. ** ** it_xml.class - XML parser / object factory */ class it_xml { /* * Add xml snippet as attribute tree to $this * @param $xmldata XML string or filehandle (result from fopen) to parse * @param $p associative array * @param $p['forcearray'] xml tags to ALWAYS return as array * @param $p['safety'] 2 causes program abort with invalid xml, 1 (default) causes error report, 0 just returns false * @param $p['encoding'] Output character encoding (utf-8, iso-8859-1 or us-ascii, default: ini_get('default_charset') * @param $p['prefix'] Optional prefix for class names * @param $p['lowercase'] Lowercase all tag and attribute names * @return XML object tree or null on failure */ function __construct($xmldata = "", $p = array()) { if ($xmldata) $this->from_xml($xmldata, $p); } /** * Factory method to return XML object tree from XML string. Params like constructor * WARNING: Returned value will not contain the $p parameters in $this->_p so writing may cause problems * Example: $root = it_xml::create("", array("b")); */ static function create($xmldata, $p = array()) { $xml = new it_xml; return $xml->from_xml($xmldata, array('factory' => true) + $p) ? $xml->_root : null; } function from_xml($xmldata, $p) { $this->_p = $p + array('encoding' => ini_get('default_charset'), 'safety' => 1); $this->_arrayforce = array_flip((array)$this->_p['forcearray']); $this->_stack = array(); unset($this->error); $parser = xml_parser_create(); xml_set_object($parser, $this); xml_set_element_handler($parser, "start_element", "end_element"); xml_set_character_data_handler($parser, "character_data"); xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0); xml_parser_set_option($parser, XML_OPTION_TARGET_ENCODING, $this->_p['encoding']); $result = true; if (is_resource($xmldata)) { while ($result && !feof($xmldata)) { $data = fread($xmldata, 1024 * 1024); while (!feof($xmldata) && preg_match('/[\x80-\xff]$/', $data)) # Make sure end of chunk is not in the middle of a UTF8 character $data .= fread($xmldata, 1); $xmlorig = $data; # Do not append to keep memory footprint down! list($data, $isutf8) = $this->_sanitize($data, $isutf8); $result = xml_parse($parser, $data); } if ($result) $result = xml_parse($parser, "", true); } else { $xmlorig = $xmldata; list($xmldata, $isutf8) = $this->_sanitize($xmldata); $result = xml_parse($parser, $xmldata, true); } if (!$result) $this->error .= sprintf("it_xml error: %s at line %d", xml_error_string(xml_get_error_code($parser)), xml_get_current_line_number($parser)); if ($this->error) { if ($this->_p['safety'] >= 2) it::fatal(array('title' => $this->error, 'body' => $xmlorig)); else if ($this->_p['safety'] >= 1) it::error(array('title' => $this->error, 'body' => $xmlorig)); if ($this->_p['factory']) $GLOBALS['IT_XML_ERROR'] = $this->error; } unset($this->_arrayforce, $this->_p['safety'], $this->_p['factory'], $this->_stack); xml_parser_free($parser); return empty($this->error); } # Use various heuristics to fix real-world XML files function _sanitize($xmldata, $isutf8 = null) { if (!isset($isutf8)) # Check if we already decided on charset yet { $xmldata = ltrim($xmldata); # Add header for charset detection (PHP5) if no header/BOM # See http://www.w3.org/TR/2006/REC-xml-20060816/#sec-guessing if (!preg_match('/^(<\?xml|\xEF\xBB\xBF|\xFE\xFF|\xFF\xFE|\x00\x00\xFE\xFF|\x00\x00\xFF\xFE)/', $xmldata)) $xmldata = '_p['encoding'] . '"?>' . $xmldata; $isutf8 = (!preg_match('/^<\?xml[^>]* encoding=/i', $xmldata) || preg_match('/^<\?xml[^>]* encoding=.utf-?8/i', $xmldata)); } # Decode illegal entities but protect semantically important ones $xmldata = html_entity_decode(preg_replace('/&(amp|lt|gt|#38|#60|#62|#x26|#x3C|#x3E);/i', '&$1;', $xmldata), ENT_NOQUOTES, $this->_p['encoding']); # If should be utf-8 and can't be decoded as such, fix it, even if mixed between both if ($isutf8 && preg_match('/[^\x80-\xff][\x80-\xff][^\x80-\xff]/', $xmldata)) $xmldata = preg_replace_callback('/[\x80-\xff]{1,4}/', function($m) { return it_xml::_utf8_fix($m[0]); }, $xmldata); # If not utf-8, remove characters illegal for latin-1 if (!$isutf8 && preg_match('/[\x00-\x08\x0b-\x0c\x0e-\x1f\x80-\x9f]/', $xmldata)) $xmldata = it_html::_cleanup($xmldata, $isutf8 ? "utf-8" : "iso-8859-1"); return array($xmldata, $isutf8); } # Encode non-utf8 characters in a string, leave utf8 alone static function _utf8_fix($str) { return preg_match('/^([\xc0-\xdf][\x80-\xbf]|[\xe0-\xef][\x80-\xbf][\x80-\xbf]|[\xf0-\xf7][\x80-\xbf][\x80-\xbf][\x80-\xbf])$/', $str) ? $str : utf8_encode($str); } function consume(/* $p */) { return false; } /** * Utility function which recursively flattens container into array * @param with_attr if true, include attributes in result array * @return Array with key/values-pairs for vals and optionally attrs of sub-objects */ function flatten($with_attr = false) { $result = $with_attr ? (array)$this->attr : array(); foreach ($this->elements() as $key => $values) { if (is_array($values)) { $result[$key] = array(); foreach ($values as $value) $result[$key][] = isset($value->val) || !get_object_vars($value) ? $value->val : $value->flatten($with_attr); } else $result[$key] = isset($values->val) || !get_object_vars($values) ? $values->val : $values->flatten($with_attr); # don' take recursion if val is set or object is empty } return $result; } # Strip namespace and convert extra characters to _ function _make_identifier($name) { $id = preg_replace(array('/^.*:/', '/\W/'), array('', '_'), $name); return $this->_p['lowercase'] ? strtolower($id) : $id; } function start_element($dummy_parser, $name, $attrs) { $name = $this->_p['prefix'] . $this->_make_identifier($name); if (!class_exists($name)) eval("class $name extends it_xml {}"); # Extending the base class caused problems with tel_xmlentry if (!$this->_stack && !$this->_p['factory']) array_unshift($this->_stack, $this); else array_unshift($this->_stack, new $name); if (!is_a($this->_stack[0], "it_xml")) $this->error .= "Class $name used in it_xml for tag $name does not extend it_xml: Name clash?\n"; if ($attrs) { foreach ($attrs as $key => $val) $this->_stack[0]->attr[$this->_make_identifier($key)] = $val; } } function end_element($dummy_parser, $name) { $name = $this->_make_identifier($name); if (!$this->_stack[0]->consume($this->_p)) { if (is_array($this->_stack[1]->$name)) array_push($this->_stack[1]->$name, $this->_stack[0]); else if (isset($this->_stack[1]->$name)) { if (is_a($this->_stack[1]->$name, "it_xml")) $this->_stack[1]->$name = array($this->_stack[1]->$name, &$this->_stack[0]); else $this->error .= "Variable $name used in it_xml for tag {$name} is already used in object: Name clash?\n"; } else if (isset($this->_arrayforce[$name])) $this->_stack[1]->$name = array(&$this->_stack[0]); else $this->_stack[1]->$name =& $this->_stack[0]; } $node = array_shift($this->_stack); if ($this->_p['factory']) $this->_root = $node; } function character_data($dummy_parser, $cdata) { if (isset($this->_stack[0]->val) || (trim($cdata) !== "")) $this->_stack[0]->val .= $cdata; } /** * Convert XML object tree back to string representation * @param $p['tag'] Name of root XML tag (otherwise determined by get_class) * @param $p['indent'] String to indent with (default: " ") * @param $p['lineseparator'] String to start new line with (default: "\n") * @param $p['prefix'] String to prefix class names with (default: "") * @param $p['sort'] value true causes to_xml() to output elements sorted * @return XML string representation of this object (only attributes with is_a == it_xml and no leading _) */ function to_xml($p = array()) { $p += array( 'tag' => substr(get_class($this), strlen($p['prefix'])), 'indent' => " ", 'lineseparator' => "\n", 'prefix' => "", 'sort' => false, 'stripempty' => false, # remove xml tags without content ); $currenttag = $p['tag']; $tag = $p['nextindentation'] . "<$currenttag"; if (empty($p['nextindentation'])) $p['nextindentation'] = $p['lineseparator']; $currentindentation = $p['nextindentation']; $p['nextindentation'] .= $p['indent']; foreach ((array)$this->attr as $key => $value) $tag .= " $key=\"" . htmlspecialchars($value, ENT_COMPAT, ini_get('default_charset')) . '"'; $vars = get_object_vars($this); if ($p['sort']) ksort($vars); foreach ($vars as $key => $values) { $p['tag'] = $key; # Manually set as PHP4 lowercases get_class() if (!preg_match('/^(_|attr$)/', $key)) # Skip implementation detail fields { if (is_array($values)) { foreach ($values as $value) { if (is_a($value, "it_xml")) { $body .= $value->to_xml($p); $indentation = $currentindentation; } } } else if (is_object($values)) { if (is_a($values, "it_xml")) { $body .= $values->to_xml($p); $indentation = $currentindentation; } } else if ($key == 'val') $body .= htmlspecialchars($values, ENT_COMPAT, ini_get('default_charset')); } } return isset($body) ? "$tag>$body$indentation" : ($p['stripempty'] ? "" : "$tag/>"); } /** * Get array containing child elements (key value pairs) * @return Array containing child elements */ function elements() { $result = array(); foreach (get_object_vars($this) as $key => $value) { if (is_a($value, "it_xml") || (is_array($value) && is_a($value[0], "it_xml"))) $result[$key] =& $this->$key; } return $result; } function error() { return is_a($this, "it_xml") ? $this->error : $GLOBALS['IT_XML_ERROR']; } /** * Set nodes as subnode of current node * @param values assoc array of key => value pairs. May contain associative or numeric subarrays. */ function set($vals, $parentprefix = null) { $prefix = $parentprefix ? $parentprefix : $this->_p['prefix']; if (is_array($vals)) { foreach ($vals as $key => $val) { $classname = $prefix . $this->_make_identifier($key); if (!class_exists($classname)) eval("class $classname extends it_xml {}"); if (is_array($val) && $val === array_values($val)) { $arr = (array)$this->$key; foreach ($val as $i => $v) { if (!is_a($arr[$i], "it_xml")) $arr[$i] = new $classname; $arr[$i]->set($v, $prefix); } $this->$key = $arr; } else { if (!is_a($this->$key, "it_xml")) $this->$key = new $classname; $this->$key->set($val, $prefix); } } } else $this->val = $vals; } function __toString() { return (string)$this->val; } } ?>