src/test/java/sample/SampleParser.java

   1 /*
   2  * sample parser
   3  *
   4  * License : The MIT License
   5  * Copyright(c) 2009 olyutorskii
   6  */
   7
   8 package sample;
   9
  10 import java.io.FileInputStream;
  11 import jp.sourceforge.jindolf.parser.*;
  12 import java.io.IOException;
  13 import java.io.InputStream;
  14 import java.nio.charset.Charset;
  15 import java.nio.charset.CharsetDecoder;
  16 import java.util.Collections;
  17 import java.util.Enumeration;
  18 import java.util.SortedMap;
  19 import java.util.TreeMap;
  20 import java.util.zip.ZipEntry;
  21 import java.util.zip.ZipFile;
  22
  23 /**
  24  * サンプルのパーサ
  25  */
  26 public class SampleParser{
  27
  28     private static final CharsetDecoder ud;
  29
  30     static{
  31         ud = Charset.forName("UTF-8").newDecoder();
  32     }
  33
  34     private SampleParser(){
  35         super();
  36         return;
  37     }
  38
  39     public static SortedMap<String, ZipEntry> createEntryMap(ZipFile file){
  40         TreeMap<String, ZipEntry> result = new TreeMap<String, ZipEntry>();
  41
  42         Enumeration<? extends ZipEntry> list = file.entries();
  43         while(list.hasMoreElements()){
  44             ZipEntry entry = list.nextElement();
  45             String name = entry.getName();
  46             result.put(name, entry);
  47         }
  48
  49         return Collections.unmodifiableSortedMap(result);
  50     }
  51
  52     public static DecodedContent contentFromStream(InputStream istream)
  53             throws IOException, DecodeException{
  54         StreamDecoder decoder = new StreamDecoder(ud);
  55         ContentBuilderUCS2 builder = new ContentBuilderUCS2();
  56
  57         decoder.setDecodeHandler(builder);
  58
  59         decoder.decode(istream);
  60
  61         DecodedContent content = builder.getContent();
  62
  63         return content;
  64     }
  65
  66     public static void parseContent(DecodedContent content)
  67             throws HtmlParseException{
  68         HtmlParser parser = new HtmlParser();
  69         HtmlHandler handler = new SampleHandler();
  70
  71         parser.setBasicHandler   (handler);
  72         parser.setTalkHandler    (handler);
  73         parser.setSysEventHandler(handler);
  74
  75         parser.parseAutomatic(content);
  76
  77         return;
  78     }
  79
  80     public static void parseStream(InputStream istream)
  81             throws IOException,
  82                    DecodeException,
  83                    HtmlParseException{
  84         DecodedContent content = contentFromStream(istream);
  85
  86         parseContent(content);
  87
  88         return;
  89     }
  90
  91     public static void main(String[] args)
  92             throws IOException,
  93                    DecodeException,
  94                    HtmlParseException {
  95         if(args.length == 0){
  96             System.out.println(
  97                      "標準入力から人狼BBSのXHTML文書の読み取りを"
  98                     +"開始します...");
  99
 100             parseStream(System.in);
 101
 102             System.exit(0);
 103
 104             return;
 105         }else if(args[0].endsWith(".zip")){
 106             System.out.println(
 107                      "ZIPアーカイブ内の*.htmlファイルから"
 108                     +"人狼BBSのXHTML文書の読み取りを開始します...");
 109
 110             ZipFile zipfile = new ZipFile(args[0]);
 111
 112             SortedMap<String, ZipEntry> map = createEntryMap(zipfile);
 113
 114             for(ZipEntry entry : map.values()){
 115                 String name = entry.getName();
 116                 if( ! name.endsWith(".html") ) continue;
 117
 118                 System.out.println(name + "のパースを開始...");
 119
 120                 InputStream istream = zipfile.getInputStream(entry);
 121                 parseStream(istream);
 122
 123                 istream.close();
 124             }
 125
 126             zipfile.close();
 127
 128             System.exit(0);
 129
 130             return;
 131         }else{
 132             System.out.println(args[0] + "のパースを開始...");
 133
 134             InputStream istream = new FileInputStream(args[0]);
 135             parseStream(istream);
 136             istream.close();
 137
 138             System.exit(0);
 139         }
 140
 141         return;
 142     }
 143
 144 }