38	"basefont",
39	"head",
40	"style",
41	"meta",
42	"title",
43	"colgroup",
44	"noembed",
45	"param",
46	"#comment"
47	];
48
49	/**
50	* The main DomDocument
51	*
52	* @see http://ca2.php.net/manual/en/ref.dom.php
53	* @var DOMDocument
54	*/
55	protected $_dom;
56
57	/**
58	* The root node of the FrameTree.
59	*
60	* @var Frame
61	*/
62	protected $_root;
63
64	/**
65	* Subtrees of absolutely positioned elements
66	*
67	* @var array of Frames
68	*/
69	protected $_absolute_frames;
70
71	/**
72	* A mapping of {@link Frame} objects to DOMNode objects
73	*
74	* @var array
75	*/
76	protected $_registry;
77
78	/**
79	* Class constructor
80	*
81	* @param DOMDocument $dom the main DomDocument object representing the current html document
82	*/
83	public function __construct(DomDocument $dom)
84	{
85	$this->_dom = $dom;
86	$this->_root = null;
87	$this->_registry = [];
88	}
89
90	/**
91	* Returns the DOMDocument object representing the current html document
92	*
93	* @return DOMDocument
94	*/
95	public function get_dom()
96	{
97	return $this->_dom;
98	}
99
100	/**
101	* Returns the root frame of the tree
102	*
103	* @return Frame
104	*/
105	public function get_root()
106	{
107	return $this->_root;
108	}
109
110	/**
111	* Returns a specific frame given its id
112	*
113	* @param string $id
114	*
115	* @return Frame\|null
116	*/
117	public function get_frame($id)
118	{
119	return isset($this->_registry[$id]) ? $this->_registry[$id] : null;
120	}
121
122	/**
123	* Returns a post-order iterator for all frames in the tree
124	*
125	* @deprecated Iterate the tree directly instead
126	* @return FrameTreeIterator
127	*/
128	public function get_frames(): FrameTreeIterator
129	{
130	return new FrameTreeIterator($this->_root);
131	}
132
133	/**
134	* Returns a post-order iterator for all frames in the tree
135	*
136	* @return FrameTreeIterator
137	*/
138	public function getIterator(): FrameTreeIterator
139	{
140	return new FrameTreeIterator($this->_root);
141	}
142
143	/**
144	* Builds the tree
145	*/
146	public function build_tree()
147	{
148	$html = $this->_dom->getElementsByTagName("html")->item(0);
149	if (is_null($html)) {
150	$html = $this->_dom->firstChild;
151	}
152
153	if (is_null($html)) {
154	throw new Exception("Requested HTML document contains no data.");
155	}
156
157	$this->fix_tables();
158
159	$this->_root = $this->_build_tree_r($html);
160	}
161
162	/**
163	* Adds missing TBODYs around TR
164	*/
165	protected function fix_tables()
166	{
167	$xp = new DOMXPath($this->_dom);
168
169	// Move table caption before the table
170	// FIXME find a better way to deal with it...
171	$captions = $xp->query('//table/caption');
172	foreach ($captions as $caption) {
173	$table = $caption->parentNode;
174	$table->parentNode->insertBefore($caption, $table);
175	}
176
177	$firstRows = $xp->query('//table/tr[1]');
178	/** @var DOMElement $tableChild */
179	foreach ($firstRows as $tableChild) {
180	$tbody = $this->_dom->createElement('tbody');
181	$tableNode = $tableChild->parentNode;
182	do {
183	if ($tableChild->nodeName === 'tr') {
184	$tmpNode = $tableChild;
185	$tableChild = $tableChild->nextSibling;
186	$tableNode->removeChild($tmpNode);
187	$tbody->appendChild($tmpNode);
188	} else {
189	if ($tbody->hasChildNodes() === true) {
190	$tableNode->insertBefore($tbody, $tableChild);
191	$tbody = $this->_dom->createElement('tbody');
192	}
193	$tableChild = $tableChild->nextSibling;
194	}
195	} while ($tableChild);
196	if ($tbody->hasChildNodes() === true) {
197	$tableNode->appendChild($tbody);
198	}
199	}
200	}
201
202	// FIXME: temporary hack, preferably we will improve rendering of sequential #text nodes
203	/**
204	* Remove a child from a node
205	*
206	* Remove a child from a node. If the removed node results in two
207	* adjacent #text nodes then combine them.
208	*
209	* @param DOMNode $node the current DOMNode being considered
210	* @param array $children an array of nodes that are the children of $node
211	* @param int $index index from the $children array of the node to remove
212	*/
213	protected function _remove_node(DOMNode $node, array &$children, $index)
214	{
215	$child = $children[$index];
216	$previousChild = $child->previousSibling;
217	$nextChild = $child->nextSibling;
218	$node->removeChild($child);
219	if (isset($previousChild, $nextChild)) {
220	if ($previousChild->nodeName === "#text" && $nextChild->nodeName === "#text") {
221	$previousChild->nodeValue .= $nextChild->nodeValue;
222	$this->_remove_node($node, $children, $index+1);
223	}
224	}
225	array_splice($children, $index, 1);
226	}
227
228	/**
229	* Recursively adds {@link Frame} objects to the tree
230	*
231	* Recursively build a tree of Frame objects based on a dom tree.
232	* No layout information is calculated at this time, although the
233	* tree may be adjusted (i.e. nodes and frames for generated content
234	* and images may be created).
235	*
236	* @param DOMNode $node the current DOMNode being considered
237	*
238	* @return Frame
239	*/
240	protected function _build_tree_r(DOMNode $node)
241	{
242	$frame = new Frame($node);
243	$id = $frame->get_id();
244	$this->_registry[$id] = $frame;
245
246	if (!$node->hasChildNodes()) {
247	return $frame;
248	}
249
250	// Store the children in an array so that the tree can be modified
251	$children = [];
252	$length = $node->childNodes->length;
253	for ($i = 0; $i < $length; $i++) {
254	$children[] = $node->childNodes->item($i);
255	}
256	$index = 0;
257	// INFO: We don't advance $index if a node is removed to avoid skipping nodes
258	while ($index < count($children)) {
259	$child = $children[$index];
260	$nodeName = strtolower($child->nodeName);
261
262	// Skip non-displaying nodes
263	if (in_array($nodeName, self::$HIDDEN_TAGS)) {
264	if ($nodeName !== "head" && $nodeName !== "style") {
265	$this->_remove_node($node, $children, $index);
266	} else {
267	$index++;
268	}
269	continue;
270	}
271	// Skip empty text nodes
272	if ($nodeName === "#text" && $child->nodeValue === "") {
273	$this->_remove_node($node, $children, $index);
274	continue;
275	}
276	// Skip empty image nodes
277	if ($nodeName === "img" && $child->getAttribute("src") === "") {
278	$this->_remove_node($node, $children, $index);
279	continue;
280	}
281
282	if (is_object($child)) {
283	$frame->append_child($this->_build_tree_r($child), false);
284	}
285	$index++;
286	}
287
288	return $frame;
289	}
290
291	/**
292	* @param DOMElement $node
293	* @param DOMElement $new_node
294	* @param string $pos
295	*
296	* @return mixed
297	*/
298	public function insert_node(DOMElement $node, DOMElement $new_node, $pos)
299	{
300	if ($pos === "after" \|\| !$node->firstChild) {
301	$node->appendChild($new_node);
302	} else {
303	$node->insertBefore($new_node, $node->firstChild);
304	}
305
306	$this->_build_tree_r($new_node);
307
308	$frame_id = $new_node->getAttribute("frame_id");
309	$frame = $this->get_frame($frame_id);
310
311	$parent_id = $node->getAttribute("frame_id");
312	$parent = $this->get_frame($parent_id);
313
314	if ($parent) {
315	if ($pos === "before") {
316	$parent->prepend_child($frame, false);
317	} else {
318	$parent->append_child($frame, false);
319	}
320	}
321
322	return $frame_id;
323	}
324	}
325