frontend/src/saccubus/conv/NicoXMLReader.java

   1 package saccubus.conv;
   2
   3 import java.util.regex.Pattern;
   4
   5 import org.xml.sax.helpers.DefaultHandler;
   6 import org.xml.sax.Attributes;
   7
   8 /**
   9  * <p>
  10  * タイトル: さきゅばす
  11  * </p>
  12  *
  13  * <p>
  14  * 説明: ニコニコ動画の動画をコメントつきで保存
  15  * </p>
  16  *
  17  * <p>
  18  * 著作権: Copyright (c) 2007 PSI
  19  * </p>
  20  *
  21  * <p>
  22  * 会社名:
  23  * </p>
  24  *
  25  * @author 未入力
  26  * @version 1.0
  27  */
  28 public class NicoXMLReader extends DefaultHandler {
  29         private final Packet Packet;
  30
  31         private Chat item;
  32
  33         private boolean item_kicked;
  34
  35         private final Pattern NG_Word;
  36
  37         private final Pattern NG_ID;
  38
  39         public NicoXMLReader(Packet packet, String ng_id, String ng_word) {
  40                 Packet = packet;
  41                 NG_Word = makePattern(ng_word);
  42                 NG_ID = makePattern(ng_id);
  43         }
  44
  45         private static Pattern makePattern(String word) {
  46                 if (word == null || word.length() <= 0) {
  47                         return null;
  48                 }
  49                 String tmp[] = word.split(" ");
  50                 String tmp2[] = new String[tmp.length];
  51                 int tmp_index = 0;
  52                 int index;
  53                 for (index = 0; index < tmp.length && tmp_index < tmp.length; index++) {
  54                         if (tmp[tmp_index].startsWith("/")) {
  55                                 String str = tmp[tmp_index];
  56                                 for (tmp_index++; tmp_index < tmp.length; tmp_index++) {
  57                                         str += " " + tmp[tmp_index];
  58                                         if (tmp[tmp_index].endsWith("/")) {
  59                                                 tmp_index++;
  60                                                 break;
  61                                         }
  62                                 }
  63                                 tmp2[index] = str;
  64                         } else if (tmp[tmp_index].startsWith("\"")) {
  65                                 String str = tmp[tmp_index];
  66                                 for (tmp_index++; tmp_index < tmp.length; tmp_index++) {
  67                                         str += " " + tmp[tmp_index];
  68                                         if (tmp[tmp_index].endsWith("\"")) {
  69                                                 tmp_index++;
  70                                                 break;
  71                                         }
  72                                 }
  73                                 tmp2[index] = str;
  74                         } else {
  75                                 tmp2[index] = tmp[tmp_index];
  76                                 tmp_index++;
  77                         }
  78                 }
  79                 String elt[] = new String[index];
  80         System.arraycopy(tmp2, 0, elt, 0, index);
  81                 String reg = "";
  82                 for (int i = 0; i < elt.length; i++) {
  83                         String e = elt[i];
  84                         System.out.println(e);
  85                         if (i > 0) {
  86                                 reg += "|";
  87                         }
  88                         if (e.indexOf("/") == 0 && e.lastIndexOf("/") == e.length() - 1) {
  89                                 reg += "(" + e.substring(1, e.length() - 1) + ")";
  90                         } else if (e.indexOf("\"") == 0
  91                                         && e.lastIndexOf("\"") == e.length() - 1) {
  92                                 reg += "(" + Pattern.quote(e.substring(1, e.length() - 1))
  93                                                 + ")";
  94                         } else {
  95                                 reg += "(.*(" + Pattern.quote(e) + ")+.*)";
  96                         }
  97                 }
  98                 System.out.println("reg:" + reg);
  99                 return Pattern.compile(reg);
 100         }
 101
 102         private static boolean match(Pattern pat, String word) {
 103                 if (word == null || word.length() <= 0 || pat == null) {
 104                         return false;
 105                 }
 106                 return pat.matcher(word).matches();
 107         }
 108
 109         /**
 110          *
 111          */
 112     @Override
 113         public void startDocument() {
 114                 System.out.println("Start converting to interval file.");
 115         }
 116
 117         /**
 118          *
 119          * @param uri
 120          *            String
 121          * @param localName
 122          *            String
 123          * @param qName
 124          *            String
 125          * @param attributes
 126          *            Attributes
 127          */
 128     @Override
 129         public void startElement(String uri, String localName, String qName,
 130                         Attributes attributes) {
 131                 if (qName.toLowerCase().equals("chat")) {
 132                         // System.out.println("----------");
 133                         item = new Chat();
 134                         item_kicked = false;
 135                         //マイメモリ削除対象
 136                         String deleted = attributes.getValue("deleted");
 137                         if(deleted != null && deleted.toLowerCase().equals("1")){
 138                                 item_kicked = true;
 139                                 return;
 140                         }
 141                         item.setDate(attributes.getValue("date"));
 142                         String mail = attributes.getValue("mail");
 143                         if (match(NG_Word, mail)) {
 144                                 item_kicked = true;
 145                                 return;
 146                         }
 147                         item.setMail(mail);
 148                         item.setNo(attributes.getValue("no"));
 149                         String user_id = attributes.getValue("user_id");
 150                         if (match(NG_ID, user_id)) {
 151                                 item_kicked = true;
 152                                 return;
 153                         }
 154                         item.setUserID(user_id);
 155                         item.setVpos(attributes.getValue("vpos"));
 156
 157                 }
 158         }
 159
 160         /**
 161          *
 162          * @param ch
 163          *            char[]
 164          * @param offset
 165          *            int
 166          * @param length
 167          *            int
 168          */
 169     @Override
 170         public void characters(char[] ch, int offset, int length) {
 171                 char input[] = (new String(ch, offset, length)).toCharArray();
 172                 for (int i = 0; i < input.length; i++) {
 173                         if (!Character.isDefined(input[i])) {
 174                                 input[i] = '?';
 175                         }
 176                 }
 177                 if (item != null) {
 178                         String com = new String(input);
 179                         if (match(NG_Word, com)) {
 180                                 item_kicked = true;
 181                                 return;
 182                         }
 183                         item.setComment(com);
 184                 }
 185         }
 186
 187         /**
 188          *
 189          * @param uri
 190          *            String
 191          * @param localName
 192          *            String
 193          * @param qName
 194          *            String
 195          */
 196     @Override
 197         public void endElement(String uri, String localName, String qName) {
 198                 if (qName.toLowerCase().equals("chat")) {
 199                         if (!item_kicked) {
 200                                 Packet.addChat(item);
 201                         }
 202                         item = null;
 203                 }
 204         }
 205
 206         /**
 207          * ドキュメント終了
 208          */
 209     @Override
 210         public void endDocument() {
 211                 // System.out.println("----------");
 212                 System.out.println("Converting finished.");
 213         }
 214
 215 }