| 1 |
<?php |
| 2 |
/** |
| 3 |
* @package dompdf |
| 4 |
* @link https://github.com/dompdf/dompdf |
| 5 |
* @license http://www.gnu.org/copyleft/lesser.html GNU Lesser General Public License |
| 6 |
*/ |
| 7 |
namespace Dompdf\Css; |
| 8 |
|
| 9 |
use DOMElement; |
| 10 |
use DOMXPath; |
| 11 |
use Dompdf\Css\Content\Url; |
| 12 |
use Dompdf\Dompdf; |
| 13 |
use Dompdf\Exception; |
| 14 |
use Dompdf\FontMetrics; |
| 15 |
use Dompdf\Frame\FrameTree; |
| 16 |
use Dompdf\Helpers; |
| 17 |
|
| 18 |
/** |
| 19 |
* The master stylesheet class |
| 20 |
* |
| 21 |
* The Stylesheet class is responsible for parsing stylesheets and style |
| 22 |
* tags/attributes. It also acts as a registry of the individual Style |
| 23 |
* objects generated by the current set of loaded CSS files and style |
| 24 |
* elements. |
| 25 |
* |
| 26 |
* @see Style |
| 27 |
* @package dompdf |
| 28 |
*/ |
| 29 |
class Stylesheet |
| 30 |
{ |
| 31 |
/** |
| 32 |
* The location of the default built-in CSS file. |
| 33 |
*/ |
| 34 |
const DEFAULT_STYLESHEET = "/lib/res/html.css"; |
| 35 |
|
| 36 |
/** |
| 37 |
* User agent stylesheet origin |
| 38 |
* |
| 39 |
* @var int |
| 40 |
*/ |
| 41 |
const ORIG_UA = 1; |
| 42 |
|
| 43 |
/** |
| 44 |
* User normal stylesheet origin |
| 45 |
* |
| 46 |
* @var int |
| 47 |
*/ |
| 48 |
const ORIG_USER = 2; |
| 49 |
|
| 50 |
/** |
| 51 |
* Author normal stylesheet origin |
| 52 |
* |
| 53 |
* @var int |
| 54 |
*/ |
| 55 |
const ORIG_AUTHOR = 3; |
| 56 |
|
| 57 |
/** |
| 58 |
* RegEx pattern representing a CSS string |
| 59 |
* |
| 60 |
* @var string |
| 61 |
*/ |
| 62 |
const PATTERN_CSS_STRING = '(?<CSS_STRING>(?<CSS_STRING_QUOTE>[\'"])(?<CSS_STRING_VALUE>.*?)(?<!\\\\)\g{CSS_STRING_QUOTE})'; |
| 63 |
|
| 64 |
/** |
| 65 |
* RegEx pattern representing the CSS var() function |
| 66 |
* |
| 67 |
* @var string |
| 68 |
*/ |
| 69 |
const PATTERN_CSS_VAR_FN = "var\((([^()]|(?R))*)\)"; |
| 70 |
|
| 71 |
/** |
| 72 |
* RegEx pattern representing the CSS url() function |
| 73 |
* |
| 74 |
* @var string |
| 75 |
*/ |
| 76 |
const PATTERN_CSS_URL_FN = '(?<CSS_URL_FN>url\(\s*(?<CSS_URL_FN_QUOTE>[\'"]?)(?<CSS_URL_FN_VALUE>.*?)(?(CSS_URL_FN_QUOTE)(?<!\\\\)\g{CSS_URL_FN_QUOTE})\s*\))'; |
| 77 |
|
| 78 |
/** |
| 79 |
* RegEx pattern representing the CSS local() function |
| 80 |
* |
| 81 |
* @var string |
| 82 |
*/ |
| 83 |
const PATTERN_CSS_LOCAL_FN = '(?<CSS_LOCAL_FN>local\(\s*(?<CSS_LOCAL_FN_QUOTE>[\'"]?)(?<CSS_LOCAL_FN_VALUE>.*?)(?(CSS_LOCAL_FN_QUOTE)(?<!\\\\)\g{CSS_LOCAL_FN_QUOTE})\s*\))'; |
| 84 |
|
| 85 |
/** |
| 86 |
* RegEx pattern representing a CSS media query |
| 87 |
* |
| 88 |
* @var string |
| 89 |
*/ |
| 90 |
const PATTERN_MEDIA_QUERY = '(?<CSS_MEDIA_QUERY>(?:(?:(?:(?<CSS_MEDIA_QUERY_OP>only|not)\s+)?(?<CSS_MEDIA_QUERY_TYPE>all|aural|bitmap|braille|dompdf|embossed|handheld|paged|print|projection|screen|speech|static|tty|tv|visual))|(?:\(\s*(?<CSS_MEDIA_QUERY_FEATURE>(?:(?:(?:min|max)-)?(?:width|height))|orientation|[^:]*?)\s*(?:\:\s*(?<CSS_MEDIA_QUERY_CONDITION>.*?)\s*)?\))))'; |
| 91 |
|
| 92 |
/* |
| 93 |
* The highest possible specificity is 0x01000000 (and that is only for author |
| 94 |
* stylesheets, as it is for inline styles). Origin precedence can be achieved by |
| 95 |
* adding multiples of 0x10000000 to the actual specificity. Important |
| 96 |
* declarations are handled in Style; though technically they should be handled |
| 97 |
* here so that user important declarations can be made to take precedence over |
| 98 |
* user important declarations, this doesn't matter in practice as Dompdf does |
| 99 |
* not support user stylesheets, and user agent stylesheets can not include |
| 100 |
* important declarations. |
| 101 |
*/ |
| 102 |
private static $_stylesheet_origins = [ |
| 103 |
self::ORIG_UA => 0x00000000, // user agent declarations |
| 104 |
self::ORIG_USER => 0x10000000, // user normal declarations |
| 105 |
self::ORIG_AUTHOR => 0x30000000, // author normal declarations |
| 106 |
]; |
| 107 |
|
| 108 |
/** |
| 109 |
* Non-CSS presentational hints (i.e. HTML 4 attributes) are handled as if added |
| 110 |
* to the beginning of an author stylesheet, i.e. anything in author stylesheets |
| 111 |
* should override them. |
| 112 |
*/ |
| 113 |
const SPEC_NON_CSS = 0x20000000; |
| 114 |
|
| 115 |
/** |
| 116 |
* Current dompdf instance |
| 117 |
* |
| 118 |
* @var Dompdf |
| 119 |
*/ |
| 120 |
private $_dompdf; |
| 121 |
|
| 122 |
/** |
| 123 |
* Array of currently defined styles |
| 124 |
* |
| 125 |
* @var array<string, Style[]> |
| 126 |
*/ |
| 127 |
private $_styles; |
| 128 |
|
| 129 |
/** |
| 130 |
* Array of embedded files (dataURIs) found in the parsed CSS |
| 131 |
* |
| 132 |
* @var array<string> |
| 133 |
*/ |
| 134 |
private $_blobs; |
| 135 |
|
| 136 |
/** |
| 137 |
* Base protocol of the document being parsed |
| 138 |
* Used to handle relative urls. |
| 139 |
* |
| 140 |
* @var string |
| 141 |
*/ |
| 142 |
private $_protocol = ""; |
| 143 |
|
| 144 |
/** |
| 145 |
* Base hostname of the document being parsed |
| 146 |
* Used to handle relative urls. |
| 147 |
* |
| 148 |
* @var string |
| 149 |
*/ |
| 150 |
private $_base_host = ""; |
| 151 |
|
| 152 |
/** |
| 153 |
* Base path of the document being parsed |
| 154 |
* Used to handle relative urls. |
| 155 |
* |
| 156 |
* @var string |
| 157 |
*/ |
| 158 |
private $_base_path = ""; |
| 159 |
|
| 160 |
/** |
| 161 |
* The styles defined by @page rules |
| 162 |
* |
| 163 |
* @var array<Style> |
| 164 |
*/ |
| 165 |
private $_page_styles; |
| 166 |
|
| 167 |
/** |
| 168 |
* List of loaded files, used to prevent recursion |
| 169 |
* |
| 170 |
* @var array |
| 171 |
*/ |
| 172 |
private $_loaded_files; |
| 173 |
|
| 174 |
/** |
| 175 |
* Current stylesheet origin |
| 176 |
* |
| 177 |
* @var int |
| 178 |
*/ |
| 179 |
private $_current_origin = self::ORIG_UA; |
| 180 |
|
| 181 |
/** |
| 182 |
* Accepted CSS media types |
| 183 |
* List of types and parsing rules for future extensions: |
| 184 |
* http://www.w3.org/TR/REC-html40/types.html |
| 185 |
* screen, tty, tv, projection, handheld, print, braille, aural, all |
| 186 |
* The following are non standard extensions for undocumented specific environments. |
| 187 |
* static, visual, bitmap, paged, dompdf |
| 188 |
* Note, even though the generated pdf file is intended for print output, |
| 189 |
* the desired content might be different (e.g. screen or projection view of html file). |
| 190 |
* Therefore allow specification of content by dompdf setting Options::defaultMediaType. |
| 191 |
* If given, replace media "print" by Options::defaultMediaType. |
| 192 |
* (Previous version $ACCEPTED_MEDIA_TYPES = $ACCEPTED_GENERIC_MEDIA_TYPES + $ACCEPTED_DEFAULT_MEDIA_TYPE) |
| 193 |
*/ |
| 194 |
static $ACCEPTED_DEFAULT_MEDIA_TYPE = "print"; |
| 195 |
static $ACCEPTED_GENERIC_MEDIA_TYPES = ["all", "static", "visual", "bitmap", "paged", "dompdf"]; |
| 196 |
static $VALID_MEDIA_TYPES = ["all", "aural", "bitmap", "braille", "dompdf", "embossed", "handheld", "paged", "print", "projection", "screen", "speech", "static", "tty", "tv", "visual"]; |
| 197 |
|
| 198 |
/** |
| 199 |
* @var FontMetrics |
| 200 |
*/ |
| 201 |
private $fontMetrics; |
| 202 |
|
| 203 |
/** |
| 204 |
* The class constructor. |
| 205 |
* |
| 206 |
* The base protocol, host & path are initialized to those of |
| 207 |
* the current script. |
| 208 |
*/ |
| 209 |
function __construct(Dompdf $dompdf) |
| 210 |
{ |
| 211 |
$this->_dompdf = $dompdf; |
| 212 |
$this->setFontMetrics($dompdf->getFontMetrics()); |
| 213 |
$this->_styles = []; |
| 214 |
$this->_loaded_files = []; |
| 215 |
$script = __FILE__; |
| 216 |
if (isset($_SERVER["SCRIPT_FILENAME"])) { |
| 217 |
$script = $_SERVER["SCRIPT_FILENAME"]; |
| 218 |
} |
| 219 |
list($this->_protocol, $this->_base_host, $this->_base_path) = Helpers::explode_url($script); |
| 220 |
$this->_page_styles = ["base" => new Style($this)]; |
| 221 |
} |
| 222 |
|
| 223 |
/** |
| 224 |
* Set the base protocol |
| 225 |
* |
| 226 |
* @param string $protocol |
| 227 |
*/ |
| 228 |
function set_protocol(string $protocol) |
| 229 |
{ |
| 230 |
$this->_protocol = $protocol; |
| 231 |
} |
| 232 |
|
| 233 |
/** |
| 234 |
* Set the base host |
| 235 |
* |
| 236 |
* @param string $host |
| 237 |
*/ |
| 238 |
function set_host(string $host) |
| 239 |
{ |
| 240 |
$this->_base_host = $host; |
| 241 |
} |
| 242 |
|
| 243 |
/** |
| 244 |
* Set the base path |
| 245 |
* |
| 246 |
* @param string $path |
| 247 |
*/ |
| 248 |
function set_base_path(string $path) |
| 249 |
{ |
| 250 |
$this->_base_path = $path; |
| 251 |
} |
| 252 |
|
| 253 |
/** |
| 254 |
* Return the Dompdf object |
| 255 |
* |
| 256 |
* @return Dompdf |
| 257 |
*/ |
| 258 |
function get_dompdf() |
| 259 |
{ |
| 260 |
return $this->_dompdf; |
| 261 |
} |
| 262 |
|
| 263 |
/** |
| 264 |
* Return the base protocol for this stylesheet |
| 265 |
* |
| 266 |
* @return string |
| 267 |
*/ |
| 268 |
function get_protocol() |
| 269 |
{ |
| 270 |
return $this->_protocol; |
| 271 |
} |
| 272 |
|
| 273 |
/** |
| 274 |
* Return the base host for this stylesheet |
| 275 |
* |
| 276 |
* @return string |
| 277 |
*/ |
| 278 |
function get_host() |
| 279 |
{ |
| 280 |
return $this->_base_host; |
| 281 |
} |
| 282 |
|
| 283 |
/** |
| 284 |
* Return the base path for this stylesheet |
| 285 |
* |
| 286 |
* @return string |
| 287 |
*/ |
| 288 |
function get_base_path() |
| 289 |
{ |
| 290 |
return $this->_base_path; |
| 291 |
} |
| 292 |
|
| 293 |
/** |
| 294 |
* Get all registered styles as an associative array, indexed by selector. |
| 295 |
* |
| 296 |
* @return array<string, Style[]> |
| 297 |
*/ |
| 298 |
public function get_styles(): array |
| 299 |
{ |
| 300 |
return $this->_styles; |
| 301 |
} |
| 302 |
|
| 303 |
/** |
| 304 |
* Return the array of page styles |
| 305 |
* |
| 306 |
* @return Style[] |
| 307 |
*/ |
| 308 |
function get_page_styles() |
| 309 |
{ |
| 310 |
return $this->_page_styles; |
| 311 |
} |
| 312 |
|
| 313 |
/** |
| 314 |
* Create a new Style object associated with this stylesheet |
| 315 |
* |
| 316 |
* @return Style |
| 317 |
*/ |
| 318 |
function create_style(): Style |
| 319 |
{ |
| 320 |
return new Style($this, $this->_current_origin); |
| 321 |
} |
| 322 |
|
| 323 |
/** |
| 324 |
* Add a new Style object to the stylesheet |
| 325 |
* |
| 326 |
* The style's origin is changed to the current origin of the stylesheet. |
| 327 |
* |
| 328 |
* @param string $key the Style's selector |
| 329 |
* @param Style $style the Style to be added |
| 330 |
*/ |
| 331 |
function add_style(string $key, Style $style): void |
| 332 |
{ |
| 333 |
if (!isset($this->_styles[$key])) { |
| 334 |
$this->_styles[$key] = []; |
| 335 |
} |
| 336 |
|
| 337 |
$style->set_origin($this->_current_origin); |
| 338 |
$this->_styles[$key][] = $style; |
| 339 |
} |
| 340 |
|
| 341 |
/** |
| 342 |
* load and parse a CSS string |
| 343 |
* |
| 344 |
* @param string $css |
| 345 |
* @param int $origin |
| 346 |
*/ |
| 347 |
function load_css(&$css, $origin = self::ORIG_AUTHOR) |
| 348 |
{ |
| 349 |
if ($origin) { |
| 350 |
$this->_current_origin = $origin; |
| 351 |
} |
| 352 |
$this->_parse_css($css); |
| 353 |
} |
| 354 |
|
| 355 |
/** |
| 356 |
* load and parse a CSS file |
| 357 |
* |
| 358 |
* @param string $file |
| 359 |
* @param int $origin |
| 360 |
*/ |
| 361 |
function load_css_file($file, $origin = self::ORIG_AUTHOR) |
| 362 |
{ |
| 363 |
if ($origin) { |
| 364 |
$this->_current_origin = $origin; |
| 365 |
} |
| 366 |
|
| 367 |
// Prevent circular references |
| 368 |
if (isset($this->_loaded_files[$file])) { |
| 369 |
return; |
| 370 |
} |
| 371 |
$this->_loaded_files[$file] = true; |
| 372 |
|
| 373 |
$parsed_url = Helpers::explode_url($file); |
| 374 |
$protocol = $parsed_url["protocol"]; |
| 375 |
|
| 376 |
if ($file !== $this->getDefaultStylesheet()) { |
| 377 |
$options = $this->_dompdf->getOptions(); |
| 378 |
$allowed_protocols = $options->getAllowedProtocols(); |
| 379 |
if (!array_key_exists($protocol, $allowed_protocols)) { |
| 380 |
Helpers::record_warnings(E_USER_WARNING, "Permission denied on $file. The communication protocol is not supported.", __FILE__, __LINE__); |
| 381 |
return; |
| 382 |
} |
| 383 |
foreach ($allowed_protocols[$protocol]["rules"] as $rule) { |
| 384 |
[$result, $message] = $rule($file); |
| 385 |
if (!$result) { |
| 386 |
Helpers::record_warnings(E_USER_WARNING, "Error loading $file: $message", __FILE__, __LINE__); |
| 387 |
return; |
| 388 |
} |
| 389 |
} |
| 390 |
} |
| 391 |
|
| 392 |
if (strpos($file, "data:") === 0) { |
| 393 |
$parsed = Helpers::parse_data_uri($file); |
| 394 |
$css = $parsed["data"]; |
| 395 |
} else { |
| 396 |
[$css, $http_response_header] = Helpers::getFileContent($file, $this->_dompdf->getHttpContext()); |
| 397 |
|
| 398 |
$good_mime_type = true; |
| 399 |
if (isset($http_response_header) && !$this->_dompdf->getQuirksmode()) { |
| 400 |
foreach ($http_response_header as $_header) { |
| 401 |
if (preg_match("@Content-Type:\s*([\w/]+)@i", $_header, $matches) && |
| 402 |
($matches[1] !== "text/css") |
| 403 |
) { |
| 404 |
$good_mime_type = false; |
| 405 |
} |
| 406 |
} |
| 407 |
} |
| 408 |
if (!$good_mime_type || $css === null) { |
| 409 |
Helpers::record_warnings(E_USER_WARNING, "Unable to load css file $file", __FILE__, __LINE__); |
| 410 |
return; |
| 411 |
} |
| 412 |
|
| 413 |
[$this->_protocol, $this->_base_host, $this->_base_path] = $parsed_url; |
| 414 |
} |
| 415 |
|
| 416 |
$this->_parse_css($css); |
| 417 |
} |
| 418 |
|
| 419 |
/** |
| 420 |
* @link https://www.w3.org/TR/CSS21/cascade.html#specificity |
| 421 |
* |
| 422 |
* @param string $selector |
| 423 |
* @param int $origin |
| 424 |
* - Stylesheet::ORIG_UA: user agent style sheet |
| 425 |
* - Stylesheet::ORIG_USER: user style sheet |
| 426 |
* - Stylesheet::ORIG_AUTHOR: author style sheet |
| 427 |
* |
| 428 |
* @return int |
| 429 |
*/ |
| 430 |
protected function specificity(string $selector, int $origin = self::ORIG_AUTHOR): int |
| 431 |
{ |
| 432 |
$a = ($selector === "!attr") ? 1 : 0; |
| 433 |
|
| 434 |
$b = min(mb_substr_count($selector, "#"), 255); |
| 435 |
|
| 436 |
$c = min(mb_substr_count($selector, ".") + |
| 437 |
mb_substr_count($selector, "[") + |
| 438 |
mb_substr_count($selector, ":") - |
| 439 |
2 * mb_substr_count($selector, "::"), 255); |
| 440 |
|
| 441 |
$d = min(mb_substr_count($selector, " ") + |
| 442 |
mb_substr_count($selector, ">") + |
| 443 |
mb_substr_count($selector, "+") + |
| 444 |
mb_substr_count($selector, "~") - |
| 445 |
mb_substr_count($selector, "~=") + |
| 446 |
mb_substr_count($selector, "::"), 255); |
| 447 |
|
| 448 |
//If a normal element name is at the beginning of the string, |
| 449 |
//a leading whitespace might have been removed on whitespace collapsing and removal |
| 450 |
//therefore there might be one whitespace less as selected element names |
| 451 |
//this can lead to a too small specificity |
| 452 |
//see selectorToXpath |
| 453 |
|
| 454 |
if (!in_array($selector[0], [" ", ">", ".", "#", "+", "~", ":", "["], true) && $selector !== "*") { |
| 455 |
$d++; |
| 456 |
} |
| 457 |
|
| 458 |
if ($this->_dompdf->getOptions()->getDebugCss()) { |
| 459 |
/*DEBUGCSS*/ |
| 460 |
print "<pre>\n"; |
| 461 |
/*DEBUGCSS*/ |
| 462 |
printf("specificity(): 0x%08x \"%s\"\n", self::$_stylesheet_origins[$origin] + (($a << 24) | ($b << 16) | ($c << 8) | ($d)), $selector); |
| 463 |
/*DEBUGCSS*/ |
| 464 |
print "</pre>"; |
| 465 |
} |
| 466 |
|
| 467 |
return self::$_stylesheet_origins[$origin] + (($a << 24) | ($b << 16) | ($c << 8) | ($d)); |
| 468 |
} |
| 469 |
|
| 470 |
/** |
| 471 |
* Converts a CSS selector to an XPath query. |
| 472 |
* |
| 473 |
* @param string $selector |
| 474 |
* @param bool $firstPass |
| 475 |
* |
| 476 |
* @return array|null |
| 477 |
*/ |
| 478 |
protected function selectorToXpath(string $selector, bool $firstPass = false): ?array |
| 479 |
{ |
| 480 |
// Collapse white space and strip whitespace around delimiters |
| 481 |
//$search = array("/\\s+/", "/\\s+([.>#+:])\\s+/"); |
| 482 |
//$replace = array(" ", "\\1"); |
| 483 |
//$selector = preg_replace($search, $replace, trim($selector)); |
| 484 |
|
| 485 |
// Initial query, always expanded to // below (non-absolute) |
| 486 |
$query = "/"; |
| 487 |
|
| 488 |
// Will contain :before and :after |
| 489 |
$pseudo_elements = []; |
| 490 |
|
| 491 |
// Parse the selector |
| 492 |
//$s = preg_split("/([ :>.#+])/", $selector, -1, PREG_SPLIT_DELIM_CAPTURE); |
| 493 |
|
| 494 |
$delimiters = [" ", ">", ".", "#", "+", "~", ":", "[", "("]; |
| 495 |
|
| 496 |
// Add an implicit space at the beginning of the selector if there is no |
| 497 |
// delimiter there already. |
| 498 |
if (!in_array($selector[0], $delimiters, true)) { |
| 499 |
$selector = " $selector"; |
| 500 |
} |
| 501 |
|
| 502 |
$name = "*"; |
| 503 |
$len = mb_strlen($selector, "UTF-8"); |
| 504 |
$i = 0; |
| 505 |
|
| 506 |
while ($i < $len) { |
| 507 |
|
| 508 |
$s = $selector[$i]; |
| 509 |
$i++; |
| 510 |
|
| 511 |
// Eat characters up to the next delimiter |
| 512 |
$tok = ""; |
| 513 |
$escape = false; |
| 514 |
$in_attr = false; |
| 515 |
$in_func = false; |
| 516 |
|
| 517 |
while ($i < $len) { |
| 518 |
$c = $selector[$i]; |
| 519 |
$c_prev = $selector[$i - 1]; |
| 520 |
|
| 521 |
if ($c_prev === "\\" && !$escape) { |
| 522 |
$escape = true; |
| 523 |
} elseif ($escape === true) { |
| 524 |
$escape = false; |
| 525 |
} |
| 526 |
|
| 527 |
if (!$escape && !$in_func && !$in_attr && in_array($c, $delimiters, true) && !($c === $c_prev && $c === ":")) { |
| 528 |
break; |
| 529 |
} |
| 530 |
|
| 531 |
if ($c_prev === "[") { |
| 532 |
$in_attr = true; |
| 533 |
} |
| 534 |
if ($c_prev === "(") { |
| 535 |
$in_func = true; |
| 536 |
} |
| 537 |
|
| 538 |
$tok .= $selector[$i++]; |
| 539 |
|
| 540 |
if (!$escape && $in_attr && $c === "]") { |
| 541 |
$in_attr = false; |
| 542 |
break; |
| 543 |
} |
| 544 |
if (!$escape && $in_func && $c === ")") { |
| 545 |
$in_func = false; |
| 546 |
break; |
| 547 |
} |
| 548 |
} |
| 549 |
$tok = $this->parse_string($tok); |
| 550 |
|
| 551 |
|
| 552 |
switch ($s) { |
| 553 |
|
| 554 |
case " ": |
| 555 |
case ">": |
| 556 |
// All elements matching the next token that are descendants |
| 557 |
// or children of the current token |
| 558 |
// https://www.w3.org/TR/selectors-3/#descendant-combinators |
| 559 |
// https://www.w3.org/TR/selectors-3/#child-combinators |
| 560 |
$expr = $s === " " ? "descendant" : "child"; |
| 561 |
|
| 562 |
// Tag names are case-insensitive |
| 563 |
$name = $tok === "" ? "*" : strtolower($tok); |
| 564 |
$query .= "/$expr::$name"; |
| 565 |
break; |
| 566 |
|
| 567 |
case "+": |
| 568 |
// Next-sibling combinator |
| 569 |
// https://www.w3.org/TR/selectors-3/#sibling-combinators |
| 570 |
|
| 571 |
// Tag names are case-insensitive |
| 572 |
$name = $tok === "" ? "*" : strtolower($tok); |
| 573 |
$query .= "/following-sibling::*[1]"; |
| 574 |
|
| 575 |
if ($name !== "*") { |
| 576 |
$query .= "[name() = '$name']"; |
| 577 |
} |
| 578 |
break; |
| 579 |
|
| 580 |
case "~": |
| 581 |
// Subsequent-sibling combinator |
| 582 |
// https://www.w3.org/TR/selectors-3/#sibling-combinators |
| 583 |
|
| 584 |
// Tag names are case-insensitive |
| 585 |
$name = $tok === "" ? "*" : strtolower($tok); |
| 586 |
$query .= "/following-sibling::$name"; |
| 587 |
break; |
| 588 |
|
| 589 |
case "#": |
| 590 |
// All elements matching the current token with id equal |
| 591 |
// to the _next_ token |
| 592 |
// https://www.w3.org/TR/selectors-3/#id-selectors |
| 593 |
if ($query === "/") { |
| 594 |
$query .= "/*"; |
| 595 |
} |
| 596 |
|
| 597 |
$query .= "[@id=\"$tok\"]"; |
| 598 |
break; |
| 599 |
|
| 600 |
case ".": |
| 601 |
// All elements matching the current token with a class |
| 602 |
// equal to the _next_ token |
| 603 |
// https://www.w3.org/TR/selectors-3/#class-html |
| 604 |
if ($query === "/") { |
| 605 |
$query .= "/*"; |
| 606 |
} |
| 607 |
|
| 608 |
// Match multiple classes: $tok contains the current selected |
| 609 |
// class. Search for class attributes with class="$tok", |
| 610 |
// class=".* $tok .*" and class=".* $tok" |
| 611 |
|
| 612 |
// This doesn't work because libxml only supports XPath 1.0... |
| 613 |
//$query .= "[matches(@$attr,\"^{$tok}\$|^{$tok}[ ]+|[ ]+{$tok}\$|[ ]+{$tok}[ ]+\")]"; |
| 614 |
|
| 615 |
$query .= "[contains(concat(' ', normalize-space(@class), ' '), concat(' ', '$tok', ' '))]"; |
| 616 |
break; |
| 617 |
|
| 618 |
case ":": |
| 619 |
if ($query === "/") { |
| 620 |
$query .= "/*"; |
| 621 |
} |
| 622 |
|
| 623 |
$last = false; |
| 624 |
|
| 625 |
// Pseudo-classes |
| 626 |
switch ($tok) { |
| 627 |
|
| 628 |
case "root": |
| 629 |
$query .= "[not(parent::*)]"; |
| 630 |
break; |
| 631 |
|
| 632 |
case "first-child": |
| 633 |
$query .= "[not(preceding-sibling::*)]"; |
| 634 |
break; |
| 635 |
|
| 636 |
case "last-child": |
| 637 |
$query .= "[not(following-sibling::*)]"; |
| 638 |
break; |
| 639 |
|
| 640 |
case "only-child": |
| 641 |
$query .= "[not(preceding-sibling::*) and not(following-sibling::*)]"; |
| 642 |
break; |
| 643 |
|
| 644 |
// https://www.w3.org/TR/selectors-3/#nth-child-pseudo |
| 645 |
/** @noinspection PhpMissingBreakStatementInspection */ |
| 646 |
case "nth-last-child": |
| 647 |
$last = true; |
| 648 |
case "nth-child": |
| 649 |
$p = $i + 1; |
| 650 |
$nth = trim(mb_substr($selector, $p, mb_strpos($selector, ")", $i, "UTF-8") - $p, "UTF-8")); |
| 651 |
$position = $last |
| 652 |
? "(count(following-sibling::*) + 1)" |
| 653 |
: "(count(preceding-sibling::*) + 1)"; |
| 654 |
|
| 655 |
$condition = $this->selectorAnPlusB($nth, $position); |
| 656 |
$query .= "[$condition]"; |
| 657 |
break; |
| 658 |
|
| 659 |
// TODO: `*:first-of-type`, `*:nth-of-type` etc. |
| 660 |
// (without fixed element name) are treated equivalent |
| 661 |
// to their `:*-child` counterparts here. They might |
| 662 |
// not be properly expressible in XPath 1.0 |
| 663 |
case "first-of-type": |
| 664 |
$query .= "[not(preceding-sibling::$name)]"; |
| 665 |
break; |
| 666 |
|
| 667 |
case "last-of-type": |
| 668 |
$query .= "[not(following-sibling::$name)]"; |
| 669 |
break; |
| 670 |
|
| 671 |
case "only-of-type": |
| 672 |
$query .= "[not(preceding-sibling::$name) and not(following-sibling::$name)]"; |
| 673 |
break; |
| 674 |
|
| 675 |
// https://www.w3.org/TR/selectors-3/#nth-of-type-pseudo |
| 676 |
/** @noinspection PhpMissingBreakStatementInspection */ |
| 677 |
case "nth-last-of-type": |
| 678 |
$last = true; |
| 679 |
case "nth-of-type": |
| 680 |
$p = $i + 1; |
| 681 |
$nth = trim(mb_substr($selector, $p, mb_strpos($selector, ")", $i, "UTF-8") - $p, "UTF-8")); |
| 682 |
$position = $last |
| 683 |
? "(count(following-sibling::$name) + 1)" |
| 684 |
: "(count(preceding-sibling::$name) + 1)"; |
| 685 |
|
| 686 |
$condition = $this->selectorAnPlusB($nth, $position); |
| 687 |
$query .= "[$condition]"; |
| 688 |
break; |
| 689 |
|
| 690 |
// https://www.w3.org/TR/selectors-4/#empty-pseudo |
| 691 |
case "empty": |
| 692 |
$query .= "[not(*) and not(normalize-space())]"; |
| 693 |
break; |
| 694 |
|
| 695 |
// TODO: bit of a hack attempt at matches support, currently only matches against elements |
| 696 |
case "matches": |
| 697 |
$p = $i + 1; |
| 698 |
$matchList = trim(mb_substr($selector, $p, mb_strpos($selector, ")", $i, "UTF-8") - $p, "UTF-8")); |
| 699 |
|
| 700 |
// Tag names are case-insensitive |
| 701 |
$elements = array_map("trim", explode(",", strtolower($matchList))); |
| 702 |
foreach ($elements as &$element) { |
| 703 |
$element = "name() = '$element'"; |
| 704 |
} |
| 705 |
|
| 706 |
$query .= "[" . implode(" or ", $elements) . "]"; |
| 707 |
break; |
| 708 |
|
| 709 |
// https://www.w3.org/TR/selectors-3/#UIstates |
| 710 |
case "disabled": |
| 711 |
case "checked": |
| 712 |
$query .= "[@$tok]"; |
| 713 |
break; |
| 714 |
|
| 715 |
case "enabled": |
| 716 |
$query .= "[not(@disabled)]"; |
| 717 |
break; |
| 718 |
|
| 719 |
// https://www.w3.org/TR/selectors-3/#dynamic-pseudos |
| 720 |
// https://www.w3.org/TR/selectors-4/#the-any-link-pseudo |
| 721 |
case "link": |
| 722 |
case "any-link": |
| 723 |
$query .= "[@href]"; |
| 724 |
break; |
| 725 |
|
| 726 |
// N/A |
| 727 |
case "visited": |
| 728 |
case "hover": |
| 729 |
case "active": |
| 730 |
case "focus": |
| 731 |
case "focus-visible": |
| 732 |
case "focus-within": |
| 733 |
$query .= "[false()]"; |
| 734 |
break; |
| 735 |
|
| 736 |
// https://www.w3.org/TR/selectors-3/#first-line |
| 737 |
// https://www.w3.org/TR/selectors-3/#first-letter |
| 738 |
case "first-line": |
| 739 |
case ":first-line": |
| 740 |
case "first-letter": |
| 741 |
case ":first-letter": |
| 742 |
// TODO |
| 743 |
$el = ltrim($tok, ":"); |
| 744 |
$pseudo_elements[$el] = true; |
| 745 |
break; |
| 746 |
|
| 747 |
// https://www.w3.org/TR/selectors-3/#gen-content |
| 748 |
case "before": |
| 749 |
case ":before": |
| 750 |
case "after": |
| 751 |
case ":after": |
| 752 |
$pos = ltrim($tok, ":"); |
| 753 |
$pseudo_elements[$pos] = true; |
| 754 |
if (!$firstPass) { |
| 755 |
$query .= "/*[@$pos]"; |
| 756 |
} |
| 757 |
break; |
| 758 |
|
| 759 |
// Invalid or unsupported pseudo-class or pseudo-element |
| 760 |
default: |
| 761 |
return null; |
| 762 |
} |
| 763 |
|
| 764 |
break; |
| 765 |
|
| 766 |
case "[": |
| 767 |
// Attribute selectors. All with an attribute matching the |
| 768 |
// following token(s) |
| 769 |
// https://www.w3.org/TR/selectors-3/#attribute-selectors |
| 770 |
if ($query === "/") { |
| 771 |
$query .= "/*"; |
| 772 |
} |
| 773 |
|
| 774 |
$attr_delimiters = ["=", "]", "~", "|", "$", "^", "*"]; |
| 775 |
$tok_len = mb_strlen($tok, "UTF-8"); |
| 776 |
$j = 0; |
| 777 |
|
| 778 |
$attr = ""; |
| 779 |
$op = ""; |
| 780 |
$value = ""; |
| 781 |
|
| 782 |
while ($j < $tok_len) { |
| 783 |
if (in_array($tok[$j], $attr_delimiters, true)) { |
| 784 |
break; |
| 785 |
} |
| 786 |
$attr .= $tok[$j++]; |
| 787 |
} |
| 788 |
|
| 789 |
if ($attr === "") { |
| 790 |
// Selector invalid: Missing attribute name |
| 791 |
return null; |
| 792 |
} |
| 793 |
|
| 794 |
if (!isset($tok[$j])) { |
| 795 |
// Selector invalid: Missing ] or operator |
| 796 |
return null; |
| 797 |
} |
| 798 |
|
| 799 |
switch ($tok[$j]) { |
| 800 |
|
| 801 |
case "~": |
| 802 |
case "|": |
| 803 |
case "^": |
| 804 |
case "$": |
| 805 |
case "*": |
| 806 |
$op .= $tok[$j++]; |
| 807 |
|
| 808 |
if (!isset($tok[$j]) || $tok[$j] !== "=") { |
| 809 |
// Selector invalid: Incomplete attribute operator |
| 810 |
return null; |
| 811 |
} |
| 812 |
|
| 813 |
$op .= $tok[$j]; |
| 814 |
break; |
| 815 |
|
| 816 |
case "=": |
| 817 |
$op = "="; |
| 818 |
break; |
| 819 |
|
| 820 |
} |
| 821 |
|
| 822 |
// Read the attribute value, if required |
| 823 |
if ($op !== "") { |
| 824 |
$j++; |
| 825 |
while ($j < $tok_len) { |
| 826 |
if ($tok[$j] === "]") { |
| 827 |
break; |
| 828 |
} |
| 829 |
$value .= $tok[$j++]; |
| 830 |
} |
| 831 |
} |
| 832 |
|
| 833 |
if (!isset($tok[$j])) { |
| 834 |
// Selector invalid: Missing ] |
| 835 |
return null; |
| 836 |
} |
| 837 |
|
| 838 |
$value = trim($value, "\"'"); |
| 839 |
|
| 840 |
switch ($op) { |
| 841 |
|
| 842 |
case "": |
| 843 |
$query .= "[@$attr]"; |
| 844 |
break; |
| 845 |
|
| 846 |
case "=": |
| 847 |
$query .= "[@$attr=\"$value\"]"; |
| 848 |
break; |
| 849 |
|
| 850 |
case "~=": |
| 851 |
// FIXME: this will break if $value contains quoted strings |
| 852 |
// (e.g. [type~="a b c" "d e f"]) |
| 853 |
$query .= $value !== "" && !preg_match("/\s+/", $value) |
| 854 |
? "[contains(concat(' ', normalize-space(@$attr), ' '), concat(' ', \"$value\", ' '))]" |
| 855 |
: "[false()]"; |
| 856 |
break; |
| 857 |
|
| 858 |
case "|=": |
| 859 |
$values = explode("-", $value); |
| 860 |
$query .= "["; |
| 861 |
|
| 862 |
foreach ($values as $val) { |
| 863 |
$query .= "starts-with(@$attr, \"$val\") or "; |
| 864 |
} |
| 865 |
|
| 866 |
$query = rtrim($query, " or ") . "]"; |
| 867 |
break; |
| 868 |
|
| 869 |
case "^=": |
| 870 |
$query .= $value !== "" |
| 871 |
? "[starts-with(@$attr,\"$value\")]" |
| 872 |
: "[false()]"; |
| 873 |
break; |
| 874 |
|
| 875 |
case "$=": |
| 876 |
$query .= $value !== "" |
| 877 |
? "[substring(@$attr, string-length(@$attr)-" . (strlen($value) - 1) . ")=\"$value\"]" |
| 878 |
: "[false()]"; |
| 879 |
break; |
| 880 |
|
| 881 |
case "*=": |
| 882 |
$query .= $value !== "" |
| 883 |
? "[contains(@$attr,\"$value\")]" |
| 884 |
: "[false()]"; |
| 885 |
break; |
| 886 |
} |
| 887 |
|
| 888 |
break; |
| 889 |
} |
| 890 |
} |
| 891 |
|
| 892 |
return ["query" => $query, "pseudo_elements" => $pseudo_elements]; |
| 893 |
} |
| 894 |
|
| 895 |
/** |
| 896 |
* Parse an `nth-child` expression of the form `an+b`, `odd`, or `even`. |
| 897 |
* |
| 898 |
* @param string $expr |
| 899 |
* @param string $position |
| 900 |
* |
| 901 |
* @return string |
| 902 |
* |
| 903 |
* @link https://www.w3.org/TR/selectors-3/#nth-child-pseudo |
| 904 |
*/ |
| 905 |
protected function selectorAnPlusB(string $expr, string $position): string |
| 906 |
{ |
| 907 |
// odd |
| 908 |
if ($expr === "odd") { |
| 909 |
return "($position mod 2) = 1"; |
| 910 |
} // even |
| 911 |
elseif ($expr === "even") { |
| 912 |
return "($position mod 2) = 0"; |
| 913 |
} // b |
| 914 |
elseif (preg_match("/^\d+$/", $expr)) { |
| 915 |
return "$position = $expr"; |
| 916 |
} |
| 917 |
|
| 918 |
// an+b |
| 919 |
// https://github.com/tenderlove/nokogiri/blob/master/lib/nokogiri/css/xpath_visitor.rb |
| 920 |
$expr = preg_replace("/\s/", "", $expr); |
| 921 |
if (!preg_match("/^(?P<a>-?[0-9]*)?n(?P<b>[-+]?[0-9]+)?$/", $expr, $matches)) { |
| 922 |
return "false()"; |
| 923 |
} |
| 924 |
|
| 925 |
$a = (isset($matches["a"]) && $matches["a"] !== "") ? ($matches["a"] !== "-" ? intval($matches["a"]) : -1) : 1; |
| 926 |
$b = (isset($matches["b"]) && $matches["b"] !== "") ? intval($matches["b"]) : 0; |
| 927 |
|
| 928 |
if ($b === 0) { |
| 929 |
return "($position mod $a) = 0"; |
| 930 |
} else { |
| 931 |
$compare = ($a < 0) ? "<=" : ">="; |
| 932 |
$b2 = -$b; |
| 933 |
if ($b2 >= 0) { |
| 934 |
$b2 = "+$b2"; |
| 935 |
} |
| 936 |
return "($position $compare $b) and ((($position $b2) mod " . abs($a) . ") = 0)"; |
| 937 |
} |
| 938 |
} |
| 939 |
|
| 940 |
/** |
| 941 |
* applies all current styles to a particular document tree |
| 942 |
* |
| 943 |
* apply_styles() applies all currently loaded styles to the provided |
| 944 |
* {@link FrameTree}. Aside from parsing CSS, this is the main purpose |
| 945 |
* of this class. |
| 946 |
* |
| 947 |
* @param FrameTree $tree |
| 948 |
*/ |
| 949 |
function apply_styles(FrameTree $tree) |
| 950 |
{ |
| 951 |
// Use XPath to select nodes. This would be easier if we could attach |
| 952 |
// Frame objects directly to DOMNodes using the setUserData() method, but |
| 953 |
// we can't do that just yet. Instead, we set a _node attribute_ in |
| 954 |
// Frame->set_id() and use that as a handle on the Frame object via |
| 955 |
// FrameTree::$_registry. |
| 956 |
|
| 957 |
// We create a scratch array of styles indexed by frame id. Once all |
| 958 |
// styles have been assigned, we order the cached styles by specificity |
| 959 |
// and create a final style object to assign to the frame. |
| 960 |
|
| 961 |
// FIXME: this is not particularly robust... |
| 962 |
|
| 963 |
$styles = []; |
| 964 |
$xp = new DOMXPath($tree->get_dom()); |
| 965 |
$DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss(); |
| 966 |
|
| 967 |
// Add generated content |
| 968 |
foreach ($this->_styles as $selector => $selector_styles) { |
| 969 |
if (strpos($selector, ":before") === false && strpos($selector, ":after") === false) { |
| 970 |
continue; |
| 971 |
} |
| 972 |
|
| 973 |
$query = $this->selectorToXpath($selector, true); |
| 974 |
if ($query === null) { |
| 975 |
Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__); |
| 976 |
continue; |
| 977 |
} |
| 978 |
|
| 979 |
// Retrieve the nodes, limit to body for generated content |
| 980 |
// TODO: If we use a context node can we remove the leading dot? |
| 981 |
$nodes = @$xp->query('.' . $query["query"]); |
| 982 |
if ($nodes === false) { |
| 983 |
Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__); |
| 984 |
continue; |
| 985 |
} |
| 986 |
|
| 987 |
foreach ($selector_styles as $style) { |
| 988 |
foreach ($nodes as $node) { |
| 989 |
// Only DOMElements get styles |
| 990 |
if (!($node instanceof DOMElement)) { |
| 991 |
continue; |
| 992 |
} |
| 993 |
|
| 994 |
foreach (array_keys($query["pseudo_elements"], true, true) as $pos) { |
| 995 |
// Do not add a new pseudo element if another one already matched |
| 996 |
if ($node->hasAttribute("dompdf_{$pos}_frame_id")) { |
| 997 |
continue; |
| 998 |
} |
| 999 |
|
| 1000 |
$content = $style->content; |
| 1001 |
|
| 1002 |
// Do not create non-displayed before/after pseudo |
| 1003 |
// elements. Since styles have not been inherited yet, |
| 1004 |
// a specified value of `inherit` will always be treated |
| 1005 |
// as `normal` here. This is fine according to the |
| 1006 |
// CSS 2.1 spec, as any value computes to `normal` on |
| 1007 |
// regular elements |
| 1008 |
// https://www.w3.org/TR/CSS21/generate.html#content |
| 1009 |
// https://www.w3.org/TR/CSS21/generate.html#undisplayed-counters |
| 1010 |
if ($content === "normal" || $content === "none") { |
| 1011 |
$specified = $style->get_specified("content"); |
| 1012 |
if (!\preg_match("/". self::PATTERN_CSS_VAR_FN . "/", $specified)) { |
| 1013 |
continue; |
| 1014 |
} else { |
| 1015 |
$content = []; |
| 1016 |
} |
| 1017 |
} |
| 1018 |
|
| 1019 |
// https://www.w3.org/TR/css-content-3/#content-property |
| 1020 |
$single = count($content) === 1 ? $content[0] : null; |
| 1021 |
|
| 1022 |
if ($single instanceof Url) { |
| 1023 |
$src = $this->resolve_url("url($single->url)", true); |
| 1024 |
$new_node = $node->ownerDocument->createElement("img_generated"); |
| 1025 |
$new_node->setAttribute("src", $src); |
| 1026 |
} else { |
| 1027 |
$new_node = $node->ownerDocument->createElement("dompdf_generated"); |
| 1028 |
} |
| 1029 |
|
| 1030 |
$new_node->setAttribute($pos, $pos); |
| 1031 |
$new_frame_id = $tree->insert_node($node, $new_node, $pos); |
| 1032 |
$node->setAttribute("dompdf_{$pos}_frame_id", $new_frame_id); |
| 1033 |
} |
| 1034 |
} |
| 1035 |
} |
| 1036 |
} |
| 1037 |
|
| 1038 |
// Apply all styles in stylesheet |
| 1039 |
foreach ($this->_styles as $selector => $selector_styles) { |
| 1040 |
$query = $this->selectorToXpath($selector); |
| 1041 |
if ($query === null) { |
| 1042 |
Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__); |
| 1043 |
continue; |
| 1044 |
} |
| 1045 |
|
| 1046 |
// Retrieve the nodes |
| 1047 |
$nodes = @$xp->query($query["query"]); |
| 1048 |
if ($nodes === false) { |
| 1049 |
Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__); |
| 1050 |
continue; |
| 1051 |
} |
| 1052 |
|
| 1053 |
foreach ($selector_styles as $style) { |
| 1054 |
$spec = $this->specificity($selector, $style->get_origin()); |
| 1055 |
|
| 1056 |
foreach ($nodes as $node) { |
| 1057 |
// Only DOMElements get styles |
| 1058 |
if (!($node instanceof DOMElement)) { |
| 1059 |
continue; |
| 1060 |
} |
| 1061 |
|
| 1062 |
$id = $node->getAttribute("frame_id"); |
| 1063 |
|
| 1064 |
// Assign the current style to the scratch array |
| 1065 |
$styles[$id][$spec][] = $style; |
| 1066 |
} |
| 1067 |
} |
| 1068 |
} |
| 1069 |
|
| 1070 |
// Set the page width, height, and orientation based on the canvas paper size |
| 1071 |
$canvas = $this->_dompdf->getCanvas(); |
| 1072 |
$paper_width = $canvas->get_width(); |
| 1073 |
$paper_height = $canvas->get_height(); |
| 1074 |
$paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait"); |
| 1075 |
|
| 1076 |
if ($this->_page_styles["base"] && is_array($this->_page_styles["base"]->size)) { |
| 1077 |
$paper_width = $this->_page_styles['base']->size[0]; |
| 1078 |
$paper_height = $this->_page_styles['base']->size[1]; |
| 1079 |
$paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait"); |
| 1080 |
} |
| 1081 |
|
| 1082 |
// Now create the styles and assign them to the appropriate frames. (We |
| 1083 |
// iterate over the tree using an implicit FrameTree iterator.) |
| 1084 |
$root_flg = false; |
| 1085 |
foreach ($tree as $frame) { |
| 1086 |
// Helpers::pre_r($frame->get_node()->nodeName . ":"); |
| 1087 |
if (!$root_flg && $this->_page_styles["base"]) { |
| 1088 |
$style = $this->_page_styles["base"]; |
| 1089 |
} else { |
| 1090 |
$style = $this->create_style(); |
| 1091 |
} |
| 1092 |
|
| 1093 |
// Find nearest DOMElement parent |
| 1094 |
$p = $frame; |
| 1095 |
while ($p = $p->get_parent()) { |
| 1096 |
if ($p->get_node()->nodeType === XML_ELEMENT_NODE) { |
| 1097 |
break; |
| 1098 |
} |
| 1099 |
} |
| 1100 |
|
| 1101 |
// Styles can only be applied directly to DOMElements; anonymous |
| 1102 |
// frames inherit from their parent |
| 1103 |
if ($frame->get_node()->nodeType !== XML_ELEMENT_NODE) { |
| 1104 |
$style->inherit($p ? $p->get_style() : null); |
| 1105 |
$frame->set_style($style); |
| 1106 |
continue; |
| 1107 |
} |
| 1108 |
|
| 1109 |
$id = $frame->get_id(); |
| 1110 |
|
| 1111 |
// Handle HTML 4.0 attributes |
| 1112 |
AttributeTranslator::translate_attributes($frame); |
| 1113 |
if (($str = $frame->get_node()->getAttribute(AttributeTranslator::$_style_attr)) !== "") { |
| 1114 |
$styles[$id][self::SPEC_NON_CSS][] = $this->_parse_properties($str); |
| 1115 |
} |
| 1116 |
|
| 1117 |
// Locate any additional style attributes |
| 1118 |
if (($str = $frame->get_node()->getAttribute("style")) !== "") { |
| 1119 |
// Destroy CSS comments |
| 1120 |
$str = preg_replace("'/\*.*?\*/'si", "", $str); |
| 1121 |
|
| 1122 |
$spec = $this->specificity("!attr", self::ORIG_AUTHOR); |
| 1123 |
$styles[$id][$spec][] = $this->_parse_properties($str); |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Grab the applicable styles |
| 1127 |
if (isset($styles[$id])) { |
| 1128 |
|
| 1129 |
/** @var array[][] $applied_styles */ |
| 1130 |
$applied_styles = $styles[$id]; |
| 1131 |
|
| 1132 |
// Sort by specificity |
| 1133 |
ksort($applied_styles); |
| 1134 |
|
| 1135 |
if ($DEBUGCSS) { |
| 1136 |
$debug_nodename = $frame->get_node()->nodeName; |
| 1137 |
print "<pre>\n$debug_nodename [\n"; |
| 1138 |
foreach ($applied_styles as $spec => $arr) { |
| 1139 |
printf(" specificity 0x%08x\n", $spec); |
| 1140 |
/** @var Style $s */ |
| 1141 |
foreach ($arr as $s) { |
| 1142 |
print " [\n"; |
| 1143 |
$s->debug_print(); |
| 1144 |
print " ]\n"; |
| 1145 |
} |
| 1146 |
} |
| 1147 |
} |
| 1148 |
|
| 1149 |
// Merge the new styles with the inherited styles |
| 1150 |
$acceptedmedia = self::$ACCEPTED_GENERIC_MEDIA_TYPES; |
| 1151 |
$acceptedmedia[] = $this->_dompdf->getOptions()->getDefaultMediaType(); |
| 1152 |
|
| 1153 |
foreach ($applied_styles as $arr) { |
| 1154 |
/** @var Style $s */ |
| 1155 |
foreach ($arr as $s) { |
| 1156 |
$media_queries = $s->get_media_queries(); |
| 1157 |
if (count($media_queries) > 0) { |
| 1158 |
$media_query_match = false; |
| 1159 |
foreach ($media_queries as $media_query_group) { |
| 1160 |
foreach ($media_query_group as $media_query) { |
| 1161 |
list($media_query_feature, $media_query_value, $media_query_operator) = $media_query; |
| 1162 |
switch ($media_query_feature) { |
| 1163 |
case "height": |
| 1164 |
$feature_match = $paper_height === (float)$style->length_in_pt($media_query_value); |
| 1165 |
break; |
| 1166 |
case "min-height": |
| 1167 |
$feature_match = $paper_height >= (float)$style->length_in_pt($media_query_value); |
| 1168 |
break; |
| 1169 |
case "max-height": |
| 1170 |
$feature_match = $paper_height <= (float)$style->length_in_pt($media_query_value); |
| 1171 |
break; |
| 1172 |
case "width": |
| 1173 |
$feature_match = $paper_width === (float)$style->length_in_pt($media_query_value); |
| 1174 |
break; |
| 1175 |
case "min-width": |
| 1176 |
$feature_match = $paper_width >= (float)$style->length_in_pt($media_query_value); |
| 1177 |
break; |
| 1178 |
case "max-width": |
| 1179 |
$feature_match = $paper_width <= (float)$style->length_in_pt($media_query_value); |
| 1180 |
break; |
| 1181 |
case "orientation": |
| 1182 |
$feature_match = $paper_orientation === $media_query_value; |
| 1183 |
break; |
| 1184 |
case "type": |
| 1185 |
$feature_match = in_array($media_query_value, $acceptedmedia, true); |
| 1186 |
break; |
| 1187 |
default: |
| 1188 |
Helpers::record_warnings(E_USER_WARNING, "Unknown media query: $media_query_feature", __FILE__, __LINE__); |
| 1189 |
continue (2); // unknown query, move to the next grouping |
| 1190 |
} |
| 1191 |
$negate = $media_query_operator === "not"; |
| 1192 |
if ($negate xor !$feature_match) { |
| 1193 |
continue (2); // failed query match, move to the next grouping |
| 1194 |
} |
| 1195 |
} |
| 1196 |
$media_query_match = true; |
| 1197 |
} |
| 1198 |
if (!$media_query_match) { |
| 1199 |
continue; |
| 1200 |
} |
| 1201 |
} |
| 1202 |
$style->merge($s); |
| 1203 |
} |
| 1204 |
} |
| 1205 |
} |
| 1206 |
|
| 1207 |
// Handle inheritance |
| 1208 |
if ($p && $DEBUGCSS) { |
| 1209 |
print " inherit [\n"; |
| 1210 |
$p->get_style()->debug_print(); |
| 1211 |
print " ]\n"; |
| 1212 |
} |
| 1213 |
|
| 1214 |
$style->inherit($p ? $p->get_style() : null); |
| 1215 |
|
| 1216 |
if ($DEBUGCSS) { |
| 1217 |
print " DomElementStyle [\n"; |
| 1218 |
$style->debug_print(); |
| 1219 |
print " ]\n"; |
| 1220 |
print "]\n</pre>"; |
| 1221 |
} |
| 1222 |
|
| 1223 |
$style->clear_important(); |
| 1224 |
$frame->set_style($style); |
| 1225 |
|
| 1226 |
if (!$root_flg && $this->_page_styles["base"]) { |
| 1227 |
$root_flg = true; |
| 1228 |
|
| 1229 |
// set the page width, height, and orientation based on the parsed page style |
| 1230 |
if ($style->size !== "auto") { |
| 1231 |
list($paper_width, $paper_height) = $style->size; |
| 1232 |
} |
| 1233 |
$paper_width = $paper_width - (float)$style->length_in_pt($style->margin_left) - (float)$style->length_in_pt($style->margin_right); |
| 1234 |
$paper_height = $paper_height - (float)$style->length_in_pt($style->margin_top) - (float)$style->length_in_pt($style->margin_bottom); |
| 1235 |
$paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait"); |
| 1236 |
} |
| 1237 |
} |
| 1238 |
|
| 1239 |
// We're done! Clean out the registry of all styles since we |
| 1240 |
// won't be needing this later. |
| 1241 |
foreach (array_keys($this->_styles) as $key) { |
| 1242 |
$this->_styles[$key] = null; |
| 1243 |
unset($this->_styles[$key]); |
| 1244 |
} |
| 1245 |
} |
| 1246 |
|
| 1247 |
/** |
| 1248 |
* parse a CSS string using a regex parser |
| 1249 |
* Called by {@link Stylesheet::parse_css()} |
| 1250 |
* |
| 1251 |
* @param string $str |
| 1252 |
*/ |
| 1253 |
private function _parse_css($str, $media_queries = []) |
| 1254 |
{ |
| 1255 |
$str = trim($str); |
| 1256 |
|
| 1257 |
// Destroy comments and remove HTML comments |
| 1258 |
$css = preg_replace([ |
| 1259 |
"'/\*.*?\*/'si", |
| 1260 |
"/^<!--/", |
| 1261 |
"/-->$/" |
| 1262 |
], "", $str); |
| 1263 |
|
| 1264 |
// shim constants for string interpolation |
| 1265 |
$pattern_atimport_string = str_replace("CSS_STRING", "CSS_ATIMPORT_STRING", self::PATTERN_CSS_STRING); |
| 1266 |
$pattern_atimport_url = str_replace("CSS_URL_FN", "CSS_ATIMPORT_URL_FN", self::PATTERN_CSS_URL_FN); |
| 1267 |
$pattern_media_query = self::PATTERN_MEDIA_QUERY; |
| 1268 |
|
| 1269 |
// Something more legible: |
| 1270 |
// ... does not handle '{' within strings, e.g. [attr="string {}"] |
| 1271 |
$re = <<<EOL |
| 1272 |
/ |
| 1273 |
# Skip leading whitespace |
| 1274 |
\s* |
| 1275 |
|
| 1276 |
# Match at-rules |
| 1277 |
(?<CSS_ATRULE>@(?<CSS_ATRULE_IDENTIFIER> |
| 1278 |
(?<CSS_ATFONT>font-face) |
| 1279 |
|(?<CSS_ATIMPORT>import) |
| 1280 |
|(?<CSS_ATMEDIA>media) |
| 1281 |
|(?<CSS_ATPAGE>page) |
| 1282 |
|(?<CSS_AT>[\w-]*) |
| 1283 |
))? |
| 1284 |
|
| 1285 |
# Branch to process segment following at-rule match |
| 1286 |
(?(CSS_ATRULE)(?: |
| 1287 |
(?(CSS_ATFONT)\s*{(?<CSS_ATFONT_BODY>.*?)}) |
| 1288 |
(?(CSS_ATIMPORT)\s*(?<CSS_ATIMPORT_RULE> |
| 1289 |
(?<CSS_ATIMPORT_URL> |
| 1290 |
{$pattern_atimport_string} |
| 1291 |
|{$pattern_atimport_url} |
| 1292 |
) |
| 1293 |
(?<CSS_ATIMPORT_MEDIA_QUERY>.*?) |
| 1294 |
);) |
| 1295 |
(?(CSS_ATMEDIA)\s*(?<CSS_ATMEDIA_RULE>[^{]*){(?<CSS_ATMEDIA_BODY> (?:[^{}]*|(?R) \s*)+? )}) |
| 1296 |
(?(CSS_ATPAGE)\s*(?<CSS_ATPAGE_RULE>[^{]*){(?<CSS_ATPAGE_BODY>.*?)}) |
| 1297 |
(?(CSS_AT)\s*([^{;]*)(;|{(?<CSS_AT_BODY> (?:[^{}]*|(?R) \s*)+? )})) |
| 1298 |
) |
| 1299 |
|
| 1300 |
# Branch to match regular rules (not preceded by '@') |
| 1301 |
|(?<CSS_RULESET>[^{]*{[^}]*})) |
| 1302 |
/isx |
| 1303 |
EOL; |
| 1304 |
|
| 1305 |
// replace data URIs with blob URIs |
| 1306 |
$offset = 0; |
| 1307 |
while ($offset < strlen($css) && ($start = strpos($css, "data:", $offset)) !== false) { |
| 1308 |
$len = 0; |
| 1309 |
$prev = $css[$start - 1]; |
| 1310 |
$pattern = "/(?<!\\\\)['\")]/"; |
| 1311 |
switch ($prev) { |
| 1312 |
case "'": |
| 1313 |
$pattern = "/(?<!\\\\)'/"; |
| 1314 |
break; |
| 1315 |
case "\"": |
| 1316 |
$pattern = "/(?<!\\\\)\"/"; |
| 1317 |
break; |
| 1318 |
case "(": |
| 1319 |
$pattern = "/(?<!\\\\)\\)/"; |
| 1320 |
break; |
| 1321 |
default: |
| 1322 |
$offset = $start + 5; |
| 1323 |
continue (2); |
| 1324 |
} |
| 1325 |
if (preg_match($pattern, $css, $matches, PREG_OFFSET_CAPTURE, $start)) { |
| 1326 |
$len = $matches[0][1] - $start; |
| 1327 |
} |
| 1328 |
if ($len <= 0) { |
| 1329 |
$offset = $start + 5; |
| 1330 |
continue; |
| 1331 |
} |
| 1332 |
$data_uri = substr($css, $start, $len); |
| 1333 |
$data_uri_hash = md5($data_uri); |
| 1334 |
$this->_blobs[$data_uri_hash] = $data_uri; |
| 1335 |
$css = substr($css, 0, $start) . "blob://" . $data_uri_hash . ($len > 0 ? substr($css, $start + $len) : ""); |
| 1336 |
$offset = $start + 7 + strlen($data_uri_hash); |
| 1337 |
} |
| 1338 |
|
| 1339 |
$matches = []; |
| 1340 |
if (preg_match_all($re, $css, $matches, PREG_SET_ORDER) === false || count($matches) === 0) { |
| 1341 |
global $_dompdf_warnings; |
| 1342 |
$_dompdf_warnings[] = "Unable to parse CSS that starts with: " . substr($str, 0, 100); |
| 1343 |
return; |
| 1344 |
} |
| 1345 |
|
| 1346 |
$media_query_regex = "/{$pattern_media_query}/isx"; |
| 1347 |
$accepted_media = self::$ACCEPTED_GENERIC_MEDIA_TYPES; |
| 1348 |
$accepted_media[] = $this->_dompdf->getOptions()->getDefaultMediaType(); |
| 1349 |
foreach ($matches as $match) { |
| 1350 |
if ($match["CSS_ATRULE_IDENTIFIER"] !== "") { |
| 1351 |
$atrule_identifier = strtolower($match["CSS_ATRULE_IDENTIFIER"]); |
| 1352 |
// Handle @rules |
| 1353 |
switch ($atrule_identifier) { |
| 1354 |
|
| 1355 |
case "import": |
| 1356 |
$this->_parse_import($match["CSS_ATIMPORT_URL"], $match["CSS_ATIMPORT_MEDIA_QUERY"]); |
| 1357 |
break; |
| 1358 |
|
| 1359 |
case "media": |
| 1360 |
$parsed_media_queries = []; |
| 1361 |
$media_query_rules = preg_split("/\s*(,|\Wor\W)\s*/", mb_strtolower(trim($match["CSS_ATMEDIA_RULE"]))); |
| 1362 |
foreach ($media_query_rules as $media_query_rule) { |
| 1363 |
$media_query_matches = []; |
| 1364 |
if (preg_match_all($media_query_regex, $media_query_rule, $media_query_matches, PREG_SET_ORDER) === false) { |
| 1365 |
continue; |
| 1366 |
} |
| 1367 |
|
| 1368 |
$mq_grouping = []; |
| 1369 |
foreach ($media_query_matches as $media_query_match) { |
| 1370 |
if (empty($media_query_match["CSS_MEDIA_QUERY_TYPE"]) === false) { |
| 1371 |
$media_query_feature = "type"; |
| 1372 |
$media_query_value = strtolower($media_query_match["CSS_MEDIA_QUERY_TYPE"]); |
| 1373 |
$media_query_operator = strtolower($media_query_match["CSS_MEDIA_QUERY_OP"]); |
| 1374 |
} elseif (empty($media_query_match["CSS_MEDIA_QUERY_FEATURE"]) === false) { |
| 1375 |
$media_query_feature = strtolower($media_query_match["CSS_MEDIA_QUERY_FEATURE"]); |
| 1376 |
$media_query_value = (array_key_exists("CSS_MEDIA_QUERY_CONDITION", $media_query_match) ? strtolower($media_query_match["CSS_MEDIA_QUERY_CONDITION"]) : null); |
| 1377 |
$media_query_operator = strtolower($media_query_match["CSS_MEDIA_QUERY_OP"]); |
| 1378 |
} else { |
| 1379 |
// partial error handling implementation per https://www.w3.org/TR/css3-mediaqueries/#error-handling |
| 1380 |
$media_query_feature = "type"; |
| 1381 |
$media_query_value = "all"; |
| 1382 |
$media_query_operator = "not"; |
| 1383 |
} |
| 1384 |
$mq_grouping[] = [$media_query_feature, $media_query_value, $media_query_operator]; |
| 1385 |
} |
| 1386 |
if (count($mq_grouping) > 0) { |
| 1387 |
$parsed_media_queries[] = $mq_grouping; |
| 1388 |
} |
| 1389 |
} |
| 1390 |
$sub_media_queries = []; |
| 1391 |
if (count($media_queries) > 0) { |
| 1392 |
foreach ($media_queries as $media_query) { |
| 1393 |
foreach ($parsed_media_queries as $parsed_media_query) { |
| 1394 |
$sub_media_queries[] = array_merge($media_query, $parsed_media_query); |
| 1395 |
} |
| 1396 |
} |
| 1397 |
} else { |
| 1398 |
$sub_media_queries = $parsed_media_queries; |
| 1399 |
} |
| 1400 |
$this->_parse_css($match["CSS_ATMEDIA_BODY"], $sub_media_queries); |
| 1401 |
break; |
| 1402 |
|
| 1403 |
case "page": |
| 1404 |
//This handles @page to be applied to page oriented media |
| 1405 |
//Note: This has a reduced syntax: |
| 1406 |
//@page { margin:1cm; color:blue; } |
| 1407 |
//Not a sequence of styles like a full.css, but only the properties |
| 1408 |
//of a single style, which is applied to the very first "root" frame before |
| 1409 |
//processing other styles of the frame. |
| 1410 |
//Working properties: |
| 1411 |
// margin (for margin around edge of paper) |
| 1412 |
// font-family (default font of pages) |
| 1413 |
// color (default text color of pages) |
| 1414 |
//Non working properties: |
| 1415 |
// border |
| 1416 |
// padding |
| 1417 |
// background-color |
| 1418 |
//Todo:Reason is unknown |
| 1419 |
//Other properties (like further font or border attributes) not tested. |
| 1420 |
//If a border or background color around each paper sheet is desired, |
| 1421 |
//assign it to the <body> tag, possibly only for the css of the correct media type. |
| 1422 |
|
| 1423 |
// If the page has a name, skip the style. |
| 1424 |
$page_selector = trim($match["CSS_ATPAGE_RULE"]); |
| 1425 |
|
| 1426 |
$key = null; |
| 1427 |
switch ($page_selector) { |
| 1428 |
case "": |
| 1429 |
$key = "base"; |
| 1430 |
break; |
| 1431 |
|
| 1432 |
case ":left": |
| 1433 |
case ":right": |
| 1434 |
case ":odd": |
| 1435 |
case ":even": |
| 1436 |
/** @noinspection PhpMissingBreakStatementInspection */ |
| 1437 |
case ":first": |
| 1438 |
$key = $page_selector; |
| 1439 |
break; |
| 1440 |
|
| 1441 |
default: |
| 1442 |
break 2; |
| 1443 |
} |
| 1444 |
|
| 1445 |
// Store the style for later... |
| 1446 |
if (empty($this->_page_styles[$key])) { |
| 1447 |
$this->_page_styles[$key] = $this->_parse_properties($match["CSS_ATPAGE_BODY"]); |
| 1448 |
} else { |
| 1449 |
$this->_page_styles[$key]->merge($this->_parse_properties($match["CSS_ATPAGE_BODY"])); |
| 1450 |
} |
| 1451 |
break; |
| 1452 |
|
| 1453 |
case "font-face": |
| 1454 |
$this->_parse_font_face($match["CSS_ATFONT_BODY"]); |
| 1455 |
break; |
| 1456 |
|
| 1457 |
default: |
| 1458 |
// ignore everything else |
| 1459 |
break; |
| 1460 |
} |
| 1461 |
|
| 1462 |
continue; |
| 1463 |
} |
| 1464 |
|
| 1465 |
if ($match["CSS_RULESET"] !== "") { |
| 1466 |
$this->_parse_sections($match["CSS_RULESET"], $media_queries); |
| 1467 |
} |
| 1468 |
} |
| 1469 |
} |
| 1470 |
|
| 1471 |
/** |
| 1472 |
* Resolve the given `url()` declaration to an absolute URL. |
| 1473 |
* |
| 1474 |
* @param string|null $val The declaration to resolve in the context of the stylesheet. |
| 1475 |
* @param bool|null $resolve_blobs Indicates whether or not to resolve blob URLs to their final value. |
| 1476 |
* @return string The resolved URL, or `none`, if the value is `none`, |
| 1477 |
* invalid, or points to a non-existent local file. |
| 1478 |
*/ |
| 1479 |
public function resolve_url($val, $resolve_blobs = false): string |
| 1480 |
{ |
| 1481 |
$DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss(); |
| 1482 |
|
| 1483 |
static $pattern = "/" . self::PATTERN_CSS_URL_FN . "/isx"; |
| 1484 |
if ($val === null || $val === "" || strcasecmp($val, "none") === 0) { |
| 1485 |
$path = "none"; |
| 1486 |
} elseif (preg_match($pattern, $val, $matches)) { |
| 1487 |
// Resolve the url now in the context of the current stylesheet |
| 1488 |
$url = $matches["CSS_URL_FN_VALUE"]; |
| 1489 |
switch ($matches["CSS_URL_FN_QUOTE"]) { |
| 1490 |
case "\"": |
| 1491 |
$url = str_replace("\\\"", "\"", $url); |
| 1492 |
break; |
| 1493 |
case "'": |
| 1494 |
$url = str_replace("\\'", "'", $url); |
| 1495 |
break; |
| 1496 |
default: |
| 1497 |
$url = str_replace(["\\(", "\\)"], ["(", ")"], $url); |
| 1498 |
break; |
| 1499 |
} |
| 1500 |
if ($resolve_blobs === true && strpos($url, "blob://") !== false) { |
| 1501 |
$url = $this->_blobs[substr($url, 7)] ?? ""; |
| 1502 |
} |
| 1503 |
$path = Helpers::build_url( |
| 1504 |
$this->_protocol, |
| 1505 |
$this->_base_host, |
| 1506 |
$this->_base_path, |
| 1507 |
$url, |
| 1508 |
$this->_dompdf->getOptions()->getChroot() |
| 1509 |
); |
| 1510 |
|
| 1511 |
if ($path !== null && strpos($path, "blob:") !== 0) { |
| 1512 |
[$protocol] = Helpers::explode_url($path); |
| 1513 |
$options = $this->_dompdf->getOptions(); |
| 1514 |
$allowed_protocols = $options->getAllowedProtocols(); |
| 1515 |
if (!array_key_exists($protocol, $allowed_protocols)) { |
| 1516 |
Helpers::record_warnings(E_USER_WARNING, "Permission denied on $path. The communication protocol is not supported.", __FILE__, __LINE__); |
| 1517 |
$path = null; |
| 1518 |
} |
| 1519 |
|
| 1520 |
foreach ($allowed_protocols[$protocol]["rules"] as $rule) { |
| 1521 |
[$result, $message] = $rule($path); |
| 1522 |
if ($result !== true) { |
| 1523 |
Helpers::record_warnings(E_USER_WARNING, "Error loading $path: $message", __FILE__, __LINE__); |
| 1524 |
$path = null; |
| 1525 |
} |
| 1526 |
} |
| 1527 |
} |
| 1528 |
|
| 1529 |
if ($path === null) { |
| 1530 |
$path = "none"; |
| 1531 |
} |
| 1532 |
} else { |
| 1533 |
$path = "none"; |
| 1534 |
} |
| 1535 |
if ($DEBUGCSS) { |
| 1536 |
$parsed_url = Helpers::explode_url($path); |
| 1537 |
print "<pre>[_image\n"; |
| 1538 |
print_r($parsed_url); |
| 1539 |
print $this->_protocol . "\n" . $this->_base_path . "\n" . $path . "\n"; |
| 1540 |
print "_image]</pre>"; |
| 1541 |
} |
| 1542 |
return $path; |
| 1543 |
} |
| 1544 |
|
| 1545 |
/** |
| 1546 |
* parse @import at-rule |
| 1547 |
* |
| 1548 |
* @param string $url the url of the imported CSS file |
| 1549 |
*/ |
| 1550 |
private function _parse_import($url, $import_media_query) |
| 1551 |
{ |
| 1552 |
// if URL is a CSS string, wrap it in the url function for parsing by the resolve_url method |
| 1553 |
if (mb_strpos($url, "url(") === false) { |
| 1554 |
$url = "url($url)"; |
| 1555 |
} |
| 1556 |
if (($url = $this->resolve_url($url, true)) === "none") { |
| 1557 |
return; |
| 1558 |
} |
| 1559 |
|
| 1560 |
// Store our current base url properties in case the new url is elsewhere |
| 1561 |
$protocol = $this->_protocol; |
| 1562 |
$host = $this->_base_host; |
| 1563 |
$path = $this->_base_path; |
| 1564 |
|
| 1565 |
$media_query_regex = "/" . self::PATTERN_MEDIA_QUERY . "/isx"; |
| 1566 |
$media_queries = preg_split("/\s*(,|\Wor\W)\s*/", mb_strtolower(trim($import_media_query ?? ""), "UTF-8")); |
| 1567 |
if (count($media_queries) === 0) { |
| 1568 |
$this->load_css_file($url, $this->_current_origin); |
| 1569 |
} else { |
| 1570 |
// Set the page width, height, and orientation based on the canvas paper size |
| 1571 |
$canvas = $this->_dompdf->getCanvas(); |
| 1572 |
$paper_width = $canvas->get_width(); |
| 1573 |
$paper_height = $canvas->get_height(); |
| 1574 |
$paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait"); |
| 1575 |
|
| 1576 |
$style = $this->_page_styles["base"] ?? new Style($this); |
| 1577 |
if (is_array($style->size)) { |
| 1578 |
$paper_width = $style->size[0]; |
| 1579 |
$paper_height = $style->size[1]; |
| 1580 |
$paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait"); |
| 1581 |
} |
| 1582 |
|
| 1583 |
$acceptedmedia = self::$ACCEPTED_GENERIC_MEDIA_TYPES; |
| 1584 |
$acceptedmedia[] = $this->_dompdf->getOptions()->getDefaultMediaType(); |
| 1585 |
|
| 1586 |
foreach ($media_queries as $media_query) { |
| 1587 |
$media_query_matches = []; |
| 1588 |
if (preg_match_all($media_query_regex, $media_query, $media_query_matches, PREG_SET_ORDER) === false) { |
| 1589 |
continue; |
| 1590 |
} |
| 1591 |
|
| 1592 |
foreach ($media_query_matches as $media_query_match) { |
| 1593 |
if (empty($media_query_match["CSS_MEDIA_QUERY_TYPE"]) === false) { |
| 1594 |
$media_query_feature = "type"; |
| 1595 |
$media_query_value = strtolower($media_query_match["CSS_MEDIA_QUERY_TYPE"]); |
| 1596 |
$media_query_operator = strtolower($media_query_match["CSS_MEDIA_QUERY_OP"]); |
| 1597 |
} elseif (empty($media_query_match["CSS_MEDIA_QUERY_FEATURE"]) === false) { |
| 1598 |
$media_query_feature = strtolower($media_query_match["CSS_MEDIA_QUERY_FEATURE"]); |
| 1599 |
$media_query_value = (array_key_exists("CSS_MEDIA_QUERY_CONDITION", $media_query_match) ? strtolower($media_query_match["CSS_MEDIA_QUERY_CONDITION"]) : null); |
| 1600 |
$media_query_operator = strtolower($media_query_match["CSS_MEDIA_QUERY_OP"]); |
| 1601 |
} else { |
| 1602 |
// partial error handling implementation per https://www.w3.org/TR/css3-mediaqueries/#error-handling |
| 1603 |
$media_query_feature = "type"; |
| 1604 |
$media_query_value = "all"; |
| 1605 |
$media_query_operator = "not"; |
| 1606 |
} |
| 1607 |
|
| 1608 |
switch ($media_query_feature) { |
| 1609 |
case "height": |
| 1610 |
$feature_match = $paper_height === (float)$style->length_in_pt($media_query_value); |
| 1611 |
break; |
| 1612 |
case "min-height": |
| 1613 |
$feature_match = $paper_height >= (float)$style->length_in_pt($media_query_value); |
| 1614 |
break; |
| 1615 |
case "max-height": |
| 1616 |
$feature_match = $paper_height <= (float)$style->length_in_pt($media_query_value); |
| 1617 |
break; |
| 1618 |
case "width": |
| 1619 |
$feature_match = $paper_width === (float)$style->length_in_pt($media_query_value); |
| 1620 |
break; |
| 1621 |
case "min-width": |
| 1622 |
$feature_match = $paper_width >= (float)$style->length_in_pt($media_query_value); |
| 1623 |
break; |
| 1624 |
case "max-width": |
| 1625 |
$feature_match = $paper_width <= (float)$style->length_in_pt($media_query_value); |
| 1626 |
break; |
| 1627 |
case "orientation": |
| 1628 |
$feature_match = $paper_orientation === $media_query_value; |
| 1629 |
break; |
| 1630 |
case "type": |
| 1631 |
$feature_match = in_array($media_query_value, $acceptedmedia, true); |
| 1632 |
break; |
| 1633 |
default: |
| 1634 |
Helpers::record_warnings(E_USER_WARNING, "Unknown media query: $media_query_feature", __FILE__, __LINE__); |
| 1635 |
continue (2); |
| 1636 |
} |
| 1637 |
$negate = $media_query_operator === "not"; |
| 1638 |
if ($negate xor !$feature_match) { |
| 1639 |
continue (2); |
| 1640 |
} |
| 1641 |
} |
| 1642 |
|
| 1643 |
//TODO: pass media queries as an argument to load_css_file and apply to all contained styles |
| 1644 |
// to better accommodate styling content in, for example, documents with varying page orientations |
| 1645 |
$this->load_css_file($url, $this->_current_origin); |
| 1646 |
break; // stop here so we don't load the same CSS more than once (at least until we implement that TODO) |
| 1647 |
} |
| 1648 |
} |
| 1649 |
|
| 1650 |
// Restore the current base url |
| 1651 |
$this->_protocol = $protocol; |
| 1652 |
$this->_base_host = $host; |
| 1653 |
$this->_base_path = $path; |
| 1654 |
} |
| 1655 |
|
| 1656 |
/** |
| 1657 |
* parse @font-face{} sections |
| 1658 |
* http://www.w3.org/TR/css3-fonts/#the-font-face-rule |
| 1659 |
* |
| 1660 |
* @param string $str CSS @font-face rules |
| 1661 |
*/ |
| 1662 |
private function _parse_font_face($str) |
| 1663 |
{ |
| 1664 |
$descriptors = $this->_parse_properties($str); |
| 1665 |
|
| 1666 |
preg_match_all("/" . self::PATTERN_CSS_LOCAL_FN . "|" . self::PATTERN_CSS_URL_FN . "\s*(?<FORMAT>format\s*\((?<FORMAT_VALUE>collection|embedded-opentype|opentype|svg|truetype|woff|woff2|" . self::PATTERN_CSS_STRING . ")\))?/i", $descriptors->src, $sources, PREG_SET_ORDER); |
| 1667 |
|
| 1668 |
$valid_sources = []; |
| 1669 |
foreach ($sources as $source) { |
| 1670 |
$urlfn = $source["CSS_URL_FN"] ?? ""; |
| 1671 |
$format = strtolower($source["CSS_STRING_VALUE"] ?? $source["FORMAT_VALUE"] ?? "truetype"); |
| 1672 |
|
| 1673 |
if ($urlfn !== "" && $format === "truetype") { |
| 1674 |
$url = $this->resolve_url($urlfn, true); |
| 1675 |
if ($url === "none") { |
| 1676 |
continue; |
| 1677 |
} |
| 1678 |
$source_info = [ |
| 1679 |
"uri" => $urlfn, |
| 1680 |
"format" => $format, |
| 1681 |
"path" => $url, |
| 1682 |
]; |
| 1683 |
$valid_sources[] = $source_info; |
| 1684 |
} |
| 1685 |
} |
| 1686 |
|
| 1687 |
if (empty($valid_sources)) { |
| 1688 |
return; |
| 1689 |
} |
| 1690 |
|
| 1691 |
$style = [ |
| 1692 |
"family" => $descriptors->get_font_family_raw(), |
| 1693 |
"weight" => $descriptors->font_weight, |
| 1694 |
"style" => $descriptors->font_style, |
| 1695 |
]; |
| 1696 |
|
| 1697 |
foreach ($valid_sources as $valid_source) { |
| 1698 |
if ($this->fontMetrics->registerFont($style, $valid_source["path"], $this->_dompdf->getHttpContext())) { |
| 1699 |
break; |
| 1700 |
} |
| 1701 |
} |
| 1702 |
} |
| 1703 |
|
| 1704 |
/** |
| 1705 |
* parse regular CSS blocks |
| 1706 |
* |
| 1707 |
* _parse_properties() creates a new Style object based on the provided |
| 1708 |
* CSS rules. |
| 1709 |
* |
| 1710 |
* @param string $str CSS rules |
| 1711 |
* @return Style |
| 1712 |
*/ |
| 1713 |
private function _parse_properties($str) |
| 1714 |
{ |
| 1715 |
$DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss(); |
| 1716 |
|
| 1717 |
if ($DEBUGCSS) { |
| 1718 |
print '[_parse_properties'; |
| 1719 |
} |
| 1720 |
|
| 1721 |
$delims = ['"' => '"', "'" => "'", '(' => ')']; |
| 1722 |
$delim = null; |
| 1723 |
$len = strlen($str); |
| 1724 |
$properties = []; |
| 1725 |
$char = null; |
| 1726 |
$prev = null; |
| 1727 |
for ($pos = 0; $pos < $len; $pos++) { |
| 1728 |
$prev = $char; |
| 1729 |
$char = $str[$pos]; |
| 1730 |
if ($delim !== null) { |
| 1731 |
if ($char === $delims[$delim] && $prev !== '\\') { |
| 1732 |
$delim = null; |
| 1733 |
} |
| 1734 |
continue; |
| 1735 |
} |
| 1736 |
if (isset($delims[$char]) && $prev !== '\\') { |
| 1737 |
$delim = $char; |
| 1738 |
continue; |
| 1739 |
} |
| 1740 |
if ($char === ';' && $prev !== '\\') { |
| 1741 |
$properties[] = substr($str, 0, $pos); |
| 1742 |
$str = substr($str, $pos+1); |
| 1743 |
$pos = 0; |
| 1744 |
$len = strlen($str); |
| 1745 |
} |
| 1746 |
} |
| 1747 |
$properties[] = $str; |
| 1748 |
$style = new Style($this, Stylesheet::ORIG_AUTHOR); |
| 1749 |
foreach ($properties as $prop) { |
| 1750 |
$prop = trim($prop); |
| 1751 |
if ($prop === "") { |
| 1752 |
continue; |
| 1753 |
} |
| 1754 |
|
| 1755 |
$important = false; |
| 1756 |
if (substr($prop, -9) === 'important') { |
| 1757 |
$prop_tmp = rtrim(substr($prop, 0, -9)); |
| 1758 |
|
| 1759 |
if (substr($prop_tmp, -1) === '!') { |
| 1760 |
$prop = rtrim(substr($prop_tmp, 0, -1)); |
| 1761 |
$important = true; |
| 1762 |
} |
| 1763 |
} |
| 1764 |
|
| 1765 |
$i = strpos($prop, ":"); |
| 1766 |
if ($i === false) { |
| 1767 |
if ($DEBUGCSS) { |
| 1768 |
print "(novalue $prop)"; |
| 1769 |
} |
| 1770 |
continue; |
| 1771 |
} |
| 1772 |
|
| 1773 |
$prop_name = rtrim(substr($prop, 0, $i)); |
| 1774 |
$value = ltrim(substr($prop, $i + 1)); |
| 1775 |
|
| 1776 |
// Regular (non-custom) properties are case-insensitive |
| 1777 |
if (strncmp($prop_name, "--", 2) !== 0) { |
| 1778 |
$prop_name = strtolower($prop_name); |
| 1779 |
} |
| 1780 |
|
| 1781 |
if ($DEBUGCSS) { |
| 1782 |
print "($prop_name:=$value" . ($important ? " !IMPORTANT" : "") . ")"; |
| 1783 |
} |
| 1784 |
|
| 1785 |
$style->set_prop($prop_name, $value, $important, false); |
| 1786 |
} |
| 1787 |
|
| 1788 |
if ($DEBUGCSS) { |
| 1789 |
print '_parse_properties]'; |
| 1790 |
} |
| 1791 |
|
| 1792 |
return $style; |
| 1793 |
} |
| 1794 |
|
| 1795 |
/** |
| 1796 |
* parse selector + rulesets |
| 1797 |
* |
| 1798 |
* @param string $str CSS selectors and rulesets |
| 1799 |
* @param array $media_queries |
| 1800 |
*/ |
| 1801 |
private function _parse_sections($str, $media_queries = []) |
| 1802 |
{ |
| 1803 |
// Pre-process selectors: collapse all whitespace and strip whitespace |
| 1804 |
// around '>', '.', ':', '+', '~', '#' |
| 1805 |
$patterns = ["/\s+/", "/\s+([>.:+~#])\s+/"]; |
| 1806 |
$replacements = [" ", "\\1"]; |
| 1807 |
$DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss(); |
| 1808 |
|
| 1809 |
$sections = explode("}", $str); |
| 1810 |
if ($DEBUGCSS) print '[_parse_sections'; |
| 1811 |
foreach ($sections as $sect) { |
| 1812 |
$i = mb_strpos($sect, "{", 0, "UTF-8"); |
| 1813 |
if ($i === false) { continue; } |
| 1814 |
|
| 1815 |
if ($DEBUGCSS) print '[section'; |
| 1816 |
|
| 1817 |
$selector_str = preg_replace($patterns, $replacements, mb_substr($sect, 0, $i, "UTF-8")); |
| 1818 |
$selectors = preg_split("/,(?![^\(]*\))/", $selector_str, 0, PREG_SPLIT_NO_EMPTY); |
| 1819 |
$style = $this->_parse_properties(trim(mb_substr($sect, $i + 1, null, "UTF-8"))); |
| 1820 |
|
| 1821 |
// Assign it to the selected elements |
| 1822 |
foreach ($selectors as $selector) { |
| 1823 |
$selector = trim($selector); |
| 1824 |
|
| 1825 |
if ($selector === "") { |
| 1826 |
if ($DEBUGCSS) print '#empty#'; |
| 1827 |
continue; |
| 1828 |
} |
| 1829 |
if ($DEBUGCSS) print '#' . $selector . '#'; |
| 1830 |
//if ($DEBUGCSS) { if (strpos($selector,'p') !== false) print '!!!p!!!#'; } |
| 1831 |
|
| 1832 |
//FIXME: tag the selector with a hash of the media query to separate it from non-conditional styles (?), xpath comments are probably not what we want to do here |
| 1833 |
if (count($media_queries) > 0) { |
| 1834 |
$style->set_media_queries($media_queries); |
| 1835 |
} |
| 1836 |
$this->add_style($selector, $style); |
| 1837 |
} |
| 1838 |
|
| 1839 |
if ($DEBUGCSS) { |
| 1840 |
print 'section]'; |
| 1841 |
} |
| 1842 |
} |
| 1843 |
|
| 1844 |
if ($DEBUGCSS) { |
| 1845 |
print "_parse_sections]\n"; |
| 1846 |
} |
| 1847 |
} |
| 1848 |
|
| 1849 |
/** |
| 1850 |
* Parses a CSS string containing quotes and escaped hex characters. |
| 1851 |
* https://www.w3.org/TR/CSS21/syndata.html#characters |
| 1852 |
* |
| 1853 |
* @param string $string The string to parse. |
| 1854 |
* |
| 1855 |
* @return string |
| 1856 |
*/ |
| 1857 |
public function parse_string(string $string): string |
| 1858 |
{ |
| 1859 |
// Strip string quotes and escapes |
| 1860 |
$string = preg_replace('/^["\']|["\']$/', "", $string); |
| 1861 |
$string = preg_replace("/\\\\([^0-9a-fA-F])/", "\\1", $string); |
| 1862 |
|
| 1863 |
// Convert escaped hex characters (e.g. \A => newline) |
| 1864 |
return preg_replace_callback( |
| 1865 |
"/\\\\([0-9a-fA-F]{1,6})\s?/", |
| 1866 |
function ($matches) { return Helpers::unichr(hexdec($matches[1])); }, |
| 1867 |
$string |
| 1868 |
) ?? ""; |
| 1869 |
} |
| 1870 |
|
| 1871 |
/** |
| 1872 |
* @return string |
| 1873 |
*/ |
| 1874 |
public function getDefaultStylesheet() |
| 1875 |
{ |
| 1876 |
$options = $this->_dompdf->getOptions(); |
| 1877 |
$rootDir = realpath($options->getRootDir()); |
| 1878 |
return Helpers::build_url("file://", "", $rootDir, $rootDir . self::DEFAULT_STYLESHEET, $options->getChroot()); |
| 1879 |
} |
| 1880 |
|
| 1881 |
/** |
| 1882 |
* @param FontMetrics $fontMetrics |
| 1883 |
* @return $this |
| 1884 |
*/ |
| 1885 |
public function setFontMetrics(FontMetrics $fontMetrics) |
| 1886 |
{ |
| 1887 |
$this->fontMetrics = $fontMetrics; |
| 1888 |
return $this; |
| 1889 |
} |
| 1890 |
|
| 1891 |
/** |
| 1892 |
* @return FontMetrics |
| 1893 |
*/ |
| 1894 |
public function getFontMetrics() |
| 1895 |
{ |
| 1896 |
return $this->fontMetrics; |
| 1897 |
} |
| 1898 |
|
| 1899 |
/** |
| 1900 |
* dumps the entire stylesheet as a string |
| 1901 |
* |
| 1902 |
* Generates a string of each selector and associated style in the |
| 1903 |
* Stylesheet. Useful for debugging. |
| 1904 |
* |
| 1905 |
* @return string |
| 1906 |
*/ |
| 1907 |
function __toString() |
| 1908 |
{ |
| 1909 |
$str = ""; |
| 1910 |
foreach ($this->_styles as $selector => $selector_styles) { |
| 1911 |
foreach ($selector_styles as $style) { |
| 1912 |
$str .= "$selector => " . $style->__toString() . "\n"; |
| 1913 |
} |
| 1914 |
} |
| 1915 |
|
| 1916 |
return $str; |
| 1917 |
} |
| 1918 |
} |
| 1919 |
|
