paquettg /
php-html-parser
@@ -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 |
@@ -16,633 +16,633 @@ |
||
| 16 | 16 | class Dom |
| 17 | 17 | { |
| 18 | 18 | |
| 19 | - /** |
|
| 20 | - * The charset we would like the output to be in. |
|
| 21 | - * |
|
| 22 | - * @var string |
|
| 23 | - */ |
|
| 24 | - protected $defaultCharset = 'UTF-8'; |
|
| 25 | - |
|
| 26 | - /** |
|
| 27 | - * Contains the root node of this dom tree. |
|
| 28 | - * |
|
| 29 | - * @var HtmlNode |
|
| 30 | - */ |
|
| 31 | - public $root; |
|
| 32 | - |
|
| 33 | - /** |
|
| 34 | - * The raw version of the document string. |
|
| 35 | - * |
|
| 36 | - * @var string |
|
| 37 | - */ |
|
| 38 | - protected $raw; |
|
| 39 | - |
|
| 40 | - /** |
|
| 41 | - * The document string. |
|
| 42 | - * |
|
| 43 | - * @var Content |
|
| 44 | - */ |
|
| 45 | - protected $content = null; |
|
| 46 | - |
|
| 47 | - /** |
|
| 48 | - * The original file size of the document. |
|
| 49 | - * |
|
| 50 | - * @var int |
|
| 51 | - */ |
|
| 52 | - protected $rawSize; |
|
| 53 | - |
|
| 54 | - /** |
|
| 55 | - * The size of the document after it is cleaned. |
|
| 56 | - * |
|
| 57 | - * @var int |
|
| 58 | - */ |
|
| 59 | - protected $size; |
|
| 60 | - |
|
| 61 | - /** |
|
| 62 | - * A global options array to be used by all load calls. |
|
| 63 | - * |
|
| 64 | - * @var array |
|
| 65 | - */ |
|
| 66 | - protected $globalOptions = []; |
|
| 67 | - |
|
| 68 | - /** |
|
| 69 | - * A persistent option object to be used for all options in the |
|
| 70 | - * parsing of the file. |
|
| 71 | - * |
|
| 72 | - * @var Options |
|
| 73 | - */ |
|
| 74 | - protected $options; |
|
| 75 | - |
|
| 76 | - /** |
|
| 77 | - * A list of tags which will always be self closing |
|
| 78 | - * |
|
| 79 | - * @var array |
|
| 80 | - */ |
|
| 81 | - protected $selfClosing = [ |
|
| 82 | - 'img', |
|
| 83 | - 'br', |
|
| 84 | - 'input', |
|
| 85 | - 'meta', |
|
| 86 | - 'link', |
|
| 87 | - 'hr', |
|
| 88 | - 'base', |
|
| 89 | - 'embed', |
|
| 90 | - 'spacer', |
|
| 91 | - ]; |
|
| 92 | - |
|
| 93 | - /** |
|
| 94 | - * Returns the inner html of the root node. |
|
| 95 | - * |
|
| 96 | - * @return string |
|
| 97 | - */ |
|
| 98 | - public function __toString() |
|
| 99 | - { |
|
| 100 | - return $this->root->innerHtml(); |
|
| 101 | - } |
|
| 102 | - |
|
| 103 | - /** |
|
| 104 | - * A simple wrapper around the root node. |
|
| 105 | - * |
|
| 106 | - * @param string $name |
|
| 107 | - * @return mixed |
|
| 108 | - */ |
|
| 109 | - public function __get($name) |
|
| 110 | - { |
|
| 111 | - return $this->root->$name; |
|
| 112 | - } |
|
| 113 | - |
|
| 114 | - /** |
|
| 115 | - * Attempts to load the dom from any resource, string, file, or URL. |
|
| 116 | - * |
|
| 117 | - * @param string $str |
|
| 118 | - * @param array $options |
|
| 119 | - * @return $this |
|
| 120 | - */ |
|
| 121 | - public function load($str, $options = []) |
|
| 122 | - { |
|
| 123 | - // check if it's a file |
|
| 124 | - if (strpos($str, "\n") === false && is_file($str)) { |
|
| 125 | - return $this->loadFromFile($str, $options); |
|
| 126 | - } |
|
| 127 | - // check if it's a url |
|
| 128 | - if (preg_match("/^https?:\/\//i", $str)) { |
|
| 129 | - return $this->loadFromUrl($str, $options); |
|
| 130 | - } |
|
| 131 | - |
|
| 132 | - return $this->loadStr($str, $options); |
|
| 133 | - } |
|
| 134 | - |
|
| 135 | - /** |
|
| 136 | - * Loads the dom from a document file/url |
|
| 137 | - * |
|
| 138 | - * @param string $file |
|
| 139 | - * @param array $options |
|
| 140 | - * @return $this |
|
| 141 | - */ |
|
| 142 | - public function loadFromFile($file, $options = []) |
|
| 143 | - { |
|
| 144 | - return $this->loadStr(file_get_contents($file), $options); |
|
| 145 | - } |
|
| 146 | - |
|
| 147 | - /** |
|
| 148 | - * Use a curl interface implementation to attempt to load |
|
| 149 | - * the content from a url. |
|
| 150 | - * |
|
| 151 | - * @param string $url |
|
| 152 | - * @param array $options |
|
| 153 | - * @param CurlInterface $curl |
|
| 154 | - * @return $this |
|
| 155 | - */ |
|
| 156 | - public function loadFromUrl($url, $options = [], CurlInterface $curl = null) |
|
| 157 | - { |
|
| 158 | - if (is_null($curl)) { |
|
| 159 | - // use the default curl interface |
|
| 160 | - $curl = new Curl; |
|
| 161 | - } |
|
| 162 | - $content = $curl->get($url); |
|
| 163 | - |
|
| 164 | - return $this->loadStr($content, $options); |
|
| 165 | - } |
|
| 166 | - |
|
| 167 | - /** |
|
| 168 | - * Parsers the html of the given string. Used for load(), loadFromFile(), |
|
| 169 | - * and loadFromUrl(). |
|
| 170 | - * |
|
| 171 | - * @param string $str |
|
| 172 | - * @param array $option |
|
| 173 | - * @return $this |
|
| 174 | - */ |
|
| 175 | - public function loadStr($str, $option) |
|
| 176 | - { |
|
| 177 | - $this->options = new Options; |
|
| 178 | - $this->options->setOptions($this->globalOptions) |
|
| 179 | - ->setOptions($option); |
|
| 180 | - |
|
| 181 | - $this->rawSize = strlen($str); |
|
| 182 | - $this->raw = $str; |
|
| 183 | - |
|
| 184 | - $html = $this->clean($str); |
|
| 185 | - |
|
| 186 | - $this->size = strlen($str); |
|
| 187 | - $this->content = new Content($html); |
|
| 188 | - |
|
| 189 | - $this->parse(); |
|
| 190 | - $this->detectCharset(); |
|
| 191 | - |
|
| 192 | - return $this; |
|
| 193 | - } |
|
| 194 | - |
|
| 195 | - /** |
|
| 196 | - * Sets a global options array to be used by all load calls. |
|
| 197 | - * |
|
| 198 | - * @param array $options |
|
| 199 | - * @return $this |
|
| 200 | - */ |
|
| 201 | - public function setOptions(array $options) |
|
| 202 | - { |
|
| 203 | - $this->globalOptions = $options; |
|
| 204 | - |
|
| 205 | - return $this; |
|
| 206 | - } |
|
| 207 | - |
|
| 208 | - /** |
|
| 209 | - * Find elements by css selector on the root node. |
|
| 210 | - * |
|
| 211 | - * @param string $selector |
|
| 212 | - * @param int $nth |
|
| 213 | - * @return array |
|
| 214 | - */ |
|
| 215 | - public function find($selector, $nth = null) |
|
| 216 | - { |
|
| 217 | - $this->isLoaded(); |
|
| 218 | - |
|
| 219 | - return $this->root->find($selector, $nth); |
|
| 220 | - } |
|
| 221 | - |
|
| 222 | - /** |
|
| 223 | - * Adds the tag (or tags in an array) to the list of tags that will always |
|
| 224 | - * be self closing. |
|
| 225 | - * |
|
| 226 | - * @param string|array $tag |
|
| 227 | - * @return $this |
|
| 228 | - */ |
|
| 229 | - public function addSelfClosingTag($tag) |
|
| 230 | - { |
|
| 231 | - if ( ! is_array($tag)) { |
|
| 232 | - $tag = [$tag]; |
|
| 233 | - } |
|
| 234 | - foreach ($tag as $value) { |
|
| 235 | - $this->selfClosing[] = $value; |
|
| 236 | - } |
|
| 237 | - |
|
| 238 | - return $this; |
|
| 239 | - } |
|
| 240 | - |
|
| 241 | - /** |
|
| 242 | - * Removes the tag (or tags in an array) from the list of tags that will |
|
| 243 | - * always be self closing. |
|
| 244 | - * |
|
| 245 | - * @param string|array $tag |
|
| 246 | - * @return $this |
|
| 247 | - */ |
|
| 248 | - public function removeSelfClosingTag($tag) |
|
| 249 | - { |
|
| 250 | - if ( ! is_array($tag)) { |
|
| 251 | - $tag = [$tag]; |
|
| 252 | - } |
|
| 253 | - $this->selfClosing = array_diff($this->selfClosing, $tag); |
|
| 254 | - |
|
| 255 | - return $this; |
|
| 256 | - } |
|
| 257 | - |
|
| 258 | - /** |
|
| 259 | - * Sets the list of self closing tags to empty. |
|
| 260 | - * |
|
| 261 | - * @return $this |
|
| 262 | - */ |
|
| 263 | - public function clearSelfClosingTags() |
|
| 264 | - { |
|
| 265 | - $this->selfClosing = []; |
|
| 266 | - |
|
| 267 | - return $this; |
|
| 268 | - } |
|
| 269 | - |
|
| 270 | - /** |
|
| 271 | - * Simple wrapper function that returns the first child. |
|
| 272 | - * |
|
| 273 | - * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 274 | - */ |
|
| 275 | - public function firstChild() |
|
| 276 | - { |
|
| 277 | - $this->isLoaded(); |
|
| 278 | - |
|
| 279 | - return $this->root->firstChild(); |
|
| 280 | - } |
|
| 281 | - |
|
| 282 | - /** |
|
| 283 | - * Simple wrapper function that returns the last child. |
|
| 284 | - * |
|
| 285 | - * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 286 | - */ |
|
| 287 | - public function lastChild() |
|
| 288 | - { |
|
| 289 | - $this->isLoaded(); |
|
| 290 | - |
|
| 291 | - return $this->root->lastChild(); |
|
| 292 | - } |
|
| 293 | - |
|
| 294 | - /** |
|
| 295 | - * Simple wrapper function that returns an element by the |
|
| 296 | - * id. |
|
| 297 | - * |
|
| 298 | - * @param string $id |
|
| 299 | - * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 300 | - */ |
|
| 301 | - public function getElementById($id) |
|
| 302 | - { |
|
| 303 | - $this->isLoaded(); |
|
| 304 | - |
|
| 305 | - return $this->find('#'.$id, 0); |
|
| 306 | - } |
|
| 307 | - |
|
| 308 | - /** |
|
| 309 | - * Simple wrapper function that returns all elements by |
|
| 310 | - * tag name. |
|
| 311 | - * |
|
| 312 | - * @param string $name |
|
| 313 | - * @return array |
|
| 314 | - */ |
|
| 315 | - public function getElementsByTag($name) |
|
| 316 | - { |
|
| 317 | - $this->isLoaded(); |
|
| 318 | - |
|
| 319 | - return $this->find($name); |
|
| 320 | - } |
|
| 321 | - |
|
| 322 | - /** |
|
| 323 | - * Simple wrapper function that returns all elements by |
|
| 324 | - * class name. |
|
| 325 | - * |
|
| 326 | - * @param string $class |
|
| 327 | - * @return array |
|
| 328 | - */ |
|
| 329 | - public function getElementsByClass($class) |
|
| 330 | - { |
|
| 331 | - $this->isLoaded(); |
|
| 332 | - |
|
| 333 | - return $this->find('.'.$class); |
|
| 334 | - } |
|
| 335 | - |
|
| 336 | - /** |
|
| 337 | - * Checks if the load methods have been called. |
|
| 338 | - * |
|
| 339 | - * @throws NotLoadedException |
|
| 340 | - */ |
|
| 341 | - protected function isLoaded() |
|
| 342 | - { |
|
| 343 | - if (is_null($this->content)) { |
|
| 344 | - throw new NotLoadedException('Content is not loaded!'); |
|
| 345 | - } |
|
| 346 | - } |
|
| 347 | - |
|
| 348 | - /** |
|
| 349 | - * Cleans the html of any none-html information. |
|
| 350 | - * |
|
| 351 | - * @param string $str |
|
| 352 | - * @return string |
|
| 353 | - */ |
|
| 354 | - protected function clean($str) |
|
| 355 | - { |
|
| 356 | - if ($this->options->get('cleanupInput') != true) { |
|
| 357 | - // skip entire cleanup step |
|
| 358 | - return $str; |
|
| 359 | - } |
|
| 360 | - |
|
| 361 | - // remove white space before closing tags |
|
| 362 | - $str = mb_eregi_replace("'\s+>", "'>", $str); |
|
| 363 | - $str = mb_eregi_replace('"\s+>', '">', $str); |
|
| 364 | - |
|
| 365 | - // clean out the \n\r |
|
| 366 | - $replace = ' '; |
|
| 367 | - if ($this->options->get('preserveLineBreaks')) { |
|
| 368 | - $replace = ' '; |
|
| 369 | - } |
|
| 370 | - $str = str_replace(["\r\n", "\r", "\n"], $replace, $str); |
|
| 371 | - |
|
| 372 | - // strip the doctype |
|
| 373 | - $str = mb_eregi_replace("<!doctype(.*?)>", '', $str); |
|
| 374 | - |
|
| 375 | - // strip out comments |
|
| 376 | - $str = mb_eregi_replace("<!--(.*?)-->", '', $str); |
|
| 377 | - |
|
| 378 | - // strip out cdata |
|
| 379 | - $str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str); |
|
| 380 | - |
|
| 381 | - // strip out <script> tags |
|
| 382 | - if ($this->options->get('removeScripts') == true) { |
|
| 383 | - $str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str); |
|
| 384 | - $str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str); |
|
| 385 | - } |
|
| 386 | - |
|
| 387 | - // strip out <style> tags |
|
| 388 | - if ($this->options->get('removeStyles') == true) { |
|
| 389 | - $str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str); |
|
| 390 | - $str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str); |
|
| 391 | - } |
|
| 392 | - |
|
| 393 | - // strip out server side scripts |
|
| 394 | - $str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str); |
|
| 395 | - |
|
| 396 | - // strip smarty scripts |
|
| 397 | - $str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str); |
|
| 398 | - |
|
| 399 | - return $str; |
|
| 400 | - } |
|
| 401 | - |
|
| 402 | - /** |
|
| 403 | - * Attempts to parse the html in content. |
|
| 404 | - */ |
|
| 405 | - protected function parse() |
|
| 406 | - { |
|
| 407 | - // add the root node |
|
| 408 | - $this->root = new HtmlNode('root'); |
|
| 409 | - $activeNode = $this->root; |
|
| 410 | - while ( ! is_null($activeNode)) { |
|
| 411 | - $str = $this->content->copyUntil('<'); |
|
| 412 | - if ($str == '') { |
|
| 413 | - $info = $this->parseTag(); |
|
| 414 | - if ( ! $info['status']) { |
|
| 415 | - // we are done here |
|
| 416 | - $activeNode = null; |
|
| 417 | - continue; |
|
| 418 | - } |
|
| 419 | - |
|
| 420 | - // check if it was a closing tag |
|
| 421 | - if ($info['closing']) { |
|
| 422 | - $originalNode = $activeNode; |
|
| 423 | - while ($activeNode->getTag()->name() != $info['tag']) { |
|
| 424 | - $activeNode = $activeNode->getParent(); |
|
| 425 | - if (is_null($activeNode)) { |
|
| 426 | - // we could not find opening tag |
|
| 427 | - $activeNode = $originalNode; |
|
| 428 | - break; |
|
| 429 | - } |
|
| 430 | - } |
|
| 431 | - if ( ! is_null($activeNode)) { |
|
| 432 | - $activeNode = $activeNode->getParent(); |
|
| 433 | - } |
|
| 434 | - continue; |
|
| 435 | - } |
|
| 436 | - |
|
| 437 | - if ( ! isset($info['node'])) { |
|
| 438 | - continue; |
|
| 439 | - } |
|
| 440 | - |
|
| 441 | - /** @var AbstractNode $node */ |
|
| 442 | - $node = $info['node']; |
|
| 443 | - $activeNode->addChild($node); |
|
| 444 | - |
|
| 445 | - // check if node is self closing |
|
| 446 | - if ( ! $node->getTag()->isSelfClosing()) { |
|
| 447 | - $activeNode = $node; |
|
| 448 | - } |
|
| 449 | - } else if ($this->options->whitespaceTextNode || |
|
| 450 | - trim($str) != '' |
|
| 451 | - ) { |
|
| 452 | - // we found text we care about |
|
| 453 | - $textNode = new TextNode($str); |
|
| 454 | - $activeNode->addChild($textNode); |
|
| 455 | - } |
|
| 456 | - } |
|
| 457 | - } |
|
| 458 | - |
|
| 459 | - /** |
|
| 460 | - * Attempt to parse a tag out of the content. |
|
| 461 | - * |
|
| 462 | - * @return array |
|
| 463 | - * @throws StrictException |
|
| 464 | - */ |
|
| 465 | - protected function parseTag() |
|
| 466 | - { |
|
| 467 | - $return = [ |
|
| 468 | - 'status' => false, |
|
| 469 | - 'closing' => false, |
|
| 470 | - 'node' => null, |
|
| 471 | - ]; |
|
| 472 | - if ($this->content->char() != '<') { |
|
| 473 | - // we are not at the beginning of a tag |
|
| 474 | - return $return; |
|
| 475 | - } |
|
| 476 | - |
|
| 477 | - // check if this is a closing tag |
|
| 478 | - if ($this->content->fastForward(1)->char() == '/') { |
|
| 479 | - // end tag |
|
| 480 | - $tag = $this->content->fastForward(1) |
|
| 481 | - ->copyByToken('slash', true); |
|
| 482 | - // move to end of tag |
|
| 483 | - $this->content->copyUntil('>'); |
|
| 484 | - $this->content->fastForward(1); |
|
| 485 | - |
|
| 486 | - // check if this closing tag counts |
|
| 487 | - $tag = strtolower($tag); |
|
| 488 | - if (in_array($tag, $this->selfClosing)) { |
|
| 489 | - $return['status'] = true; |
|
| 490 | - |
|
| 491 | - return $return; |
|
| 492 | - } else { |
|
| 493 | - $return['status'] = true; |
|
| 494 | - $return['closing'] = true; |
|
| 495 | - $return['tag'] = strtolower($tag); |
|
| 496 | - } |
|
| 497 | - |
|
| 498 | - return $return; |
|
| 499 | - } |
|
| 500 | - |
|
| 501 | - $tag = strtolower($this->content->copyByToken('slash', true)); |
|
| 502 | - $node = new HtmlNode($tag); |
|
| 503 | - |
|
| 504 | - // attributes |
|
| 505 | - while ($this->content->char() != '>' && |
|
| 506 | - $this->content->char() != '/') { |
|
| 507 | - $space = $this->content->skipByToken('blank', true); |
|
| 508 | - if (empty($space)) { |
|
| 509 | - $this->content->fastForward(1); |
|
| 510 | - continue; |
|
| 511 | - } |
|
| 512 | - |
|
| 513 | - $name = $this->content->copyByToken('equal', true); |
|
| 514 | - if ($name == '/') { |
|
| 515 | - break; |
|
| 516 | - } |
|
| 517 | - |
|
| 518 | - if (empty($name)) { |
|
| 519 | - $this->content->fastForward(1); |
|
| 520 | - continue; |
|
| 521 | - } |
|
| 522 | - |
|
| 523 | - $this->content->skipByToken('blank'); |
|
| 524 | - if ($this->content->char() == '=') { |
|
| 525 | - $attr = []; |
|
| 526 | - $this->content->fastForward(1) |
|
| 527 | - ->skipByToken('blank'); |
|
| 528 | - switch ($this->content->char()) { |
|
| 529 | - case '"': |
|
| 530 | - $attr['doubleQuote'] = true; |
|
| 531 | - $this->content->fastForward(1); |
|
| 532 | - $string = $this->content->copyUntil('"', true, true); |
|
| 533 | - do { |
|
| 534 | - $moreString = $this->content->copyUntilUnless('"', '=>'); |
|
| 535 | - $string .= $moreString; |
|
| 536 | - } while ( ! empty($moreString)); |
|
| 537 | - $attr['value'] = $string; |
|
| 538 | - $this->content->fastForward(1); |
|
| 539 | - $node->getTag()->$name = $attr; |
|
| 540 | - break; |
|
| 541 | - case "'": |
|
| 542 | - $attr['doubleQuote'] = false; |
|
| 543 | - $this->content->fastForward(1); |
|
| 544 | - $string = $this->content->copyUntil("'", true, true); |
|
| 545 | - do { |
|
| 546 | - $moreString = $this->content->copyUntilUnless("'", '=>'); |
|
| 547 | - $string .= $moreString; |
|
| 548 | - } while ( ! empty($moreString)); |
|
| 549 | - $attr['value'] = $string; |
|
| 550 | - $this->content->fastForward(1); |
|
| 551 | - $node->getTag()->$name = $attr; |
|
| 552 | - break; |
|
| 553 | - default: |
|
| 554 | - $attr['doubleQuote'] = true; |
|
| 555 | - $attr['value'] = $this->content->copyByToken('attr', true); |
|
| 556 | - $node->getTag()->$name = $attr; |
|
| 557 | - break; |
|
| 558 | - } |
|
| 559 | - } else { |
|
| 560 | - // no value attribute |
|
| 561 | - if ($this->options->strict) { |
|
| 562 | - // can't have this in strict html |
|
| 563 | - $character = $this->content->getPosition(); |
|
| 564 | - throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)"); |
|
| 565 | - } |
|
| 566 | - $node->getTag()->$name = [ |
|
| 567 | - 'value' => null, |
|
| 568 | - 'doubleQuote' => true, |
|
| 569 | - ]; |
|
| 570 | - if ($this->content->char() != '>') { |
|
| 571 | - $this->content->rewind(1); |
|
| 572 | - } |
|
| 573 | - } |
|
| 574 | - } |
|
| 575 | - |
|
| 576 | - $this->content->skipByToken('blank'); |
|
| 577 | - if ($this->content->char() == '/') { |
|
| 578 | - // self closing tag |
|
| 579 | - $node->getTag()->selfClosing(); |
|
| 580 | - $this->content->fastForward(1); |
|
| 581 | - } elseif (in_array($tag, $this->selfClosing)) { |
|
| 582 | - |
|
| 583 | - // Should be a self closing tag, check if we are strict |
|
| 584 | - if ($this->options->strict) { |
|
| 585 | - $character = $this->content->getPosition(); |
|
| 586 | - throw new StrictException("Tag '$tag' is not self closing! (character #$character)"); |
|
| 587 | - } |
|
| 588 | - |
|
| 589 | - // We force self closing on this tag. |
|
| 590 | - $node->getTag()->selfClosing(); |
|
| 591 | - } |
|
| 592 | - |
|
| 593 | - $this->content->fastForward(1); |
|
| 594 | - |
|
| 595 | - $return['status'] = true; |
|
| 596 | - $return['node'] = $node; |
|
| 597 | - |
|
| 598 | - return $return; |
|
| 599 | - } |
|
| 600 | - |
|
| 601 | - /** |
|
| 602 | - * Attempts to detect the charset that the html was sent in. |
|
| 603 | - * |
|
| 604 | - * @return bool |
|
| 605 | - */ |
|
| 606 | - protected function detectCharset() |
|
| 607 | - { |
|
| 608 | - // set the default |
|
| 609 | - $encode = new Encode; |
|
| 610 | - $encode->from($this->defaultCharset); |
|
| 611 | - $encode->to($this->defaultCharset); |
|
| 612 | - |
|
| 613 | - if ( ! is_null($this->options->enforceEncoding)) { |
|
| 614 | - // they want to enforce the given encoding |
|
| 615 | - $encode->from($this->options->enforceEncoding); |
|
| 616 | - $encode->to($this->options->enforceEncoding); |
|
| 617 | - |
|
| 618 | - return false; |
|
| 619 | - } |
|
| 620 | - |
|
| 621 | - $meta = $this->root->find('meta[http-equiv=Content-Type]', 0); |
|
| 622 | - if (is_null($meta)) { |
|
| 623 | - // could not find meta tag |
|
| 624 | - $this->root->propagateEncoding($encode); |
|
| 625 | - |
|
| 626 | - return false; |
|
| 627 | - } |
|
| 628 | - $content = $meta->content; |
|
| 629 | - if (empty($content)) { |
|
| 630 | - // could not find content |
|
| 631 | - $this->root->propagateEncoding($encode); |
|
| 632 | - |
|
| 633 | - return false; |
|
| 634 | - } |
|
| 635 | - $matches = []; |
|
| 636 | - if (preg_match('/charset=(.+)/', $content, $matches)) { |
|
| 637 | - $encode->from(trim($matches[1])); |
|
| 638 | - $this->root->propagateEncoding($encode); |
|
| 639 | - |
|
| 640 | - return true; |
|
| 641 | - } |
|
| 642 | - |
|
| 643 | - // no charset found |
|
| 644 | - $this->root->propagateEncoding($encode); |
|
| 645 | - |
|
| 646 | - return false; |
|
| 647 | - } |
|
| 19 | + /** |
|
| 20 | + * The charset we would like the output to be in. |
|
| 21 | + * |
|
| 22 | + * @var string |
|
| 23 | + */ |
|
| 24 | + protected $defaultCharset = 'UTF-8'; |
|
| 25 | + |
|
| 26 | + /** |
|
| 27 | + * Contains the root node of this dom tree. |
|
| 28 | + * |
|
| 29 | + * @var HtmlNode |
|
| 30 | + */ |
|
| 31 | + public $root; |
|
| 32 | + |
|
| 33 | + /** |
|
| 34 | + * The raw version of the document string. |
|
| 35 | + * |
|
| 36 | + * @var string |
|
| 37 | + */ |
|
| 38 | + protected $raw; |
|
| 39 | + |
|
| 40 | + /** |
|
| 41 | + * The document string. |
|
| 42 | + * |
|
| 43 | + * @var Content |
|
| 44 | + */ |
|
| 45 | + protected $content = null; |
|
| 46 | + |
|
| 47 | + /** |
|
| 48 | + * The original file size of the document. |
|
| 49 | + * |
|
| 50 | + * @var int |
|
| 51 | + */ |
|
| 52 | + protected $rawSize; |
|
| 53 | + |
|
| 54 | + /** |
|
| 55 | + * The size of the document after it is cleaned. |
|
| 56 | + * |
|
| 57 | + * @var int |
|
| 58 | + */ |
|
| 59 | + protected $size; |
|
| 60 | + |
|
| 61 | + /** |
|
| 62 | + * A global options array to be used by all load calls. |
|
| 63 | + * |
|
| 64 | + * @var array |
|
| 65 | + */ |
|
| 66 | + protected $globalOptions = []; |
|
| 67 | + |
|
| 68 | + /** |
|
| 69 | + * A persistent option object to be used for all options in the |
|
| 70 | + * parsing of the file. |
|
| 71 | + * |
|
| 72 | + * @var Options |
|
| 73 | + */ |
|
| 74 | + protected $options; |
|
| 75 | + |
|
| 76 | + /** |
|
| 77 | + * A list of tags which will always be self closing |
|
| 78 | + * |
|
| 79 | + * @var array |
|
| 80 | + */ |
|
| 81 | + protected $selfClosing = [ |
|
| 82 | + 'img', |
|
| 83 | + 'br', |
|
| 84 | + 'input', |
|
| 85 | + 'meta', |
|
| 86 | + 'link', |
|
| 87 | + 'hr', |
|
| 88 | + 'base', |
|
| 89 | + 'embed', |
|
| 90 | + 'spacer', |
|
| 91 | + ]; |
|
| 92 | + |
|
| 93 | + /** |
|
| 94 | + * Returns the inner html of the root node. |
|
| 95 | + * |
|
| 96 | + * @return string |
|
| 97 | + */ |
|
| 98 | + public function __toString() |
|
| 99 | + { |
|
| 100 | + return $this->root->innerHtml(); |
|
| 101 | + } |
|
| 102 | + |
|
| 103 | + /** |
|
| 104 | + * A simple wrapper around the root node. |
|
| 105 | + * |
|
| 106 | + * @param string $name |
|
| 107 | + * @return mixed |
|
| 108 | + */ |
|
| 109 | + public function __get($name) |
|
| 110 | + { |
|
| 111 | + return $this->root->$name; |
|
| 112 | + } |
|
| 113 | + |
|
| 114 | + /** |
|
| 115 | + * Attempts to load the dom from any resource, string, file, or URL. |
|
| 116 | + * |
|
| 117 | + * @param string $str |
|
| 118 | + * @param array $options |
|
| 119 | + * @return $this |
|
| 120 | + */ |
|
| 121 | + public function load($str, $options = []) |
|
| 122 | + { |
|
| 123 | + // check if it's a file |
|
| 124 | + if (strpos($str, "\n") === false && is_file($str)) { |
|
| 125 | + return $this->loadFromFile($str, $options); |
|
| 126 | + } |
|
| 127 | + // check if it's a url |
|
| 128 | + if (preg_match("/^https?:\/\//i", $str)) { |
|
| 129 | + return $this->loadFromUrl($str, $options); |
|
| 130 | + } |
|
| 131 | + |
|
| 132 | + return $this->loadStr($str, $options); |
|
| 133 | + } |
|
| 134 | + |
|
| 135 | + /** |
|
| 136 | + * Loads the dom from a document file/url |
|
| 137 | + * |
|
| 138 | + * @param string $file |
|
| 139 | + * @param array $options |
|
| 140 | + * @return $this |
|
| 141 | + */ |
|
| 142 | + public function loadFromFile($file, $options = []) |
|
| 143 | + { |
|
| 144 | + return $this->loadStr(file_get_contents($file), $options); |
|
| 145 | + } |
|
| 146 | + |
|
| 147 | + /** |
|
| 148 | + * Use a curl interface implementation to attempt to load |
|
| 149 | + * the content from a url. |
|
| 150 | + * |
|
| 151 | + * @param string $url |
|
| 152 | + * @param array $options |
|
| 153 | + * @param CurlInterface $curl |
|
| 154 | + * @return $this |
|
| 155 | + */ |
|
| 156 | + public function loadFromUrl($url, $options = [], CurlInterface $curl = null) |
|
| 157 | + { |
|
| 158 | + if (is_null($curl)) { |
|
| 159 | + // use the default curl interface |
|
| 160 | + $curl = new Curl; |
|
| 161 | + } |
|
| 162 | + $content = $curl->get($url); |
|
| 163 | + |
|
| 164 | + return $this->loadStr($content, $options); |
|
| 165 | + } |
|
| 166 | + |
|
| 167 | + /** |
|
| 168 | + * Parsers the html of the given string. Used for load(), loadFromFile(), |
|
| 169 | + * and loadFromUrl(). |
|
| 170 | + * |
|
| 171 | + * @param string $str |
|
| 172 | + * @param array $option |
|
| 173 | + * @return $this |
|
| 174 | + */ |
|
| 175 | + public function loadStr($str, $option) |
|
| 176 | + { |
|
| 177 | + $this->options = new Options; |
|
| 178 | + $this->options->setOptions($this->globalOptions) |
|
| 179 | + ->setOptions($option); |
|
| 180 | + |
|
| 181 | + $this->rawSize = strlen($str); |
|
| 182 | + $this->raw = $str; |
|
| 183 | + |
|
| 184 | + $html = $this->clean($str); |
|
| 185 | + |
|
| 186 | + $this->size = strlen($str); |
|
| 187 | + $this->content = new Content($html); |
|
| 188 | + |
|
| 189 | + $this->parse(); |
|
| 190 | + $this->detectCharset(); |
|
| 191 | + |
|
| 192 | + return $this; |
|
| 193 | + } |
|
| 194 | + |
|
| 195 | + /** |
|
| 196 | + * Sets a global options array to be used by all load calls. |
|
| 197 | + * |
|
| 198 | + * @param array $options |
|
| 199 | + * @return $this |
|
| 200 | + */ |
|
| 201 | + public function setOptions(array $options) |
|
| 202 | + { |
|
| 203 | + $this->globalOptions = $options; |
|
| 204 | + |
|
| 205 | + return $this; |
|
| 206 | + } |
|
| 207 | + |
|
| 208 | + /** |
|
| 209 | + * Find elements by css selector on the root node. |
|
| 210 | + * |
|
| 211 | + * @param string $selector |
|
| 212 | + * @param int $nth |
|
| 213 | + * @return array |
|
| 214 | + */ |
|
| 215 | + public function find($selector, $nth = null) |
|
| 216 | + { |
|
| 217 | + $this->isLoaded(); |
|
| 218 | + |
|
| 219 | + return $this->root->find($selector, $nth); |
|
| 220 | + } |
|
| 221 | + |
|
| 222 | + /** |
|
| 223 | + * Adds the tag (or tags in an array) to the list of tags that will always |
|
| 224 | + * be self closing. |
|
| 225 | + * |
|
| 226 | + * @param string|array $tag |
|
| 227 | + * @return $this |
|
| 228 | + */ |
|
| 229 | + public function addSelfClosingTag($tag) |
|
| 230 | + { |
|
| 231 | + if ( ! is_array($tag)) { |
|
| 232 | + $tag = [$tag]; |
|
| 233 | + } |
|
| 234 | + foreach ($tag as $value) { |
|
| 235 | + $this->selfClosing[] = $value; |
|
| 236 | + } |
|
| 237 | + |
|
| 238 | + return $this; |
|
| 239 | + } |
|
| 240 | + |
|
| 241 | + /** |
|
| 242 | + * Removes the tag (or tags in an array) from the list of tags that will |
|
| 243 | + * always be self closing. |
|
| 244 | + * |
|
| 245 | + * @param string|array $tag |
|
| 246 | + * @return $this |
|
| 247 | + */ |
|
| 248 | + public function removeSelfClosingTag($tag) |
|
| 249 | + { |
|
| 250 | + if ( ! is_array($tag)) { |
|
| 251 | + $tag = [$tag]; |
|
| 252 | + } |
|
| 253 | + $this->selfClosing = array_diff($this->selfClosing, $tag); |
|
| 254 | + |
|
| 255 | + return $this; |
|
| 256 | + } |
|
| 257 | + |
|
| 258 | + /** |
|
| 259 | + * Sets the list of self closing tags to empty. |
|
| 260 | + * |
|
| 261 | + * @return $this |
|
| 262 | + */ |
|
| 263 | + public function clearSelfClosingTags() |
|
| 264 | + { |
|
| 265 | + $this->selfClosing = []; |
|
| 266 | + |
|
| 267 | + return $this; |
|
| 268 | + } |
|
| 269 | + |
|
| 270 | + /** |
|
| 271 | + * Simple wrapper function that returns the first child. |
|
| 272 | + * |
|
| 273 | + * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 274 | + */ |
|
| 275 | + public function firstChild() |
|
| 276 | + { |
|
| 277 | + $this->isLoaded(); |
|
| 278 | + |
|
| 279 | + return $this->root->firstChild(); |
|
| 280 | + } |
|
| 281 | + |
|
| 282 | + /** |
|
| 283 | + * Simple wrapper function that returns the last child. |
|
| 284 | + * |
|
| 285 | + * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 286 | + */ |
|
| 287 | + public function lastChild() |
|
| 288 | + { |
|
| 289 | + $this->isLoaded(); |
|
| 290 | + |
|
| 291 | + return $this->root->lastChild(); |
|
| 292 | + } |
|
| 293 | + |
|
| 294 | + /** |
|
| 295 | + * Simple wrapper function that returns an element by the |
|
| 296 | + * id. |
|
| 297 | + * |
|
| 298 | + * @param string $id |
|
| 299 | + * @return \PHPHtmlParser\Dom\AbstractNode |
|
| 300 | + */ |
|
| 301 | + public function getElementById($id) |
|
| 302 | + { |
|
| 303 | + $this->isLoaded(); |
|
| 304 | + |
|
| 305 | + return $this->find('#'.$id, 0); |
|
| 306 | + } |
|
| 307 | + |
|
| 308 | + /** |
|
| 309 | + * Simple wrapper function that returns all elements by |
|
| 310 | + * tag name. |
|
| 311 | + * |
|
| 312 | + * @param string $name |
|
| 313 | + * @return array |
|
| 314 | + */ |
|
| 315 | + public function getElementsByTag($name) |
|
| 316 | + { |
|
| 317 | + $this->isLoaded(); |
|
| 318 | + |
|
| 319 | + return $this->find($name); |
|
| 320 | + } |
|
| 321 | + |
|
| 322 | + /** |
|
| 323 | + * Simple wrapper function that returns all elements by |
|
| 324 | + * class name. |
|
| 325 | + * |
|
| 326 | + * @param string $class |
|
| 327 | + * @return array |
|
| 328 | + */ |
|
| 329 | + public function getElementsByClass($class) |
|
| 330 | + { |
|
| 331 | + $this->isLoaded(); |
|
| 332 | + |
|
| 333 | + return $this->find('.'.$class); |
|
| 334 | + } |
|
| 335 | + |
|
| 336 | + /** |
|
| 337 | + * Checks if the load methods have been called. |
|
| 338 | + * |
|
| 339 | + * @throws NotLoadedException |
|
| 340 | + */ |
|
| 341 | + protected function isLoaded() |
|
| 342 | + { |
|
| 343 | + if (is_null($this->content)) { |
|
| 344 | + throw new NotLoadedException('Content is not loaded!'); |
|
| 345 | + } |
|
| 346 | + } |
|
| 347 | + |
|
| 348 | + /** |
|
| 349 | + * Cleans the html of any none-html information. |
|
| 350 | + * |
|
| 351 | + * @param string $str |
|
| 352 | + * @return string |
|
| 353 | + */ |
|
| 354 | + protected function clean($str) |
|
| 355 | + { |
|
| 356 | + if ($this->options->get('cleanupInput') != true) { |
|
| 357 | + // skip entire cleanup step |
|
| 358 | + return $str; |
|
| 359 | + } |
|
| 360 | + |
|
| 361 | + // remove white space before closing tags |
|
| 362 | + $str = mb_eregi_replace("'\s+>", "'>", $str); |
|
| 363 | + $str = mb_eregi_replace('"\s+>', '">', $str); |
|
| 364 | + |
|
| 365 | + // clean out the \n\r |
|
| 366 | + $replace = ' '; |
|
| 367 | + if ($this->options->get('preserveLineBreaks')) { |
|
| 368 | + $replace = ' '; |
|
| 369 | + } |
|
| 370 | + $str = str_replace(["\r\n", "\r", "\n"], $replace, $str); |
|
| 371 | + |
|
| 372 | + // strip the doctype |
|
| 373 | + $str = mb_eregi_replace("<!doctype(.*?)>", '', $str); |
|
| 374 | + |
|
| 375 | + // strip out comments |
|
| 376 | + $str = mb_eregi_replace("<!--(.*?)-->", '', $str); |
|
| 377 | + |
|
| 378 | + // strip out cdata |
|
| 379 | + $str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str); |
|
| 380 | + |
|
| 381 | + // strip out <script> tags |
|
| 382 | + if ($this->options->get('removeScripts') == true) { |
|
| 383 | + $str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str); |
|
| 384 | + $str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str); |
|
| 385 | + } |
|
| 386 | + |
|
| 387 | + // strip out <style> tags |
|
| 388 | + if ($this->options->get('removeStyles') == true) { |
|
| 389 | + $str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str); |
|
| 390 | + $str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str); |
|
| 391 | + } |
|
| 392 | + |
|
| 393 | + // strip out server side scripts |
|
| 394 | + $str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str); |
|
| 395 | + |
|
| 396 | + // strip smarty scripts |
|
| 397 | + $str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str); |
|
| 398 | + |
|
| 399 | + return $str; |
|
| 400 | + } |
|
| 401 | + |
|
| 402 | + /** |
|
| 403 | + * Attempts to parse the html in content. |
|
| 404 | + */ |
|
| 405 | + protected function parse() |
|
| 406 | + { |
|
| 407 | + // add the root node |
|
| 408 | + $this->root = new HtmlNode('root'); |
|
| 409 | + $activeNode = $this->root; |
|
| 410 | + while ( ! is_null($activeNode)) { |
|
| 411 | + $str = $this->content->copyUntil('<'); |
|
| 412 | + if ($str == '') { |
|
| 413 | + $info = $this->parseTag(); |
|
| 414 | + if ( ! $info['status']) { |
|
| 415 | + // we are done here |
|
| 416 | + $activeNode = null; |
|
| 417 | + continue; |
|
| 418 | + } |
|
| 419 | + |
|
| 420 | + // check if it was a closing tag |
|
| 421 | + if ($info['closing']) { |
|
| 422 | + $originalNode = $activeNode; |
|
| 423 | + while ($activeNode->getTag()->name() != $info['tag']) { |
|
| 424 | + $activeNode = $activeNode->getParent(); |
|
| 425 | + if (is_null($activeNode)) { |
|
| 426 | + // we could not find opening tag |
|
| 427 | + $activeNode = $originalNode; |
|
| 428 | + break; |
|
| 429 | + } |
|
| 430 | + } |
|
| 431 | + if ( ! is_null($activeNode)) { |
|
| 432 | + $activeNode = $activeNode->getParent(); |
|
| 433 | + } |
|
| 434 | + continue; |
|
| 435 | + } |
|
| 436 | + |
|
| 437 | + if ( ! isset($info['node'])) { |
|
| 438 | + continue; |
|
| 439 | + } |
|
| 440 | + |
|
| 441 | + /** @var AbstractNode $node */ |
|
| 442 | + $node = $info['node']; |
|
| 443 | + $activeNode->addChild($node); |
|
| 444 | + |
|
| 445 | + // check if node is self closing |
|
| 446 | + if ( ! $node->getTag()->isSelfClosing()) { |
|
| 447 | + $activeNode = $node; |
|
| 448 | + } |
|
| 449 | + } else if ($this->options->whitespaceTextNode || |
|
| 450 | + trim($str) != '' |
|
| 451 | + ) { |
|
| 452 | + // we found text we care about |
|
| 453 | + $textNode = new TextNode($str); |
|
| 454 | + $activeNode->addChild($textNode); |
|
| 455 | + } |
|
| 456 | + } |
|
| 457 | + } |
|
| 458 | + |
|
| 459 | + /** |
|
| 460 | + * Attempt to parse a tag out of the content. |
|
| 461 | + * |
|
| 462 | + * @return array |
|
| 463 | + * @throws StrictException |
|
| 464 | + */ |
|
| 465 | + protected function parseTag() |
|
| 466 | + { |
|
| 467 | + $return = [ |
|
| 468 | + 'status' => false, |
|
| 469 | + 'closing' => false, |
|
| 470 | + 'node' => null, |
|
| 471 | + ]; |
|
| 472 | + if ($this->content->char() != '<') { |
|
| 473 | + // we are not at the beginning of a tag |
|
| 474 | + return $return; |
|
| 475 | + } |
|
| 476 | + |
|
| 477 | + // check if this is a closing tag |
|
| 478 | + if ($this->content->fastForward(1)->char() == '/') { |
|
| 479 | + // end tag |
|
| 480 | + $tag = $this->content->fastForward(1) |
|
| 481 | + ->copyByToken('slash', true); |
|
| 482 | + // move to end of tag |
|
| 483 | + $this->content->copyUntil('>'); |
|
| 484 | + $this->content->fastForward(1); |
|
| 485 | + |
|
| 486 | + // check if this closing tag counts |
|
| 487 | + $tag = strtolower($tag); |
|
| 488 | + if (in_array($tag, $this->selfClosing)) { |
|
| 489 | + $return['status'] = true; |
|
| 490 | + |
|
| 491 | + return $return; |
|
| 492 | + } else { |
|
| 493 | + $return['status'] = true; |
|
| 494 | + $return['closing'] = true; |
|
| 495 | + $return['tag'] = strtolower($tag); |
|
| 496 | + } |
|
| 497 | + |
|
| 498 | + return $return; |
|
| 499 | + } |
|
| 500 | + |
|
| 501 | + $tag = strtolower($this->content->copyByToken('slash', true)); |
|
| 502 | + $node = new HtmlNode($tag); |
|
| 503 | + |
|
| 504 | + // attributes |
|
| 505 | + while ($this->content->char() != '>' && |
|
| 506 | + $this->content->char() != '/') { |
|
| 507 | + $space = $this->content->skipByToken('blank', true); |
|
| 508 | + if (empty($space)) { |
|
| 509 | + $this->content->fastForward(1); |
|
| 510 | + continue; |
|
| 511 | + } |
|
| 512 | + |
|
| 513 | + $name = $this->content->copyByToken('equal', true); |
|
| 514 | + if ($name == '/') { |
|
| 515 | + break; |
|
| 516 | + } |
|
| 517 | + |
|
| 518 | + if (empty($name)) { |
|
| 519 | + $this->content->fastForward(1); |
|
| 520 | + continue; |
|
| 521 | + } |
|
| 522 | + |
|
| 523 | + $this->content->skipByToken('blank'); |
|
| 524 | + if ($this->content->char() == '=') { |
|
| 525 | + $attr = []; |
|
| 526 | + $this->content->fastForward(1) |
|
| 527 | + ->skipByToken('blank'); |
|
| 528 | + switch ($this->content->char()) { |
|
| 529 | + case '"': |
|
| 530 | + $attr['doubleQuote'] = true; |
|
| 531 | + $this->content->fastForward(1); |
|
| 532 | + $string = $this->content->copyUntil('"', true, true); |
|
| 533 | + do { |
|
| 534 | + $moreString = $this->content->copyUntilUnless('"', '=>'); |
|
| 535 | + $string .= $moreString; |
|
| 536 | + } while ( ! empty($moreString)); |
|
| 537 | + $attr['value'] = $string; |
|
| 538 | + $this->content->fastForward(1); |
|
| 539 | + $node->getTag()->$name = $attr; |
|
| 540 | + break; |
|
| 541 | + case "'": |
|
| 542 | + $attr['doubleQuote'] = false; |
|
| 543 | + $this->content->fastForward(1); |
|
| 544 | + $string = $this->content->copyUntil("'", true, true); |
|
| 545 | + do { |
|
| 546 | + $moreString = $this->content->copyUntilUnless("'", '=>'); |
|
| 547 | + $string .= $moreString; |
|
| 548 | + } while ( ! empty($moreString)); |
|
| 549 | + $attr['value'] = $string; |
|
| 550 | + $this->content->fastForward(1); |
|
| 551 | + $node->getTag()->$name = $attr; |
|
| 552 | + break; |
|
| 553 | + default: |
|
| 554 | + $attr['doubleQuote'] = true; |
|
| 555 | + $attr['value'] = $this->content->copyByToken('attr', true); |
|
| 556 | + $node->getTag()->$name = $attr; |
|
| 557 | + break; |
|
| 558 | + } |
|
| 559 | + } else { |
|
| 560 | + // no value attribute |
|
| 561 | + if ($this->options->strict) { |
|
| 562 | + // can't have this in strict html |
|
| 563 | + $character = $this->content->getPosition(); |
|
| 564 | + throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)"); |
|
| 565 | + } |
|
| 566 | + $node->getTag()->$name = [ |
|
| 567 | + 'value' => null, |
|
| 568 | + 'doubleQuote' => true, |
|
| 569 | + ]; |
|
| 570 | + if ($this->content->char() != '>') { |
|
| 571 | + $this->content->rewind(1); |
|
| 572 | + } |
|
| 573 | + } |
|
| 574 | + } |
|
| 575 | + |
|
| 576 | + $this->content->skipByToken('blank'); |
|
| 577 | + if ($this->content->char() == '/') { |
|
| 578 | + // self closing tag |
|
| 579 | + $node->getTag()->selfClosing(); |
|
| 580 | + $this->content->fastForward(1); |
|
| 581 | + } elseif (in_array($tag, $this->selfClosing)) { |
|
| 582 | + |
|
| 583 | + // Should be a self closing tag, check if we are strict |
|
| 584 | + if ($this->options->strict) { |
|
| 585 | + $character = $this->content->getPosition(); |
|
| 586 | + throw new StrictException("Tag '$tag' is not self closing! (character #$character)"); |
|
| 587 | + } |
|
| 588 | + |
|
| 589 | + // We force self closing on this tag. |
|
| 590 | + $node->getTag()->selfClosing(); |
|
| 591 | + } |
|
| 592 | + |
|
| 593 | + $this->content->fastForward(1); |
|
| 594 | + |
|
| 595 | + $return['status'] = true; |
|
| 596 | + $return['node'] = $node; |
|
| 597 | + |
|
| 598 | + return $return; |
|
| 599 | + } |
|
| 600 | + |
|
| 601 | + /** |
|
| 602 | + * Attempts to detect the charset that the html was sent in. |
|
| 603 | + * |
|
| 604 | + * @return bool |
|
| 605 | + */ |
|
| 606 | + protected function detectCharset() |
|
| 607 | + { |
|
| 608 | + // set the default |
|
| 609 | + $encode = new Encode; |
|
| 610 | + $encode->from($this->defaultCharset); |
|
| 611 | + $encode->to($this->defaultCharset); |
|
| 612 | + |
|
| 613 | + if ( ! is_null($this->options->enforceEncoding)) { |
|
| 614 | + // they want to enforce the given encoding |
|
| 615 | + $encode->from($this->options->enforceEncoding); |
|
| 616 | + $encode->to($this->options->enforceEncoding); |
|
| 617 | + |
|
| 618 | + return false; |
|
| 619 | + } |
|
| 620 | + |
|
| 621 | + $meta = $this->root->find('meta[http-equiv=Content-Type]', 0); |
|
| 622 | + if (is_null($meta)) { |
|
| 623 | + // could not find meta tag |
|
| 624 | + $this->root->propagateEncoding($encode); |
|
| 625 | + |
|
| 626 | + return false; |
|
| 627 | + } |
|
| 628 | + $content = $meta->content; |
|
| 629 | + if (empty($content)) { |
|
| 630 | + // could not find content |
|
| 631 | + $this->root->propagateEncoding($encode); |
|
| 632 | + |
|
| 633 | + return false; |
|
| 634 | + } |
|
| 635 | + $matches = []; |
|
| 636 | + if (preg_match('/charset=(.+)/', $content, $matches)) { |
|
| 637 | + $encode->from(trim($matches[1])); |
|
| 638 | + $this->root->propagateEncoding($encode); |
|
| 639 | + |
|
| 640 | + return true; |
|
| 641 | + } |
|
| 642 | + |
|
| 643 | + // no charset found |
|
| 644 | + $this->root->propagateEncoding($encode); |
|
| 645 | + |
|
| 646 | + return false; |
|
| 647 | + } |
|
| 648 | 648 | } |
@@ -9,97 +9,97 @@ |
||
| 9 | 9 | class TextNode extends LeafNode |
| 10 | 10 | { |
| 11 | 11 | |
| 12 | - /** |
|
| 13 | - * This is a text node. |
|
| 14 | - * |
|
| 15 | - * @var Tag |
|
| 16 | - */ |
|
| 17 | - protected $tag; |
|
| 12 | + /** |
|
| 13 | + * This is a text node. |
|
| 14 | + * |
|
| 15 | + * @var Tag |
|
| 16 | + */ |
|
| 17 | + protected $tag; |
|
| 18 | 18 | |
| 19 | - /** |
|
| 20 | - * This is the text in this node. |
|
| 21 | - * |
|
| 22 | - * @var string |
|
| 23 | - */ |
|
| 24 | - protected $text; |
|
| 19 | + /** |
|
| 20 | + * This is the text in this node. |
|
| 21 | + * |
|
| 22 | + * @var string |
|
| 23 | + */ |
|
| 24 | + protected $text; |
|
| 25 | 25 | |
| 26 | - /** |
|
| 27 | - * This is the converted version of the text. |
|
| 28 | - * |
|
| 29 | - * @var string |
|
| 30 | - */ |
|
| 31 | - protected $convertedText = null; |
|
| 26 | + /** |
|
| 27 | + * This is the converted version of the text. |
|
| 28 | + * |
|
| 29 | + * @var string |
|
| 30 | + */ |
|
| 31 | + protected $convertedText = null; |
|
| 32 | 32 | |
| 33 | - /** |
|
| 34 | - * Sets the text for this node. |
|
| 35 | - * |
|
| 36 | - * @param string $text |
|
| 37 | - */ |
|
| 38 | - public function __construct($text) |
|
| 39 | - { |
|
| 40 | - // remove double spaces |
|
| 41 | - $text = mb_ereg_replace('\s+', ' ', $text); |
|
| 33 | + /** |
|
| 34 | + * Sets the text for this node. |
|
| 35 | + * |
|
| 36 | + * @param string $text |
|
| 37 | + */ |
|
| 38 | + public function __construct($text) |
|
| 39 | + { |
|
| 40 | + // remove double spaces |
|
| 41 | + $text = mb_ereg_replace('\s+', ' ', $text); |
|
| 42 | 42 | |
| 43 | - // restore line breaks |
|
| 44 | - $text = str_replace(' ', "\n", $text); |
|
| 43 | + // restore line breaks |
|
| 44 | + $text = str_replace(' ', "\n", $text); |
|
| 45 | 45 | |
| 46 | - $this->text = $text; |
|
| 47 | - $this->tag = new Tag('text'); |
|
| 48 | - parent::__construct(); |
|
| 49 | - } |
|
| 46 | + $this->text = $text; |
|
| 47 | + $this->tag = new Tag('text'); |
|
| 48 | + parent::__construct(); |
|
| 49 | + } |
|
| 50 | 50 | |
| 51 | - /** |
|
| 52 | - * Returns the text of this node. |
|
| 53 | - * |
|
| 54 | - * @return string |
|
| 55 | - */ |
|
| 56 | - public function text() |
|
| 57 | - { |
|
| 58 | - // convert charset |
|
| 59 | - if ( ! is_null($this->encode)) { |
|
| 60 | - if ( ! is_null($this->convertedText)) { |
|
| 61 | - // we already know the converted value |
|
| 62 | - return $this->convertedText; |
|
| 63 | - } |
|
| 64 | - $text = $this->encode->convert($this->text); |
|
| 51 | + /** |
|
| 52 | + * Returns the text of this node. |
|
| 53 | + * |
|
| 54 | + * @return string |
|
| 55 | + */ |
|
| 56 | + public function text() |
|
| 57 | + { |
|
| 58 | + // convert charset |
|
| 59 | + if ( ! is_null($this->encode)) { |
|
| 60 | + if ( ! is_null($this->convertedText)) { |
|
| 61 | + // we already know the converted value |
|
| 62 | + return $this->convertedText; |
|
| 63 | + } |
|
| 64 | + $text = $this->encode->convert($this->text); |
|
| 65 | 65 | |
| 66 | - // remember the conversion |
|
| 67 | - $this->convertedText = $text; |
|
| 66 | + // remember the conversion |
|
| 67 | + $this->convertedText = $text; |
|
| 68 | 68 | |
| 69 | - return $text; |
|
| 70 | - } else { |
|
| 71 | - return $this->text; |
|
| 72 | - } |
|
| 73 | - } |
|
| 69 | + return $text; |
|
| 70 | + } else { |
|
| 71 | + return $this->text; |
|
| 72 | + } |
|
| 73 | + } |
|
| 74 | 74 | |
| 75 | - /** |
|
| 76 | - * This node has no html, just return the text. |
|
| 77 | - * |
|
| 78 | - * @return string |
|
| 79 | - * @uses $this->text() |
|
| 80 | - */ |
|
| 81 | - public function innerHtml() |
|
| 82 | - { |
|
| 83 | - return $this->text(); |
|
| 84 | - } |
|
| 75 | + /** |
|
| 76 | + * This node has no html, just return the text. |
|
| 77 | + * |
|
| 78 | + * @return string |
|
| 79 | + * @uses $this->text() |
|
| 80 | + */ |
|
| 81 | + public function innerHtml() |
|
| 82 | + { |
|
| 83 | + return $this->text(); |
|
| 84 | + } |
|
| 85 | 85 | |
| 86 | - /** |
|
| 87 | - * This node has no html, just return the text. |
|
| 88 | - * |
|
| 89 | - * @return string |
|
| 90 | - * @uses $this->text() |
|
| 91 | - */ |
|
| 92 | - public function outerHtml() |
|
| 93 | - { |
|
| 94 | - return $this->text(); |
|
| 95 | - } |
|
| 86 | + /** |
|
| 87 | + * This node has no html, just return the text. |
|
| 88 | + * |
|
| 89 | + * @return string |
|
| 90 | + * @uses $this->text() |
|
| 91 | + */ |
|
| 92 | + public function outerHtml() |
|
| 93 | + { |
|
| 94 | + return $this->text(); |
|
| 95 | + } |
|
| 96 | 96 | |
| 97 | - /** |
|
| 98 | - * Call this when something in the node tree has changed. Like a child has been added |
|
| 99 | - * or a parent has been changed. |
|
| 100 | - */ |
|
| 101 | - protected function clear() |
|
| 102 | - { |
|
| 103 | - $this->convertedText = null; |
|
| 104 | - } |
|
| 97 | + /** |
|
| 98 | + * Call this when something in the node tree has changed. Like a child has been added |
|
| 99 | + * or a parent has been changed. |
|
| 100 | + */ |
|
| 101 | + protected function clear() |
|
| 102 | + { |
|
| 103 | + $this->convertedText = null; |
|
| 104 | + } |
|
| 105 | 105 | } |
@@ -16,474 +16,474 @@ |
||
| 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 $this; |
|
| 153 | - } |
|
| 154 | - |
|
| 155 | - /** |
|
| 156 | - * Removes this node and all its children from the |
|
| 157 | - * DOM tree. |
|
| 158 | - * |
|
| 159 | - * @return void |
|
| 160 | - */ |
|
| 161 | - public function delete() |
|
| 162 | - { |
|
| 163 | - if ( ! is_null($this->parent)) { |
|
| 164 | - $this->parent->removeChild($this->id); |
|
| 165 | - } |
|
| 166 | - |
|
| 167 | - $this->parent = null; |
|
| 168 | - } |
|
| 169 | - |
|
| 170 | - /** |
|
| 171 | - * Sets the encoding class to this node. |
|
| 172 | - * |
|
| 173 | - * @param Encode $encode |
|
| 174 | - * @return void |
|
| 175 | - */ |
|
| 176 | - public function propagateEncoding(Encode $encode) |
|
| 177 | - { |
|
| 178 | - $this->encode = $encode; |
|
| 179 | - $this->tag->setEncoding($encode); |
|
| 180 | - } |
|
| 181 | - |
|
| 182 | - /** |
|
| 183 | - * Checks if the given node id is an ancestor of |
|
| 184 | - * the current node. |
|
| 185 | - * |
|
| 186 | - * @param int $id |
|
| 187 | - * @return bool |
|
| 188 | - */ |
|
| 189 | - public function isAncestor($id) |
|
| 190 | - { |
|
| 191 | - if ( ! is_null($this->getAncestor($id))) { |
|
| 192 | - return true; |
|
| 193 | - } |
|
| 194 | - |
|
| 195 | - return false; |
|
| 196 | - } |
|
| 197 | - |
|
| 198 | - /** |
|
| 199 | - * Attempts to get an ancestor node by the given id. |
|
| 200 | - * |
|
| 201 | - * @param int $id |
|
| 202 | - * @return null|AbstractNode |
|
| 203 | - */ |
|
| 204 | - public function getAncestor($id) |
|
| 205 | - { |
|
| 206 | - if ( ! is_null($this->parent)) { |
|
| 207 | - if ($this->parent->id() == $id) { |
|
| 208 | - return $this->parent; |
|
| 209 | - } |
|
| 210 | - |
|
| 211 | - return $this->parent->getAncestor($id); |
|
| 212 | - } |
|
| 213 | - |
|
| 214 | - return null; |
|
| 215 | - } |
|
| 216 | - |
|
| 217 | - /** |
|
| 218 | - * Shortcut to return the first child. |
|
| 219 | - * |
|
| 220 | - * @return AbstractNode |
|
| 221 | - * @uses $this->getChild() |
|
| 222 | - */ |
|
| 223 | - public function firstChild() |
|
| 224 | - { |
|
| 225 | - reset($this->children); |
|
| 226 | - $key = key($this->children); |
|
| 227 | - |
|
| 228 | - return $this->getChild($key); |
|
| 229 | - } |
|
| 230 | - |
|
| 231 | - /** |
|
| 232 | - * Attempts to get the last child. |
|
| 233 | - * |
|
| 234 | - * @return AbstractNode |
|
| 235 | - */ |
|
| 236 | - public function lastChild() |
|
| 237 | - { |
|
| 238 | - end($this->children); |
|
| 239 | - $key = key($this->children); |
|
| 240 | - |
|
| 241 | - return $this->getChild($key); |
|
| 242 | - } |
|
| 243 | - |
|
| 244 | - /** |
|
| 245 | - * Attempts to get the next sibling. |
|
| 246 | - * |
|
| 247 | - * @return AbstractNode |
|
| 248 | - * @throws ParentNotFoundException |
|
| 249 | - */ |
|
| 250 | - public function nextSibling() |
|
| 251 | - { |
|
| 252 | - if (is_null($this->parent)) { |
|
| 253 | - throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 254 | - } |
|
| 255 | - |
|
| 256 | - return $this->parent->nextChild($this->id); |
|
| 257 | - } |
|
| 258 | - |
|
| 259 | - /** |
|
| 260 | - * Attempts to get the previous sibling |
|
| 261 | - * |
|
| 262 | - * @return AbstractNode |
|
| 263 | - * @throws ParentNotFoundException |
|
| 264 | - */ |
|
| 265 | - public function previousSibling() |
|
| 266 | - { |
|
| 267 | - if (is_null($this->parent)) { |
|
| 268 | - throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 269 | - } |
|
| 270 | - |
|
| 271 | - return $this->parent->previousChild($this->id); |
|
| 272 | - } |
|
| 273 | - |
|
| 274 | - /** |
|
| 275 | - * Gets the tag object of this node. |
|
| 276 | - * |
|
| 277 | - * @return Tag |
|
| 278 | - */ |
|
| 279 | - public function getTag() |
|
| 280 | - { |
|
| 281 | - return $this->tag; |
|
| 282 | - } |
|
| 283 | - |
|
| 284 | - /** |
|
| 285 | - * A wrapper method that simply calls the getAttribute method |
|
| 286 | - * on the tag of this node. |
|
| 287 | - * |
|
| 288 | - * @return array |
|
| 289 | - */ |
|
| 290 | - public function getAttributes() |
|
| 291 | - { |
|
| 292 | - $attributes = $this->tag->getAttributes(); |
|
| 293 | - foreach ($attributes as $name => $info) { |
|
| 294 | - $attributes[$name] = $info['value']; |
|
| 295 | - } |
|
| 296 | - |
|
| 297 | - return $attributes; |
|
| 298 | - } |
|
| 299 | - |
|
| 300 | - /** |
|
| 301 | - * A wrapper method that simply calls the getAttribute method |
|
| 302 | - * on the tag of this node. |
|
| 303 | - * |
|
| 304 | - * @param string $key |
|
| 305 | - * @return mixed |
|
| 306 | - */ |
|
| 307 | - public function getAttribute($key) |
|
| 308 | - { |
|
| 309 | - $attribute = $this->tag->getAttribute($key); |
|
| 310 | - if ( ! is_null($attribute)) { |
|
| 311 | - $attribute = $attribute['value']; |
|
| 312 | - } |
|
| 313 | - |
|
| 314 | - return $attribute; |
|
| 315 | - } |
|
| 316 | - |
|
| 317 | - /** |
|
| 318 | - * A wrapper method that simply calls the setAttribute method |
|
| 319 | - * on the tag of this node. |
|
| 320 | - * |
|
| 321 | - * @param string $key |
|
| 322 | - * @param string $value |
|
| 323 | - * @return $this |
|
| 324 | - */ |
|
| 325 | - public function setAttribute($key, $value) |
|
| 326 | - { |
|
| 327 | - $this->tag->setAttribute($key, $value); |
|
| 328 | - |
|
| 329 | - return $this; |
|
| 330 | - } |
|
| 331 | - |
|
| 332 | - /** |
|
| 333 | - * Function to locate a specific ancestor tag in the path to the root. |
|
| 334 | - * |
|
| 335 | - * @param string $tag |
|
| 336 | - * @return AbstractNode |
|
| 337 | - * @throws ParentNotFoundException |
|
| 338 | - */ |
|
| 339 | - public function ancestorByTag($tag) |
|
| 340 | - { |
|
| 341 | - // Start by including ourselves in the comparison. |
|
| 342 | - $node = $this; |
|
| 343 | - |
|
| 344 | - while ( ! is_null($node)) { |
|
| 345 | - if ($node->tag->name() == $tag) { |
|
| 346 | - return $node; |
|
| 347 | - } |
|
| 348 | - |
|
| 349 | - $node = $node->getParent(); |
|
| 350 | - } |
|
| 351 | - |
|
| 352 | - throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag'); |
|
| 353 | - } |
|
| 354 | - |
|
| 355 | - /** |
|
| 356 | - * Find elements by css selector |
|
| 357 | - * |
|
| 358 | - * @param string $selector |
|
| 359 | - * @param int $nth |
|
| 360 | - * @return array|AbstractNode |
|
| 361 | - */ |
|
| 362 | - public function find($selector, $nth = null) |
|
| 363 | - { |
|
| 364 | - $selector = new Selector($selector); |
|
| 365 | - $nodes = $selector->find($this); |
|
| 366 | - |
|
| 367 | - if ( ! is_null($nth)) { |
|
| 368 | - // return nth-element or array |
|
| 369 | - if (isset($nodes[$nth])) { |
|
| 370 | - return $nodes[$nth]; |
|
| 371 | - } |
|
| 372 | - |
|
| 373 | - return null; |
|
| 374 | - } |
|
| 375 | - |
|
| 376 | - return $nodes; |
|
| 377 | - } |
|
| 378 | - |
|
| 379 | - /** |
|
| 380 | - * Function to try a few tricks to determine the displayed size of an img on the page. |
|
| 381 | - * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types. |
|
| 382 | - * |
|
| 383 | - * Future enhancement: |
|
| 384 | - * Look in the tag to see if there is a class or id specified that has a height or width attribute to it. |
|
| 385 | - * |
|
| 386 | - * Far future enhancement |
|
| 387 | - * 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 |
|
| 388 | - * Note that in this case, the class or id will have the img sub-selector for it to apply to the image. |
|
| 389 | - * |
|
| 390 | - * ridiculously far future development |
|
| 391 | - * 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. |
|
| 392 | - * |
|
| 393 | - * @author John Schlick |
|
| 394 | - * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out. |
|
| 395 | - */ |
|
| 396 | - public function get_display_size() |
|
| 397 | - { |
|
| 398 | - $width = -1; |
|
| 399 | - $height = -1; |
|
| 400 | - |
|
| 401 | - if ($this->tag->name() != 'img') { |
|
| 402 | - return false; |
|
| 403 | - } |
|
| 404 | - |
|
| 405 | - // See if there is a height or width attribute in the tag itself. |
|
| 406 | - if ( ! is_null($this->tag->getAttribute('width'))) { |
|
| 407 | - $width = $this->tag->getAttribute('width'); |
|
| 408 | - } |
|
| 409 | - |
|
| 410 | - if ( ! is_null($this->tag->getAttribute('height'))) { |
|
| 411 | - $height = $this->tag->getAttribute('height'); |
|
| 412 | - } |
|
| 413 | - |
|
| 414 | - // Now look for an inline style. |
|
| 415 | - if ( ! is_null($this->tag->getAttribute('style'))) { |
|
| 416 | - // Thanks to user 'gnarf' from stackoverflow for this regular expression. |
|
| 417 | - $attributes = []; |
|
| 418 | - preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches, |
|
| 419 | - PREG_SET_ORDER); |
|
| 420 | - foreach ($matches as $match) { |
|
| 421 | - $attributes[$match[1]] = $match[2]; |
|
| 422 | - } |
|
| 423 | - |
|
| 424 | - $width = $this->getLength($attributes, $width, 'width'); |
|
| 425 | - $height = $this->getLength($attributes, $width, 'height'); |
|
| 426 | - } |
|
| 427 | - |
|
| 428 | - $result = [ |
|
| 429 | - 'height' => $height, |
|
| 430 | - 'width' => $width, |
|
| 431 | - ]; |
|
| 432 | - |
|
| 433 | - return $result; |
|
| 434 | - } |
|
| 435 | - |
|
| 436 | - /** |
|
| 437 | - * If there is a length in the style attributes use it. |
|
| 438 | - * |
|
| 439 | - * @param array $attributes |
|
| 440 | - * @param int $length |
|
| 441 | - * @param string $key |
|
| 442 | - * @return int |
|
| 443 | - */ |
|
| 444 | - protected function getLength(array $attributes, $length, $key) |
|
| 445 | - { |
|
| 446 | - if (isset($attributes[$key]) && $length == -1) { |
|
| 447 | - // check that the last two characters are px (pixels) |
|
| 448 | - if (strtolower(substr($attributes[$key], -2)) == 'px') { |
|
| 449 | - $proposed_length = substr($attributes[$key], 0, -2); |
|
| 450 | - // Now make sure that it's an integer and not something stupid. |
|
| 451 | - if (filter_var($proposed_length, FILTER_VALIDATE_INT)) { |
|
| 452 | - $length = $proposed_length; |
|
| 453 | - } |
|
| 454 | - } |
|
| 455 | - } |
|
| 456 | - |
|
| 457 | - return $length; |
|
| 458 | - } |
|
| 459 | - |
|
| 460 | - /** |
|
| 461 | - * Gets the inner html of this node. |
|
| 462 | - * |
|
| 463 | - * @return string |
|
| 464 | - */ |
|
| 465 | - abstract public function innerHtml(); |
|
| 466 | - |
|
| 467 | - /** |
|
| 468 | - * Gets the html of this node, including it's own |
|
| 469 | - * tag. |
|
| 470 | - * |
|
| 471 | - * @return string |
|
| 472 | - */ |
|
| 473 | - abstract public function outerHtml(); |
|
| 474 | - |
|
| 475 | - /** |
|
| 476 | - * Gets the text of this node (if there is any text). |
|
| 477 | - * |
|
| 478 | - * @return string |
|
| 479 | - */ |
|
| 480 | - abstract public function text(); |
|
| 481 | - |
|
| 482 | - /** |
|
| 483 | - * Call this when something in the node tree has changed. Like a child has been added |
|
| 484 | - * or a parent has been changed. |
|
| 485 | - * |
|
| 486 | - * @return void |
|
| 487 | - */ |
|
| 488 | - 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 $this; |
|
| 153 | + } |
|
| 154 | + |
|
| 155 | + /** |
|
| 156 | + * Removes this node and all its children from the |
|
| 157 | + * DOM tree. |
|
| 158 | + * |
|
| 159 | + * @return void |
|
| 160 | + */ |
|
| 161 | + public function delete() |
|
| 162 | + { |
|
| 163 | + if ( ! is_null($this->parent)) { |
|
| 164 | + $this->parent->removeChild($this->id); |
|
| 165 | + } |
|
| 166 | + |
|
| 167 | + $this->parent = null; |
|
| 168 | + } |
|
| 169 | + |
|
| 170 | + /** |
|
| 171 | + * Sets the encoding class to this node. |
|
| 172 | + * |
|
| 173 | + * @param Encode $encode |
|
| 174 | + * @return void |
|
| 175 | + */ |
|
| 176 | + public function propagateEncoding(Encode $encode) |
|
| 177 | + { |
|
| 178 | + $this->encode = $encode; |
|
| 179 | + $this->tag->setEncoding($encode); |
|
| 180 | + } |
|
| 181 | + |
|
| 182 | + /** |
|
| 183 | + * Checks if the given node id is an ancestor of |
|
| 184 | + * the current node. |
|
| 185 | + * |
|
| 186 | + * @param int $id |
|
| 187 | + * @return bool |
|
| 188 | + */ |
|
| 189 | + public function isAncestor($id) |
|
| 190 | + { |
|
| 191 | + if ( ! is_null($this->getAncestor($id))) { |
|
| 192 | + return true; |
|
| 193 | + } |
|
| 194 | + |
|
| 195 | + return false; |
|
| 196 | + } |
|
| 197 | + |
|
| 198 | + /** |
|
| 199 | + * Attempts to get an ancestor node by the given id. |
|
| 200 | + * |
|
| 201 | + * @param int $id |
|
| 202 | + * @return null|AbstractNode |
|
| 203 | + */ |
|
| 204 | + public function getAncestor($id) |
|
| 205 | + { |
|
| 206 | + if ( ! is_null($this->parent)) { |
|
| 207 | + if ($this->parent->id() == $id) { |
|
| 208 | + return $this->parent; |
|
| 209 | + } |
|
| 210 | + |
|
| 211 | + return $this->parent->getAncestor($id); |
|
| 212 | + } |
|
| 213 | + |
|
| 214 | + return null; |
|
| 215 | + } |
|
| 216 | + |
|
| 217 | + /** |
|
| 218 | + * Shortcut to return the first child. |
|
| 219 | + * |
|
| 220 | + * @return AbstractNode |
|
| 221 | + * @uses $this->getChild() |
|
| 222 | + */ |
|
| 223 | + public function firstChild() |
|
| 224 | + { |
|
| 225 | + reset($this->children); |
|
| 226 | + $key = key($this->children); |
|
| 227 | + |
|
| 228 | + return $this->getChild($key); |
|
| 229 | + } |
|
| 230 | + |
|
| 231 | + /** |
|
| 232 | + * Attempts to get the last child. |
|
| 233 | + * |
|
| 234 | + * @return AbstractNode |
|
| 235 | + */ |
|
| 236 | + public function lastChild() |
|
| 237 | + { |
|
| 238 | + end($this->children); |
|
| 239 | + $key = key($this->children); |
|
| 240 | + |
|
| 241 | + return $this->getChild($key); |
|
| 242 | + } |
|
| 243 | + |
|
| 244 | + /** |
|
| 245 | + * Attempts to get the next sibling. |
|
| 246 | + * |
|
| 247 | + * @return AbstractNode |
|
| 248 | + * @throws ParentNotFoundException |
|
| 249 | + */ |
|
| 250 | + public function nextSibling() |
|
| 251 | + { |
|
| 252 | + if (is_null($this->parent)) { |
|
| 253 | + throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 254 | + } |
|
| 255 | + |
|
| 256 | + return $this->parent->nextChild($this->id); |
|
| 257 | + } |
|
| 258 | + |
|
| 259 | + /** |
|
| 260 | + * Attempts to get the previous sibling |
|
| 261 | + * |
|
| 262 | + * @return AbstractNode |
|
| 263 | + * @throws ParentNotFoundException |
|
| 264 | + */ |
|
| 265 | + public function previousSibling() |
|
| 266 | + { |
|
| 267 | + if (is_null($this->parent)) { |
|
| 268 | + throw new ParentNotFoundException('Parent is not set for this node.'); |
|
| 269 | + } |
|
| 270 | + |
|
| 271 | + return $this->parent->previousChild($this->id); |
|
| 272 | + } |
|
| 273 | + |
|
| 274 | + /** |
|
| 275 | + * Gets the tag object of this node. |
|
| 276 | + * |
|
| 277 | + * @return Tag |
|
| 278 | + */ |
|
| 279 | + public function getTag() |
|
| 280 | + { |
|
| 281 | + return $this->tag; |
|
| 282 | + } |
|
| 283 | + |
|
| 284 | + /** |
|
| 285 | + * A wrapper method that simply calls the getAttribute method |
|
| 286 | + * on the tag of this node. |
|
| 287 | + * |
|
| 288 | + * @return array |
|
| 289 | + */ |
|
| 290 | + public function getAttributes() |
|
| 291 | + { |
|
| 292 | + $attributes = $this->tag->getAttributes(); |
|
| 293 | + foreach ($attributes as $name => $info) { |
|
| 294 | + $attributes[$name] = $info['value']; |
|
| 295 | + } |
|
| 296 | + |
|
| 297 | + return $attributes; |
|
| 298 | + } |
|
| 299 | + |
|
| 300 | + /** |
|
| 301 | + * A wrapper method that simply calls the getAttribute method |
|
| 302 | + * on the tag of this node. |
|
| 303 | + * |
|
| 304 | + * @param string $key |
|
| 305 | + * @return mixed |
|
| 306 | + */ |
|
| 307 | + public function getAttribute($key) |
|
| 308 | + { |
|
| 309 | + $attribute = $this->tag->getAttribute($key); |
|
| 310 | + if ( ! is_null($attribute)) { |
|
| 311 | + $attribute = $attribute['value']; |
|
| 312 | + } |
|
| 313 | + |
|
| 314 | + return $attribute; |
|
| 315 | + } |
|
| 316 | + |
|
| 317 | + /** |
|
| 318 | + * A wrapper method that simply calls the setAttribute method |
|
| 319 | + * on the tag of this node. |
|
| 320 | + * |
|
| 321 | + * @param string $key |
|
| 322 | + * @param string $value |
|
| 323 | + * @return $this |
|
| 324 | + */ |
|
| 325 | + public function setAttribute($key, $value) |
|
| 326 | + { |
|
| 327 | + $this->tag->setAttribute($key, $value); |
|
| 328 | + |
|
| 329 | + return $this; |
|
| 330 | + } |
|
| 331 | + |
|
| 332 | + /** |
|
| 333 | + * Function to locate a specific ancestor tag in the path to the root. |
|
| 334 | + * |
|
| 335 | + * @param string $tag |
|
| 336 | + * @return AbstractNode |
|
| 337 | + * @throws ParentNotFoundException |
|
| 338 | + */ |
|
| 339 | + public function ancestorByTag($tag) |
|
| 340 | + { |
|
| 341 | + // Start by including ourselves in the comparison. |
|
| 342 | + $node = $this; |
|
| 343 | + |
|
| 344 | + while ( ! is_null($node)) { |
|
| 345 | + if ($node->tag->name() == $tag) { |
|
| 346 | + return $node; |
|
| 347 | + } |
|
| 348 | + |
|
| 349 | + $node = $node->getParent(); |
|
| 350 | + } |
|
| 351 | + |
|
| 352 | + throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag'); |
|
| 353 | + } |
|
| 354 | + |
|
| 355 | + /** |
|
| 356 | + * Find elements by css selector |
|
| 357 | + * |
|
| 358 | + * @param string $selector |
|
| 359 | + * @param int $nth |
|
| 360 | + * @return array|AbstractNode |
|
| 361 | + */ |
|
| 362 | + public function find($selector, $nth = null) |
|
| 363 | + { |
|
| 364 | + $selector = new Selector($selector); |
|
| 365 | + $nodes = $selector->find($this); |
|
| 366 | + |
|
| 367 | + if ( ! is_null($nth)) { |
|
| 368 | + // return nth-element or array |
|
| 369 | + if (isset($nodes[$nth])) { |
|
| 370 | + return $nodes[$nth]; |
|
| 371 | + } |
|
| 372 | + |
|
| 373 | + return null; |
|
| 374 | + } |
|
| 375 | + |
|
| 376 | + return $nodes; |
|
| 377 | + } |
|
| 378 | + |
|
| 379 | + /** |
|
| 380 | + * Function to try a few tricks to determine the displayed size of an img on the page. |
|
| 381 | + * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types. |
|
| 382 | + * |
|
| 383 | + * Future enhancement: |
|
| 384 | + * Look in the tag to see if there is a class or id specified that has a height or width attribute to it. |
|
| 385 | + * |
|
| 386 | + * Far future enhancement |
|
| 387 | + * 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 |
|
| 388 | + * Note that in this case, the class or id will have the img sub-selector for it to apply to the image. |
|
| 389 | + * |
|
| 390 | + * ridiculously far future development |
|
| 391 | + * 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. |
|
| 392 | + * |
|
| 393 | + * @author John Schlick |
|
| 394 | + * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out. |
|
| 395 | + */ |
|
| 396 | + public function get_display_size() |
|
| 397 | + { |
|
| 398 | + $width = -1; |
|
| 399 | + $height = -1; |
|
| 400 | + |
|
| 401 | + if ($this->tag->name() != 'img') { |
|
| 402 | + return false; |
|
| 403 | + } |
|
| 404 | + |
|
| 405 | + // See if there is a height or width attribute in the tag itself. |
|
| 406 | + if ( ! is_null($this->tag->getAttribute('width'))) { |
|
| 407 | + $width = $this->tag->getAttribute('width'); |
|
| 408 | + } |
|
| 409 | + |
|
| 410 | + if ( ! is_null($this->tag->getAttribute('height'))) { |
|
| 411 | + $height = $this->tag->getAttribute('height'); |
|
| 412 | + } |
|
| 413 | + |
|
| 414 | + // Now look for an inline style. |
|
| 415 | + if ( ! is_null($this->tag->getAttribute('style'))) { |
|
| 416 | + // Thanks to user 'gnarf' from stackoverflow for this regular expression. |
|
| 417 | + $attributes = []; |
|
| 418 | + preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches, |
|
| 419 | + PREG_SET_ORDER); |
|
| 420 | + foreach ($matches as $match) { |
|
| 421 | + $attributes[$match[1]] = $match[2]; |
|
| 422 | + } |
|
| 423 | + |
|
| 424 | + $width = $this->getLength($attributes, $width, 'width'); |
|
| 425 | + $height = $this->getLength($attributes, $width, 'height'); |
|
| 426 | + } |
|
| 427 | + |
|
| 428 | + $result = [ |
|
| 429 | + 'height' => $height, |
|
| 430 | + 'width' => $width, |
|
| 431 | + ]; |
|
| 432 | + |
|
| 433 | + return $result; |
|
| 434 | + } |
|
| 435 | + |
|
| 436 | + /** |
|
| 437 | + * If there is a length in the style attributes use it. |
|
| 438 | + * |
|
| 439 | + * @param array $attributes |
|
| 440 | + * @param int $length |
|
| 441 | + * @param string $key |
|
| 442 | + * @return int |
|
| 443 | + */ |
|
| 444 | + protected function getLength(array $attributes, $length, $key) |
|
| 445 | + { |
|
| 446 | + if (isset($attributes[$key]) && $length == -1) { |
|
| 447 | + // check that the last two characters are px (pixels) |
|
| 448 | + if (strtolower(substr($attributes[$key], -2)) == 'px') { |
|
| 449 | + $proposed_length = substr($attributes[$key], 0, -2); |
|
| 450 | + // Now make sure that it's an integer and not something stupid. |
|
| 451 | + if (filter_var($proposed_length, FILTER_VALIDATE_INT)) { |
|
| 452 | + $length = $proposed_length; |
|
| 453 | + } |
|
| 454 | + } |
|
| 455 | + } |
|
| 456 | + |
|
| 457 | + return $length; |
|
| 458 | + } |
|
| 459 | + |
|
| 460 | + /** |
|
| 461 | + * Gets the inner html of this node. |
|
| 462 | + * |
|
| 463 | + * @return string |
|
| 464 | + */ |
|
| 465 | + abstract public function innerHtml(); |
|
| 466 | + |
|
| 467 | + /** |
|
| 468 | + * Gets the html of this node, including it's own |
|
| 469 | + * tag. |
|
| 470 | + * |
|
| 471 | + * @return string |
|
| 472 | + */ |
|
| 473 | + abstract public function outerHtml(); |
|
| 474 | + |
|
| 475 | + /** |
|
| 476 | + * Gets the text of this node (if there is any text). |
|
| 477 | + * |
|
| 478 | + * @return string |
|
| 479 | + */ |
|
| 480 | + abstract public function text(); |
|
| 481 | + |
|
| 482 | + /** |
|
| 483 | + * Call this when something in the node tree has changed. Like a child has been added |
|
| 484 | + * or a parent has been changed. |
|
| 485 | + * |
|
| 486 | + * @return void |
|
| 487 | + */ |
|
| 488 | + abstract protected function clear(); |
|
| 489 | 489 | } |
