实现MHT文件格式的解析和内容抽取

由于我们的业务系统中有大量的MHT格式的资料，需要对其建立索引，搜索很久了一直没有找到相关解析的类库，只好自己动手丰衣足食了。已实现内容的提取以及和lucene的整合，稍后会完善编码检测及其他内容的提取，做一个完整的parser出来。  

 

• 文本内容提取：  首先提取html部分的内容，解码之后使用nekoHtml提取文本内容；

public class MhtDocHandler extends HtmDocHandler {

    private DOMFragmentParser parser = new DOMFragmentParser();

    public Document getDocument(InputStream is) throws DocumentHandlerException {

        DocumentFragment node = new HTMLDocumentImpl().createDocumentFragment();
        try {

            String mhts = IOUtils.toString(is);
            int a1 = mhts.indexOf("<HTML");
            int a2 = mhts.indexOf("</HTML>");

            String html = mhts.substring(a1, a2 + 8);

            //在mht中文本按照QuotedPrintable格式编码
        html = decodeQuotedPrintable(html, "UTF-8");

            StringReader r = new StringReader(html);

            parser.parse(new InputSource(r), node);
        }
        catch (Exception e) {
            throw new DocumentHandlerException("Cannot parse MHT document: ", e);
        }

        Document doc = new Document();

        StringBuffer sb = new StringBuffer();
        getText(sb, node, "title");
        String title = sb.toString().trim();

        sb.setLength(0);
        getText(sb, node, "body");
        String text = sb.toString().trim();

        if (!title.equals("")) {
            doc.add(new Field(WikiDOC.DOC_TITLE, title, 
                                  Field.Store.YES, Field.Index.TOKENIZED, 
                                       Field.TermVector.WITH_POSITIONS_OFFSETS));
        }
        if (!text.equals("")) {
            doc.add(new Field(WikiDOC.DOC_CONTENT, text, 
                                       Field.Store.COMPRESS, Field.Index.TOKENIZED, 
                                           Field.TermVector.WITH_POSITIONS_OFFSETS));
        }

        return doc;
    }

    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;
        }
    }

}

 <!---->

good job