|
1
2
3
4 package net.nutch.parse.mp3;
5
6
7 import net.nutch.parse.*;
8 import net.nutch.protocol.Content;
9 import org.farng.mp3.MP3File;
10 import org.farng.mp3.TagException;
11 import org.farng.mp3.id3.AbstractID3v2;
12 import org.farng.mp3.id3.AbstractID3v2Frame;
13 import org.farng.mp3.id3.ID3v1;
14 import org.farng.mp3.object.AbstractMP3Object;
15
16 import java.io.File  ;
17 import java.io.FileOutputStream ;
18 import java.io.IOException ;
19 import java.net.MalformedURLException ;
20 import java.util.Iterator ;
21
22
26
27 public class MP3Parser implements Parser {
28
29 private MetadataCollector metadataCollector = new MetadataCollector();
30
31 public Parse getParse(Content content) throws ParseException {
32 Parse parse = null;
33 metadataCollector.putAll(content.getMetadata());
34
35 byte[] raw = content.getContent();
36
37 File tmp = null;
38 try {
39 tmp = File.createTempFile("nutch", ".mp3");
40 FileOutputStream fos = new FileOutputStream (tmp);
41 fos.write(raw);
42 fos.close();
43 MP3File mp3 = new MP3File(tmp);
44
45 if (mp3.hasID3v2Tag()) {
46 parse = getID3v2Parse(mp3);
47 } else if (mp3.hasID3v1Tag()) {
48 parse = getID3v1Parse(mp3);
49 } else {
50 throw new ParseException("No textual content available");
51 }
52
53
54 } catch (IOException e) {
55 throw new ParseException("Couldn't create temporary file", e);
56 } catch (TagException e) {
57 throw new ParseException("ID3 Tags could not be parsed", e);
58 } finally{
59 tmp.delete();
60 }
61 return parse;
62 }
63
64 private Parse getID3v1Parse(MP3File mp3) throws MalformedURLException {
65 ID3v1 tag = mp3.getID3v1Tag();
66 metadataCollector.notifyProperty("TALB-Text", tag.getAlbum());
67 metadataCollector.notifyProperty("TPE1-Text", tag.getArtist());
68 metadataCollector.notifyProperty("COMM-Text", tag.getComment());
69 metadataCollector.notifyProperty("TCON-Text", "(" + tag.getGenre() + ")");
70 metadataCollector.notifyProperty("TIT2-Text", tag.getTitle());
71 metadataCollector.notifyProperty("TYER-Text", tag.getYear());
72 ParseData parseData = new ParseData(metadataCollector.getTitle(),
73 metadataCollector.getOutlinks(),
74 metadataCollector.getData());
75 return new ParseImpl(metadataCollector.getText(), parseData);
76 }
77
78 public Parse getID3v2Parse(MP3File mp3) throws IOException {
79 AbstractID3v2 tag = mp3.getID3v2Tag();
80 Iterator it = tag.iterator();
81 while (it.hasNext()) {
82 AbstractID3v2Frame frame = (AbstractID3v2Frame) it.next();
83 String name = frame.getIdentifier().trim();
84 if (!name.equals("APIC")) {
85 Iterator itBody = frame.getBody().iterator();
86 while (itBody.hasNext()) {
87 AbstractMP3Object mp3Obj = (AbstractMP3Object) itBody.next();
88 String bodyName = mp3Obj.getIdentifier();
89 if (!bodyName.equals("Picture data")) {
90 String bodyValue = mp3Obj.getValue().toString();
91 metadataCollector.notifyProperty(name + "-" + bodyName, bodyValue);
92 }
93 }
94 }
95 }
96 ParseData parseData = new ParseData(metadataCollector.getTitle(),
97 metadataCollector.getOutlinks(),
98 metadataCollector.getData());
99 return new ParseImpl(metadataCollector.getText(), parseData);
100 }
101
102
103 }
104 |
Popular Tags |
Java API By Example, From Geeks To Geeks.