[347] | 1 | <?php
|
---|
| 2 | /**
|
---|
| 3 | * Html2Pdf Library - parsing Html class
|
---|
| 4 | *
|
---|
| 5 | * HTML => PDF converter
|
---|
| 6 | * distributed under the OSL-3.0 License
|
---|
| 7 | *
|
---|
| 8 | * @package Html2pdf
|
---|
| 9 | * @author Laurent MINGUET <webmaster@html2pdf.fr>
|
---|
| 10 | * @copyright 2017 Laurent MINGUET
|
---|
| 11 | */
|
---|
| 12 | namespace Spipu\Html2Pdf\Parsing;
|
---|
| 13 |
|
---|
| 14 | use Spipu\Html2Pdf\Exception\HtmlParsingException;
|
---|
| 15 |
|
---|
| 16 | /**
|
---|
| 17 | * Class TagParser
|
---|
| 18 | */
|
---|
| 19 | class TagParser
|
---|
| 20 | {
|
---|
| 21 | protected $level; // table level
|
---|
| 22 | protected $num = 0; // table number
|
---|
| 23 | protected $textParser;
|
---|
| 24 |
|
---|
| 25 | public function __construct(TextParser $textParser)
|
---|
| 26 | {
|
---|
| 27 | $this->textParser = $textParser;
|
---|
| 28 | $this->level = array($this->num);
|
---|
| 29 | }
|
---|
| 30 |
|
---|
| 31 | /**
|
---|
| 32 | * analyze a HTML tag
|
---|
| 33 | *
|
---|
| 34 | * @param string $code HTML code to analyze
|
---|
| 35 | *
|
---|
| 36 | * @return Node corresponding action
|
---|
| 37 | * @throws HtmlParsingException
|
---|
| 38 | */
|
---|
| 39 | public function analyzeTag($code)
|
---|
| 40 | {
|
---|
| 41 | // name of the tag, opening, closure, autoclosure
|
---|
| 42 | $tag = '<([\/]{0,1})([_a-z0-9]+)([\/>\s]+)';
|
---|
| 43 | if (!preg_match('/'.$tag.'/isU', $code, $match)) {
|
---|
| 44 | $e = new HtmlParsingException('The HTML tag ['.$code.'] provided is invalid');
|
---|
| 45 | $e->setInvalidTag($code);
|
---|
| 46 | throw $e;
|
---|
| 47 | }
|
---|
| 48 | $close = ($match[1] === '/' ? true : false);
|
---|
| 49 | $autoclose = preg_match('/\/>$/isU', $code);
|
---|
| 50 | $name = strtolower($match[2]);
|
---|
| 51 |
|
---|
| 52 | // required parameters (depends on the tag name)
|
---|
| 53 | $defaultParams = array();
|
---|
| 54 | $defaultParams['style'] = '';
|
---|
| 55 | if ($name === 'img') {
|
---|
| 56 | $defaultParams['alt'] = '';
|
---|
| 57 | $defaultParams['src'] = '';
|
---|
| 58 | } elseif ($name === 'a') {
|
---|
| 59 | $defaultParams['href'] = '';
|
---|
| 60 | }
|
---|
| 61 |
|
---|
| 62 | $param = array_merge($defaultParams, $this->extractTagAttributes($code));
|
---|
| 63 | $param['style'] = trim($param['style']);
|
---|
| 64 | if (strlen($param['style']) > 0 && substr($param['style'], -1) !== ';') {
|
---|
| 65 | $param['style'].= ';';
|
---|
| 66 | }
|
---|
| 67 |
|
---|
| 68 | // compliance of each parameter
|
---|
| 69 | $color = "#000000";
|
---|
| 70 | $border = null;
|
---|
| 71 | foreach ($param as $key => $val) {
|
---|
| 72 | switch ($key) {
|
---|
| 73 | case 'width':
|
---|
| 74 | unset($param[$key]);
|
---|
| 75 | $param['style'] .= 'width: '.$val.'px; ';
|
---|
| 76 | break;
|
---|
| 77 |
|
---|
| 78 | case 'align':
|
---|
| 79 | if ($name === 'img') {
|
---|
| 80 | unset($param[$key]);
|
---|
| 81 | $param['style'] .= 'float: '.$val.'; ';
|
---|
| 82 | } elseif ($name !== 'table') {
|
---|
| 83 | unset($param[$key]);
|
---|
| 84 | $param['style'] .= 'text-align: '.$val.'; ';
|
---|
| 85 | }
|
---|
| 86 | break;
|
---|
| 87 |
|
---|
| 88 | case 'valign':
|
---|
| 89 | unset($param[$key]);
|
---|
| 90 | $param['style'] .= 'vertical-align: '.$val.'; ';
|
---|
| 91 | break;
|
---|
| 92 |
|
---|
| 93 | case 'height':
|
---|
| 94 | unset($param[$key]);
|
---|
| 95 | $param['style'] .= 'height: '.$val.'px; ';
|
---|
| 96 | break;
|
---|
| 97 |
|
---|
| 98 | case 'bgcolor':
|
---|
| 99 | unset($param[$key]);
|
---|
| 100 | $param['style'] .= 'background: '.$val.'; ';
|
---|
| 101 | break;
|
---|
| 102 |
|
---|
| 103 | case 'bordercolor':
|
---|
| 104 | unset($param[$key]);
|
---|
| 105 | $color = $val;
|
---|
| 106 | break;
|
---|
| 107 |
|
---|
| 108 | case 'border':
|
---|
| 109 | unset($param[$key]);
|
---|
| 110 | if (preg_match('/^[0-9]+$/isU', $val)) {
|
---|
| 111 | $val = $val.'px';
|
---|
| 112 | }
|
---|
| 113 | $border = $val;
|
---|
| 114 | break;
|
---|
| 115 |
|
---|
| 116 | case 'cellpadding':
|
---|
| 117 | case 'cellspacing':
|
---|
| 118 | if (preg_match('/^([0-9]+)$/isU', $val)) {
|
---|
| 119 | $param[$key] = $val.'px';
|
---|
| 120 | }
|
---|
| 121 | break;
|
---|
| 122 |
|
---|
| 123 | case 'colspan':
|
---|
| 124 | case 'rowspan':
|
---|
| 125 | $val = preg_replace('/[^0-9]/isU', '', $val);
|
---|
| 126 | if (!$val) {
|
---|
| 127 | $val = 1;
|
---|
| 128 | }
|
---|
| 129 | $param[$key] = (int) $val;
|
---|
| 130 | break;
|
---|
| 131 |
|
---|
| 132 | case 'color':
|
---|
| 133 | if ($name === 'font') {
|
---|
| 134 | unset($param[$key]);
|
---|
| 135 | $param['style'] .= 'color: '.$val.'; ';
|
---|
| 136 | }
|
---|
| 137 | break;
|
---|
| 138 | }
|
---|
| 139 | }
|
---|
| 140 |
|
---|
| 141 | // compliance of the border
|
---|
| 142 | if ($border !== null) {
|
---|
| 143 | if ($border && $border !== '0px') {
|
---|
| 144 | $border = 'solid '.$border.' '.$color;
|
---|
| 145 | } else {
|
---|
| 146 | $border = 'none';
|
---|
| 147 | }
|
---|
| 148 |
|
---|
| 149 | $param['style'] .= 'border: '.$border.'; ';
|
---|
| 150 | $param['border'] = $border;
|
---|
| 151 | }
|
---|
| 152 |
|
---|
| 153 | // reading styles: decomposition and standardization
|
---|
| 154 | $styles = explode(';', $param['style']);
|
---|
| 155 | $param['style'] = array();
|
---|
| 156 | foreach ($styles as $style) {
|
---|
| 157 | $tmp = explode(':', $style);
|
---|
| 158 | if (count($tmp) > 1) {
|
---|
| 159 | $cod = $tmp[0];
|
---|
| 160 | unset($tmp[0]);
|
---|
| 161 | $tmp = implode(':', $tmp);
|
---|
| 162 | $param['style'][trim(strtolower($cod))] = preg_replace('/[\s]+/isU', ' ', trim($tmp));
|
---|
| 163 | }
|
---|
| 164 | }
|
---|
| 165 |
|
---|
| 166 | // determining the level of table opening, with an added level
|
---|
| 167 | if (in_array($name, array('ul', 'ol', 'table')) && !$close) {
|
---|
| 168 | $this->num++;
|
---|
| 169 | array_push($this->level, $this->num);
|
---|
| 170 | }
|
---|
| 171 |
|
---|
| 172 | // get the level of the table containing the element
|
---|
| 173 | if (!isset($param['num'])) {
|
---|
| 174 | $param['num'] = end($this->level);
|
---|
| 175 | }
|
---|
| 176 |
|
---|
| 177 | // for closures table: remove a level
|
---|
| 178 | if (in_array($name, array('ul', 'ol', 'table')) && $close) {
|
---|
| 179 | array_pop($this->level);
|
---|
| 180 | }
|
---|
| 181 |
|
---|
| 182 | // prepare the parameters
|
---|
| 183 | if (isset($param['value'])) {
|
---|
| 184 | $keepSpaces = in_array($name, array('qrcode', 'barcode'));
|
---|
| 185 | $param['value'] = $this->textParser->prepareTxt($param['value'], !$keepSpaces);
|
---|
| 186 | }
|
---|
| 187 | if (isset($param['alt'])) {
|
---|
| 188 | $param['alt'] = $this->textParser->prepareTxt($param['alt']);
|
---|
| 189 | }
|
---|
| 190 | if (isset($param['title'])) {
|
---|
| 191 | $param['title'] = $this->textParser->prepareTxt($param['title']);
|
---|
| 192 | }
|
---|
| 193 | if (isset($param['class'])) {
|
---|
| 194 | $param['class'] = $this->textParser->prepareTxt($param['class']);
|
---|
| 195 | }
|
---|
| 196 |
|
---|
| 197 | // return the new action to do
|
---|
| 198 | return new Node($name, $param, $close, $autoclose);
|
---|
| 199 | }
|
---|
| 200 |
|
---|
| 201 | /**
|
---|
| 202 | * Extract the list of attribute => value inside an HTML tag
|
---|
| 203 | *
|
---|
| 204 | * @param string $code The full HTML tag to parse
|
---|
| 205 | *
|
---|
| 206 | * @return array
|
---|
| 207 | */
|
---|
| 208 | public function extractTagAttributes($code)
|
---|
| 209 | {
|
---|
| 210 | $param = array();
|
---|
| 211 | $regexes = array(
|
---|
| 212 | '([a-zA-Z0-9_]+)=([^"\'\s>]+)', // read the parameters : name=value
|
---|
| 213 | '([a-zA-Z0-9_]+)=["]([^"]*)["]', // read the parameters : name="value"
|
---|
| 214 | "([a-zA-Z0-9_]+)=[']([^']*)[']" // read the parameters : name='value'
|
---|
| 215 | );
|
---|
| 216 |
|
---|
| 217 | foreach ($regexes as $regex) {
|
---|
| 218 | preg_match_all('/'.$regex.'/is', $code, $match);
|
---|
| 219 | $amountMatch = count($match[0]);
|
---|
| 220 | for ($k = 0; $k < $amountMatch; $k++) {
|
---|
| 221 | $param[trim(strtolower($match[1][$k]))] = trim($match[2][$k]);
|
---|
| 222 | }
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 | return $param;
|
---|
| 226 | }
|
---|
| 227 | }
|
---|