|
1 package org.enhydra.snapper.parsers;
2
3 import org.enhydra.snapper.SnapperManager;
4 import org.enhydra.snapper.utils.*;
5
6 import java.io.*;
7 import java.util.Vector  ;
8
9 import javax.swing.text.DefaultStyledDocument ;
10 import javax.swing.text.Document ;
11 import javax.swing.text.rtf.RTFEditorKit ;
12
13
14 public class Parser {
15 private String fileName;
16 private String parsedText;
17 private File temp;
18
19 public void parse() {};
20 public Vector parse(File originalFile) {
21 Vector retVal = new Vector ();
22
23 try {
24 if (Utils.isOO(originalFile.getPath())) {
25 File metatmp = new File(System.getProperty("user.dir") + File.separator + "metasearch.txt");
27 metatmp.createNewFile();
28 OoToText oc = new OoToText(originalFile.getPath(), metatmp.getPath());
29 oc.parse();
30 String title = originalFile.getName();
31 retVal.add(new Long (originalFile.lastModified()));
32 retVal.add(originalFile.getPath());
33 retVal.add(oc.getParsedText());
34 retVal.add("oo");
35 retVal.add(title);
36 retVal.add("");
37 if (metatmp.exists())
38 metatmp.delete();
39 return retVal;
40 }
41 else if (Utils.isDoc(originalFile.getPath())) {
42
43
44 WordParser wd = new WordParser();
45 wd.setFileName(originalFile.getPath());
46 wd.parse();
47 String title;
48 if ( wd.getTitle() == null || wd.getTitle().equals("") )
49 title = originalFile.getName();
50 else
51 title = wd.getTitle();
52 retVal.add(new Long (originalFile.lastModified()));
53 retVal.add(originalFile.getPath());
54 retVal.add(wd.getParsedText());
55 retVal.add("doc");
56 retVal.add(title);
57 retVal.add(wd.getProperties());
58 return retVal;
59
60 }
61
62 else if (Utils.isPPT(originalFile.getPath())) {
63 PowerParser pp = new PowerParser();
64 pp.setFileName(originalFile.getPath());
65 pp.parse();
66 String title;
67 if ( pp.getTitle() == null || pp.getTitle().equals("") )
68 title = originalFile.getName();
69 else
70 title = pp.getTitle();
71 retVal.add(new Long (originalFile.lastModified()));
72 retVal.add(originalFile.getPath());
73 retVal.add(pp.getParsedText());
74 retVal.add("ppt");
75 retVal.add(title);
76 retVal.add(pp.getProperties());
77 return retVal;
78
79
80 }
81 else if (Utils.isPPS(originalFile.getPath())) {
82 PowerParser pp = new PowerParser();
83 pp.setFileName(originalFile.getPath());
84 pp.parse();
85 String title;
86 if ( pp.getTitle() == null || pp.getTitle().equals("") )
87 title = originalFile.getName();
88 else
89 title = pp.getTitle();
90 retVal.add(new Long (originalFile.lastModified()));
91 retVal.add(originalFile.getPath());
92 retVal.add(pp.getParsedText());
93 retVal.add("pps");
94 retVal.add(title);
95 retVal.add(pp.getProperties());
96 return retVal;
97
98 }
99
100 else if (Utils.isText(originalFile.getPath())) {
101
102 String contents = ReadWriteTextFile.getContents(originalFile);
103 retVal.add(new Long (originalFile.lastModified()));
104 retVal.add(originalFile.getPath());
105 retVal.add(contents);
106 retVal.add("txt");
107 retVal.add(originalFile.getName());
108 retVal.add("");
109 return retVal;
110
111
112 }
113 else if (Utils.isEML(originalFile.getPath())) {
114
115 String contents = ReadWriteTextFile.getContents(originalFile);
116 retVal.add(new Long (originalFile.lastModified()));
117 retVal.add(originalFile.getPath());
118 retVal.add(contents);
119 retVal.add("eml");
120 retVal.add(originalFile.getName());
121 retVal.add("");
122 return retVal;
123
124
125 }
126 else if (Utils.isExcel(originalFile.getPath())) {
127
128 ExcelParser ep = new ExcelParser();
129 ep.setFileName(originalFile.getPath());
130 ep.parse();
131 String title;
132 if (ep.getTitle() == null || ep.getTitle().equals(""))
133 title = originalFile.getName();
134 else
135 title = ep.getTitle();
136 retVal.add(new Long (originalFile.lastModified()));
137 retVal.add(originalFile.getPath());
138 retVal.add(ep.getParsedText());
139 retVal.add("xls");
140 retVal.add(title);
141 retVal.add("");
142 originalFile = null;
144 return retVal;
145
146
147 }
148 else if (Utils.isPDF(originalFile.getPath())) {
149
150 SnapperPDFParser pdf = new SnapperPDFParser();
151 pdf.setFileName(originalFile.getPath());
152 pdf.parse();
153 String title;
154 if ( pdf.getTitle() == null || pdf.getTitle().equals("") )
155 title = originalFile.getName();
156 else
157 title = pdf.getTitle();
158 retVal.add(new Long (originalFile.lastModified()));
159 retVal.add(originalFile.getPath());
160 retVal.add(pdf.getParsedText());
161 retVal.add("pdf");
162 retVal.add(title);
163 retVal.add("");
164 return retVal;
165
166 }
167 else if (Utils.isHTML(originalFile.getPath())) {
168
169 HTMLParser html = new HTMLParser();
170 html.setFileName(originalFile.getPath());
171 html.parse();
172 String title;
173 if ( html.getTitle() == null || html.getTitle().equals("") )
174 title = originalFile.getName();
175 else
176 title = html.getTitle();
177 retVal.add(new Long (originalFile.lastModified()));
178 retVal.add(originalFile.getPath());
179 retVal.add(html.getParsedText());
180 retVal.add("html");
181 retVal.add(originalFile.getName());
182 retVal.add("");
183 return retVal;
184
185 }
186 else if (Utils.isMSG(originalFile.getPath())) {
187
188 MsgParser msgp = new MsgParser();
189 msgp.setFileName(originalFile.getPath());
190 msgp.parse();
191 String title;
192 if ( msgp.getTitle() == null || msgp.getTitle().equals("") )
193 title = originalFile.getName();
194 else
195 title = msgp.getTitle();
196 retVal.add(new Long (originalFile.lastModified()));
197 retVal.add(originalFile.getPath());
198 retVal.add(msgp.getParsedText());
199 retVal.add("msg");
200 retVal.add(title);
201 retVal.add("");
202 originalFile = null;
204 return retVal;
205
206
207 }
208 else if (Utils.isRTF(originalFile.getPath())) {
209 DefaultStyledDocument styledDocument = new DefaultStyledDocument ();
210 String contents = "";
211
212 FileInputStream in = null;
213 try{
214 in = new FileInputStream(originalFile);
215 RTFEditorKit kit = new RTFEditorKit ();
218 Document doc = kit.createDefaultDocument();
219 kit.read(in, doc, 0);
220
221 contents = doc.getText(0, doc.getLength());
222
225 String title = originalFile.getName();
226 retVal.add(new Long (originalFile.lastModified()));
227 retVal.add(originalFile.getPath());
228 retVal.add(contents);
229 retVal.add("rtf");
230 retVal.add(originalFile.getName());
231 retVal.add("");
232 in.close();
233 return retVal;
234 }
235 catch (Exception ex) {
236 ParserManager.logger.debug("File could not be parsed: " + originalFile.getName());
237 try{
239 in.close();
240 }catch (Exception e) {}
241 retVal.removeAllElements();
242 retVal.add(null);
243 retVal.add(originalFile.getPath());
244 retVal.add(null);
245 retVal.add(null);
246 retVal.add(null);
247 retVal.add(null);
248 return retVal;
249 }
250
251 }
252 else if (Utils.isZIP(originalFile.getPath())) {
253 ZIPParser zipp = new ZIPParser(SnapperManager.getInstance().getTempDir() + File.separator);
254 zipp.setFileName(originalFile.getPath());
255 zipp.parse();
256 return zipp.getFiles();
258
259 }
266 else
267 retVal.add(new Long (originalFile.lastModified()));
269 retVal.add(originalFile.getPath());
270 retVal.add("");
271 retVal.add("other");
273 retVal.add(originalFile.getName());
274 retVal.add("");
275 return retVal;
276
277 }
278 catch (Exception ex) {
281 System.out.println("Exception while parsing...: " + ex);
282 ParserManager.logger.debug("Exception while parsing...: " + ex);
283 retVal.removeAllElements();
284 retVal.add(null);
285 retVal.add(originalFile.getPath());
286 retVal.add(null);
287 retVal.add(null);
288 retVal.add(null);
289 retVal.add(null);
290 return retVal;
291 }
292
294 }
295
296 public File getFile()
297 {
298 return temp;
299 }
300 }
301
302 |
Popular Tags |
Java API By Example, From Geeks To Geeks.