/**
*
* get tag and attrs
* @param string $value
*/
public function getTagData($value)
{
$mixed = $this->getInstance("Fl_Html_TagToken", $value)->run();
$attrs = $mixed['attrs'];
$data = array($this->options['tag'] => $mixed['tag']);
if (!empty($attrs)) {
foreach ($attrs as $aItem) {
$count = count($aItem);
if ($count === 1) {
$attrsData[$aItem[0]] = "";
} elseif ($count === 3) {
if ($this->options['attrs_remove_quote']) {
$valueDetail = Fl_Html_Static::getUnquoteText($aItem[2]);
$attrsData[$aItem[0]] = $valueDetail['text'];
} else {
$attrsData[$aItem[0]] = $aItem[2];
}
}
}
$data[$this->options['attrs']] = $attrsData;
}
return $data;
}
/**
*
* @param array $token
*/
public function checkHtmlTokens()
{
$tokens = $this->getTokens('html');
foreach ($tokens as $item) {
if ($item['type'] === FL_TOKEN_HTML_TAG_START) {
$attrTokens = $this->getInstance("Fl_Html_TagToken", $item['value'])->run();
$tagName = strtolower($attrTokens['tag']);
$tag = '<' . $attrTokens['tag'] . FL_SPACE;
$attrTokens = $attrTokens['attrs'];
foreach ($attrTokens as $attrItem) {
$count = count($attrItem);
$attr = strtolower($attrItem[0]);
if ($count == 1) {
$tag .= $this->checkIt(array_merge($item, array('value' => $attrItem[0])), 'html') . FL_SPACE;
} elseif ($count === 3) {
if ($attr && strpos($attr, 'on') === 0) {
$type = 'event';
} elseif ($attr === 'src' || $attr === 'href' || $tagName === 'form' && $attr === 'action') {
$type = 'url';
} else {
$type = 'html';
}
$value = $this->checkIt(array_merge($item, array('value' => $attrItem[2])), $type);
$tag .= $attrItem[0] . '=' . $value . FL_SPACE;
}
}
$tag = trim($tag) . ">";
$item['value'] = $tag;
} else {
if ($item['type'] === FL_TOKEN_HTML_SCRIPT_TAG) {
//这里要判断是否是前端模版
$detail = Fl_Html_Static::splitSpecialValue($item['value'], 'script', $this);
$tagInfo = Fl_Html_Static::getScriptTagInfo($detail['tag_start'], $this);
//前端模版用HTML转义
if ($tagInfo['tpl']) {
$item['value'] = $this->checkIt($item, 'html');
} else {
$item['value'] = $this->checkIt($item, 'js');
}
} else {
$item['value'] = $this->checkIt($item, 'html');
}
}
$this->addOutput($item);
}
}
/**
*
* beautify
* @param array $ast
*/
public function beautifyAst($ast, $parentType = '')
{
$result = '';
$first = true;
foreach ($ast as $item) {
if (!$first) {
$result = rtrim($result, FL_NEWLINE) . FL_NEWLINE;
}
$hasComment = false;
if (count($item['value']['commentBefore'])) {
$hasComment = true;
}
$comment = $this->beautifyComment($item['value']);
if ($comment['remove']) {
$result = rtrim($result, FL_NEWLINE);
}
$result .= $comment['value'];
//
if ($item['type'] === FL_TOKEN_HTML_TAG_END) {
continue;
}
if ($first) {
$first = false;
}
$indent = $newline = false;
if ($item['type'] === FL_TOKEN_HTML_TAG) {
$count = count($item['children']);
if ($count > 1) {
$indent = true;
$newline = true;
} elseif ($count === 1) {
$c = $item['children'][$count - 1];
if ($c['type'] !== FL_TOKEN_HTML_TEXT) {
$indent = true;
$newline = true;
}
}
} else {
if (count($ast) > 1) {
$newline = true;
}
if ($item['type'] === FL_TOKEN_HTML_DOCTYPE || $item['type'] === FL_TOKEN_HTML_SINGLE_TAG) {
$newline = true;
}
if ($item['type'] === FL_TOKEN_HTML_STYLE_TAG || $item['type'] === FL_TOKEN_HTML_SCRIPT_TAG) {
$newline = true;
$indent = true;
}
}
if ($item['type'] !== FL_TOKEN_HTML_TEXT) {
$result .= $this->getIndentString();
} else {
if (count($ast) > 1) {
$result .= $this->getIndentString();
}
}
if ($parentType === FL_TOKEN_HTML_STYLE_TAG) {
Fl::loadClass("Fl_Css_Static");
$value = Fl_Css_Static::getStyleDetail($item['value']['value']);
if ($value['prefix']) {
$result .= $value['prefix'] . FL_NEWLINE;
}
$result .= $this->beautify_special($value['value'], 'css');
if ($value['suffix']) {
$result .= FL_NEWLINE . $value['suffix'] . FL_NEWLINE;
}
} elseif ($parentType === FL_TOKEN_HTML_SCRIPT_TAG) {
$result .= $this->beautify_special($item['value']['value'], 'js');
} else {
$result .= $item['value']['value'];
}
if ($newline) {
$result .= FL_NEWLINE;
}
if ($indent) {
$this->indent++;
}
if (count($item['children'])) {
$type = $item['type'];
if ($type === FL_TOKEN_HTML_SCRIPT_TAG) {
$tagInfo = Fl_Html_Static::getScriptTagInfo($item['value']['value'], $this);
//虽然是script标签，但不一定是Js
if (!$tagInfo['script'] || $tagInfo['external']) {
$type = '';
}
}
$children = $this->beautifyAst($item['children'], $type);
if ($tagInfo['external'] && !strlen(trim($children))) {
$result = rtrim($result, FL_NEWLINE);
$newline = false;
$indent = 2;
} else {
$result .= $children;
}
}
$types = array(FL_TOKEN_HTML_TAG => 1, FL_TOKEN_HTML_SCRIPT_TAG => 1, FL_TOKEN_HTML_PRE_TAG => 1, FL_TOKEN_HTML_STYLE_TAG => 1, FL_TOKEN_HTML_TEXTAREA_TAG => 1);
$this->preToken = $item['value'];
if (isset($types[$item['type']])) {
if ($newline) {
$result = rtrim($result, FL_NEWLINE);
$result .= FL_NEWLINE;
}
if ($indent) {
$this->indent--;
if ($indent === true) {
$result .= $this->getIndentString();
}
}
if (!empty($item['end'])) {
$this->preToken = $item['end'];
$comment = $this->beautifyComment($item['end']);
$result .= rtrim($comment['value'], FL_NEWLINE);
}
$result .= '</' . $item['tag'] . '>';
}
}
return $result;
}
/**
*
* 判断当前的text是否可删除
* @param string $text
* @param array $nextToken
*/
public function textCanRemove($text, $preToken = array(), $nextToken = array())
{
if ($this->options['removeBlockBlank'] && preg_match('/^\\s+$/', $text)) {
$pregTag = $pregToken['lowerTag'];
$nextTag = $nextToken['lowerTag'];
if ($pregTag && Fl_Html_Static::isBlockTag($pregTag, $this->options['blockBlankList'])) {
return true;
}
if ($nextTag && Fl_Html_Static::isBlockTag($nextTag, $this->options['blockBlankList'])) {
return true;
}
}
return false;
}
/**
*
* 获取文本节点
*/
public function getTextToken($char)
{
/*if (! isset ( $this->text {$this->pos + 1} )) {
return false;
}*/
$next = $this->text[$this->pos + 1];
$renext = $this->text[$this->pos + 2];
/*
* return when next token is tpl
*/
if ($this->hasTplToken && $this->ld === substr($this->text, $this->pos + 1, strlen($this->ld))) {
return false;
}
/*
* 如果下一个字符是“<”, 并且下下个字符不是<
* 需要兼容<div>welefen<<</div>这样的情况
* Chrome下： 对于<div>welefen< 会被解析成<div>welefen, 最后的<字符会被忽略
*/
if ($next === Fl_Html_Static::LEFT && $renext !== Fl_Html_Static::LEFT && Fl_Html_Static::isTagFirstChar($renext)) {
return false;
}
}
/**
*
* 获取文本节点
*/
public function getTextToken($char)
{
$next = $this->getPosChar($this->pos + 1);
$renext = $this->getPosChar($this->pos + 2);
if ($next === false) {
return false;
}
/*
* 如果下一个是模板语法的token，则返回
* 这里用find有严重的性能问题，故直接使用substr比较
*/
if ($this->ld && $this->ld === substr($this->text, $this->pos + 1, strlen($this->ld))) {
return false;
}
/*
* 如果下一个字符是“<”, 并且下下个字符不是<
* 需要兼容<div>welefen<<</div>这样的情况
* Chrome下： 对于<div>welefen< 会被解析成<div>welefen, 最后的<字符会被忽略
*/
if ($next === Fl_Html_Static::LEFT && $renext !== Fl_Html_Static::LEFT && Fl_Html_Static::isTagFirstChar($renext)) {
return false;
}
}
/**
*
* 压缩闭合标签
* @param array $token
*/
public function compressEndTag($token)
{
if ($this->options['remove_optional_end_tag']) {
if (Fl_Html_Static::isOptionalEndTag($token['lowerTag'], $this->options['remove_optional_end_tag_list'])) {
return '';
}
}
$tag = $this->options['tag_to_lower'] ? $token['lowerTag'] : $token['tag'];
return '</' . $tag . '>';
}
/**
*
* 过滤开始标签
* @param array $token
*/
public function filterTag($token, $notCheckECss = false)
{
$value = $token['value'];
if (!empty($value)) {
$instance = $this->getInstance('Fl_Html_TagToken', $value);
$result = $instance->run();
} else {
$result = $token;
}
$tag = strtolower($result['tag']);
//外链的css
if (!$notCheckECss && $this->isExternalCss($result)) {
return $this->filterExternalCss($result);
}
if ($this->options['use_blank_tag_filter']) {
if (!in_array($tag, $this->blankTagList)) {
return false;
}
}
$attrs = $result['attrs'];
$attrResult = array();
foreach ($attrs as $item) {
$name = strtolower($item[0]);
//过滤事件
if ($this->options['remove_tag_event']) {
if (strpos($name, 'on') === 0) {
continue;
}
}
//标签属性白名单
if ($this->options['use_blank_tag_property_filter']) {
if (!in_array($name, $this->blankTagPropertyList)) {
//标签的特殊属性过滤
if ($this->options['use_special_tag_filter']) {
if (isset($this->allowSpecialTagProperty[$tag])) {
$propList = $this->allowSpecialTagProperty[$tag];
if (isset($propList[$name])) {
$value = $propList[$name];
$values = Fl_Html_Static::getUnquoteText($item[2]);
//正则
if (substr($value, 0, 1) === '/') {
if (preg_match($value, $values['text'])) {
$attrResult[] = $item;
}
} else {
if ($value === $values['text']) {
$attrResult[] = $item;
}
}
}
}
}
continue;
}
}
//a链接修复和过滤
if ($tag == 'a' && $this->options['filter_a_href_value'] && $name == 'href') {
if (count($item) == 3 && $item[1] == '=') {
$values = Fl_Html_Static::getUnquoteText($item[2]);
$url = Fl_Static::getFixedUrl($values['text'], $this->url);
if ($this->options['url_max_length']) {
$url = substr($url, 0, $this->options['url_max_length']);
}
$item[2] = $values['quote'] . $url . $values['quote'];
} else {
continue;
}
}
//图片连接的修复和过滤
if ($tag == 'img' && $this->options['filter_img_src_value'] && $name == 'src') {
if (count($item) == 3 && $item[1] == '=') {
$values = Fl_Html_Static::getUnquoteText($item[2]);
$url = Fl_Static::getFixedUrl($values['text'], $this->url);
if ($this->options['url_max_length']) {
$url = substr($url, 0, $this->options['url_max_length']);
}
$item[2] = $values['quote'] . $url . $values['quote'];
} else {
continue;
}
}
//style value
if ($this->options['filter_tag_style_value'] && $name == 'style') {
if (count($item) == 3 && $item[1] == '=') {
$values = Fl_Html_Static::getUnquoteText($item[2]);
$text = 'a{' . $values['text'] . '}';
$instance = $this->getInstance("Fl_Css_Filter", $text);
$instance->url = $this->url;
$instance->getResourceContentFn = $this->getResourceContentFn;
$text = $instance->run($this->options);
$item[2] = $values['quote'] . substr($text, 2, strlen($text) - 3) . $values['quote'];
} else {
continue;
}
}
$attrResult[] = $item;
}
$attrsJoin = array();
foreach ($attrResult as $item) {
$attrsJoin[] = join("", $item);
}
if (empty($attrsJoin)) {
return '<' . $tag . '>';
}
return '<' . $tag . ' ' . join(" ", $attrsJoin) . ">";
}
/**
*
* 获取特定属性的值
* @param array $attrs
* @param string $name
*/
public static function getAttrValue($attrs, $name = '')
{
foreach ($attrs as $item) {
if (count($item) === 3) {
if (strtolower($item[0]) === $name) {
$value = $item[2];
$valueInfo = Fl_Html_Static::getUnquoteText($value);
return $valueInfo['text'];
}
}
}
return false;
}
/**
*
* special
*/
public function specialStatement()
{
$tag = strtolower(Fl_Html_Static::getTagName($this->currentToken['value'], $this));
$special = Fl_Html_Static::splitSpecialValue($this->currentToken['value'], $tag, $this);
$this->currentToken['value'] = $special['tag_start'];
return array("type" => $this->currentToken['type'], "tag" => $tag, "value" => $this->getValue($this->currentToken), "children" => array(array("type" => FL_TOKEN_HTML_TEXT, "value" => $this->getValue(array_merge($this->currentToken, array("value" => $special["content"]))))));
}
