|
1
2
17
18 package org.apache.poi.hwpf;
19
20 import org.apache.poi.hwpf.HWPFDocument;
21 import org.apache.poi.hwpf.usermodel.*;
22 import org.apache.poi.hwpf.model.*;
23
24 import java.io.*;
25
26 public class Word2Forrest
27 {
28 Writer _out;
29 HWPFDocument _doc;
30
31 public Word2Forrest(HWPFDocument doc, OutputStream stream)
32 throws IOException, UnsupportedEncodingException
33 {
34 OutputStreamWriter out = new OutputStreamWriter (stream, "UTF-8");
35 _out = out;
36 _doc = doc;
37
38 init ();
39 openDocument ();
40 openBody ();
41
42 Range r = doc.getRange ();
43 StyleSheet styleSheet = doc.getStyleSheet ();
44
45 int sectionLevel = 0;
46 int lenParagraph = r.numParagraphs ();
47 boolean inCode = false;
48 for (int x = 0; x < lenParagraph; x++)
49 {
50 Paragraph p = r.getParagraph (x);
51 String  text = p.text ();
52 if (text.trim ().length () == 0)
53 {
54 continue;
55 }
56 StyleDescription paragraphStyle = styleSheet.getStyleDescription (p.
57 getStyleIndex ());
58 String styleName = paragraphStyle.getName();
59 if (styleName.startsWith ("Heading"))
60 {
61 if (inCode)
62 {
63 closeSource();
64 inCode = false;
65 }
66
67 int headerLevel = Integer.parseInt (styleName.substring (8));
68 if (headerLevel > sectionLevel)
69 {
70 openSection ();
71 }
72 else
73 {
74 for (int y = 0; y < (sectionLevel - headerLevel) + 1; y++)
75 {
76 closeSection ();
77 }
78 openSection ();
79 }
80 sectionLevel = headerLevel;
81 openTitle ();
82 writePlainText (text);
83 closeTitle ();
84 }
85 else
86 {
87 int cruns = p.numCharacterRuns ();
88 CharacterRun run = p.getCharacterRun (0);
89 String fontName = run.getFontName();
90 if (fontName.startsWith ("Courier"))
91 {
92 if (!inCode)
93 {
94 openSource ();
95 inCode = true;
96 }
97 writePlainText (p.text());
98 }
99 else
100 {
101 if (inCode)
102 {
103 inCode = false;
104 closeSource();
105 }
106 openParagraph();
107 writePlainText(p.text());
108 closeParagraph();
109 }
110 }
111 }
112 for (int x = 0; x < sectionLevel; x++)
113 {
114 closeSection();
115 }
116 closeBody();
117 closeDocument();
118 _out.flush();
119
120 }
121
122 public void init ()
123 throws IOException
124 {
125 _out.write ("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\r\n");
126 _out.write ("<!DOCTYPE document PUBLIC \"-//APACHE//DTD Documentation V1.1//EN\" \"./dtd/document-v11.dtd\">\r\n");
127 }
128
129 public void openDocument ()
130 throws IOException
131 {
132 _out.write ("<document>\r\n");
133 }
134 public void closeDocument ()
135 throws IOException
136 {
137 _out.write ("</document>\r\n");
138 }
139
140
141 public void openBody ()
142 throws IOException
143 {
144 _out.write ("<body>\r\n");
145 }
146
147 public void closeBody ()
148 throws IOException
149 {
150 _out.write ("</body>\r\n");
151 }
152
153
154 public void openSection ()
155 throws IOException
156 {
157 _out.write ("<section>");
158
159 }
160
161 public void closeSection ()
162 throws IOException
163 {
164 _out.write ("</section>");
165
166 }
167
168 public void openTitle ()
169 throws IOException
170 {
171 _out.write ("<title>");
172 }
173
174 public void closeTitle ()
175 throws IOException
176 {
177 _out.write ("</title>");
178 }
179
180 public void writePlainText (String text)
181 throws IOException
182 {
183 _out.write (text);
184 }
185
186 public void openParagraph ()
187 throws IOException
188 {
189 _out.write ("<p>");
190 }
191
192 public void closeParagraph ()
193 throws IOException
194 {
195 _out.write ("</p>");
196 }
197
198 public void openSource ()
199 throws IOException
200 {
201 _out.write ("<source><![CDATA[");
202 }
203 public void closeSource ()
204 throws IOException
205 {
206 _out.write ("]]></source>");
207 }
208
209
210 public static void main(String [] args)
211 {
212 try
213 {
214 OutputStream out = new FileOutputStream("c:\\test.xml");
215
216 new Word2Forrest(new HWPFDocument(new FileInputStream(args[0])), out);
217 out.close();
218 }
219 catch (Throwable t)
220 {
221 t.printStackTrace();
222 }
223
224 }
225 }
226 |
Popular Tags |
Java API By Example, From Geeks To Geeks.