:
$query .= "[contains(concat(' ', @$attr, ' '), concat(' ', '$tok', ' '))]";
$tok = "";
break;
case "+":
// All sibling elements that folow the current token
if ( mb_substr($query, -1, 1) != "/" )
$query .= "/";
$query .= "following-sibling::$tok";
$tok = "";
break;
case ":":
// Pseudo-classes
switch ($tok) {
case "first-child":
break;
case "link":
$query .= "[@href]";
$tok = "";
break;
case "first-line":
break;
case "first-letter":
break;
case "before":
break;
case "after":
break;
}
break;
case "[":
// Attribute selectors. All with an attribute matching the following token(s)
$attr_delimiters = array("=", "]", "~", "|");
$tok_len = mb_strlen($tok);
$j = 0;
$attr = "";
$op = "";
$value = "";
while ( $j < $tok_len ) {
if ( in_array($tok{$j}, $attr_delimiters) )
break;
$attr .= $tok{$j++};
}
switch ( $tok{$j} ) {
case "~":
case "|":
$op .= $tok{$j++};
if ( $tok{$j} != "=" )
throw new DOMPDF_Exception("Invalid CSS selector syntax: invalid attribute selector: $selector");
$op .= $tok{$j};
break;
case "=":
$op = "=";
break;
}
// Read the attribute value, if required
if ( $op != "" ) {
$j++;
while ( $j < $tok_len ) {
if ( $tok{$j} == "]" )
break;
$value .= $tok{$j++};
}
}
if ( $attr == "" )
throw new DOMPDF_Exception("Invalid CSS selector syntax: missing attribute name");
switch ( $op ) {
case "":
$query .= "[@$attr]";
break;
case "=":
$query .= "[@$attr$op\"$value\"]";
break;
case "~=":
// FIXME: this will break if $value contains quoted strings
// (e.g. [type~="a b c" "d e f"])
$values = explode(" ", $value);
$query .= "[";
foreach ( $values as $val )
$query .= "@$attr=\"$val\" or ";
$query = rtrim($query, " or ") . "]";
break;
case "|=":
$values = explode("-", $value);
$query .= "[";
foreach ($values as $val)
$query .= "starts-with(@$attr, \"$val\") or ";
$query = rtrim($query, " or ") . "]";
break;
}
break;
}
}
$i++;
// case ":":
// // Pseudo selectors: ignore for now. Partially handled directly
// // below.
// // Skip until the next special character, leaving the token as-is
// while ( $i < $len ) {
// if ( in_array($selector{$i}, $delimiters) )
// break;
// $i++;
// }
// break;
// default:
// // Add the character to the token
// $tok .= $selector{$i++};
// break;
// }
// }
// Trim the trailing '/' from the query
if ( mb_strlen($query) > 2 )
$query = rtrim($query, "/");
return $query;
}
/**
* applies all current styles to a particular document tree
*
* apply_styles() applies all currently loaded styles to the provided
* {@link Frame_Tree}. Aside from parsing CSS, this is the main purpose
* of this class.
*
* @param Frame_Tree $tree
*/
function apply_styles(Frame_Tree $tree) {
// Use XPath to select nodes. This would be easier if we could attach
// Frame objects directly to DOMNodes using the setUserData() method, but
// we can't do that just yet. Instead, we set a _node attribute_ in
// Frame->set_id() and use that as a handle on the Frame object via
// Frame_Tree::$_registry.
// We create a scratch array of styles indexed by frame id. Once all
// styles have been assigned, we order the cached styles by specificity
// and create a final style object to assign to the frame.
// FIXME: this is not particularly robust...
$styles = array();
$xp = new DOMXPath($tree->get_dom());
// Apply all styles in stylesheet
foreach ($this->_styles as $selector => $style) {
$query = $this->_css_selector_to_xpath($selector);
// pre_var_dump($selector);
// pre_var_dump($query);
// echo ($style);
// Retrieve the nodes
$nodes = $xp->query($query);
foreach ($nodes as $node) {
//echo $node->nodeName . "\n";
// Retrieve the node id
if ( $node->nodeType != 1 ) // Only DOMElements get styles
continue;
$id = $node->getAttribute("frame_id");
// Assign the current style to the scratch array
$spec = $this->_specificity($selector);
$styles[$id][$spec][] = $style;
}
}
// Now create the styles and assign them to the appropriate frames. (We
// iterate over the tree using an implicit Frame_Tree iterator.)
$root_flg = false;
foreach ($tree->get_frames() as $frame) {
// pre_r($frame->get_node()->nodeName . ":");
if ( !$root_flg && $this->_page_style ) {
$style = $this->_page_style;
$root_flg = true;
} else
$style = $this->create_style();
// Find nearest DOMElement parent
$p = $frame;
while ( $p = $p->get_parent() )
if ($p->get_node()->nodeType == 1 )
break;
// Styles can only be applied directly to DOMElements; anonymous
// frames inherit from their parent
if ( $frame->get_node()->nodeType != 1 ) {
if ( $p )
$style->inherit($p->get_style());
$frame->set_style($style);
continue;
}
$id = $frame->get_id();
// Handle HTML 4.0 attributes
Attribute_Translator::translate_attributes($frame);
// Locate any additional style attributes
if ( ($str = $frame->get_node()->getAttribute("style")) !== "" ) {
$spec = $this->_specificity("!style attribute");
$styles[$id][$spec][] = $this->_parse_properties($str);
}
// Grab the applicable styles
if ( isset($styles[$id]) ) {
$applied_styles = $styles[ $frame->get_id() ];
// Sort by specificity
ksort($applied_styles);
// Merge the new styles with the inherited styles
foreach ($applied_styles as $arr) {
foreach ($arr as $s)
$style->merge($s);
}
}
// Inherit parent's styles if required
if ( $p ) {
$style->inherit( $p->get_style() );
}
// pre_r($frame->get_node()->nodeName . ":");
// echo "";
// echo $style;
// echo "
";
$frame->set_style($style);
}
// We're done! Clean out the registry of all styles since we
// won't be needing this later.
foreach ( array_keys($this->_styles) as $key ) {
unset($this->_styles[$key]);
}
}
/**
* parse a CSS string using a regex parser
*
* Called by {@link Stylesheet::parse_css()}
*
* @param string $str
*/
private function _parse_css($str) {
// Destroy comments
$css = preg_replace("'/\*.*?\*/'si", "", $str);
// FIXME: handle '{' within strings, e.g. [attr="string {}"]
// Something more legible:
$re =
"/\s* # Skip leading whitespace \n".
"( @([^\s]+)\s+([^{;]*) (?:;|({)) )? # Match @rules followed by ';' or '{' \n".
"(?(1) # Only parse sub-sections if we're in an @rule... \n".
" (?(4) # ...and if there was a leading '{' \n".
" \s*( (?:(?>[^{}]+) ({)? # Parse rulesets and individual @page rules \n".
" (?(6) (?>[^}]*) }) \s*)+? \n".
" ) \n".
" }) # Balancing '}' \n".
"| # Branch to match regular rules (not preceeded by '@')\n".
"([^{]*{[^}]*})) # Parse normal rulesets\n".
"/xs";
if ( preg_match_all($re, $css, $matches, PREG_SET_ORDER) === false )
// An error occured
throw new DOMPDF_Exception("Error parsing css file: preg_match_all() failed.");
// After matching, the array indicies are set as follows:
//
// [0] => complete text of match
// [1] => contains '@import ...;' or '@media {' if applicable
// [2] => text following @ for cases where [1] is set
// [3] => media types or full text following '@import ...;'
// [4] => '{', if present
// [5] => rulesets within media rules
// [6] => '{', within media rules
// [7] => individual rules, outside of media rules
//
//pre_r($matches);
foreach ( $matches as $match ) {
$match[2] = trim($match[2]);
if ( $match[2] !== "" ) {
// Handle @rules
switch ($match[2]) {
case "import":
$this->_parse_import($match[3]);
break;
case "media":
if ( in_array(mb_strtolower(trim($match[3])), self::$ACCEPTED_MEDIA_TYPES ) ) {
$this->_parse_sections($match[5]);
}
break;
case "page":
// Store the style for later...
if ( is_null($this->_page_style) )
$this->_page_style = $this->_parse_properties($match[5]);
else
$this->_page_style->merge($this->_parse_properties($match[5]));
break;
default:
// ignore everything else
break;
}
continue;
}
if ( $match[7] !== "" )
$this->_parse_sections($match[7]);
}
}
/**
* parse @import{} sections
*
* @param string $url the url of the imported CSS file
*/
private function _parse_import($url) {
$arr = preg_split("/[\s\n]/", $url);
$url = array_pop($arr);
$accept = false;
if ( count($arr) > 0 ) {
// @import url media_type [media_type...]
foreach ( $arr as $type ) {
if ( in_array($type, self::$ACCEPTED_MEDIA_TYPES) ) {
$accept = true;
break;
}
}
} else
// unconditional import
$accept = true;
if ( $accept ) {
$url = str_replace(array('"',"url", "(", ")"), "", $url);
// Store our current base url properties in case the new url is elsewhere
$protocol = $this->_protocol;
$host = $this->_base_host;
$path = $this->_base_path;
// If the protocol is php, assume that we will import using file://
$url = build_url($protocol == "php://" ? "file://" : $protocol, $host, $path, $url);
$this->load_css_file($url);
// Restore the current base url
$this->_protocol = $protocol;
$this->_base_host = $host;
$this->_base_path = $path;
}
}
/**
* parse regular CSS blocks
*
* _parse_properties() creates a new Style object based on the provided
* CSS rules.
*
* @param string $str CSS rules
* @return Style
*/
private function _parse_properties($str) {
$properties = explode(";", $str);
// Create the style
$style = new Style($this);
foreach ($properties as $prop) {
$prop = trim($prop);
if ($prop == "")
continue;
$i = mb_strpos($prop, ":");
if ( $i === false )
continue;
$prop_name = mb_strtolower(mb_substr($prop, 0, $i));
$value = mb_substr($prop, $i+1);
$style->$prop_name = $value;
}
return $style;
}
/**
* parse selector + rulesets
*
* @param string $str CSS selectors and rulesets
*/
private function _parse_sections($str) {
// Pre-process: collapse all whitespace and strip whitespace around '>',
// '.', ':', '+', '#'
$patterns = array("/[\\s\n]+/", "/\\s+([>.:+#])\\s+/");
$replacements = array(" ", "\\1");
$str = preg_replace($patterns, $replacements, $str);
$sections = explode("}", $str);
foreach ($sections as $sect) {
$i = mb_strpos($sect, "{");
$selectors = explode(",", mb_substr($sect, 0, $i));
$style = $this->_parse_properties(trim(mb_substr($sect, $i+1)));
// Assign it to the selected elements
foreach ($selectors as $selector) {
$selector = trim($selector);
if ($selector == "")
continue;
$this->add_style($selector, $style);
}
}
}
/**
* dumps the entire stylesheet as a string
*
* Generates a string of each selector and associated style in the
* Stylesheet. Useful for debugging.
*
* @return string
*/
function __toString() {
$str = "";
foreach ($this->_styles as $selector => $style)
$str .= "$selector => " . $style->__toString() . "\n";
return $str;
}
}
?>