|
1 package org.enhydra.snapper.parsers;
2
3 import java.io.File  ;
4 import java.io.FileInputStream ;
5 import java.io.InputStream ;
6 import java.util.Collection ;
7 import java.util.Iterator ;
8 import java.util.Map ;
9 import java.util.Set ;
10
11
12 import org.enhydra.snapper.api.*;
13 import org.textmining.text.extraction.WordExtractor;
14
15 import org.apache.poi.poifs.eventfilesystem.POIFSReader;
16 import org.apache.poi.poifs.eventfilesystem.POIFSReaderEvent;
17 import org.apache.poi.poifs.eventfilesystem.POIFSReaderListener;
18 import org.apache.poi.hpsf.DocumentSummaryInformation;
20 import org.apache.poi.hpsf.PropertySetFactory;
21 import org.apache.poi.hpsf.SummaryInformation;
22
23
24 public class WordParser implements org.enhydra.snapper.api.Parser {
25 private String fileName;
26 private String parsedText;
27 public String title;
28 private String properties = "";
29 public Map customproperties;
30
31 public String parse(InputStream is) throws java.io.IOException { return "";}
32
33 public void parse () {
34 FileInputStream fis1 = null, fis = null;
35 try {
36
37 FileInputStream in = new FileInputStream (new File (fileName));
38 parsedText = new WordExtractor().extractText(in);
39
46
47 in.close();
48 POIFSReader r = new POIFSReader();
50 MyPOIFSReaderListener mypoi = new MyPOIFSReaderListener();
51 r.registerListener(mypoi,
52 "\005SummaryInformation");
53 fis = new FileInputStream (fileName);
54 r.read(fis);
55 title = mypoi.getTitle();
56 fis.close();
57 fis = null;
58
59
60 POIFSReader r1 = new POIFSReader();
61 DocumentPOIFSReaderListener docpoi = new DocumentPOIFSReaderListener();
62 r1.registerListener(docpoi,
63 "\005DocumentSummaryInformation");
64
65 fis1 = new FileInputStream (fileName);
66
68 r1.read(fis1);
69 if (docpoi.getCustomProperties() == null){
70 properties = "";
71 return;
72 }
73 customproperties = docpoi.getCustomProperties();
74 createString(customproperties);
75 fis1.close();
76 fis1 = null;
77
78
79
80
81 } catch (Throwable e) {
82
83
84 try{
85
86 ParserManager.logger.debug("***** File could not be parsed: " + fileName);
87 } catch (Exception ex) {
88 System.out.println("***** File could not be parsed: " + fileName);
89 }
90
91 }
92
93 finally {
94 if (fis != null){
95 try{
96 fis.close();
97 } catch (Exception ex){};
98 fis = null;
99 }
100 if (fis1 != null){
101 try{
102 fis1.close();
103 } catch (Exception ex){};
104 fis1 = null;
105 }
106 }
107
108
109 }
110
111
112 public void setFileName(String fileName) {
113 this.fileName = fileName;
114 }
115
116 public String getParsedText() {
117 return parsedText;
118 }
119
120 public String getTitle() {
121 return title;
122 }
123
124 public String getProperties(){
125 return properties;
126 }
127
128 private void createString( Map customproperties){
129 Collection values = customproperties.values();
130 Iterator valuesIterator = values.iterator();
131 properties = "";
132 Set keys = customproperties.keySet();
133 Iterator keysIterator = keys.iterator();
134
135 for (Iterator it= customproperties.entrySet().iterator(); it.hasNext(); ) {
136 Map.Entry entry = (Map.Entry )it.next();
137 properties += entry.getKey().toString() + " = " + entry.getValue() + " , ";
138 }
139
140
141 }
142 }
143
144 class MyPOIFSReaderListener implements POIFSReaderListener
145 {
146 String title;
147 public void processPOIFSReaderEvent(POIFSReaderEvent event)
148 {
149 SummaryInformation si = null;
150 try
151 {
152 si = (SummaryInformation)
153 PropertySetFactory.create(event.getStream());
154 }
155 catch (Exception ex)
156 {
157 throw new RuntimeException
158 ("Property set stream \"" + event.getPath() +
159 event.getName() + "\": " + ex);
160 }
161 title = si.getTitle();
162 }
163
164 public String getTitle(){
165 return title;
166 }
167 }
168
169 class DocumentPOIFSReaderListener implements POIFSReaderListener
170 { Map customproperties;
171 public void processPOIFSReaderEvent(POIFSReaderEvent event)
172 {
173 DocumentSummaryInformation dsi = null;
174 try
175 {
176 dsi = (DocumentSummaryInformation)
177 PropertySetFactory.create(event.getStream());
178 }
179 catch (Exception ex)
180 {
181 throw new RuntimeException
182 ("Property set stream \"" + event.getPath() +
183 event.getName() + "\": " + ex);
184 }
185 customproperties = dsi.getCustomProperties();
186 }
187
188 public Map getCustomProperties(){
189 return customproperties;
190 }
191
192
193
194 }
195 |
Popular Tags |
Java API By Example, From Geeks To Geeks.