Mht格式文本提取，mht文本提取,protected St

Mht格式文本提取，mht文本提取,protected St

protected String getText(InputStream is)  {    String mhts = super.readString(is);//读取文件数据    int a1 = mhts.indexOf("<HTML");    int a2 = mhts.indexOf("</HTML>");    String html = mhts.substring(a1, a2 + 7);    html = decodeQuotedPrintable(html, "GBK");//todo 实际编码    return  super.dumpText(is);//使用Jsoup读取html中的文本}public static String decodeQuotedPrintable(String str, String encoding) {    if (str == null) {        return null;    }    try {        //str = str.replaceAll("=\\n", "");//??        byte[] bytes = str.getBytes("US-ASCII");        ByteArrayOutputStream buffer = new ByteArrayOutputStream();        for (int i = 0; i < bytes.length; i++) {            int b = bytes[i];            if (b == '=') {                int u = Character.digit((char) bytes[++i], 16);                int l = Character.digit((char) bytes[++i], 16);                if (u == -1 || l == -1) {//??                    continue;                }                buffer.write((char) ((u << 4) + l));            }            else {                buffer.write(b);            }        }        return buffer.toString(encoding);    }    catch (Exception e) {        e.printStackTrace();        return str;    }}//该片段来自于http://byrx.net