Completed
Push — development ( 29b82e...e01cc9 )
by Ashutosh
10:05
created

Stylesheet::_css_selector_to_xpath()   F

Complexity

Conditions 93
Paths > 20000

Size

Total Lines 443
Code Lines 250

Duplication

Lines 0
Ratio 0 %

Importance

Changes 0
Metric Value
eloc 250
dl 0
loc 443
rs 0
c 0
b 0
f 0
cc 93
nc 103272
nop 2

How to fix   Long Method    Complexity   

Long Method

Small methods make your code easier to understand, in particular if combined with a good name. Besides, if your method is small, finding a good name is usually much easier.

For example, if you find yourself adding comments to a method's body, this is usually a good sign to extract the commented part to a new method, and use the comment as a starting point when coming up with a good name for this new method.

Commonly applied refactorings include:

1
<?php
2
/**
3
 * @package dompdf
4
 * @link    http://dompdf.github.com/
5
 * @author  Benj Carson <[email protected]>
6
 * @author  Helmut Tischer <[email protected]>
7
 * @author  Fabien Ménager <[email protected]>
8
 * @license http://www.gnu.org/copyleft/lesser.html GNU Lesser General Public License
9
 */
10
namespace Dompdf\Css;
11
12
use DOMElement;
13
use DOMXPath;
14
use Dompdf\Dompdf;
15
use Dompdf\Helpers;
16
use Dompdf\Exception;
17
use Dompdf\FontMetrics;
18
use Dompdf\Frame\FrameTree;
19
20
/**
21
 * The master stylesheet class
22
 *
23
 * The Stylesheet class is responsible for parsing stylesheets and style
24
 * tags/attributes.  It also acts as a registry of the individual Style
25
 * objects generated by the current set of loaded CSS files and style
26
 * elements.
27
 *
28
 * @see Style
29
 * @package dompdf
30
 */
31
class Stylesheet
32
{
33
    /**
34
     * The location of the default built-in CSS file.
35
     */
36
    const DEFAULT_STYLESHEET = "/lib/res/html.css";
37
38
    /**
39
     * User agent stylesheet origin
40
     *
41
     * @var int
42
     */
43
    const ORIG_UA = 1;
44
45
    /**
46
     * User normal stylesheet origin
47
     *
48
     * @var int
49
     */
50
    const ORIG_USER = 2;
51
52
    /**
53
     * Author normal stylesheet origin
54
     *
55
     * @var int
56
     */
57
    const ORIG_AUTHOR = 3;
58
59
    /*
60
     * The highest possible specificity is 0x01000000 (and that is only for author
61
     * stylesheets, as it is for inline styles). Origin precedence can be achieved by
62
     * adding multiples of 0x10000000 to the actual specificity. Important
63
     * declarations are handled in Style; though technically they should be handled
64
     * here so that user important declarations can be made to take precedence over
65
     * user important declarations, this doesn't matter in practice as Dompdf does
66
     * not support user stylesheets, and user agent stylesheets can not include
67
     * important declarations.
68
     */
69
    private static $_stylesheet_origins = array(
70
        self::ORIG_UA => 0x00000000, // user agent declarations
71
        self::ORIG_USER => 0x10000000, // user normal declarations
72
        self::ORIG_AUTHOR => 0x30000000, // author normal declarations
73
    );
74
75
    /*
76
     * Non-CSS presentational hints (i.e. HTML 4 attributes) are handled as if added
77
     * to the beginning of an author stylesheet, i.e. anything in author stylesheets
78
     * should override them.
79
     */
80
    const SPEC_NON_CSS = 0x20000000;
81
82
    /**
83
     * Current dompdf instance
84
     *
85
     * @var Dompdf
86
     */
87
    private $_dompdf;
88
89
    /**
90
     * Array of currently defined styles
91
     *
92
     * @var Style[]
93
     */
94
    private $_styles;
95
96
    /**
97
     * Base protocol of the document being parsed
98
     * Used to handle relative urls.
99
     *
100
     * @var string
101
     */
102
    private $_protocol;
103
104
    /**
105
     * Base hostname of the document being parsed
106
     * Used to handle relative urls.
107
     *
108
     * @var string
109
     */
110
    private $_base_host;
111
112
    /**
113
     * Base path of the document being parsed
114
     * Used to handle relative urls.
115
     *
116
     * @var string
117
     */
118
    private $_base_path;
119
120
    /**
121
     * The styles defined by @page rules
122
     *
123
     * @var array<Style>
124
     */
125
    private $_page_styles;
126
127
    /**
128
     * List of loaded files, used to prevent recursion
129
     *
130
     * @var array
131
     */
132
    private $_loaded_files;
133
134
    /**
135
     * Current stylesheet origin
136
     *
137
     * @var int
138
     */
139
    private $_current_origin = self::ORIG_UA;
140
141
    /**
142
     * Accepted CSS media types
143
     * List of types and parsing rules for future extensions:
144
     * http://www.w3.org/TR/REC-html40/types.html
145
     *   screen, tty, tv, projection, handheld, print, braille, aural, all
146
     * The following are non standard extensions for undocumented specific environments.
147
     *   static, visual, bitmap, paged, dompdf
148
     * Note, even though the generated pdf file is intended for print output,
149
     * the desired content might be different (e.g. screen or projection view of html file).
150
     * Therefore allow specification of content by dompdf setting Options::defaultMediaType.
151
     * If given, replace media "print" by Options::defaultMediaType.
152
     * (Previous version $ACCEPTED_MEDIA_TYPES = $ACCEPTED_GENERIC_MEDIA_TYPES + $ACCEPTED_DEFAULT_MEDIA_TYPE)
153
     */
154
    static $ACCEPTED_DEFAULT_MEDIA_TYPE = "print";
155
    static $ACCEPTED_GENERIC_MEDIA_TYPES = array("all", "static", "visual", "bitmap", "paged", "dompdf");
156
    static $VALID_MEDIA_TYPES = array("all", "aural", "bitmap", "braille", "dompdf", "embossed", "handheld", "paged", "print", "projection", "screen", "speech", "static", "tty", "tv", "visual");
157
158
    /**
159
     * @var FontMetrics
160
     */
161
    private $fontMetrics;
162
163
    /**
164
     * The class constructor.
165
     *
166
     * The base protocol, host & path are initialized to those of
167
     * the current script.
168
     */
169
    function __construct(Dompdf $dompdf)
170
    {
171
        $this->_dompdf = $dompdf;
172
        $this->setFontMetrics($dompdf->getFontMetrics());
173
        $this->_styles = array();
174
        $this->_loaded_files = array();
175
        list($this->_protocol, $this->_base_host, $this->_base_path) = Helpers::explode_url($_SERVER["SCRIPT_FILENAME"]);
176
        $this->_page_styles = array("base" => null);
177
    }
178
179
    /**
180
     * Set the base protocol
181
     *
182
     * @param string $protocol
183
     */
184
    function set_protocol($protocol)
185
    {
186
        $this->_protocol = $protocol;
187
    }
188
189
    /**
190
     * Set the base host
191
     *
192
     * @param string $host
193
     */
194
    function set_host($host)
195
    {
196
        $this->_base_host = $host;
197
    }
198
199
    /**
200
     * Set the base path
201
     *
202
     * @param string $path
203
     */
204
    function set_base_path($path)
205
    {
206
        $this->_base_path = $path;
207
    }
208
209
    /**
210
     * Return the Dompdf object
211
     *
212
     * @return Dompdf
213
     */
214
    function get_dompdf()
215
    {
216
        return $this->_dompdf;
217
    }
218
219
    /**
220
     * Return the base protocol for this stylesheet
221
     *
222
     * @return string
223
     */
224
    function get_protocol()
225
    {
226
        return $this->_protocol;
227
    }
228
229
    /**
230
     * Return the base host for this stylesheet
231
     *
232
     * @return string
233
     */
234
    function get_host()
235
    {
236
        return $this->_base_host;
237
    }
238
239
    /**
240
     * Return the base path for this stylesheet
241
     *
242
     * @return string
243
     */
244
    function get_base_path()
245
    {
246
        return $this->_base_path;
247
    }
248
249
    /**
250
     * Return the array of page styles
251
     *
252
     * @return Style[]
253
     */
254
    function get_page_styles()
255
    {
256
        return $this->_page_styles;
257
    }
258
259
    /**
260
     * Add a new Style object to the stylesheet
261
     * add_style() adds a new Style object to the current stylesheet, or
262
     * merges a new Style with an existing one.
263
     *
264
     * @param string $key the Style's selector
265
     * @param Style $style the Style to be added
266
     *
267
     * @throws \Dompdf\Exception
268
     */
269
    function add_style($key, Style $style)
270
    {
271
        if (!is_string($key)) {
272
            throw new Exception("CSS rule must be keyed by a string.");
273
        }
274
275
        if (!isset($this->_styles[$key])) {
276
            $this->_styles[$key] = array();
277
        }
278
        $new_style = clone $style;
279
        $new_style->set_origin($this->_current_origin);
280
        $this->_styles[$key][] = $new_style;
281
    }
282
283
    /**
284
     * lookup a specifc Style collection
285
     *
286
     * lookup() returns the Style collection specified by $key, or null if the Style is
287
     * not found.
288
     *
289
     * @param string $key the selector of the requested Style
290
     * @return Style
291
     *
292
     * @Fixme _styles is a two dimensional array. It should produce wrong results
293
     */
294
    function lookup($key)
295
    {
296
        if (!isset($this->_styles[$key])) {
297
            return null;
298
        }
299
300
        return $this->_styles[$key];
301
    }
302
303
    /**
304
     * create a new Style object associated with this stylesheet
305
     *
306
     * @param Style $parent The style of this style's parent in the DOM tree
307
     * @return Style
308
     */
309
    function create_style(Style $parent = null)
310
    {
311
        return new Style($this, $this->_current_origin);
312
    }
313
314
    /**
315
     * load and parse a CSS string
316
     *
317
     * @param string $css
318
     * @param int $origin
319
     */
320
    function load_css(&$css, $origin = self::ORIG_AUTHOR)
321
    {
322
        if ($origin) {
323
            $this->_current_origin = $origin;
324
        }
325
        $this->_parse_css($css);
326
    }
327
328
329
    /**
330
     * load and parse a CSS file
331
     *
332
     * @param string $file
333
     * @param int $origin
334
     */
335
    function load_css_file($file, $origin = self::ORIG_AUTHOR)
336
    {
337
        if ($origin) {
338
            $this->_current_origin = $origin;
339
        }
340
341
        // Prevent circular references
342
        if (isset($this->_loaded_files[$file])) {
343
            return;
344
        }
345
346
        $this->_loaded_files[$file] = true;
347
348
        if (strpos($file, "data:") === 0) {
349
            $parsed = Helpers::parse_data_uri($file);
350
            $css = $parsed["data"];
351
        } else {
352
            $parsed_url = Helpers::explode_url($file);
353
354
            list($this->_protocol, $this->_base_host, $this->_base_path, $filename) = $parsed_url;
355
356
            // Fix submitted by Nick Oostveen for aliased directory support:
357
            if ($this->_protocol == "") {
358
                $file = $this->_base_path . $filename;
359
            } else {
360
                $file = Helpers::build_url($this->_protocol, $this->_base_host, $this->_base_path, $filename);
361
            }
362
363
            list($css, $http_response_header) = Helpers::getFileContent($file, $this->_dompdf->getHttpContext());
364
365
            $good_mime_type = true;
366
367
            // See http://the-stickman.com/web-development/php/getting-http-response-headers-when-using-file_get_contents/
368
            if (isset($http_response_header) && !$this->_dompdf->getQuirksmode()) {
369
                foreach ($http_response_header as $_header) {
370
                    if (preg_match("@Content-Type:\s*([\w/]+)@i", $_header, $matches) &&
371
                        ($matches[1] !== "text/css")
372
                    ) {
373
                        $good_mime_type = false;
374
                    }
375
                }
376
            }
377
378
            if (!$good_mime_type || $css == "") {
379
                Helpers::record_warnings(E_USER_WARNING, "Unable to load css file $file", __FILE__, __LINE__);
380
                return;
381
            }
382
        }
383
384
        $this->_parse_css($css);
385
    }
386
387
    /**
388
     * @link http://www.w3.org/TR/CSS21/cascade.html#specificity
389
     *
390
     * @param string $selector
391
     * @param int $origin :
392
     *    - Stylesheet::ORIG_UA: user agent style sheet
393
     *    - Stylesheet::ORIG_USER: user style sheet
394
     *    - Stylesheet::ORIG_AUTHOR: author style sheet
395
     *
396
     * @return int
397
     */
398
    private function _specificity($selector, $origin = self::ORIG_AUTHOR)
399
    {
400
        // http://www.w3.org/TR/CSS21/cascade.html#specificity
401
        // ignoring the ":" pseudoclass modifiers
402
        // also ignored in _css_selector_to_xpath
403
404
        $a = ($selector === "!attr") ? 1 : 0;
405
406
        $b = min(mb_substr_count($selector, "#"), 255);
407
408
        $c = min(mb_substr_count($selector, ".") +
409
            mb_substr_count($selector, "["), 255);
410
411
        $d = min(mb_substr_count($selector, " ") +
412
            mb_substr_count($selector, ">") +
413
            mb_substr_count($selector, "+"), 255);
414
415
        //If a normal element name is at the beginning of the string,
416
        //a leading whitespace might have been removed on whitespace collapsing and removal
417
        //therefore there might be one whitespace less as selected element names
418
        //this can lead to a too small specificity
419
        //see _css_selector_to_xpath
420
421
        if (!in_array($selector[0], array(" ", ">", ".", "#", "+", ":", "[")) && $selector !== "*") {
422
            $d++;
423
        }
424
425
        if ($this->_dompdf->getOptions()->getDebugCss()) {
426
            /*DEBUGCSS*/
427
            print "<pre>\n";
428
            /*DEBUGCSS*/
429
            printf("_specificity(): 0x%08x \"%s\"\n", self::$_stylesheet_origins[$origin] + (($a << 24) | ($b << 16) | ($c << 8) | ($d)), $selector);
430
            /*DEBUGCSS*/
431
            print "</pre>";
432
        }
433
434
        return self::$_stylesheet_origins[$origin] + (($a << 24) | ($b << 16) | ($c << 8) | ($d));
435
    }
436
437
    /**
438
     * Converts a CSS selector to an XPath query.
439
     *
440
     * @param string $selector
441
     * @param bool $first_pass
442
     *
443
     * @throws Exception
444
     * @return string
445
     */
446
    private function _css_selector_to_xpath($selector, $first_pass = false)
447
    {
448
449
        // Collapse white space and strip whitespace around delimiters
450
        //$search = array("/\\s+/", "/\\s+([.>#+:])\\s+/");
451
        //$replace = array(" ", "\\1");
452
        //$selector = preg_replace($search, $replace, trim($selector));
453
454
        // Initial query (non-absolute)
455
        $query = "//";
456
457
        // Will contain :before and :after
458
        $pseudo_elements = array();
459
460
        // Will contain :link, etc
461
        $pseudo_classes = array();
462
463
        // Parse the selector
464
        //$s = preg_split("/([ :>.#+])/", $selector, -1, PREG_SPLIT_DELIM_CAPTURE);
465
466
        $delimiters = array(" ", ">", ".", "#", "+", ":", "[", "(");
467
468
        // Add an implicit * at the beginning of the selector
469
        // if it begins with an attribute selector
470
        if ($selector[0] === "[") {
471
            $selector = "*$selector";
472
        }
473
474
        // Add an implicit space at the beginning of the selector if there is no
475
        // delimiter there already.
476
        if (!in_array($selector[0], $delimiters)) {
477
            $selector = " $selector";
478
        }
479
480
        $tok = "";
481
        $len = mb_strlen($selector);
482
        $i = 0;
483
484
        while ($i < $len) {
485
486
            $s = $selector[$i];
487
            $i++;
488
489
            // Eat characters up to the next delimiter
490
            $tok = "";
491
            $in_attr = false;
492
            $in_func = false;
493
494
            while ($i < $len) {
495
                $c = $selector[$i];
496
                $c_prev = $selector[$i - 1];
497
498
                if (!$in_func && !$in_attr && in_array($c, $delimiters) && !(($c == $c_prev) == ":")) {
499
                    break;
500
                }
501
502
                if ($c_prev === "[") {
503
                    $in_attr = true;
504
                }
505
                if ($c_prev === "(") {
506
                    $in_func = true;
507
                }
508
509
                $tok .= $selector[$i++];
510
511
                if ($in_attr && $c === "]") {
512
                    $in_attr = false;
513
                    break;
514
                }
515
                if ($in_func && $c === ")") {
516
                    $in_func = false;
517
                    break;
518
                }
519
            }
520
521
            switch ($s) {
522
523
                case " ":
524
                case ">":
525
                    // All elements matching the next token that are direct children of
526
                    // the current token
527
                    $expr = $s === " " ? "descendant" : "child";
528
529
                    if (mb_substr($query, -1, 1) !== "/") {
530
                        $query .= "/";
531
                    }
532
533
                    // Tag names are case-insensitive
534
                    $tok = strtolower($tok);
535
536
                    if (!$tok) {
537
                        $tok = "*";
538
                    }
539
540
                    $query .= "$expr::$tok";
541
                    $tok = "";
542
                    break;
543
544
                case ".":
545
                case "#":
546
                    // All elements matching the current token with a class/id equal to
547
                    // the _next_ token.
548
549
                    $attr = $s === "." ? "class" : "id";
550
551
                    // empty class/id == *
552
                    if (mb_substr($query, -1, 1) === "/") {
553
                        $query .= "*";
554
                    }
555
556
                    // Match multiple classes: $tok contains the current selected
557
                    // class.  Search for class attributes with class="$tok",
558
                    // class=".* $tok .*" and class=".* $tok"
559
560
                    // This doesn't work because libxml only supports XPath 1.0...
561
                    //$query .= "[matches(@$attr,\"^${tok}\$|^${tok}[ ]+|[ ]+${tok}\$|[ ]+${tok}[ ]+\")]";
562
563
                    // Query improvement by Michael Sheakoski <[email protected]>:
564
                    $query .= "[contains(concat(' ', @$attr, ' '), concat(' ', '$tok', ' '))]";
565
                    $tok = "";
566
                    break;
567
568
                case "+":
569
                    // All sibling elements that folow the current token
570
                    if (mb_substr($query, -1, 1) !== "/") {
571
                        $query .= "/";
572
                    }
573
574
                    $query .= "following-sibling::$tok";
575
                    $tok = "";
576
                    break;
577
578
                case ":":
579
                    $i2 = $i - strlen($tok) - 2; // the char before ":"
580
                    if (($i2 < 0 || !isset($selector[$i2]) || (in_array($selector[$i2], $delimiters) && $selector[$i2] != ":")) && substr($query, -1) != "*") {
581
                        $query .= "*";
582
                    }
583
584
                    $last = false;
585
586
                    // Pseudo-classes
587
                    switch ($tok) {
588
589
                        case "first-child":
590
                            $query .= "[1]";
591
                            $tok = "";
592
                            break;
593
594
                        case "last-child":
595
                            $query .= "[not(following-sibling::*)]";
596
                            $tok = "";
597
                            break;
598
599
                        case "first-of-type":
600
                            $query .= "[position() = 1]";
601
                            $tok = "";
602
                            break;
603
604
                        case "last-of-type":
605
                            $query .= "[position() = last()]";
606
                            $tok = "";
607
                            break;
608
609
                        // an+b, n, odd, and even
610
                        /** @noinspection PhpMissingBreakStatementInspection */
611
                        case "nth-last-of-type":
612
                            $last = true;
613
                        case "nth-of-type":
614
                            //FIXME: this fix-up is pretty ugly, would parsing the selector in reverse work better generally?
615
                            $descendant_delimeter = strrpos($query, "::");
616
                            $isChild = substr($query, $descendant_delimeter-5, 5) == "child";
617
                            $el = substr($query, $descendant_delimeter+2);
618
                            $query = substr($query, 0, strrpos($query, "/")) . ($isChild ? "/" : "//") . $el;
619
620
                            $pseudo_classes[$tok] = true;
621
                            $p = $i + 1;
622
                            $nth = trim(mb_substr($selector, $p, strpos($selector, ")", $i) - $p));
623
624
                            // 1
625
                            if (preg_match("/^\d+$/", $nth)) {
626
                                $condition = "position() = $nth";
627
                            } // odd
628
                            elseif ($nth === "odd") {
629
                                $condition = "(position() mod 2) = 1";
630
                            } // even
631
                            elseif ($nth === "even") {
632
                                $condition = "(position() mod 2) = 0";
633
                            } // an+b
634
                            else {
635
                                $condition = $this->_selector_an_plus_b($nth, $last);
636
                            }
637
638
                            $query .= "[$condition]";
639
                            $tok = "";
640
                            break;
641
                        /** @noinspection PhpMissingBreakStatementInspection */
642
                        case "nth-last-child":
643
                            $last = true;
644
                        case "nth-child":
645
                            //FIXME: this fix-up is pretty ugly, would parsing the selector in reverse work better generally?
646
                            $descendant_delimeter = strrpos($query, "::");
647
                            $isChild = substr($query, $descendant_delimeter-5, 5) == "child";
648
                            $el = substr($query, $descendant_delimeter+2);
649
                            $query = substr($query, 0, strrpos($query, "/")) . ($isChild ? "/" : "//") . "*";
650
651
                            $pseudo_classes[$tok] = true;
652
                            $p = $i + 1;
653
                            $nth = trim(mb_substr($selector, $p, strpos($selector, ")", $i) - $p));
654
655
                            // 1
656
                            if (preg_match("/^\d+$/", $nth)) {
657
                                $condition = "position() = $nth";
658
                            } // odd
659
                            elseif ($nth === "odd") {
660
                                $condition = "(position() mod 2) = 1";
661
                            } // even
662
                            elseif ($nth === "even") {
663
                                $condition = "(position() mod 2) = 0";
664
                            } // an+b
665
                            else {
666
                                $condition = $this->_selector_an_plus_b($nth, $last);
667
                            }
668
669
                            $query .= "[$condition]";
670
                            if ($el != "*") {
671
                                $query .= "[name() = '$el']";
672
                            }
673
                            $tok = "";
674
                            break;
675
676
                        //TODO: bit of a hack attempt at matches support, currently only matches against elements
677
                        case "matches":
678
                            $pseudo_classes[$tok] = true;
679
                            $p = $i + 1;
680
                            $matchList = trim(mb_substr($selector, $p, strpos($selector, ")", $i) - $p));
681
682
                            // Tag names are case-insensitive
683
                            $elements = array_map("trim", explode(",", strtolower($matchList)));
684
                            foreach ($elements as &$element) {
685
                                $element = "name() = '$element'";
686
                            }
687
688
                            $query .= "[" . implode(" or ", $elements) . "]";
689
                            $tok = "";
690
                            break;
691
692
                        case "link":
693
                            $query .= "[@href]";
694
                            $tok = "";
695
                            break;
696
697
                        case "first-line":
698
                        case ":first-line":
699
                        case "first-letter":
700
                        case ":first-letter":
701
                            // TODO
702
                            $el = trim($tok, ":");
703
                            $pseudo_elements[$el] = true;
704
                            break;
705
706
                            // N/A
707
                        case "focus":
708
                        case "active":
709
                        case "hover":
710
                        case "visited":
711
                            $query .= "[false()]";
712
                            $tok = "";
713
                            break;
714
715
                        /* Pseudo-elements */
716
                        case "before":
717
                        case ":before":
718
                        case "after":
719
                        case ":after":
720
                            $pos = trim($tok, ":");
721
                            $pseudo_elements[$pos] = true;
722
                            if (!$first_pass) {
723
                                $query .= "/*[@$pos]";
724
                            }
725
726
                            $tok = "";
727
                            break;
728
729
                        case "empty":
730
                            $query .= "[not(*) and not(normalize-space())]";
731
                            $tok = "";
732
                            break;
733
734
                        case "disabled":
735
                        case "checked":
736
                            $query .= "[@$tok]";
737
                            $tok = "";
738
                            break;
739
740
                        case "enabled":
741
                            $query .= "[not(@disabled)]";
742
                            $tok = "";
743
                            break;
744
745
                        // the selector is not handled, until we support all possible selectors force an empty set (silent failure)
746
                        default:
747
                            $query = "/../.."; // go up two levels because generated content starts on the body element
748
                            $tok = "";
749
                            break;
750
                    }
751
752
                    break;
753
754
                case "[":
755
                    // Attribute selectors.  All with an attribute matching the following token(s)
756
                    $attr_delimiters = array("=", "]", "~", "|", "$", "^", "*");
757
                    $tok_len = mb_strlen($tok);
758
                    $j = 0;
759
760
                    $attr = "";
761
                    $op = "";
762
                    $value = "";
763
764
                    while ($j < $tok_len) {
765
                        if (in_array($tok[$j], $attr_delimiters)) {
766
                            break;
767
                        }
768
                        $attr .= $tok[$j++];
769
                    }
770
771
                    switch ($tok[$j]) {
772
773
                        case "~":
774
                        case "|":
775
                        case "$":
776
                        case "^":
777
                        case "*":
778
                            $op .= $tok[$j++];
779
780
                            if ($tok[$j] !== "=") {
781
                                throw new Exception("Invalid CSS selector syntax: invalid attribute selector: $selector");
782
                            }
783
784
                            $op .= $tok[$j];
785
                            break;
786
787
                        case "=":
788
                            $op = "=";
789
                            break;
790
791
                    }
792
793
                    // Read the attribute value, if required
794
                    if ($op != "") {
795
                        $j++;
796
                        while ($j < $tok_len) {
797
                            if ($tok[$j] === "]") {
798
                                break;
799
                            }
800
                            $value .= $tok[$j++];
801
                        }
802
                    }
803
804
                    if ($attr == "") {
805
                        throw new Exception("Invalid CSS selector syntax: missing attribute name");
806
                    }
807
808
                    $value = trim($value, "\"'");
809
810
                    switch ($op) {
811
812
                        case "":
813
                            $query .= "[@$attr]";
814
                            break;
815
816
                        case "=":
817
                            $query .= "[@$attr=\"$value\"]";
818
                            break;
819
820
                        case "~=":
821
                            // FIXME: this will break if $value contains quoted strings
822
                            // (e.g. [type~="a b c" "d e f"])
823
                            $values = explode(" ", $value);
824
                            $query .= "[";
825
826
                            foreach ($values as $val) {
827
                                $query .= "@$attr=\"$val\" or ";
828
                            }
829
830
                            $query = rtrim($query, " or ") . "]";
831
                            break;
832
833
                        case "|=":
834
                            $values = explode("-", $value);
835
                            $query .= "[";
836
837
                            foreach ($values as $val) {
838
                                $query .= "starts-with(@$attr, \"$val\") or ";
839
                            }
840
841
                            $query = rtrim($query, " or ") . "]";
842
                            break;
843
844
                        case "$=":
845
                            $query .= "[substring(@$attr, string-length(@$attr)-" . (strlen($value) - 1) . ")=\"$value\"]";
846
                            break;
847
848
                        case "^=":
849
                            $query .= "[starts-with(@$attr,\"$value\")]";
850
                            break;
851
852
                        case "*=":
853
                            $query .= "[contains(@$attr,\"$value\")]";
854
                            break;
855
                    }
856
857
                    break;
858
            }
859
        }
860
        $i++;
861
862
//       case ":":
863
//         // Pseudo selectors: ignore for now.  Partially handled directly
864
//         // below.
865
866
//         // Skip until the next special character, leaving the token as-is
867
//         while ( $i < $len ) {
868
//           if ( in_array($selector[$i], $delimiters) )
869
//             break;
870
//           $i++;
871
//         }
872
//         break;
873
874
//       default:
875
//         // Add the character to the token
876
//         $tok .= $selector[$i++];
877
//         break;
878
//       }
879
880
//    }
881
882
883
        // Trim the trailing '/' from the query
884
        if (mb_strlen($query) > 2) {
885
            $query = rtrim($query, "/");
886
        }
887
888
        return array("query" => $query, "pseudo_elements" => $pseudo_elements);
889
    }
890
891
    /**
892
     * https://github.com/tenderlove/nokogiri/blob/master/lib/nokogiri/css/xpath_visitor.rb
893
     *
894
     * @param $expr
895
     * @param bool $last
896
     * @return string
897
     */
898
    protected function _selector_an_plus_b($expr, $last = false)
899
    {
900
        $expr = preg_replace("/\s/", "", $expr);
901
        if (!preg_match("/^(?P<a>-?[0-9]*)?n(?P<b>[-+]?[0-9]+)?$/", $expr, $matches)) {
902
            return "false()";
903
        }
904
905
        $a = ((isset($matches["a"]) && $matches["a"] !== "") ? intval($matches["a"]) : 1);
906
        $b = ((isset($matches["b"]) && $matches["b"] !== "") ? intval($matches["b"]) : 0);
907
908
        $position = ($last ? "(last()-position()+1)" : "position()");
909
910
        if ($b == 0) {
911
            return "($position mod $a) = 0";
912
        } else {
913
            $compare = (($a < 0) ? "<=" : ">=");
914
            $b2 = -$b;
915
            if ($b2 >= 0) {
916
                $b2 = "+$b2";
917
            }
918
            return "($position $compare $b) and ((($position $b2) mod " . abs($a) . ") = 0)";
919
        }
920
    }
921
922
    /**
923
     * applies all current styles to a particular document tree
924
     *
925
     * apply_styles() applies all currently loaded styles to the provided
926
     * {@link FrameTree}.  Aside from parsing CSS, this is the main purpose
927
     * of this class.
928
     *
929
     * @param \Dompdf\Frame\FrameTree $tree
930
     */
931
    function apply_styles(FrameTree $tree)
932
    {
933
        // Use XPath to select nodes.  This would be easier if we could attach
934
        // Frame objects directly to DOMNodes using the setUserData() method, but
935
        // we can't do that just yet.  Instead, we set a _node attribute_ in
936
        // Frame->set_id() and use that as a handle on the Frame object via
937
        // FrameTree::$_registry.
938
939
        // We create a scratch array of styles indexed by frame id.  Once all
940
        // styles have been assigned, we order the cached styles by specificity
941
        // and create a final style object to assign to the frame.
942
943
        // FIXME: this is not particularly robust...
944
945
        $styles = array();
946
        $xp = new DOMXPath($tree->get_dom());
947
        $DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss();
948
949
        // Add generated content
950
        foreach ($this->_styles as $selector => $selector_styles) {
951
            /** @var Style $style */
952
            foreach ($selector_styles as $style) {
953
                if (strpos($selector, ":before") === false && strpos($selector, ":after") === false) {
954
                    continue;
955
                }
956
957
                $query = $this->_css_selector_to_xpath($selector, true);
958
959
                // Retrieve the nodes, limit to body for generated content
960
                //TODO: If we use a context node can we remove the leading dot?
961
                $nodes = @$xp->query('.' . $query["query"]);
962
                if ($nodes == null) {
963
                    Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__);
964
                    continue;
965
                }
966
967
                /** @var \DOMElement $node */
968
                foreach ($nodes as $node) {
969
                    // Only DOMElements get styles
970
                    if ($node->nodeType != XML_ELEMENT_NODE) {
971
                        continue;
972
                    }
973
974
                    foreach (array_keys($query["pseudo_elements"], true, true) as $pos) {
975
                        // Do not add a new pseudo element if another one already matched
976
                        if ($node->hasAttribute("dompdf_{$pos}_frame_id")) {
977
                            continue;
978
                        }
979
980
                        if (($src = $this->_image($style->get_prop('content'))) !== "none") {
981
                            $new_node = $node->ownerDocument->createElement("img_generated");
982
                            $new_node->setAttribute("src", $src);
983
                        } else {
984
                            $new_node = $node->ownerDocument->createElement("dompdf_generated");
985
                        }
986
987
                        $new_node->setAttribute($pos, $pos);
988
                        $new_frame_id = $tree->insert_node($node, $new_node, $pos);
989
                        $node->setAttribute("dompdf_{$pos}_frame_id", $new_frame_id);
990
                    }
991
                }
992
            }
993
        }
994
995
        // Apply all styles in stylesheet
996
        foreach ($this->_styles as $selector => $selector_styles) {
997
            /** @var Style $style */
998
            foreach ($selector_styles as $style) {
999
                $query = $this->_css_selector_to_xpath($selector);
1000
1001
                // Retrieve the nodes
1002
                $nodes = @$xp->query($query["query"]);
1003
                if ($nodes == null) {
1004
                    Helpers::record_warnings(E_USER_WARNING, "The CSS selector '$selector' is not valid", __FILE__, __LINE__);
1005
                    continue;
1006
                }
1007
1008
                $spec = $this->_specificity($selector, $style->get_origin());
1009
1010
                foreach ($nodes as $node) {
1011
                    // Retrieve the node id
1012
                    // Only DOMElements get styles
1013
                    if ($node->nodeType != XML_ELEMENT_NODE) {
1014
                        continue;
1015
                    }
1016
1017
                    $id = $node->getAttribute("frame_id");
1018
1019
                    // Assign the current style to the scratch array
1020
                    $styles[$id][$spec][] = $style;
1021
                }
1022
            }
1023
        }
1024
1025
        // Set the page width, height, and orientation based on the canvas paper size
1026
        $canvas = $this->_dompdf->getCanvas();
1027
        $paper_width = $canvas->get_width();
1028
        $paper_height = $canvas->get_height();
1029
        $paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait");
1030
1031
        if ($this->_page_styles["base"] && is_array($this->_page_styles["base"]->size)) {
1032
            $paper_width = $this->_page_styles['base']->size[0];
1033
            $paper_height = $this->_page_styles['base']->size[1];
1034
            $paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait");
1035
        }
1036
1037
        // Now create the styles and assign them to the appropriate frames. (We
1038
        // iterate over the tree using an implicit FrameTree iterator.)
1039
        $root_flg = false;
1040
        foreach ($tree->get_frames() as $frame) {
1041
            // Helpers::pre_r($frame->get_node()->nodeName . ":");
1042
            if (!$root_flg && $this->_page_styles["base"]) {
1043
                $style = $this->_page_styles["base"];
1044
            } else {
1045
                $style = $this->create_style();
1046
            }
1047
1048
            // Find nearest DOMElement parent
1049
            $p = $frame;
1050
            while ($p = $p->get_parent()) {
1051
                if ($p->get_node()->nodeType == XML_ELEMENT_NODE) {
1052
                    break;
1053
                }
1054
            }
1055
1056
            // Styles can only be applied directly to DOMElements; anonymous
1057
            // frames inherit from their parent
1058
            if ($frame->get_node()->nodeType != XML_ELEMENT_NODE) {
1059
                if ($p) {
1060
                    $style->inherit($p->get_style());
1061
                }
1062
1063
                $frame->set_style($style);
1064
                continue;
1065
            }
1066
1067
            $id = $frame->get_id();
1068
1069
            // Handle HTML 4.0 attributes
1070
            AttributeTranslator::translate_attributes($frame);
1071
            if (($str = $frame->get_node()->getAttribute(AttributeTranslator::$_style_attr)) !== "") {
1072
                $styles[$id][self::SPEC_NON_CSS][] = $this->_parse_properties($str);
1073
            }
1074
1075
            // Locate any additional style attributes
1076
            if (($str = $frame->get_node()->getAttribute("style")) !== "") {
1077
                // Destroy CSS comments
1078
                $str = preg_replace("'/\*.*?\*/'si", "", $str);
1079
1080
                $spec = $this->_specificity("!attr", self::ORIG_AUTHOR);
1081
                $styles[$id][$spec][] = $this->_parse_properties($str);
1082
            }
1083
1084
            // Grab the applicable styles
1085
            if (isset($styles[$id])) {
1086
1087
                /** @var array[][] $applied_styles */
1088
                $applied_styles = $styles[$frame->get_id()];
1089
1090
                // Sort by specificity
1091
                ksort($applied_styles);
1092
1093
                if ($DEBUGCSS) {
1094
                    $debug_nodename = $frame->get_node()->nodeName;
1095
                    print "<pre>\n[$debug_nodename\n";
1096
                    foreach ($applied_styles as $spec => $arr) {
1097
                        printf("specificity: 0x%08x\n", $spec);
1098
                        /** @var Style $s */
1099
                        foreach ($arr as $s) {
1100
                            print "[\n";
1101
                            $s->debug_print();
1102
                            print "]\n";
1103
                        }
1104
                    }
1105
                }
1106
1107
                // Merge the new styles with the inherited styles
1108
                $acceptedmedia = self::$ACCEPTED_GENERIC_MEDIA_TYPES;
1109
                $acceptedmedia[] = $this->_dompdf->getOptions()->getDefaultMediaType();
1110
                foreach ($applied_styles as $arr) {
1111
                    /** @var Style $s */
1112
                    foreach ($arr as $s) {
1113
                        $media_queries = $s->get_media_queries();
1114
                        foreach ($media_queries as $media_query) {
1115
                            list($media_query_feature, $media_query_value) = $media_query;
1116
                            // if any of the Style's media queries fail then do not apply the style
1117
                            //TODO: When the media query logic is fully developed we should not apply the Style when any of the media queries fail or are bad, per https://www.w3.org/TR/css3-mediaqueries/#error-handling
1118
                            if (in_array($media_query_feature, self::$VALID_MEDIA_TYPES)) {
1119
                                if ((strlen($media_query_feature) === 0 && !in_array($media_query, $acceptedmedia)) || (in_array($media_query, $acceptedmedia) && $media_query_value == "not")) {
1120
                                    continue (3);
1121
                                }
1122
                            } else {
1123
                                switch ($media_query_feature) {
1124
                                    case "height":
1125
                                        if ($paper_height !== (float)$style->length_in_pt($media_query_value)) {
1126
                                            continue (3);
1127
                                        }
1128
                                        break;
1129
                                    case "min-height":
1130
                                        if ($paper_height < (float)$style->length_in_pt($media_query_value)) {
1131
                                            continue (3);
1132
                                        }
1133
                                        break;
1134
                                    case "max-height":
1135
                                        if ($paper_height > (float)$style->length_in_pt($media_query_value)) {
1136
                                            continue (3);
1137
                                        }
1138
                                        break;
1139
                                    case "width":
1140
                                        if ($paper_width !== (float)$style->length_in_pt($media_query_value)) {
1141
                                            continue (3);
1142
                                        }
1143
                                        break;
1144
                                    case "min-width":
1145
                                        //if (min($paper_width, $media_query_width) === $paper_width) {
1146
                                        if ($paper_width < (float)$style->length_in_pt($media_query_value)) {
1147
                                            continue (3);
1148
                                        }
1149
                                        break;
1150
                                    case "max-width":
1151
                                        //if (max($paper_width, $media_query_width) === $paper_width) {
1152
                                        if ($paper_width > (float)$style->length_in_pt($media_query_value)) {
1153
                                            continue (3);
1154
                                        }
1155
                                        break;
1156
                                    case "orientation":
1157
                                        if ($paper_orientation !== $media_query_value) {
1158
                                            continue (3);
1159
                                        }
1160
                                        break;
1161
                                    default:
1162
                                        Helpers::record_warnings(E_USER_WARNING, "Unknown media query: $media_query_feature", __FILE__, __LINE__);
1163
                                        break;
1164
                                }
1165
                            }
1166
                        }
1167
1168
                        $style->merge($s);
1169
                    }
1170
                }
1171
            }
1172
1173
            // Inherit parent's styles if required
1174
            if ($p) {
1175
1176
                if ($DEBUGCSS) {
1177
                    print "inherit:\n";
1178
                    print "[\n";
1179
                    $p->get_style()->debug_print();
1180
                    print "]\n";
1181
                }
1182
1183
                $style->inherit($p->get_style());
1184
            }
1185
1186
            if ($DEBUGCSS) {
1187
                print "DomElementStyle:\n";
1188
                print "[\n";
1189
                $style->debug_print();
1190
                print "]\n";
1191
                print "/$debug_nodename]\n</pre>";
1192
            }
1193
1194
            /*DEBUGCSS print: see below different print debugging method
1195
            Helpers::pre_r($frame->get_node()->nodeName . ":");
1196
            echo "<pre>";
1197
            echo $style;
1198
            echo "</pre>";*/
1199
            $frame->set_style($style);
1200
1201
            if (!$root_flg && $this->_page_styles["base"]) {
1202
                $root_flg = true;
1203
1204
                // set the page width, height, and orientation based on the parsed page style
1205
                if ($style->size !== "auto") {
1206
                    list($paper_width, $paper_height) = $style->size;
1207
                }
1208
                $paper_width = $paper_width - (float)$style->length_in_pt($style->margin_left) - (float)$style->length_in_pt($style->margin_right);
1209
                $paper_height = $paper_height - (float)$style->length_in_pt($style->margin_top) - (float)$style->length_in_pt($style->margin_bottom);
1210
                $paper_orientation = ($paper_width > $paper_height ? "landscape" : "portrait");
1211
            }
1212
        }
1213
1214
        // We're done!  Clean out the registry of all styles since we
1215
        // won't be needing this later.
1216
        foreach (array_keys($this->_styles) as $key) {
1217
            $this->_styles[$key] = null;
1218
            unset($this->_styles[$key]);
1219
        }
1220
1221
    }
1222
1223
    /**
1224
     * parse a CSS string using a regex parser
1225
     * Called by {@link Stylesheet::parse_css()}
1226
     *
1227
     * @param string $str
1228
     *
1229
     * @throws Exception
1230
     */
1231
    private function _parse_css($str)
1232
    {
1233
1234
        $str = trim($str);
1235
1236
        // Destroy comments and remove HTML comments
1237
        $css = preg_replace(array(
1238
            "'/\*.*?\*/'si",
1239
            "/^<!--/",
1240
            "/-->$/"
1241
        ), "", $str);
1242
1243
        // FIXME: handle '{' within strings, e.g. [attr="string {}"]
1244
1245
        // Something more legible:
1246
        $re =
1247
            "/\s*                                   # Skip leading whitespace                             \n" .
1248
            "( @([^\s{]+)\s*([^{;]*) (?:;|({)) )?   # Match @rules followed by ';' or '{'                 \n" .
1249
            "(?(1)                                  # Only parse sub-sections if we're in an @rule...     \n" .
1250
            "  (?(4)                                # ...and if there was a leading '{'                   \n" .
1251
            "    \s*( (?:(?>[^{}]+) ({)?            # Parse rulesets and individual @page rules           \n" .
1252
            "            (?(6) (?>[^}]*) }) \s*)+?                                                        \n" .
1253
            "       )                                                                                     \n" .
1254
            "   })                                  # Balancing '}'                                       \n" .
1255
            "|                                      # Branch to match regular rules (not preceded by '@') \n" .
1256
            "([^{]*{[^}]*}))                        # Parse normal rulesets                               \n" .
1257
            "/xs";
1258
1259
        if (preg_match_all($re, $css, $matches, PREG_SET_ORDER) === false) {
1260
            // An error occurred
1261
            throw new Exception("Error parsing css file: preg_match_all() failed.");
1262
        }
1263
1264
        // After matching, the array indicies are set as follows:
1265
        //
1266
        // [0] => complete text of match
1267
        // [1] => contains '@import ...;' or '@media {' if applicable
1268
        // [2] => text following @ for cases where [1] is set
1269
        // [3] => media types or full text following '@import ...;'
1270
        // [4] => '{', if present
1271
        // [5] => rulesets within media rules
1272
        // [6] => '{', within media rules
1273
        // [7] => individual rules, outside of media rules
1274
        //
1275
1276
        $media_query_regex = "/(?:((only|not)?\s*(" . implode("|", self::$VALID_MEDIA_TYPES) . "))|(\s*\(\s*((?:(min|max)-)?([\w\-]+))\s*(?:\:\s*(.*?)\s*)?\)))/isx";
1277
1278
        //Helpers::pre_r($matches);
1279
        foreach ($matches as $match) {
1280
            $match[2] = trim($match[2]);
1281
1282
            if ($match[2] !== "") {
1283
                // Handle @rules
1284
                switch ($match[2]) {
1285
1286
                    case "import":
1287
                        $this->_parse_import($match[3]);
1288
                        break;
1289
1290
                    case "media":
1291
                        $acceptedmedia = self::$ACCEPTED_GENERIC_MEDIA_TYPES;
1292
                        $acceptedmedia[] = $this->_dompdf->getOptions()->getDefaultMediaType();
1293
1294
                        $media_queries = preg_split("/\s*,\s*/", mb_strtolower(trim($match[3])));
1295
                        foreach ($media_queries as $media_query) {
1296
                            if (in_array($media_query, $acceptedmedia)) {
1297
                                //if we have a media type match go ahead and parse the stylesheet
1298
                                $this->_parse_sections($match[5]);
1299
                                break;
1300
                            } elseif (!in_array($media_query, self::$VALID_MEDIA_TYPES)) {
1301
                                // otherwise conditionally parse the stylesheet assuming there are parseable media queries
1302
                                if (preg_match_all($media_query_regex, $media_query, $media_query_matches, PREG_SET_ORDER) !== false) {
1303
                                    $mq = array();
1304
                                    foreach ($media_query_matches as $media_query_match) {
1305
                                        if (empty($media_query_match[1]) === false) {
1306
                                            $media_query_feature = strtolower($media_query_match[3]);
1307
                                            $media_query_value = strtolower($media_query_match[2]);
1308
                                            $mq[] = array($media_query_feature, $media_query_value);
1309
                                        } else if (empty($media_query_match[4]) === false) {
1310
                                            $media_query_feature = strtolower($media_query_match[5]);
1311
                                            $media_query_value = (array_key_exists(8, $media_query_match) ? strtolower($media_query_match[8]) : null);
1312
                                            $mq[] = array($media_query_feature, $media_query_value);
1313
                                        }
1314
                                    }
1315
                                    $this->_parse_sections($match[5], $mq);
1316
                                    break;
1317
                                }
1318
                            }
1319
                        }
1320
                        break;
1321
1322
                    case "page":
1323
                        //This handles @page to be applied to page oriented media
1324
                        //Note: This has a reduced syntax:
1325
                        //@page { margin:1cm; color:blue; }
1326
                        //Not a sequence of styles like a full.css, but only the properties
1327
                        //of a single style, which is applied to the very first "root" frame before
1328
                        //processing other styles of the frame.
1329
                        //Working properties:
1330
                        // margin (for margin around edge of paper)
1331
                        // font-family (default font of pages)
1332
                        // color (default text color of pages)
1333
                        //Non working properties:
1334
                        // border
1335
                        // padding
1336
                        // background-color
1337
                        //Todo:Reason is unknown
1338
                        //Other properties (like further font or border attributes) not tested.
1339
                        //If a border or background color around each paper sheet is desired,
1340
                        //assign it to the <body> tag, possibly only for the css of the correct media type.
1341
1342
                        // If the page has a name, skip the style.
1343
                        $page_selector = trim($match[3]);
1344
1345
                        $key = null;
1346
                        switch ($page_selector) {
1347
                            case "":
1348
                                $key = "base";
1349
                                break;
1350
1351
                            case ":left":
1352
                            case ":right":
1353
                            case ":odd":
1354
                            case ":even":
1355
                            /** @noinspection PhpMissingBreakStatementInspection */
1356
                            case ":first":
1357
                                $key = $page_selector;
1358
1359
                            default:
1360
                                continue;
1361
                        }
1362
1363
                        // Store the style for later...
1364
                        if (empty($this->_page_styles[$key])) {
1365
                            $this->_page_styles[$key] = $this->_parse_properties($match[5]);
1366
                        } else {
1367
                            $this->_page_styles[$key]->merge($this->_parse_properties($match[5]));
1368
                        }
1369
                        break;
1370
1371
                    case "font-face":
1372
                        $this->_parse_font_face($match[5]);
1373
                        break;
1374
1375
                    default:
1376
                        // ignore everything else
1377
                        break;
1378
                }
1379
1380
                continue;
1381
            }
1382
1383
            if ($match[7] !== "") {
1384
                $this->_parse_sections($match[7]);
1385
            }
1386
1387
        }
1388
    }
1389
1390
    /**
1391
     * See also style.cls Style::_image(), refactoring?, works also for imported css files
1392
     *
1393
     * @param $val
1394
     * @return string
1395
     */
1396
    protected function _image($val)
1397
    {
1398
        $DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss();
1399
        $parsed_url = "none";
1400
1401
        if (mb_strpos($val, "url") === false) {
1402
            $path = "none"; //Don't resolve no image -> otherwise would prefix path and no longer recognize as none
1403
        } else {
1404
            $val = preg_replace("/url\(\s*['\"]?([^'\")]+)['\"]?\s*\)/", "\\1", trim($val));
1405
1406
            // Resolve the url now in the context of the current stylesheet
1407
            $parsed_url = Helpers::explode_url($val);
1408
            if ($parsed_url["protocol"] == "" && $this->get_protocol() == "") {
1409
                if ($parsed_url["path"][0] === '/' || $parsed_url["path"][0] === '\\') {
1410
                    $path = $_SERVER["DOCUMENT_ROOT"] . '/';
1411
                } else {
1412
                    $path = $this->get_base_path();
1413
                }
1414
1415
                $path .= $parsed_url["path"] . $parsed_url["file"];
1416
                $path = realpath($path);
1417
                // If realpath returns FALSE then specifically state that there is no background image
1418
                // FIXME: Is this causing problems for imported CSS files? There are some './none' references when running the test cases.
1419
                if (!$path) {
1420
                    $path = 'none';
1421
                }
1422
            } else {
1423
                $path = Helpers::build_url($this->get_protocol(),
1424
                    $this->get_host(),
1425
                    $this->get_base_path(),
1426
                    $val);
1427
            }
1428
        }
1429
1430
        if ($DEBUGCSS) {
1431
            print "<pre>[_image\n";
1432
            print_r($parsed_url);
1433
            print $this->get_protocol() . "\n" . $this->get_base_path() . "\n" . $path . "\n";
1434
            print "_image]</pre>";;
1435
        }
1436
1437
        return $path;
1438
    }
1439
1440
    /**
1441
     * parse @import{} sections
1442
     *
1443
     * @param string $url the url of the imported CSS file
1444
     */
1445
    private function _parse_import($url)
1446
    {
1447
        $arr = preg_split("/[\s\n,]/", $url, -1, PREG_SPLIT_NO_EMPTY);
1448
        $url = array_shift($arr);
1449
        $accept = false;
1450
1451
        if (count($arr) > 0) {
1452
            $acceptedmedia = self::$ACCEPTED_GENERIC_MEDIA_TYPES;
1453
            $acceptedmedia[] = $this->_dompdf->getOptions()->getDefaultMediaType();
1454
1455
            // @import url media_type [media_type...]
1456
            foreach ($arr as $type) {
1457
                if (in_array(mb_strtolower(trim($type)), $acceptedmedia)) {
1458
                    $accept = true;
1459
                    break;
1460
                }
1461
            }
1462
1463
        } else {
1464
            // unconditional import
1465
            $accept = true;
1466
        }
1467
1468
        if ($accept) {
1469
            // Store our current base url properties in case the new url is elsewhere
1470
            $protocol = $this->_protocol;
1471
            $host = $this->_base_host;
1472
            $path = $this->_base_path;
1473
1474
            // $url = str_replace(array('"',"url", "(", ")"), "", $url);
1475
            // If the protocol is php, assume that we will import using file://
1476
            // $url = Helpers::build_url($protocol == "php://" ? "file://" : $protocol, $host, $path, $url);
1477
            // Above does not work for subfolders and absolute urls.
1478
            // Todo: As above, do we need to replace php or file to an empty protocol for local files?
1479
1480
            $url = $this->_image($url);
1481
1482
            $this->load_css_file($url);
1483
1484
            // Restore the current base url
1485
            $this->_protocol = $protocol;
1486
            $this->_base_host = $host;
1487
            $this->_base_path = $path;
1488
        }
1489
1490
    }
1491
1492
    /**
1493
     * parse @font-face{} sections
1494
     * http://www.w3.org/TR/css3-fonts/#the-font-face-rule
1495
     *
1496
     * @param string $str CSS @font-face rules
1497
     */
1498
    private function _parse_font_face($str)
1499
    {
1500
        $descriptors = $this->_parse_properties($str);
1501
1502
        preg_match_all("/(url|local)\s*\([\"\']?([^\"\'\)]+)[\"\']?\)\s*(format\s*\([\"\']?([^\"\'\)]+)[\"\']?\))?/i", $descriptors->src, $src);
1503
1504
        $sources = array();
1505
        $valid_sources = array();
1506
1507
        foreach ($src[0] as $i => $value) {
1508
            $source = array(
1509
                "local" => strtolower($src[1][$i]) === "local",
1510
                "uri" => $src[2][$i],
1511
                "format" => strtolower($src[4][$i]),
1512
                "path" => Helpers::build_url($this->_protocol, $this->_base_host, $this->_base_path, $src[2][$i]),
1513
            );
1514
1515
            if (!$source["local"] && in_array($source["format"], array("", "truetype"))) {
1516
                $valid_sources[] = $source;
1517
            }
1518
1519
            $sources[] = $source;
1520
        }
1521
1522
        // No valid sources
1523
        if (empty($valid_sources)) {
1524
            return;
1525
        }
1526
1527
        $style = array(
1528
            "family" => $descriptors->get_font_family_raw(),
1529
            "weight" => $descriptors->font_weight,
1530
            "style" => $descriptors->font_style,
1531
        );
1532
1533
        $this->getFontMetrics()->registerFont($style, $valid_sources[0]["path"], $this->_dompdf->getHttpContext());
1534
    }
1535
1536
    /**
1537
     * parse regular CSS blocks
1538
     *
1539
     * _parse_properties() creates a new Style object based on the provided
1540
     * CSS rules.
1541
     *
1542
     * @param string $str CSS rules
1543
     * @return Style
1544
     */
1545
    private function _parse_properties($str)
1546
    {
1547
        $properties = preg_split("/;(?=(?:[^\(]*\([^\)]*\))*(?![^\)]*\)))/", $str);
1548
        $DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss();
1549
1550
        if ($DEBUGCSS) {
1551
            print '[_parse_properties';
1552
        }
1553
1554
        // Create the style
1555
        $style = new Style($this, Stylesheet::ORIG_AUTHOR);
1556
1557
        foreach ($properties as $prop) {
1558
            // If the $prop contains an url, the regex may be wrong
1559
            // @todo: fix the regex so that it works everytime
1560
            /*if (strpos($prop, "url(") === false) {
1561
              if (preg_match("/([a-z-]+)\s*:\s*[^:]+$/i", $prop, $m))
1562
                $prop = $m[0];
1563
            }*/
1564
            //A css property can have " ! important" appended (whitespace optional)
1565
            //strip this off to decode core of the property correctly.
1566
            //Pass on in the style to allow proper handling:
1567
            //!important properties can only be overridden by other !important ones.
1568
            //$style->$prop_name = is a shortcut of $style->__set($prop_name,$value);.
1569
            //If no specific set function available, set _props["prop_name"]
1570
            //style is always copied completely, or $_props handled separately
1571
            //Therefore set a _important_props["prop_name"]=true to indicate the modifier
1572
1573
            /* Instead of short code, prefer the typical case with fast code
1574
          $important = preg_match("/(.*?)!\s*important/",$prop,$match);
1575
            if ( $important ) {
1576
              $prop = $match[1];
1577
            }
1578
            $prop = trim($prop);
1579
            */
1580
            if ($DEBUGCSS) print '(';
1581
1582
            $important = false;
1583
            $prop = trim($prop);
1584
1585
            if (substr($prop, -9) === 'important') {
1586
                $prop_tmp = rtrim(substr($prop, 0, -9));
1587
1588
                if (substr($prop_tmp, -1) === '!') {
1589
                    $prop = rtrim(substr($prop_tmp, 0, -1));
1590
                    $important = true;
1591
                }
1592
            }
1593
1594
            if ($prop === "") {
1595
                if ($DEBUGCSS) print 'empty)';
1596
                continue;
1597
            }
1598
1599
            $i = mb_strpos($prop, ":");
1600
            if ($i === false) {
1601
                if ($DEBUGCSS) print 'novalue' . $prop . ')';
1602
                continue;
1603
            }
1604
1605
            $prop_name = rtrim(mb_strtolower(mb_substr($prop, 0, $i)));
1606
            $value = ltrim(mb_substr($prop, $i + 1));
1607
            if ($DEBUGCSS) print $prop_name . ':=' . $value . ($important ? '!IMPORTANT' : '') . ')';
1608
            //New style, anyway empty
1609
            //if ($important || !$style->important_get($prop_name) ) {
1610
            //$style->$prop_name = array($value,$important);
1611
            //assignment might be replaced by overloading through __set,
1612
            //and overloaded functions might check _important_props,
1613
            //therefore set _important_props first.
1614
            if ($important) {
1615
                $style->important_set($prop_name);
1616
            }
1617
            //For easier debugging, don't use overloading of assignments with __set
1618
            $style->$prop_name = $value;
1619
            //$style->props_set($prop_name, $value);
1620
        }
1621
        if ($DEBUGCSS) print '_parse_properties]';
1622
1623
        return $style;
1624
    }
1625
1626
    /**
1627
     * parse selector + rulesets
1628
     *
1629
     * @param string $str CSS selectors and rulesets
1630
     * @param array $media_queries
1631
     */
1632
    private function _parse_sections($str, $media_queries = array())
1633
    {
1634
        // Pre-process: collapse all whitespace and strip whitespace around '>',
1635
        // '.', ':', '+', '#'
1636
1637
        $patterns = array("/[\\s\n]+/", "/\\s+([>.:+#])\\s+/");
1638
        $replacements = array(" ", "\\1");
1639
        $str = preg_replace($patterns, $replacements, $str);
1640
        $DEBUGCSS = $this->_dompdf->getOptions()->getDebugCss();
1641
1642
        $sections = explode("}", $str);
1643
        if ($DEBUGCSS) print '[_parse_sections';
1644
        foreach ($sections as $sect) {
1645
            $i = mb_strpos($sect, "{");
1646
            if ($i === false) { continue; }
1647
1648
            //$selectors = explode(",", mb_substr($sect, 0, $i));
1649
            $selectors = preg_split("/,(?![^\(]*\))/", mb_substr($sect, 0, $i),0, PREG_SPLIT_NO_EMPTY);
1650
            if ($DEBUGCSS) print '[section';
1651
1652
            $style = $this->_parse_properties(trim(mb_substr($sect, $i + 1)));
1653
1654
            // Assign it to the selected elements
1655
            foreach ($selectors as $selector) {
1656
                $selector = trim($selector);
1657
1658
                if ($selector == "") {
1659
                    if ($DEBUGCSS) print '#empty#';
1660
                    continue;
1661
                }
1662
                if ($DEBUGCSS) print '#' . $selector . '#';
1663
                //if ($DEBUGCSS) { if (strpos($selector,'p') !== false) print '!!!p!!!#'; }
1664
1665
                //FIXME: tag the selector with a hash of the media query to separate it from non-conditional styles (?), xpath comments are probably not what we want to do here
1666
                if (count($media_queries) > 0) {
1667
                    $style->set_media_queries($media_queries);
1668
                }
1669
                $this->add_style($selector, $style);
1670
            }
1671
1672
            if ($DEBUGCSS) {
1673
                print 'section]';
1674
            }
1675
        }
1676
1677
        if ($DEBUGCSS) {
1678
            print '_parse_sections]';
1679
        }
1680
    }
1681
1682
    /**
1683
     * @return string
1684
     */
1685
    public static function getDefaultStylesheet()
1686
    {
1687
        $dir = realpath(__DIR__ . "/../..");
1688
        return $dir . self::DEFAULT_STYLESHEET;
1689
    }
1690
1691
    /**
1692
     * @param FontMetrics $fontMetrics
1693
     * @return $this
1694
     */
1695
    public function setFontMetrics(FontMetrics $fontMetrics)
1696
    {
1697
        $this->fontMetrics = $fontMetrics;
1698
        return $this;
1699
    }
1700
1701
    /**
1702
     * @return FontMetrics
1703
     */
1704
    public function getFontMetrics()
1705
    {
1706
        return $this->fontMetrics;
1707
    }
1708
1709
    /**
1710
     * dumps the entire stylesheet as a string
1711
     *
1712
     * Generates a string of each selector and associated style in the
1713
     * Stylesheet.  Useful for debugging.
1714
     *
1715
     * @return string
1716
     */
1717
    function __toString()
1718
    {
1719
        $str = "";
1720
        foreach ($this->_styles as $selector => $selector_styles) {
1721
            /** @var Style $style */
1722
            foreach ($selector_styles as $style) {
1723
                $str .= "$selector => " . $style->__toString() . "\n";
1724
            }
1725
        }
1726
1727
        return $str;
1728
    }
1729
}