paquettg /
php-html-parser
@@ -12,30 +12,30 @@ |
||
| 12 | 12 | abstract class ArrayNode extends AbstractNode implements IteratorAggregate, Countable |
| 13 | 13 | { |
| 14 | 14 | |
| 15 | - /** |
|
| 16 | - * Gets the iterator |
|
| 17 | - * |
|
| 18 | - * @return ArrayIterator |
|
| 19 | - */ |
|
| 20 | - public function getIterator() |
|
| 21 | - { |
|
| 22 | - return new ArrayIterator($this->getIteratorArray()); |
|
| 23 | - } |
|
| 15 | + /** |
|
| 16 | + * Gets the iterator |
|
| 17 | + * |
|
| 18 | + * @return ArrayIterator |
|
| 19 | + */ |
|
| 20 | + public function getIterator() |
|
| 21 | + { |
|
| 22 | + return new ArrayIterator($this->getIteratorArray()); |
|
| 23 | + } |
|
| 24 | 24 | |
| 25 | - /** |
|
| 26 | - * Returns the count of the iterator array. |
|
| 27 | - * |
|
| 28 | - * @return int |
|
| 29 | - */ |
|
| 30 | - public function count() |
|
| 31 | - { |
|
| 32 | - return count($this->getIteratorArray()); |
|
| 33 | - } |
|
| 25 | + /** |
|
| 26 | + * Returns the count of the iterator array. |
|
| 27 | + * |
|
| 28 | + * @return int |
|
| 29 | + */ |
|
| 30 | + public function count() |
|
| 31 | + { |
|
| 32 | + return count($this->getIteratorArray()); |
|
| 33 | + } |
|
| 34 | 34 | |
| 35 | - /** |
|
| 36 | - * Returns the array to be used the the iterator. |
|
| 37 | - * |
|
| 38 | - * @return array |
|
| 39 | - */ |
|
| 40 | - abstract protected function getIteratorArray(); |
|
| 35 | + /** |
|
| 36 | + * Returns the array to be used the the iterator. |
|
| 37 | + * |
|
| 38 | + * @return array |
|
| 39 | + */ |
|
| 40 | + abstract protected function getIteratorArray(); |
|
| 41 | 41 | } |
@@ -11,44 +11,44 @@ |
||
| 11 | 11 | class MockNode extends InnerNode |
| 12 | 12 | { |
| 13 | 13 | |
| 14 | - /** |
|
| 15 | - * Mock of innner html. |
|
| 16 | - */ |
|
| 17 | - public function innerHtml() |
|
| 18 | - { |
|
| 19 | - } |
|
| 14 | + /** |
|
| 15 | + * Mock of innner html. |
|
| 16 | + */ |
|
| 17 | + public function innerHtml() |
|
| 18 | + { |
|
| 19 | + } |
|
| 20 | 20 | |
| 21 | - /** |
|
| 22 | - * Mock of outer html. |
|
| 23 | - */ |
|
| 24 | - public function outerHtml() |
|
| 25 | - { |
|
| 26 | - } |
|
| 21 | + /** |
|
| 22 | + * Mock of outer html. |
|
| 23 | + */ |
|
| 24 | + public function outerHtml() |
|
| 25 | + { |
|
| 26 | + } |
|
| 27 | 27 | |
| 28 | - /** |
|
| 29 | - * Mock of text. |
|
| 30 | - */ |
|
| 31 | - public function text() |
|
| 32 | - { |
|
| 33 | - } |
|
| 28 | + /** |
|
| 29 | + * Mock of text. |
|
| 30 | + */ |
|
| 31 | + public function text() |
|
| 32 | + { |
|
| 33 | + } |
|
| 34 | 34 | |
| 35 | - /** |
|
| 36 | - * Clear content of this node |
|
| 37 | - */ |
|
| 38 | - protected function clear() |
|
| 39 | - { |
|
| 40 | - $this->innerHtml = null; |
|
| 41 | - $this->outerHtml = null; |
|
| 42 | - $this->text = null; |
|
| 43 | - } |
|
| 35 | + /** |
|
| 36 | + * Clear content of this node |
|
| 37 | + */ |
|
| 38 | + protected function clear() |
|
| 39 | + { |
|
| 40 | + $this->innerHtml = null; |
|
| 41 | + $this->outerHtml = null; |
|
| 42 | + $this->text = null; |
|
| 43 | + } |
|
| 44 | 44 | |
| 45 | - /** |
|
| 46 | - * Returns all children of this html node. |
|
| 47 | - * |
|
| 48 | - * @return array |
|
| 49 | - */ |
|
| 50 | - protected function getIteratorArray() |
|
| 51 | - { |
|
| 52 | - return $this->getChildren(); |
|
| 53 | - } |
|
| 45 | + /** |
|
| 46 | + * Returns all children of this html node. |
|
| 47 | + * |
|
| 48 | + * @return array |
|
| 49 | + */ |
|
| 50 | + protected function getIteratorArray() |
|
| 51 | + { |
|
| 52 | + return $this->getChildren(); |
|
| 53 | + } |
|
| 54 | 54 | } |
@@ -13,257 +13,257 @@ |
||
| 13 | 13 | abstract class InnerNode extends ArrayNode |
| 14 | 14 | {
|
| 15 | 15 | |
| 16 | - /** |
|
| 17 | - * An array of all the children. |
|
| 18 | - * |
|
| 19 | - * @var array |
|
| 20 | - */ |
|
| 21 | - protected $children = []; |
|
| 22 | - |
|
| 23 | - /** |
|
| 24 | - * Sets the encoding class to this node and propagates it |
|
| 25 | - * to all its children. |
|
| 26 | - * |
|
| 27 | - * @param Encode $encode |
|
| 28 | - * @return void |
|
| 29 | - */ |
|
| 30 | - public function propagateEncoding(Encode $encode) |
|
| 31 | - {
|
|
| 32 | - $this->encode = $encode; |
|
| 33 | - $this->tag->setEncoding($encode); |
|
| 34 | - // check children |
|
| 35 | - foreach ($this->children as $id => $child) {
|
|
| 36 | - /** @var AbstractNode $node */ |
|
| 37 | - $node = $child['node']; |
|
| 38 | - $node->propagateEncoding($encode); |
|
| 39 | - } |
|
| 40 | - } |
|
| 41 | - |
|
| 42 | - /** |
|
| 43 | - * Checks if this node has children. |
|
| 44 | - * |
|
| 45 | - * @return bool |
|
| 46 | - */ |
|
| 47 | - public function hasChildren() |
|
| 48 | - {
|
|
| 49 | - return ! empty($this->children); |
|
| 50 | - } |
|
| 51 | - |
|
| 52 | - /** |
|
| 53 | - * Returns the child by id. |
|
| 54 | - * |
|
| 55 | - * @param int $id |
|
| 56 | - * @return AbstractNode |
|
| 57 | - * @throws ChildNotFoundException |
|
| 58 | - */ |
|
| 59 | - public function getChild($id) |
|
| 60 | - {
|
|
| 61 | - if ( ! isset($this->children[$id])) {
|
|
| 62 | - throw new ChildNotFoundException("Child '$id' not found in this node.");
|
|
| 63 | - } |
|
| 64 | - |
|
| 65 | - return $this->children[$id]['node']; |
|
| 66 | - } |
|
| 67 | - |
|
| 68 | - /** |
|
| 69 | - * Returns a new array of child nodes |
|
| 70 | - * |
|
| 71 | - * @return array |
|
| 72 | - */ |
|
| 73 | - public function getChildren() |
|
| 74 | - {
|
|
| 75 | - $nodes = []; |
|
| 76 | - try {
|
|
| 77 | - $child = $this->firstChild(); |
|
| 78 | - do {
|
|
| 79 | - $nodes[] = $child; |
|
| 80 | - $child = $this->nextChild($child->id()); |
|
| 81 | - } while ( ! is_null($child)); |
|
| 82 | - } catch (ChildNotFoundException $e) {
|
|
| 83 | - // we are done looking for children |
|
| 84 | - } |
|
| 85 | - |
|
| 86 | - return $nodes; |
|
| 87 | - } |
|
| 88 | - |
|
| 89 | - /** |
|
| 90 | - * Counts children |
|
| 91 | - * |
|
| 92 | - * @return int |
|
| 93 | - */ |
|
| 94 | - public function countChildren() |
|
| 95 | - {
|
|
| 96 | - return count($this->children); |
|
| 97 | - } |
|
| 98 | - |
|
| 99 | - /** |
|
| 100 | - * Adds a child node to this node and returns the id of the child for this |
|
| 101 | - * parent. |
|
| 102 | - * |
|
| 103 | - * @param AbstractNode $child |
|
| 104 | - * @return bool |
|
| 105 | - * @throws CircularException |
|
| 106 | - */ |
|
| 107 | - public function addChild(AbstractNode $child) |
|
| 108 | - {
|
|
| 109 | - $key = null; |
|
| 110 | - |
|
| 111 | - // check integrity |
|
| 112 | - if ($this->isAncestor($child->id())) {
|
|
| 113 | - throw new CircularException('Can not add child. It is my ancestor.');
|
|
| 114 | - } |
|
| 115 | - |
|
| 116 | - // check if child is itself |
|
| 117 | - if ($child->id() == $this->id) {
|
|
| 118 | - throw new CircularException('Can not set itself as a child.');
|
|
| 119 | - } |
|
| 120 | - |
|
| 121 | - if ($this->hasChildren()) {
|
|
| 122 | - if (isset($this->children[$child->id()])) {
|
|
| 123 | - // we already have this child |
|
| 124 | - return false; |
|
| 125 | - } |
|
| 126 | - $sibling = $this->lastChild(); |
|
| 127 | - $key = $sibling->id(); |
|
| 128 | - $this->children[$key]['next'] = $child->id(); |
|
| 129 | - } |
|
| 130 | - |
|
| 131 | - // add the child |
|
| 132 | - $this->children[$child->id()] = [ |
|
| 133 | - 'node' => $child, |
|
| 134 | - 'next' => null, |
|
| 135 | - 'prev' => $key, |
|
| 136 | - ]; |
|
| 137 | - |
|
| 138 | - // tell child I am the new parent |
|
| 139 | - $child->setParent($this); |
|
| 140 | - |
|
| 141 | - //clear any cache |
|
| 142 | - $this->clear(); |
|
| 143 | - |
|
| 144 | - return true; |
|
| 145 | - } |
|
| 146 | - |
|
| 147 | - /** |
|
| 148 | - * Removes the child by id. |
|
| 149 | - * |
|
| 150 | - * @param int $id |
|
| 151 | - * @return $this |
|
| 152 | - */ |
|
| 153 | - public function removeChild($id) |
|
| 154 | - {
|
|
| 155 | - if ( ! isset($this->children[$id])) {
|
|
| 156 | - return $this; |
|
| 157 | - } |
|
| 158 | - |
|
| 159 | - // handle moving next and previous assignments. |
|
| 160 | - $next = $this->children[$id]['next']; |
|
| 161 | - $prev = $this->children[$id]['prev']; |
|
| 162 | - if ( ! is_null($next)) {
|
|
| 163 | - $this->children[$next]['prev'] = $prev; |
|
| 164 | - } |
|
| 165 | - if ( ! is_null($prev)) {
|
|
| 166 | - $this->children[$prev]['next'] = $next; |
|
| 167 | - } |
|
| 168 | - |
|
| 169 | - // remove the child |
|
| 170 | - unset($this->children[$id]); |
|
| 171 | - |
|
| 172 | - //clear any cache |
|
| 173 | - $this->clear(); |
|
| 174 | - |
|
| 175 | - return $this; |
|
| 176 | - } |
|
| 177 | - |
|
| 178 | - /** |
|
| 179 | - * Attempts to get the next child. |
|
| 180 | - * |
|
| 181 | - * @param int $id |
|
| 182 | - * @return AbstractNode |
|
| 183 | - * @uses $this->getChild() |
|
| 184 | - */ |
|
| 185 | - public function nextChild($id) |
|
| 186 | - {
|
|
| 187 | - $child = $this->getChild($id); |
|
| 188 | - $next = $this->children[$child->id()]['next']; |
|
| 189 | - |
|
| 190 | - return $this->getChild($next); |
|
| 191 | - } |
|
| 192 | - |
|
| 193 | - /** |
|
| 194 | - * Attempts to get the previous child. |
|
| 195 | - * |
|
| 196 | - * @param int $id |
|
| 197 | - * @return AbstractNode |
|
| 198 | - * @uses $this->getChild() |
|
| 199 | - */ |
|
| 200 | - public function previousChild($id) |
|
| 201 | - {
|
|
| 202 | - $child = $this->getchild($id); |
|
| 203 | - $next = $this->children[$child->id()]['prev']; |
|
| 204 | - |
|
| 205 | - return $this->getChild($next); |
|
| 206 | - } |
|
| 207 | - |
|
| 208 | - /** |
|
| 209 | - * Checks if the given node id is a child of the |
|
| 210 | - * current node. |
|
| 211 | - * |
|
| 212 | - * @param int $id |
|
| 213 | - * @return bool |
|
| 214 | - */ |
|
| 215 | - public function isChild($id) |
|
| 216 | - {
|
|
| 217 | - foreach ($this->children as $childId => $child) {
|
|
| 218 | - if ($id == $childId) {
|
|
| 219 | - return true; |
|
| 220 | - } |
|
| 221 | - } |
|
| 222 | - |
|
| 223 | - return false; |
|
| 224 | - } |
|
| 225 | - |
|
| 226 | - /** |
|
| 227 | - * Checks if the given node id is a descendant of the |
|
| 228 | - * current node. |
|
| 229 | - * |
|
| 230 | - * @param int $id |
|
| 231 | - * @return bool |
|
| 232 | - */ |
|
| 233 | - public function isDescendant($id) |
|
| 234 | - {
|
|
| 235 | - if ($this->isChild($id)) {
|
|
| 236 | - return true; |
|
| 237 | - } |
|
| 238 | - |
|
| 239 | - foreach ($this->children as $childId => $child) {
|
|
| 240 | - /** @var InnerNode $node */ |
|
| 241 | - $node = $child['node']; |
|
| 242 | - if ($node instanceof InnerNode && |
|
| 243 | - $node->hasChildren() && |
|
| 244 | - $node->isDescendant($id) |
|
| 245 | - ) {
|
|
| 246 | - return true; |
|
| 247 | - } |
|
| 248 | - } |
|
| 249 | - |
|
| 250 | - return false; |
|
| 251 | - } |
|
| 252 | - |
|
| 253 | - /** |
|
| 254 | - * Sets the parent node. |
|
| 255 | - * |
|
| 256 | - * @param InnerNode $parent |
|
| 257 | - * @return $this |
|
| 258 | - * @throws CircularException |
|
| 259 | - */ |
|
| 260 | - public function setParent(InnerNode $parent) |
|
| 261 | - {
|
|
| 262 | - // check integrity |
|
| 263 | - if ($this->isDescendant($parent->id())) {
|
|
| 264 | - throw new CircularException('Can not add descendant "'.$parent->id().'" as my parent.');
|
|
| 265 | - } |
|
| 266 | - |
|
| 267 | - return parent::setParent($parent); |
|
| 268 | - } |
|
| 16 | + /** |
|
| 17 | + * An array of all the children. |
|
| 18 | + * |
|
| 19 | + * @var array |
|
| 20 | + */ |
|
| 21 | + protected $children = []; |
|
| 22 | + |
|
| 23 | + /** |
|
| 24 | + * Sets the encoding class to this node and propagates it |
|
| 25 | + * to all its children. |
|
| 26 | + * |
|
| 27 | + * @param Encode $encode |
|
| 28 | + * @return void |
|
| 29 | + */ |
|
| 30 | + public function propagateEncoding(Encode $encode) |
|
| 31 | + {
|
|
| 32 | + $this->encode = $encode; |
|
| 33 | + $this->tag->setEncoding($encode); |
|
| 34 | + // check children |
|
| 35 | + foreach ($this->children as $id => $child) {
|
|
| 36 | + /** @var AbstractNode $node */ |
|
| 37 | + $node = $child['node']; |
|
| 38 | + $node->propagateEncoding($encode); |
|
| 39 | + } |
|
| 40 | + } |
|
| 41 | + |
|
| 42 | + /** |
|
| 43 | + * Checks if this node has children. |
|
| 44 | + * |
|
| 45 | + * @return bool |
|
| 46 | + */ |
|
| 47 | + public function hasChildren() |
|
| 48 | + {
|
|
| 49 | + return ! empty($this->children); |
|
| 50 | + } |
|
| 51 | + |
|
| 52 | + /** |
|
| 53 | + * Returns the child by id. |
|
| 54 | + * |
|
| 55 | + * @param int $id |
|
| 56 | + * @return AbstractNode |
|
| 57 | + * @throws ChildNotFoundException |
|
| 58 | + */ |
|
| 59 | + public function getChild($id) |
|
| 60 | + {
|
|
| 61 | + if ( ! isset($this->children[$id])) {
|
|
| 62 | + throw new ChildNotFoundException("Child '$id' not found in this node.");
|
|
| 63 | + } |
|
| 64 | + |
|
| 65 | + return $this->children[$id]['node']; |
|
| 66 | + } |
|
| 67 | + |
|
| 68 | + /** |
|
| 69 | + * Returns a new array of child nodes |
|
| 70 | + * |
|
| 71 | + * @return array |
|
| 72 | + */ |
|
| 73 | + public function getChildren() |
|
| 74 | + {
|
|
| 75 | + $nodes = []; |
|
| 76 | + try {
|
|
| 77 | + $child = $this->firstChild(); |
|
| 78 | + do {
|
|
| 79 | + $nodes[] = $child; |
|
| 80 | + $child = $this->nextChild($child->id()); |
|
| 81 | + } while ( ! is_null($child)); |
|
| 82 | + } catch (ChildNotFoundException $e) {
|
|
| 83 | + // we are done looking for children |
|
| 84 | + } |
|
| 85 | + |
|
| 86 | + return $nodes; |
|
| 87 | + } |
|
| 88 | + |
|
| 89 | + /** |
|
| 90 | + * Counts children |
|
| 91 | + * |
|
| 92 | + * @return int |
|
| 93 | + */ |
|
| 94 | + public function countChildren() |
|
| 95 | + {
|
|
| 96 | + return count($this->children); |
|
| 97 | + } |
|
| 98 | + |
|
| 99 | + /** |
|
| 100 | + * Adds a child node to this node and returns the id of the child for this |
|
| 101 | + * parent. |
|
| 102 | + * |
|
| 103 | + * @param AbstractNode $child |
|
| 104 | + * @return bool |
|
| 105 | + * @throws CircularException |
|
| 106 | + */ |
|
| 107 | + public function addChild(AbstractNode $child) |
|
| 108 | + {
|
|
| 109 | + $key = null; |
|
| 110 | + |
|
| 111 | + // check integrity |
|
| 112 | + if ($this->isAncestor($child->id())) {
|
|
| 113 | + throw new CircularException('Can not add child. It is my ancestor.');
|
|
| 114 | + } |
|
| 115 | + |
|
| 116 | + // check if child is itself |
|
| 117 | + if ($child->id() == $this->id) {
|
|
| 118 | + throw new CircularException('Can not set itself as a child.');
|
|
| 119 | + } |
|
| 120 | + |
|
| 121 | + if ($this->hasChildren()) {
|
|
| 122 | + if (isset($this->children[$child->id()])) {
|
|
| 123 | + // we already have this child |
|
| 124 | + return false; |
|
| 125 | + } |
|
| 126 | + $sibling = $this->lastChild(); |
|
| 127 | + $key = $sibling->id(); |
|
| 128 | + $this->children[$key]['next'] = $child->id(); |
|
| 129 | + } |
|
| 130 | + |
|
| 131 | + // add the child |
|
| 132 | + $this->children[$child->id()] = [ |
|
| 133 | + 'node' => $child, |
|
| 134 | + 'next' => null, |
|
| 135 | + 'prev' => $key, |
|
| 136 | + ]; |
|
| 137 | + |
|
| 138 | + // tell child I am the new parent |
|
| 139 | + $child->setParent($this); |
|
| 140 | + |
|
| 141 | + //clear any cache |
|
| 142 | + $this->clear(); |
|
| 143 | + |
|
| 144 | + return true; |
|
| 145 | + } |
|
| 146 | + |
|
| 147 | + /** |
|
| 148 | + * Removes the child by id. |
|
| 149 | + * |
|
| 150 | + * @param int $id |
|
| 151 | + * @return $this |
|
| 152 | + */ |
|
| 153 | + public function removeChild($id) |
|
| 154 | + {
|
|
| 155 | + if ( ! isset($this->children[$id])) {
|
|
| 156 | + return $this; |
|
| 157 | + } |
|
| 158 | + |
|
| 159 | + // handle moving next and previous assignments. |
|
| 160 | + $next = $this->children[$id]['next']; |
|
| 161 | + $prev = $this->children[$id]['prev']; |
|
| 162 | + if ( ! is_null($next)) {
|
|
| 163 | + $this->children[$next]['prev'] = $prev; |
|
| 164 | + } |
|
| 165 | + if ( ! is_null($prev)) {
|
|
| 166 | + $this->children[$prev]['next'] = $next; |
|
| 167 | + } |
|
| 168 | + |
|
| 169 | + // remove the child |
|
| 170 | + unset($this->children[$id]); |
|
| 171 | + |
|
| 172 | + //clear any cache |
|
| 173 | + $this->clear(); |
|
| 174 | + |
|
| 175 | + return $this; |
|
| 176 | + } |
|
| 177 | + |
|
| 178 | + /** |
|
| 179 | + * Attempts to get the next child. |
|
| 180 | + * |
|
| 181 | + * @param int $id |
|
| 182 | + * @return AbstractNode |
|
| 183 | + * @uses $this->getChild() |
|
| 184 | + */ |
|
| 185 | + public function nextChild($id) |
|
| 186 | + {
|
|
| 187 | + $child = $this->getChild($id); |
|
| 188 | + $next = $this->children[$child->id()]['next']; |
|
| 189 | + |
|
| 190 | + return $this->getChild($next); |
|
| 191 | + } |
|
| 192 | + |
|
| 193 | + /** |
|
| 194 | + * Attempts to get the previous child. |
|
| 195 | + * |
|
| 196 | + * @param int $id |
|
| 197 | + * @return AbstractNode |
|
| 198 | + * @uses $this->getChild() |
|
| 199 | + */ |
|
| 200 | + public function previousChild($id) |
|
| 201 | + {
|
|
| 202 | + $child = $this->getchild($id); |
|
| 203 | + $next = $this->children[$child->id()]['prev']; |
|
| 204 | + |
|
| 205 | + return $this->getChild($next); |
|
| 206 | + } |
|
| 207 | + |
|
| 208 | + /** |
|
| 209 | + * Checks if the given node id is a child of the |
|
| 210 | + * current node. |
|
| 211 | + * |
|
| 212 | + * @param int $id |
|
| 213 | + * @return bool |
|
| 214 | + */ |
|
| 215 | + public function isChild($id) |
|
| 216 | + {
|
|
| 217 | + foreach ($this->children as $childId => $child) {
|
|
| 218 | + if ($id == $childId) {
|
|
| 219 | + return true; |
|
| 220 | + } |
|
| 221 | + } |
|
| 222 | + |
|
| 223 | + return false; |
|
| 224 | + } |
|
| 225 | + |
|
| 226 | + /** |
|
| 227 | + * Checks if the given node id is a descendant of the |
|
| 228 | + * current node. |
|
| 229 | + * |
|
| 230 | + * @param int $id |
|
| 231 | + * @return bool |
|
| 232 | + */ |
|
| 233 | + public function isDescendant($id) |
|
| 234 | + {
|
|
| 235 | + if ($this->isChild($id)) {
|
|
| 236 | + return true; |
|
| 237 | + } |
|
| 238 | + |
|
| 239 | + foreach ($this->children as $childId => $child) {
|
|
| 240 | + /** @var InnerNode $node */ |
|
| 241 | + $node = $child['node']; |
|
| 242 | + if ($node instanceof InnerNode && |
|
| 243 | + $node->hasChildren() && |
|
| 244 | + $node->isDescendant($id) |
|
| 245 | + ) {
|
|
| 246 | + return true; |
|
| 247 | + } |
|
| 248 | + } |
|
| 249 | + |
|
| 250 | + return false; |
|
| 251 | + } |
|
| 252 | + |
|
| 253 | + /** |
|
| 254 | + * Sets the parent node. |
|
| 255 | + * |
|
| 256 | + * @param InnerNode $parent |
|
| 257 | + * @return $this |
|
| 258 | + * @throws CircularException |
|
| 259 | + */ |
|
| 260 | + public function setParent(InnerNode $parent) |
|
| 261 | + {
|
|
| 262 | + // check integrity |
|
| 263 | + if ($this->isDescendant($parent->id())) {
|
|
| 264 | + throw new CircularException('Can not add descendant "'.$parent->id().'" as my parent.');
|
|
| 265 | + } |
|
| 266 | + |
|
| 267 | + return parent::setParent($parent); |
|
| 268 | + } |
|
| 269 | 269 | } |
| 270 | 270 | \ No newline at end of file |
@@ -16,459 +16,459 @@ |
||
| 16 | 16 | abstract class AbstractNode |
| 17 | 17 | { |
| 18 | 18 | |
| 19 | - /** |
|
| 20 | - * Contains the tag name/type |
|
| 21 | - * |
|
| 22 | - * @var \PHPHtmlParser\Dom\Tag |
|
| 23 | - */ |
|
| 24 | - protected $tag; |
|
| 25 | - |
|
| 26 | - /** |
|
| 27 | - * Contains a list of attributes on this tag. |
|
| 28 | - * |
|
| 29 | - * @var array |
|
| 30 | - */ |
|
| 31 | - protected $attr = []; |
|
| 32 | - |
|
| 33 | - /** |
|
| 34 | - * Contains the parent Node. |
|
| 35 | - * |
|
| 36 | - * @var InnerNode |
|
| 37 | - */ |
|
| 38 | - protected $parent = null; |
|
| 39 | - |
|
| 40 | - /** |
|
| 41 | - * The unique id of the class. Given by PHP. |
|
| 42 | - * |
|
| 43 | - * @var string |
|
| 44 | - */ |
|
| 45 | - protected $id; |
|
| 46 | - |
|
| 47 | - /** |
|
| 48 | - * The encoding class used to encode strings. |
|
| 49 | - * |
|
| 50 | - * @var mixed |
|
| 51 | - */ |
|
| 52 | - protected $encode; |
|
| 53 | - |
|
| 54 | - /** |
|
| 55 | - * Creates a unique spl hash for this node. |
|
| 56 | - */ |
|
| 57 | - public function __construct() |
|
| 58 | - { |
|
| 59 | - $this->id = spl_object_hash($this); |
|
| 60 | - } |
|
| 61 | - |
|
| 62 | - /** |
|
| 63 | - * Magic get method for attributes and certain methods. |
|
| 64 | - * |
|
| 65 | - * @param string $key |
|
| 66 | - * @return mixed |
|
| 67 | - */ |
|
| 68 | - public function __get($key) |
|
| 69 | - { |
|
| 70 | - // check attribute first |
|
| 71 | - if ( ! is_null($this->getAttribute($key))) { |
|
| 72 | - return $this->getAttribute($key); |
|
| 73 | - } |
|
| 74 | - switch (strtolower($key)) { |
|
| 75 | - case 'outerhtml': |
|
| 76 | - return $this->outerHtml(); |
|
| 77 | - case 'innerhtml': |
|
| 78 | - return $this->innerHtml(); |
|
| 79 | - case 'text': |
|
| 80 | - return $this->text(); |
|
| 81 | - } |
|
| 82 | - |
|
| 83 | - return null; |
|
| 84 | - } |
|
| 85 | - |
|
| 86 | - /** |
|
| 87 | - * Attempts to clear out any object references. |
|
| 88 | - */ |
|
| 89 | - public function __destruct() |
|
| 90 | - { |
|
| 91 | - $this->tag = null; |
|
| 92 | - $this->attr = []; |
|
| 93 | - $this->parent = null; |
|
| 94 | - $this->children = []; |
|
| 95 | - } |
|
| 96 | - |
|
| 97 | - /** |
|
| 98 | - * Simply calls the outer text method. |
|
| 99 | - * |
|
| 100 | - * @return string |
|
| 101 | - */ |
|
| 102 | - public function __toString() |
|
| 103 | - { |
|
| 104 | - return $this->outerHtml(); |
|
| 105 | - } |
|
| 106 | - |
|
| 107 | - /** |
|
| 108 | - * Returns the id of this object. |
|
| 109 | - */ |
|
| 110 | - public function id() |
|
| 111 | - { |
|
| 112 | - return $this->id; |
|
| 113 | - } |
|
| 114 | - |
|
| 115 | - /** |
|
| 116 | - * Returns the parent of node. |
|
| 117 | - * |
|
| 118 | - * @return AbstractNode |
|
| 119 | - */ |
|
| 120 | - public function getParent() |
|
| 121 | - { |
|
| 122 | - return $this->parent; |
|
| 123 | - } |
|
| 124 | - |
|
| 125 | - /** |
|
| 126 | - * Sets the parent node. |
|
| 127 | - * |
|
| 128 | - * @param InnerNode $parent |
|
| 129 | - * @return $this |
|
| 130 | - * @throws CircularException |
|
| 131 | - */ |
|
| 132 | - public function setParent(InnerNode $parent) |
|
| 133 | - { |
|
| 134 | - // remove from old parent |
|
| 135 | - if ( ! is_null($this->parent)) { |
|
| 136 | - if ($this->parent->id() == $parent->id()) { |
|
| 137 | - // already the parent |
|
| 138 | - return $this; |
|
| 139 | - } |
|
| 140 | - |
|
| 141 | - $this->parent->removeChild($this->id); |
|
| 142 | - } |
|
| 143 | - |
|
| 144 | - $this->parent = $parent; |
|
| 145 | - |
|
| 146 | - // assign child to parent |
|
| 147 | - $this->parent->addChild($this); |
|
| 148 | - |
|
| 149 | - //clear any cache |
|
| 150 | - $this->clear(); |
|
| 151 | - |
|
| 152 | - return $this; |
|
| 153 | - } |
|
| 154 | - |
|
| 155 | - /** |
|
| 156 | - * Sets the encoding class to this node. |
|
| 157 | - * |
|
| 158 | - * @param Encode $encode |
|
| 159 | - * @return void |
|
| 160 | - */ |
|
| 161 | - public function propagateEncoding(Encode $encode) |
|
| 162 | - { |
|
| 163 | - $this->encode = $encode; |
|
| 164 | - $this->tag->setEncoding($encode); |
|
| 165 | - } |
|
| 166 | - |
|
| 167 | - /** |
|
| 168 | - * Checks if the given node id is an ancestor of |
|
| 169 | - * the current node. |
|
| 170 | - * |
|
| 171 | - * @param int $id |
|
| 172 | - * @return bool |
|
| 173 | - */ |
|
| 174 | - public function isAncestor($id) |
|
| 175 | - { |
|
| 176 | - if ( ! is_null($this->getAncestor($id))) { |
|
| 177 | - return true; |
|
| 178 | - } |
|
| 179 | - |
|
| 180 | - return false; |
|
| 181 | - } |
|
| 182 | - |
|
| 183 | - /** |
|
| 184 | - * Attempts to get an ancestor node by the given id. |
|
| 185 | - * |
|
| 186 | - * @param int $id |
|
| 187 | - * @return null|AbstractNode |
|
| 188 | - */ |
|
| 189 | - public function getAncestor($id) |
|
| 190 | - { |
|
| 191 | - if ( ! is_null($this->parent)) { |
|
| 192 | - if ($this->parent->id() == $id) { |
|
| 193 | - return $this->parent; |
|
| 194 | - } |
|
| 195 | - |
|
| 196 | - return $this->parent->getAncestor($id); |
|
| 197 | - } |
|
| 198 | - |
|
| 199 | - return null; |
|
| 200 | - } |
|
| 201 | - |
|
| 202 | - /** |
|
| 203 | - * Shortcut to return the first child. |
|
| 204 | - * |
|
| 205 | - * @return AbstractNode |
|
| 206 | - * @uses $this->getChild() |
|
| 207 | - */ |
|
| 208 | - public function firstChild() |
|
| 209 | - { |
|
| 210 | - reset($this->children); |
|
| 211 | - $key = key($this->children); |
|
| 212 | - |
|
| 213 | - return $this->getChild($key); |
|
| 214 | - } |
|
| 215 | - |
|
| 216 | - /** |
|
| 217 | - * Attempts to get the last child. |
|
| 218 | - * |
|
| 219 | - * @return AbstractNode |
|
| 220 | - */ |
|
| 221 | - public function lastChild() |
|
| 222 | - { |
|
| 223 | - end($this->children); |
|
| 224 | - $key = key($this->children); |
|
| 225 | - |
|
| 226 | - return $this->getChild($key); |
|
| 227 | - } |
|
| 228 | - |
|
| 229 | - /** |
|
| 230 | - * Attempts to get the next sibling. |
|
| 231 | - * |
|
| 232 | - * @return AbstractNode |
|
| 233 | - * @throws ParentNotFoundException |
|
| 234 | - */ |
|
| 235 | - public function nextSibling() |
|
| 236 | - { |
|
| 237 | - if (is_null($this->parent)) { |
|
| 238 | - throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 239 | - } |
|
| 240 | - |
|
| 241 | - return $this->parent->nextChild($this->id); |
|
| 242 | - } |
|
| 243 | - |
|
| 244 | - /** |
|
| 245 | - * Attempts to get the previous sibling |
|
| 246 | - * |
|
| 247 | - * @return AbstractNode |
|
| 248 | - * @throws ParentNotFoundException |
|
| 249 | - */ |
|
| 250 | - public function previousSibling() |
|
| 251 | - { |
|
| 252 | - if (is_null($this->parent)) { |
|
| 253 | - throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 254 | - } |
|
| 255 | - |
|
| 256 | - return $this->parent->previousChild($this->id); |
|
| 257 | - } |
|
| 258 | - |
|
| 259 | - /** |
|
| 260 | - * Gets the tag object of this node. |
|
| 261 | - * |
|
| 262 | - * @return Tag |
|
| 263 | - */ |
|
| 264 | - public function getTag() |
|
| 265 | - { |
|
| 266 | - return $this->tag; |
|
| 267 | - } |
|
| 268 | - |
|
| 269 | - /** |
|
| 270 | - * A wrapper method that simply calls the getAttribute method |
|
| 271 | - * on the tag of this node. |
|
| 272 | - * |
|
| 273 | - * @return array |
|
| 274 | - */ |
|
| 275 | - public function getAttributes() |
|
| 276 | - { |
|
| 277 | - $attributes = $this->tag->getAttributes(); |
|
| 278 | - foreach ($attributes as $name => $info) { |
|
| 279 | - $attributes[$name] = $info['value']; |
|
| 280 | - } |
|
| 281 | - |
|
| 282 | - return $attributes; |
|
| 283 | - } |
|
| 284 | - |
|
| 285 | - /** |
|
| 286 | - * A wrapper method that simply calls the getAttribute method |
|
| 287 | - * on the tag of this node. |
|
| 288 | - * |
|
| 289 | - * @param string $key |
|
| 290 | - * @return mixed |
|
| 291 | - */ |
|
| 292 | - public function getAttribute($key) |
|
| 293 | - { |
|
| 294 | - $attribute = $this->tag->getAttribute($key); |
|
| 295 | - if ( ! is_null($attribute)) { |
|
| 296 | - $attribute = $attribute['value']; |
|
| 297 | - } |
|
| 298 | - |
|
| 299 | - return $attribute; |
|
| 300 | - } |
|
| 301 | - |
|
| 302 | - /** |
|
| 303 | - * A wrapper method that simply calls the setAttribute method |
|
| 304 | - * on the tag of this node. |
|
| 305 | - * |
|
| 306 | - * @param string $key |
|
| 307 | - * @param string $value |
|
| 308 | - * @return $this |
|
| 309 | - */ |
|
| 310 | - public function setAttribute($key, $value) |
|
| 311 | - { |
|
| 312 | - $this->tag->setAttribute($key, $value); |
|
| 313 | - |
|
| 314 | - return $this; |
|
| 315 | - } |
|
| 316 | - |
|
| 317 | - /** |
|
| 318 | - * Function to locate a specific ancestor tag in the path to the root. |
|
| 319 | - * |
|
| 320 | - * @param string $tag |
|
| 321 | - * @return AbstractNode |
|
| 322 | - * @throws ParentNotFoundException |
|
| 323 | - */ |
|
| 324 | - public function ancestorByTag($tag) |
|
| 325 | - { |
|
| 326 | - // Start by including ourselves in the comparison. |
|
| 327 | - $node = $this; |
|
| 328 | - |
|
| 329 | - while ( ! is_null($node)) { |
|
| 330 | - if ($node->tag->name() == $tag) { |
|
| 331 | - return $node; |
|
| 332 | - } |
|
| 333 | - |
|
| 334 | - $node = $node->getParent(); |
|
| 335 | - } |
|
| 336 | - |
|
| 337 | - throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag'); |
|
| 338 | - } |
|
| 339 | - |
|
| 340 | - /** |
|
| 341 | - * Find elements by css selector |
|
| 342 | - * |
|
| 343 | - * @param string $selector |
|
| 344 | - * @param int $nth |
|
| 345 | - * @return array|AbstractNode |
|
| 346 | - */ |
|
| 347 | - public function find($selector, $nth = null) |
|
| 348 | - { |
|
| 349 | - $selector = new Selector($selector); |
|
| 350 | - $nodes = $selector->find($this); |
|
| 351 | - |
|
| 352 | - if ( ! is_null($nth)) { |
|
| 353 | - // return nth-element or array |
|
| 354 | - if (isset($nodes[$nth])) { |
|
| 355 | - return $nodes[$nth]; |
|
| 356 | - } |
|
| 357 | - |
|
| 358 | - return null; |
|
| 359 | - } |
|
| 360 | - |
|
| 361 | - return $nodes; |
|
| 362 | - } |
|
| 363 | - |
|
| 364 | - /** |
|
| 365 | - * Function to try a few tricks to determine the displayed size of an img on the page. |
|
| 366 | - * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types. |
|
| 367 | - * |
|
| 368 | - * Future enhancement: |
|
| 369 | - * Look in the tag to see if there is a class or id specified that has a height or width attribute to it. |
|
| 370 | - * |
|
| 371 | - * Far future enhancement |
|
| 372 | - * Look at all the parent tags of this image to see if they specify a class or id that has an img selector that specifies a height or width |
|
| 373 | - * Note that in this case, the class or id will have the img sub-selector for it to apply to the image. |
|
| 374 | - * |
|
| 375 | - * ridiculously far future development |
|
| 376 | - * If the class or id is specified in a SEPARATE css file that's not on the page, go get it and do what we were just doing for the ones on the page. |
|
| 377 | - * |
|
| 378 | - * @author John Schlick |
|
| 379 | - * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out. |
|
| 380 | - */ |
|
| 381 | - public function get_display_size() |
|
| 382 | - { |
|
| 383 | - $width = -1; |
|
| 384 | - $height = -1; |
|
| 385 | - |
|
| 386 | - if ($this->tag->name() != 'img') { |
|
| 387 | - return false; |
|
| 388 | - } |
|
| 389 | - |
|
| 390 | - // See if there is a height or width attribute in the tag itself. |
|
| 391 | - if ( ! is_null($this->tag->getAttribute('width'))) { |
|
| 392 | - $width = $this->tag->getAttribute('width'); |
|
| 393 | - } |
|
| 394 | - |
|
| 395 | - if ( ! is_null($this->tag->getAttribute('height'))) { |
|
| 396 | - $height = $this->tag->getAttribute('height'); |
|
| 397 | - } |
|
| 398 | - |
|
| 399 | - // Now look for an inline style. |
|
| 400 | - if ( ! is_null($this->tag->getAttribute('style'))) { |
|
| 401 | - // Thanks to user 'gnarf' from stackoverflow for this regular expression. |
|
| 402 | - $attributes = []; |
|
| 403 | - preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches, |
|
| 404 | - PREG_SET_ORDER); |
|
| 405 | - foreach ($matches as $match) { |
|
| 406 | - $attributes[$match[1]] = $match[2]; |
|
| 407 | - } |
|
| 408 | - |
|
| 409 | - $width = $this->getLength($attributes, $width, 'width'); |
|
| 410 | - $height = $this->getLength($attributes, $width, 'height'); |
|
| 411 | - } |
|
| 412 | - |
|
| 413 | - $result = [ |
|
| 414 | - 'height' => $height, |
|
| 415 | - 'width' => $width, |
|
| 416 | - ]; |
|
| 417 | - |
|
| 418 | - return $result; |
|
| 419 | - } |
|
| 420 | - |
|
| 421 | - /** |
|
| 422 | - * If there is a length in the style attributes use it. |
|
| 423 | - * |
|
| 424 | - * @param array $attributes |
|
| 425 | - * @param int $length |
|
| 426 | - * @param string $key |
|
| 427 | - * @return int |
|
| 428 | - */ |
|
| 429 | - protected function getLength(array $attributes, $length, $key) |
|
| 430 | - { |
|
| 431 | - if (isset($attributes[$key]) && $length == -1) { |
|
| 432 | - // check that the last two characters are px (pixels) |
|
| 433 | - if (strtolower(substr($attributes[$key], -2)) == 'px') { |
|
| 434 | - $proposed_length = substr($attributes[$key], 0, -2); |
|
| 435 | - // Now make sure that it's an integer and not something stupid. |
|
| 436 | - if (filter_var($proposed_length, FILTER_VALIDATE_INT)) { |
|
| 437 | - $length = $proposed_length; |
|
| 438 | - } |
|
| 439 | - } |
|
| 440 | - } |
|
| 441 | - |
|
| 442 | - return $length; |
|
| 443 | - } |
|
| 444 | - |
|
| 445 | - /** |
|
| 446 | - * Gets the inner html of this node. |
|
| 447 | - * |
|
| 448 | - * @return string |
|
| 449 | - */ |
|
| 450 | - abstract public function innerHtml(); |
|
| 451 | - |
|
| 452 | - /** |
|
| 453 | - * Gets the html of this node, including it's own |
|
| 454 | - * tag. |
|
| 455 | - * |
|
| 456 | - * @return string |
|
| 457 | - */ |
|
| 458 | - abstract public function outerHtml(); |
|
| 459 | - |
|
| 460 | - /** |
|
| 461 | - * Gets the text of this node (if there is any text). |
|
| 462 | - * |
|
| 463 | - * @return string |
|
| 464 | - */ |
|
| 465 | - abstract public function text(); |
|
| 466 | - |
|
| 467 | - /** |
|
| 468 | - * Call this when something in the node tree has changed. Like a child has been added |
|
| 469 | - * or a parent has been changed. |
|
| 470 | - * |
|
| 471 | - * @return void |
|
| 472 | - */ |
|
| 473 | - abstract protected function clear(); |
|
| 19 | + /** |
|
| 20 | + * Contains the tag name/type |
|
| 21 | + * |
|
| 22 | + * @var \PHPHtmlParser\Dom\Tag |
|
| 23 | + */ |
|
| 24 | + protected $tag; |
|
| 25 | + |
|
| 26 | + /** |
|
| 27 | + * Contains a list of attributes on this tag. |
|
| 28 | + * |
|
| 29 | + * @var array |
|
| 30 | + */ |
|
| 31 | + protected $attr = []; |
|
| 32 | + |
|
| 33 | + /** |
|
| 34 | + * Contains the parent Node. |
|
| 35 | + * |
|
| 36 | + * @var InnerNode |
|
| 37 | + */ |
|
| 38 | + protected $parent = null; |
|
| 39 | + |
|
| 40 | + /** |
|
| 41 | + * The unique id of the class. Given by PHP. |
|
| 42 | + * |
|
| 43 | + * @var string |
|
| 44 | + */ |
|
| 45 | + protected $id; |
|
| 46 | + |
|
| 47 | + /** |
|
| 48 | + * The encoding class used to encode strings. |
|
| 49 | + * |
|
| 50 | + * @var mixed |
|
| 51 | + */ |
|
| 52 | + protected $encode; |
|
| 53 | + |
|
| 54 | + /** |
|
| 55 | + * Creates a unique spl hash for this node. |
|
| 56 | + */ |
|
| 57 | + public function __construct() |
|
| 58 | + { |
|
| 59 | + $this->id = spl_object_hash($this); |
|
| 60 | + } |
|
| 61 | + |
|
| 62 | + /** |
|
| 63 | + * Magic get method for attributes and certain methods. |
|
| 64 | + * |
|
| 65 | + * @param string $key |
|
| 66 | + * @return mixed |
|
| 67 | + */ |
|
| 68 | + public function __get($key) |
|
| 69 | + { |
|
| 70 | + // check attribute first |
|
| 71 | + if ( ! is_null($this->getAttribute($key))) { |
|
| 72 | + return $this->getAttribute($key); |
|
| 73 | + } |
|
| 74 | + switch (strtolower($key)) { |
|
| 75 | + case 'outerhtml': |
|
| 76 | + return $this->outerHtml(); |
|
| 77 | + case 'innerhtml': |
|
| 78 | + return $this->innerHtml(); |
|
| 79 | + case 'text': |
|
| 80 | + return $this->text(); |
|
| 81 | + } |
|
| 82 | + |
|
| 83 | + return null; |
|
| 84 | + } |
|
| 85 | + |
|
| 86 | + /** |
|
| 87 | + * Attempts to clear out any object references. |
|
| 88 | + */ |
|
| 89 | + public function __destruct() |
|
| 90 | + { |
|
| 91 | + $this->tag = null; |
|
| 92 | + $this->attr = []; |
|
| 93 | + $this->parent = null; |
|
| 94 | + $this->children = []; |
|
| 95 | + } |
|
| 96 | + |
|
| 97 | + /** |
|
| 98 | + * Simply calls the outer text method. |
|
| 99 | + * |
|
| 100 | + * @return string |
|
| 101 | + */ |
|
| 102 | + public function __toString() |
|
| 103 | + { |
|
| 104 | + return $this->outerHtml(); |
|
| 105 | + } |
|
| 106 | + |
|
| 107 | + /** |
|
| 108 | + * Returns the id of this object. |
|
| 109 | + */ |
|
| 110 | + public function id() |
|
| 111 | + { |
|
| 112 | + return $this->id; |
|
| 113 | + } |
|
| 114 | + |
|
| 115 | + /** |
|
| 116 | + * Returns the parent of node. |
|
| 117 | + * |
|
| 118 | + * @return AbstractNode |
|
| 119 | + */ |
|
| 120 | + public function getParent() |
|
| 121 | + { |
|
| 122 | + return $this->parent; |
|
| 123 | + } |
|
| 124 | + |
|
| 125 | + /** |
|
| 126 | + * Sets the parent node. |
|
| 127 | + * |
|
| 128 | + * @param InnerNode $parent |
|
| 129 | + * @return $this |
|
| 130 | + * @throws CircularException |
|
| 131 | + */ |
|
| 132 | + public function setParent(InnerNode $parent) |
|
| 133 | + { |
|
| 134 | + // remove from old parent |
|
| 135 | + if ( ! is_null($this->parent)) { |
|
| 136 | + if ($this->parent->id() == $parent->id()) { |
|
| 137 | + // already the parent |
|
| 138 | + return $this; |
|
| 139 | + } |
|
| 140 | + |
|
| 141 | + $this->parent->removeChild($this->id); |
|
| 142 | + } |
|
| 143 | + |
|
| 144 | + $this->parent = $parent; |
|
| 145 | + |
|
| 146 | + // assign child to parent |
|
| 147 | + $this->parent->addChild($this); |
|
| 148 | + |
|
| 149 | + //clear any cache |
|
| 150 | + $this->clear(); |
|
| 151 | + |
|
| 152 | + return $this; |
|
| 153 | + } |
|
| 154 | + |
|
| 155 | + /** |
|
| 156 | + * Sets the encoding class to this node. |
|
| 157 | + * |
|
| 158 | + * @param Encode $encode |
|
| 159 | + * @return void |
|
| 160 | + */ |
|
| 161 | + public function propagateEncoding(Encode $encode) |
|
| 162 | + { |
|
| 163 | + $this->encode = $encode; |
|
| 164 | + $this->tag->setEncoding($encode); |
|
| 165 | + } |
|
| 166 | + |
|
| 167 | + /** |
|
| 168 | + * Checks if the given node id is an ancestor of |
|
| 169 | + * the current node. |
|
| 170 | + * |
|
| 171 | + * @param int $id |
|
| 172 | + * @return bool |
|
| 173 | + */ |
|
| 174 | + public function isAncestor($id) |
|
| 175 | + { |
|
| 176 | + if ( ! is_null($this->getAncestor($id))) { |
|
| 177 | + return true; |
|
| 178 | + } |
|
| 179 | + |
|
| 180 | + return false; |
|
| 181 | + } |
|
| 182 | + |
|
| 183 | + /** |
|
| 184 | + * Attempts to get an ancestor node by the given id. |
|
| 185 | + * |
|
| 186 | + * @param int $id |
|
| 187 | + * @return null|AbstractNode |
|
| 188 | + */ |
|
| 189 | + public function getAncestor($id) |
|
| 190 | + { |
|
| 191 | + if ( ! is_null($this->parent)) { |
|
| 192 | + if ($this->parent->id() == $id) { |
|
| 193 | + return $this->parent; |
|
| 194 | + } |
|
| 195 | + |
|
| 196 | + return $this->parent->getAncestor($id); |
|
| 197 | + } |
|
| 198 | + |
|
| 199 | + return null; |
|
| 200 | + } |
|
| 201 | + |
|
| 202 | + /** |
|
| 203 | + * Shortcut to return the first child. |
|
| 204 | + * |
|
| 205 | + * @return AbstractNode |
|
| 206 | + * @uses $this->getChild() |
|
| 207 | + */ |
|
| 208 | + public function firstChild() |
|
| 209 | + { |
|
| 210 | + reset($this->children); |
|
| 211 | + $key = key($this->children); |
|
| 212 | + |
|
| 213 | + return $this->getChild($key); |
|
| 214 | + } |
|
| 215 | + |
|
| 216 | + /** |
|
| 217 | + * Attempts to get the last child. |
|
| 218 | + * |
|
| 219 | + * @return AbstractNode |
|
| 220 | + */ |
|
| 221 | + public function lastChild() |
|
| 222 | + { |
|
| 223 | + end($this->children); |
|
| 224 | + $key = key($this->children); |
|
| 225 | + |
|
| 226 | + return $this->getChild($key); |
|
| 227 | + } |
|
| 228 | + |
|
| 229 | + /** |
|
| 230 | + * Attempts to get the next sibling. |
|
| 231 | + * |
|
| 232 | + * @return AbstractNode |
|
| 233 | + * @throws ParentNotFoundException |
|
| 234 | + */ |
|
| 235 | + public function nextSibling() |
|
| 236 | + { |
|
| 237 | + if (is_null($this->parent)) { |
|
| 238 | + throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 239 | + } |
|
| 240 | + |
|
| 241 | + return $this->parent->nextChild($this->id); |
|
| 242 | + } |
|
| 243 | + |
|
| 244 | + /** |
|
| 245 | + * Attempts to get the previous sibling |
|
| 246 | + * |
|
| 247 | + * @return AbstractNode |
|
| 248 | + * @throws ParentNotFoundException |
|
| 249 | + */ |
|
| 250 | + public function previousSibling() |
|
| 251 | + { |
|
| 252 | + if (is_null($this->parent)) { |
|
| 253 | + throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 254 | + } |
|
| 255 | + |
|
| 256 | + return $this->parent->previousChild($this->id); |
|
| 257 | + } |
|
| 258 | + |
|
| 259 | + /** |
|
| 260 | + * Gets the tag object of this node. |
|
| 261 | + * |
|
| 262 | + * @return Tag |
|
| 263 | + */ |
|
| 264 | + public function getTag() |
|
| 265 | + { |
|
| 266 | + return $this->tag; |
|
| 267 | + } |
|
| 268 | + |
|
| 269 | + /** |
|
| 270 | + * A wrapper method that simply calls the getAttribute method |
|
| 271 | + * on the tag of this node. |
|
| 272 | + * |
|
| 273 | + * @return array |
|
| 274 | + */ |
|
| 275 | + public function getAttributes() |
|
| 276 | + { |
|
| 277 | + $attributes = $this->tag->getAttributes(); |
|
| 278 | + foreach ($attributes as $name => $info) { |
|
| 279 | + $attributes[$name] = $info['value']; |
|
| 280 | + } |
|
| 281 | + |
|
| 282 | + return $attributes; |
|
| 283 | + } |
|
| 284 | + |
|
| 285 | + /** |
|
| 286 | + * A wrapper method that simply calls the getAttribute method |
|
| 287 | + * on the tag of this node. |
|
| 288 | + * |
|
| 289 | + * @param string $key |
|
| 290 | + * @return mixed |
|
| 291 | + */ |
|
| 292 | + public function getAttribute($key) |
|
| 293 | + { |
|
| 294 | + $attribute = $this->tag->getAttribute($key); |
|
| 295 | + if ( ! is_null($attribute)) { |
|
| 296 | + $attribute = $attribute['value']; |
|
| 297 | + } |
|
| 298 | + |
|
| 299 | + return $attribute; |
|
| 300 | + } |
|
| 301 | + |
|
| 302 | + /** |
|
| 303 | + * A wrapper method that simply calls the setAttribute method |
|
| 304 | + * on the tag of this node. |
|
| 305 | + * |
|
| 306 | + * @param string $key |
|
| 307 | + * @param string $value |
|
| 308 | + * @return $this |
|
| 309 | + */ |
|
| 310 | + public function setAttribute($key, $value) |
|
| 311 | + { |
|
| 312 | + $this->tag->setAttribute($key, $value); |
|
| 313 | + |
|
| 314 | + return $this; |
|
| 315 | + } |
|
| 316 | + |
|
| 317 | + /** |
|
| 318 | + * Function to locate a specific ancestor tag in the path to the root. |
|
| 319 | + * |
|
| 320 | + * @param string $tag |
|
| 321 | + * @return AbstractNode |
|
| 322 | + * @throws ParentNotFoundException |
|
| 323 | + */ |
|
| 324 | + public function ancestorByTag($tag) |
|
| 325 | + { |
|
| 326 | + // Start by including ourselves in the comparison. |
|
| 327 | + $node = $this; |
|
| 328 | + |
|
| 329 | + while ( ! is_null($node)) { |
|
| 330 | + if ($node->tag->name() == $tag) { |
|
| 331 | + return $node; |
|
| 332 | + } |
|
| 333 | + |
|
| 334 | + $node = $node->getParent(); |
|
| 335 | + } |
|
| 336 | + |
|
| 337 | + throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag'); |
|
| 338 | + } |
|
| 339 | + |
|
| 340 | + /** |
|
| 341 | + * Find elements by css selector |
|
| 342 | + * |
|
| 343 | + * @param string $selector |
|
| 344 | + * @param int $nth |
|
| 345 | + * @return array|AbstractNode |
|
| 346 | + */ |
|
| 347 | + public function find($selector, $nth = null) |
|
| 348 | + { |
|
| 349 | + $selector = new Selector($selector); |
|
| 350 | + $nodes = $selector->find($this); |
|
| 351 | + |
|
| 352 | + if ( ! is_null($nth)) { |
|
| 353 | + // return nth-element or array |
|
| 354 | + if (isset($nodes[$nth])) { |
|
| 355 | + return $nodes[$nth]; |
|
| 356 | + } |
|
| 357 | + |
|
| 358 | + return null; |
|
| 359 | + } |
|
| 360 | + |
|
| 361 | + return $nodes; |
|
| 362 | + } |
|
| 363 | + |
|
| 364 | + /** |
|
| 365 | + * Function to try a few tricks to determine the displayed size of an img on the page. |
|
| 366 | + * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types. |
|
| 367 | + * |
|
| 368 | + * Future enhancement: |
|
| 369 | + * Look in the tag to see if there is a class or id specified that has a height or width attribute to it. |
|
| 370 | + * |
|
| 371 | + * Far future enhancement |
|
| 372 | + * Look at all the parent tags of this image to see if they specify a class or id that has an img selector that specifies a height or width |
|
| 373 | + * Note that in this case, the class or id will have the img sub-selector for it to apply to the image. |
|
| 374 | + * |
|
| 375 | + * ridiculously far future development |
|
| 376 | + * If the class or id is specified in a SEPARATE css file that's not on the page, go get it and do what we were just doing for the ones on the page. |
|
| 377 | + * |
|
| 378 | + * @author John Schlick |
|
| 379 | + * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out. |
|
| 380 | + */ |
|
| 381 | + public function get_display_size() |
|
| 382 | + { |
|
| 383 | + $width = -1; |
|
| 384 | + $height = -1; |
|
| 385 | + |
|
| 386 | + if ($this->tag->name() != 'img') { |
|
| 387 | + return false; |
|
| 388 | + } |
|
| 389 | + |
|
| 390 | + // See if there is a height or width attribute in the tag itself. |
|
| 391 | + if ( ! is_null($this->tag->getAttribute('width'))) { |
|
| 392 | + $width = $this->tag->getAttribute('width'); |
|
| 393 | + } |
|
| 394 | + |
|
| 395 | + if ( ! is_null($this->tag->getAttribute('height'))) { |
|
| 396 | + $height = $this->tag->getAttribute('height'); |
|
| 397 | + } |
|
| 398 | + |
|
| 399 | + // Now look for an inline style. |
|
| 400 | + if ( ! is_null($this->tag->getAttribute('style'))) { |
|
| 401 | + // Thanks to user 'gnarf' from stackoverflow for this regular expression. |
|
| 402 | + $attributes = []; |
|
| 403 | + preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches, |
|
| 404 | + PREG_SET_ORDER); |
|
| 405 | + foreach ($matches as $match) { |
|
| 406 | + $attributes[$match[1]] = $match[2]; |
|
| 407 | + } |
|
| 408 | + |
|
| 409 | + $width = $this->getLength($attributes, $width, 'width'); |
|
| 410 | + $height = $this->getLength($attributes, $width, 'height'); |
|
| 411 | + } |
|
| 412 | + |
|
| 413 | + $result = [ |
|
| 414 | + 'height' => $height, |
|
| 415 | + 'width' => $width, |
|
| 416 | + ]; |
|
| 417 | + |
|
| 418 | + return $result; |
|
| 419 | + } |
|
| 420 | + |
|
| 421 | + /** |
|
| 422 | + * If there is a length in the style attributes use it. |
|
| 423 | + * |
|
| 424 | + * @param array $attributes |
|
| 425 | + * @param int $length |
|
| 426 | + * @param string $key |
|
| 427 | + * @return int |
|
| 428 | + */ |
|
| 429 | + protected function getLength(array $attributes, $length, $key) |
|
| 430 | + { |
|
| 431 | + if (isset($attributes[$key]) && $length == -1) { |
|
| 432 | + // check that the last two characters are px (pixels) |
|
| 433 | + if (strtolower(substr($attributes[$key], -2)) == 'px') { |
|
| 434 | + $proposed_length = substr($attributes[$key], 0, -2); |
|
| 435 | + // Now make sure that it's an integer and not something stupid. |
|
| 436 | + if (filter_var($proposed_length, FILTER_VALIDATE_INT)) { |
|
| 437 | + $length = $proposed_length; |
|
| 438 | + } |
|
| 439 | + } |
|
| 440 | + } |
|
| 441 | + |
|
| 442 | + return $length; |
|
| 443 | + } |
|
| 444 | + |
|
| 445 | + /** |
|
| 446 | + * Gets the inner html of this node. |
|
| 447 | + * |
|
| 448 | + * @return string |
|
| 449 | + */ |
|
| 450 | + abstract public function innerHtml(); |
|
| 451 | + |
|
| 452 | + /** |
|
| 453 | + * Gets the html of this node, including it's own |
|
| 454 | + * tag. |
|
| 455 | + * |
|
| 456 | + * @return string |
|
| 457 | + */ |
|
| 458 | + abstract public function outerHtml(); |
|
| 459 | + |
|
| 460 | + /** |
|
| 461 | + * Gets the text of this node (if there is any text). |
|
| 462 | + * |
|
| 463 | + * @return string |
|
| 464 | + */ |
|
| 465 | + abstract public function text(); |
|
| 466 | + |
|
| 467 | + /** |
|
| 468 | + * Call this when something in the node tree has changed. Like a child has been added |
|
| 469 | + * or a parent has been changed. |
|
| 470 | + * |
|
| 471 | + * @return void |
|
| 472 | + */ |
|
| 473 | + abstract protected function clear(); |
|
| 474 | 474 | } |
@@ -15,361 +15,361 @@ |
||
| 15 | 15 | class Selector |
| 16 | 16 | { |
| 17 | 17 | |
| 18 | - /** |
|
| 19 | - * Pattern of CSS selectors, modified from 'mootools' |
|
| 20 | - * |
|
| 21 | - * @var string |
|
| 22 | - */ |
|
| 23 | - protected $pattern = "/([\w-:\*>]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is"; |
|
| 24 | - |
|
| 25 | - protected $selectors = []; |
|
| 26 | - |
|
| 27 | - /** |
|
| 28 | - * Constructs with the selector string |
|
| 29 | - * |
|
| 30 | - * @param string $selector |
|
| 31 | - */ |
|
| 32 | - public function __construct($selector) |
|
| 33 | - { |
|
| 34 | - $this->parseSelectorString($selector); |
|
| 35 | - } |
|
| 36 | - |
|
| 37 | - /** |
|
| 38 | - * Returns the selectors that where found in __construct |
|
| 39 | - * |
|
| 40 | - * @return array |
|
| 41 | - */ |
|
| 42 | - public function getSelectors() |
|
| 43 | - { |
|
| 44 | - return $this->selectors; |
|
| 45 | - } |
|
| 46 | - |
|
| 47 | - /** |
|
| 48 | - * Attempts to find the selectors starting from the given |
|
| 49 | - * node object. |
|
| 50 | - * |
|
| 51 | - * @param AbstractNode $node |
|
| 52 | - * @return array|Collection |
|
| 53 | - */ |
|
| 54 | - public function find(AbstractNode $node) |
|
| 55 | - { |
|
| 56 | - $results = new Collection; |
|
| 57 | - foreach ($this->selectors as $selector) { |
|
| 58 | - $nodes = [$node]; |
|
| 59 | - if (count($selector) == 0) { |
|
| 60 | - continue; |
|
| 61 | - } |
|
| 62 | - |
|
| 63 | - $options = []; |
|
| 64 | - foreach ($selector as $rule) { |
|
| 65 | - if ($rule['alterNext']) { |
|
| 66 | - $options[] = $this->alterNext($rule); |
|
| 67 | - continue; |
|
| 68 | - } |
|
| 69 | - $nodes = $this->seek($nodes, $rule, $options); |
|
| 70 | - // clear the options |
|
| 71 | - $options = []; |
|
| 72 | - } |
|
| 73 | - |
|
| 74 | - // this is the final set of nodes |
|
| 75 | - foreach ($nodes as $result) { |
|
| 76 | - $results[] = $result; |
|
| 77 | - } |
|
| 78 | - } |
|
| 79 | - |
|
| 80 | - return $results; |
|
| 81 | - } |
|
| 82 | - |
|
| 83 | - /** |
|
| 84 | - * Parses the selector string |
|
| 85 | - * |
|
| 86 | - * @param string $selector |
|
| 87 | - */ |
|
| 88 | - protected function parseSelectorString($selector) |
|
| 89 | - { |
|
| 90 | - $matches = []; |
|
| 91 | - preg_match_all($this->pattern, trim($selector).' ', $matches, PREG_SET_ORDER); |
|
| 92 | - |
|
| 93 | - // skip tbody |
|
| 94 | - $result = []; |
|
| 95 | - foreach ($matches as $match) { |
|
| 96 | - // default values |
|
| 97 | - $tag = strtolower(trim($match[1])); |
|
| 98 | - $operator = '='; |
|
| 99 | - $key = null; |
|
| 100 | - $value = null; |
|
| 101 | - $noKey = false; |
|
| 102 | - $alterNext = false; |
|
| 103 | - |
|
| 104 | - // check for elements that alter the behavior of the next element |
|
| 105 | - if ($tag == '>') { |
|
| 106 | - $alterNext = true; |
|
| 107 | - } |
|
| 108 | - |
|
| 109 | - // check for id selector |
|
| 110 | - if ( ! empty($match[2])) { |
|
| 111 | - $key = 'id'; |
|
| 112 | - $value = $match[2]; |
|
| 113 | - } |
|
| 114 | - |
|
| 115 | - // check for class selector |
|
| 116 | - if ( ! empty($match[3])) { |
|
| 117 | - $key = 'class'; |
|
| 118 | - $value = $match[3]; |
|
| 119 | - } |
|
| 120 | - |
|
| 121 | - // and final attribute selector |
|
| 122 | - if ( ! empty($match[4])) { |
|
| 123 | - $key = strtolower($match[4]); |
|
| 124 | - } |
|
| 125 | - if ( ! empty($match[5])) { |
|
| 126 | - $operator = $match[5]; |
|
| 127 | - } |
|
| 128 | - if ( ! empty($match[6])) { |
|
| 129 | - $value = $match[6]; |
|
| 130 | - } |
|
| 131 | - |
|
| 132 | - // check for elements that do not have a specified attribute |
|
| 133 | - if (isset($key[0]) && $key[0] == '!') { |
|
| 134 | - $key = substr($key, 1); |
|
| 135 | - $noKey = true; |
|
| 136 | - } |
|
| 137 | - |
|
| 138 | - $result[] = [ |
|
| 139 | - 'tag' => $tag, |
|
| 140 | - 'key' => $key, |
|
| 141 | - 'value' => $value, |
|
| 142 | - 'operator' => $operator, |
|
| 143 | - 'noKey' => $noKey, |
|
| 144 | - 'alterNext' => $alterNext, |
|
| 145 | - ]; |
|
| 146 | - if (trim($match[7]) == ',') { |
|
| 147 | - $this->selectors[] = $result; |
|
| 148 | - $result = []; |
|
| 149 | - } |
|
| 150 | - } |
|
| 151 | - |
|
| 152 | - // save last results |
|
| 153 | - if (count($result) > 0) { |
|
| 154 | - $this->selectors[] = $result; |
|
| 155 | - } |
|
| 156 | - } |
|
| 157 | - |
|
| 158 | - /** |
|
| 159 | - * Attempts to find all children that match the rule |
|
| 160 | - * given. |
|
| 161 | - * |
|
| 162 | - * @param array $nodes |
|
| 163 | - * @param array $rule |
|
| 164 | - * @param array $options |
|
| 165 | - * @return array |
|
| 166 | - * @recursive |
|
| 167 | - */ |
|
| 168 | - protected function seek(array $nodes, array $rule, array $options) |
|
| 169 | - { |
|
| 170 | - // XPath index |
|
| 171 | - if (count($rule['tag']) > 0 && |
|
| 172 | - count($rule['key']) > 0 && |
|
| 173 | - is_numeric($rule['key']) |
|
| 174 | - ) { |
|
| 175 | - $count = 0; |
|
| 176 | - /** @var AbstractNode $node */ |
|
| 177 | - foreach ($nodes as $node) { |
|
| 178 | - if ($rule['tag'] == '*' || |
|
| 179 | - $rule['tag'] == $node->getTag()->name()) { |
|
| 180 | - ++$count; |
|
| 181 | - if ($count == $rule['key']) { |
|
| 182 | - // found the node we wanted |
|
| 183 | - return [$node]; |
|
| 184 | - } |
|
| 185 | - } |
|
| 186 | - } |
|
| 187 | - |
|
| 188 | - return []; |
|
| 189 | - } |
|
| 190 | - |
|
| 191 | - $options = $this->flattenOptions($options); |
|
| 192 | - |
|
| 193 | - $return = []; |
|
| 194 | - /** @var InnerNode $node */ |
|
| 195 | - foreach ($nodes as $node) { |
|
| 196 | - // check if we are a leaf |
|
| 197 | - if ($node instanceof LeafNode || |
|
| 198 | - ! $node->hasChildren()) { |
|
| 199 | - continue; |
|
| 200 | - } |
|
| 201 | - |
|
| 202 | - $children = []; |
|
| 203 | - $child = $node->firstChild(); |
|
| 204 | - while ( ! is_null($child)) { |
|
| 205 | - // wild card, grab all |
|
| 206 | - if ($rule['tag'] == '*' && is_null($rule['key'])) { |
|
| 207 | - $return[] = $child; |
|
| 208 | - try { |
|
| 209 | - $child = $node->nextChild($child->id()); |
|
| 210 | - } catch (ChildNotFoundException $e) { |
|
| 211 | - // no more children |
|
| 212 | - $child = null; |
|
| 213 | - } |
|
| 214 | - continue; |
|
| 215 | - } |
|
| 216 | - |
|
| 217 | - $pass = true; |
|
| 218 | - // check tag |
|
| 219 | - if ( ! empty($rule['tag']) && $rule['tag'] != $child->getTag()->name() && |
|
| 220 | - $rule['tag'] != '*' |
|
| 221 | - ) { |
|
| 222 | - // child failed tag check |
|
| 223 | - $pass = false; |
|
| 224 | - } |
|
| 225 | - |
|
| 226 | - // check key |
|
| 227 | - if ($pass && ! is_null($rule['key'])) { |
|
| 228 | - if ($rule['noKey']) { |
|
| 229 | - if ( ! is_null($child->getAttribute($rule['key']))) { |
|
| 230 | - $pass = false; |
|
| 231 | - } |
|
| 232 | - } else { |
|
| 233 | - if ($rule['key'] != 'plaintext' && |
|
| 234 | - is_null($child->getAttribute($rule['key'])) |
|
| 235 | - ) { |
|
| 236 | - $pass = false; |
|
| 237 | - } |
|
| 238 | - } |
|
| 239 | - } |
|
| 240 | - |
|
| 241 | - // compare values |
|
| 242 | - if ($pass && ! is_null($rule['key']) && |
|
| 243 | - ! is_null($rule['value']) && $rule['value'] != '*' |
|
| 244 | - ) { |
|
| 245 | - if ($rule['key'] == 'plaintext') { |
|
| 246 | - // plaintext search |
|
| 247 | - $nodeValue = $child->text(); |
|
| 248 | - } else { |
|
| 249 | - // normal search |
|
| 250 | - $nodeValue = $child->getAttribute($rule['key']); |
|
| 251 | - } |
|
| 252 | - |
|
| 253 | - $check = $this->match($rule['operator'], $rule['value'], $nodeValue); |
|
| 254 | - |
|
| 255 | - // handle multiple classes |
|
| 256 | - if ( ! $check && $rule['key'] == 'class') { |
|
| 257 | - $childClasses = explode(' ', $child->getAttribute('class')); |
|
| 258 | - foreach ($childClasses as $class) { |
|
| 259 | - if ( ! empty($class)) { |
|
| 260 | - $check = $this->match($rule['operator'], $rule['value'], $class); |
|
| 261 | - } |
|
| 262 | - if ($check) { |
|
| 263 | - break; |
|
| 264 | - } |
|
| 265 | - } |
|
| 266 | - } |
|
| 267 | - |
|
| 268 | - if ( ! $check) { |
|
| 269 | - $pass = false; |
|
| 270 | - } |
|
| 271 | - } |
|
| 272 | - |
|
| 273 | - if ($pass) { |
|
| 274 | - // it passed all checks |
|
| 275 | - $return[] = $child; |
|
| 276 | - } else { |
|
| 277 | - // this child failed to be matched |
|
| 278 | - if ($child instanceof InnerNode && |
|
| 279 | - $child->hasChildren()) { |
|
| 280 | - // we still want to check its children |
|
| 281 | - $children[] = $child; |
|
| 282 | - } |
|
| 283 | - } |
|
| 284 | - |
|
| 285 | - try { |
|
| 286 | - // get next child |
|
| 287 | - $child = $node->nextChild($child->id()); |
|
| 288 | - } catch (ChildNotFoundException $e) { |
|
| 289 | - // no more children |
|
| 290 | - $child = null; |
|
| 291 | - } |
|
| 292 | - } |
|
| 293 | - |
|
| 294 | - if (( ! isset($options['checkGrandChildren']) || |
|
| 295 | - $options['checkGrandChildren']) |
|
| 296 | - && count($children) > 0 |
|
| 297 | - ) { |
|
| 298 | - // we have children that failed but are not leaves. |
|
| 299 | - $matches = $this->seek($children, $rule, $options); |
|
| 300 | - foreach ($matches as $match) { |
|
| 301 | - $return[] = $match; |
|
| 302 | - } |
|
| 303 | - } |
|
| 304 | - } |
|
| 305 | - |
|
| 306 | - return $return; |
|
| 307 | - } |
|
| 308 | - |
|
| 309 | - /** |
|
| 310 | - * Attempts to match the given arguments with the given operator. |
|
| 311 | - * |
|
| 312 | - * @param string $operator |
|
| 313 | - * @param string $pattern |
|
| 314 | - * @param string $value |
|
| 315 | - * @return bool |
|
| 316 | - */ |
|
| 317 | - protected function match($operator, $pattern, $value) |
|
| 318 | - { |
|
| 319 | - $value = strtolower($value); |
|
| 320 | - $pattern = strtolower($pattern); |
|
| 321 | - switch ($operator) { |
|
| 322 | - case '=': |
|
| 323 | - return $value === $pattern; |
|
| 324 | - case '!=': |
|
| 325 | - return $value !== $pattern; |
|
| 326 | - case '^=': |
|
| 327 | - return preg_match('/^'.preg_quote($pattern, '/').'/', $value); |
|
| 328 | - case '$=': |
|
| 329 | - return preg_match('/'.preg_quote($pattern, '/').'$/', $value); |
|
| 330 | - case '*=': |
|
| 331 | - if ($pattern[0] == '/') { |
|
| 332 | - return preg_match($pattern, $value); |
|
| 333 | - } |
|
| 334 | - |
|
| 335 | - return preg_match("/".$pattern."/i", $value); |
|
| 336 | - } |
|
| 337 | - |
|
| 338 | - return false; |
|
| 339 | - } |
|
| 340 | - |
|
| 341 | - /** |
|
| 342 | - * Attempts to figure out what the alteration will be for |
|
| 343 | - * the next element. |
|
| 344 | - * |
|
| 345 | - * @param array $rule |
|
| 346 | - * @return array |
|
| 347 | - */ |
|
| 348 | - protected function alterNext($rule) |
|
| 349 | - { |
|
| 350 | - $options = []; |
|
| 351 | - if ($rule['tag'] == '>') { |
|
| 352 | - $options['checkGrandChildren'] = false; |
|
| 353 | - } |
|
| 354 | - |
|
| 355 | - return $options; |
|
| 356 | - } |
|
| 357 | - |
|
| 358 | - /** |
|
| 359 | - * Flattens the option array. |
|
| 360 | - * |
|
| 361 | - * @param array $optionsArray |
|
| 362 | - * @return array |
|
| 363 | - */ |
|
| 364 | - protected function flattenOptions(array $optionsArray) |
|
| 365 | - { |
|
| 366 | - $options = []; |
|
| 367 | - foreach ($optionsArray as $optionArray) { |
|
| 368 | - foreach ($optionArray as $key => $option) { |
|
| 369 | - $options[$key] = $option; |
|
| 370 | - } |
|
| 371 | - } |
|
| 372 | - |
|
| 373 | - return $options; |
|
| 374 | - } |
|
| 18 | + /** |
|
| 19 | + * Pattern of CSS selectors, modified from 'mootools' |
|
| 20 | + * |
|
| 21 | + * @var string |
|
| 22 | + */ |
|
| 23 | + protected $pattern = "/([\w-:\*>]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is"; |
|
| 24 | + |
|
| 25 | + protected $selectors = []; |
|
| 26 | + |
|
| 27 | + /** |
|
| 28 | + * Constructs with the selector string |
|
| 29 | + * |
|
| 30 | + * @param string $selector |
|
| 31 | + */ |
|
| 32 | + public function __construct($selector) |
|
| 33 | + { |
|
| 34 | + $this->parseSelectorString($selector); |
|
| 35 | + } |
|
| 36 | + |
|
| 37 | + /** |
|
| 38 | + * Returns the selectors that where found in __construct |
|
| 39 | + * |
|
| 40 | + * @return array |
|
| 41 | + */ |
|
| 42 | + public function getSelectors() |
|
| 43 | + { |
|
| 44 | + return $this->selectors; |
|
| 45 | + } |
|
| 46 | + |
|
| 47 | + /** |
|
| 48 | + * Attempts to find the selectors starting from the given |
|
| 49 | + * node object. |
|
| 50 | + * |
|
| 51 | + * @param AbstractNode $node |
|
| 52 | + * @return array|Collection |
|
| 53 | + */ |
|
| 54 | + public function find(AbstractNode $node) |
|
| 55 | + { |
|
| 56 | + $results = new Collection; |
|
| 57 | + foreach ($this->selectors as $selector) { |
|
| 58 | + $nodes = [$node]; |
|
| 59 | + if (count($selector) == 0) { |
|
| 60 | + continue; |
|
| 61 | + } |
|
| 62 | + |
|
| 63 | + $options = []; |
|
| 64 | + foreach ($selector as $rule) { |
|
| 65 | + if ($rule['alterNext']) { |
|
| 66 | + $options[] = $this->alterNext($rule); |
|
| 67 | + continue; |
|
| 68 | + } |
|
| 69 | + $nodes = $this->seek($nodes, $rule, $options); |
|
| 70 | + // clear the options |
|
| 71 | + $options = []; |
|
| 72 | + } |
|
| 73 | + |
|
| 74 | + // this is the final set of nodes |
|
| 75 | + foreach ($nodes as $result) { |
|
| 76 | + $results[] = $result; |
|
| 77 | + } |
|
| 78 | + } |
|
| 79 | + |
|
| 80 | + return $results; |
|
| 81 | + } |
|
| 82 | + |
|
| 83 | + /** |
|
| 84 | + * Parses the selector string |
|
| 85 | + * |
|
| 86 | + * @param string $selector |
|
| 87 | + */ |
|
| 88 | + protected function parseSelectorString($selector) |
|
| 89 | + { |
|
| 90 | + $matches = []; |
|
| 91 | + preg_match_all($this->pattern, trim($selector).' ', $matches, PREG_SET_ORDER); |
|
| 92 | + |
|
| 93 | + // skip tbody |
|
| 94 | + $result = []; |
|
| 95 | + foreach ($matches as $match) { |
|
| 96 | + // default values |
|
| 97 | + $tag = strtolower(trim($match[1])); |
|
| 98 | + $operator = '='; |
|
| 99 | + $key = null; |
|
| 100 | + $value = null; |
|
| 101 | + $noKey = false; |
|
| 102 | + $alterNext = false; |
|
| 103 | + |
|
| 104 | + // check for elements that alter the behavior of the next element |
|
| 105 | + if ($tag == '>') { |
|
| 106 | + $alterNext = true; |
|
| 107 | + } |
|
| 108 | + |
|
| 109 | + // check for id selector |
|
| 110 | + if ( ! empty($match[2])) { |
|
| 111 | + $key = 'id'; |
|
| 112 | + $value = $match[2]; |
|
| 113 | + } |
|
| 114 | + |
|
| 115 | + // check for class selector |
|
| 116 | + if ( ! empty($match[3])) { |
|
| 117 | + $key = 'class'; |
|
| 118 | + $value = $match[3]; |
|
| 119 | + } |
|
| 120 | + |
|
| 121 | + // and final attribute selector |
|
| 122 | + if ( ! empty($match[4])) { |
|
| 123 | + $key = strtolower($match[4]); |
|
| 124 | + } |
|
| 125 | + if ( ! empty($match[5])) { |
|
| 126 | + $operator = $match[5]; |
|
| 127 | + } |
|
| 128 | + if ( ! empty($match[6])) { |
|
| 129 | + $value = $match[6]; |
|
| 130 | + } |
|
| 131 | + |
|
| 132 | + // check for elements that do not have a specified attribute |
|
| 133 | + if (isset($key[0]) && $key[0] == '!') { |
|
| 134 | + $key = substr($key, 1); |
|
| 135 | + $noKey = true; |
|
| 136 | + } |
|
| 137 | + |
|
| 138 | + $result[] = [ |
|
| 139 | + 'tag' => $tag, |
|
| 140 | + 'key' => $key, |
|
| 141 | + 'value' => $value, |
|
| 142 | + 'operator' => $operator, |
|
| 143 | + 'noKey' => $noKey, |
|
| 144 | + 'alterNext' => $alterNext, |
|
| 145 | + ]; |
|
| 146 | + if (trim($match[7]) == ',') { |
|
| 147 | + $this->selectors[] = $result; |
|
| 148 | + $result = []; |
|
| 149 | + } |
|
| 150 | + } |
|
| 151 | + |
|
| 152 | + // save last results |
|
| 153 | + if (count($result) > 0) { |
|
| 154 | + $this->selectors[] = $result; |
|
| 155 | + } |
|
| 156 | + } |
|
| 157 | + |
|
| 158 | + /** |
|
| 159 | + * Attempts to find all children that match the rule |
|
| 160 | + * given. |
|
| 161 | + * |
|
| 162 | + * @param array $nodes |
|
| 163 | + * @param array $rule |
|
| 164 | + * @param array $options |
|
| 165 | + * @return array |
|
| 166 | + * @recursive |
|
| 167 | + */ |
|
| 168 | + protected function seek(array $nodes, array $rule, array $options) |
|
| 169 | + { |
|
| 170 | + // XPath index |
|
| 171 | + if (count($rule['tag']) > 0 && |
|
| 172 | + count($rule['key']) > 0 && |
|
| 173 | + is_numeric($rule['key']) |
|
| 174 | + ) { |
|
| 175 | + $count = 0; |
|
| 176 | + /** @var AbstractNode $node */ |
|
| 177 | + foreach ($nodes as $node) { |
|
| 178 | + if ($rule['tag'] == '*' || |
|
| 179 | + $rule['tag'] == $node->getTag()->name()) { |
|
| 180 | + ++$count; |
|
| 181 | + if ($count == $rule['key']) { |
|
| 182 | + // found the node we wanted |
|
| 183 | + return [$node]; |
|
| 184 | + } |
|
| 185 | + } |
|
| 186 | + } |
|
| 187 | + |
|
| 188 | + return []; |
|
| 189 | + } |
|
| 190 | + |
|
| 191 | + $options = $this->flattenOptions($options); |
|
| 192 | + |
|
| 193 | + $return = []; |
|
| 194 | + /** @var InnerNode $node */ |
|
| 195 | + foreach ($nodes as $node) { |
|
| 196 | + // check if we are a leaf |
|
| 197 | + if ($node instanceof LeafNode || |
|
| 198 | + ! $node->hasChildren()) { |
|
| 199 | + continue; |
|
| 200 | + } |
|
| 201 | + |
|
| 202 | + $children = []; |
|
| 203 | + $child = $node->firstChild(); |
|
| 204 | + while ( ! is_null($child)) { |
|
| 205 | + // wild card, grab all |
|
| 206 | + if ($rule['tag'] == '*' && is_null($rule['key'])) { |
|
| 207 | + $return[] = $child; |
|
| 208 | + try { |
|
| 209 | + $child = $node->nextChild($child->id()); |
|
| 210 | + } catch (ChildNotFoundException $e) { |
|
| 211 | + // no more children |
|
| 212 | + $child = null; |
|
| 213 | + } |
|
| 214 | + continue; |
|
| 215 | + } |
|
| 216 | + |
|
| 217 | + $pass = true; |
|
| 218 | + // check tag |
|
| 219 | + if ( ! empty($rule['tag']) && $rule['tag'] != $child->getTag()->name() && |
|
| 220 | + $rule['tag'] != '*' |
|
| 221 | + ) { |
|
| 222 | + // child failed tag check |
|
| 223 | + $pass = false; |
|
| 224 | + } |
|
| 225 | + |
|
| 226 | + // check key |
|
| 227 | + if ($pass && ! is_null($rule['key'])) { |
|
| 228 | + if ($rule['noKey']) { |
|
| 229 | + if ( ! is_null($child->getAttribute($rule['key']))) { |
|
| 230 | + $pass = false; |
|
| 231 | + } |
|
| 232 | + } else { |
|
| 233 | + if ($rule['key'] != 'plaintext' && |
|
| 234 | + is_null($child->getAttribute($rule['key'])) |
|
| 235 | + ) { |
|
| 236 | + $pass = false; |
|
| 237 | + } |
|
| 238 | + } |
|
| 239 | + } |
|
| 240 | + |
|
| 241 | + // compare values |
|
| 242 | + if ($pass && ! is_null($rule['key']) && |
|
| 243 | + ! is_null($rule['value']) && $rule['value'] != '*' |
|
| 244 | + ) { |
|
| 245 | + if ($rule['key'] == 'plaintext') { |
|
| 246 | + // plaintext search |
|
| 247 | + $nodeValue = $child->text(); |
|
| 248 | + } else { |
|
| 249 | + // normal search |
|
| 250 | + $nodeValue = $child->getAttribute($rule['key']); |
|
| 251 | + } |
|
| 252 | + |
|
| 253 | + $check = $this->match($rule['operator'], $rule['value'], $nodeValue); |
|
| 254 | + |
|
| 255 | + // handle multiple classes |
|
| 256 | + if ( ! $check && $rule['key'] == 'class') { |
|
| 257 | + $childClasses = explode(' ', $child->getAttribute('class')); |
|
| 258 | + foreach ($childClasses as $class) { |
|
| 259 | + if ( ! empty($class)) { |
|
| 260 | + $check = $this->match($rule['operator'], $rule['value'], $class); |
|
| 261 | + } |
|
| 262 | + if ($check) { |
|
| 263 | + break; |
|
| 264 | + } |
|
| 265 | + } |
|
| 266 | + } |
|
| 267 | + |
|
| 268 | + if ( ! $check) { |
|
| 269 | + $pass = false; |
|
| 270 | + } |
|
| 271 | + } |
|
| 272 | + |
|
| 273 | + if ($pass) { |
|
| 274 | + // it passed all checks |
|
| 275 | + $return[] = $child; |
|
| 276 | + } else { |
|
| 277 | + // this child failed to be matched |
|
| 278 | + if ($child instanceof InnerNode && |
|
| 279 | + $child->hasChildren()) { |
|
| 280 | + // we still want to check its children |
|
| 281 | + $children[] = $child; |
|
| 282 | + } |
|
| 283 | + } |
|
| 284 | + |
|
| 285 | + try { |
|
| 286 | + // get next child |
|
| 287 | + $child = $node->nextChild($child->id()); |
|
| 288 | + } catch (ChildNotFoundException $e) { |
|
| 289 | + // no more children |
|
| 290 | + $child = null; |
|
| 291 | + } |
|
| 292 | + } |
|
| 293 | + |
|
| 294 | + if (( ! isset($options['checkGrandChildren']) || |
|
| 295 | + $options['checkGrandChildren']) |
|
| 296 | + && count($children) > 0 |
|
| 297 | + ) { |
|
| 298 | + // we have children that failed but are not leaves. |
|
| 299 | + $matches = $this->seek($children, $rule, $options); |
|
| 300 | + foreach ($matches as $match) { |
|
| 301 | + $return[] = $match; |
|
| 302 | + } |
|
| 303 | + } |
|
| 304 | + } |
|
| 305 | + |
|
| 306 | + return $return; |
|
| 307 | + } |
|
| 308 | + |
|
| 309 | + /** |
|
| 310 | + * Attempts to match the given arguments with the given operator. |
|
| 311 | + * |
|
| 312 | + * @param string $operator |
|
| 313 | + * @param string $pattern |
|
| 314 | + * @param string $value |
|
| 315 | + * @return bool |
|
| 316 | + */ |
|
| 317 | + protected function match($operator, $pattern, $value) |
|
| 318 | + { |
|
| 319 | + $value = strtolower($value); |
|
| 320 | + $pattern = strtolower($pattern); |
|
| 321 | + switch ($operator) { |
|
| 322 | + case '=': |
|
| 323 | + return $value === $pattern; |
|
| 324 | + case '!=': |
|
| 325 | + return $value !== $pattern; |
|
| 326 | + case '^=': |
|
| 327 | + return preg_match('/^'.preg_quote($pattern, '/').'/', $value); |
|
| 328 | + case '$=': |
|
| 329 | + return preg_match('/'.preg_quote($pattern, '/').'$/', $value); |
|
| 330 | + case '*=': |
|
| 331 | + if ($pattern[0] == '/') { |
|
| 332 | + return preg_match($pattern, $value); |
|
| 333 | + } |
|
| 334 | + |
|
| 335 | + return preg_match("/".$pattern."/i", $value); |
|
| 336 | + } |
|
| 337 | + |
|
| 338 | + return false; |
|
| 339 | + } |
|
| 340 | + |
|
| 341 | + /** |
|
| 342 | + * Attempts to figure out what the alteration will be for |
|
| 343 | + * the next element. |
|
| 344 | + * |
|
| 345 | + * @param array $rule |
|
| 346 | + * @return array |
|
| 347 | + */ |
|
| 348 | + protected function alterNext($rule) |
|
| 349 | + { |
|
| 350 | + $options = []; |
|
| 351 | + if ($rule['tag'] == '>') { |
|
| 352 | + $options['checkGrandChildren'] = false; |
|
| 353 | + } |
|
| 354 | + |
|
| 355 | + return $options; |
|
| 356 | + } |
|
| 357 | + |
|
| 358 | + /** |
|
| 359 | + * Flattens the option array. |
|
| 360 | + * |
|
| 361 | + * @param array $optionsArray |
|
| 362 | + * @return array |
|
| 363 | + */ |
|
| 364 | + protected function flattenOptions(array $optionsArray) |
|
| 365 | + { |
|
| 366 | + $options = []; |
|
| 367 | + foreach ($optionsArray as $optionArray) { |
|
| 368 | + foreach ($optionArray as $key => $option) { |
|
| 369 | + $options[$key] = $option; |
|
| 370 | + } |
|
| 371 | + } |
|
| 372 | + |
|
| 373 | + return $options; |
|
| 374 | + } |
|
| 375 | 375 | } |
