<?php
//---------------------------------------
// Html 標(biāo)記WAP語言
//----------------------------------------
function html2wml($content)
{
  //保留圖片
  preg_match_all("/<img([^>]*)>/isU", $content, $imgarr);
  if(isset($imgarr[0]) && count($imgarr[0])>0 )
  {
   foreach($imgarr[0] as $k=>$v) $content = str_replace($v, "WAP-IMG::{$k}", $content);
  }
  // 過濾掉樣式表和腳本
  $content = preg_replace("/<style .*?<\\/style>/is", "", $content);
  $content = preg_replace("/<script .*?<\\/script>/is", "", $content);
  // 首先將各種可以引起換行的標(biāo)簽（如<br />、<p> 之類）替換成換行符"\\n"
  $content = preg_replace("/<br \\s*\\/?\\/>/i", "\\n", $content);
  $content = preg_replace("/<\\/?p>/i", "\\n", $content);
  $content = preg_replace("/<\\/?td>/i", "\\n", $content);
  $content = preg_replace("/<\\/?div>/i", "\\n", $content);
  $content = preg_replace("/<\\/?blockquote>/i", "\\n", $content);
  $content = preg_replace("/<\\/?li>/i", "\\n", $content);
  // 將"&nbsp;"替換為空格
  $content = preg_replace("/\\&nbsp\\;/i", " ", $content);
  $content = preg_replace("/\\&nbsp/i", " ", $content);
  // 過濾掉剩下的 HTML 標(biāo)簽
  $content = strip_tags($content);
  // 將 HTML 中的實體（entity）轉(zhuǎn)化為它所對應(yīng)的字符
  $content = html_entity_decode($content, ENT_QUOTES, "GB2312");
  // 過濾掉不能轉(zhuǎn)化的實體（entity）
  $content = preg_replace('/\\&\\#.*?\\;/i', '', $content);
  // 上面是將 HTML 網(wǎng)頁內(nèi)容轉(zhuǎn)化為帶換行的純文本，下面是將這些純文本轉(zhuǎn)化為 WML。
  $content = str_replace('$', '$$', $content);
  $content = str_replace("\\r\\n", "\\n", htmlspecialchars($content));
  $content = explode("\\n", $content);
  for ($i = 0; $i < count($content); $i++)
  {
  $content[$i] = trim($content[$i]);
  // 如果去掉全角空格為空行，則設(shè)為空行，否則不對全角空格過濾。
  if (str_replace('　', '', $content[$i]) == '') $content[$i] = '';
  }
  $content = str_replace("<p><br /></p>\\n", "", '<p>'.implode("<br /></p>\\n<p>", $content)."<br /></p>\\n");
  //還原圖片
  if(isset($imgarr[0]) && count($imgarr[0])>0 )
  {
    foreach($imgarr[0] as $k=>$v)
    {
     $attstr = (preg_match('#/$#', $imgarr[1][$k])) ? '<img '.$imgarr[1][$k].'>' : '<img '.$imgarr[1][$k].' />';
     $content = str_replace("WAP-IMG::{$k}", $attstr, $content);
    }
  }
  $content = preg_replace("/&amp;[a-z]{3,10};/isU", ' ', $content);
  return $content;
}
function text2wml($content)
{
  $content = str_replace('$', '$$', $content);
  $content = str_replace("\\r\\n", "\\n", htmlspecialchars($content));
  $content = explode("\\n", $content);
  for ($i = 0; $i < count($content); $i++)
  {
  // 過濾首尾空格
  $content[$i] = trim($content[$i]);
  // 如果去掉全角空格為空行，則設(shè)為空行，否則不對全角空格過濾。
  if (str_replace("　", "", $content[$i]) == "") $content[$i] = "";
  }
  //合并各行，轉(zhuǎn)化為 WML，并過濾掉空行
  $content = str_replace("<p><br /></p>\\n", "", "<p>".implode("<br /></p>\\n<p>", $content)."<br /></p>\\n");
  return $content;
}
?>