|
1
31 package org.pdfbox;
32
33 import java.awt.image.BufferedImage  ;
34 import java.io.File ;
35 import java.util.Iterator ;
36 import java.util.List ;
37
38 import javax.imageio.IIOException ;
39 import javax.imageio.IIOImage ;
40 import javax.imageio.ImageIO ;
41 import javax.imageio.ImageWriteParam ;
42 import javax.imageio.ImageWriter ;
43 import javax.imageio.stream.ImageOutputStream ;
44
45 import org.pdfbox.exceptions.InvalidPasswordException;
46
47 import org.pdfbox.pdmodel.PDDocument;
48 import org.pdfbox.pdmodel.PDPage;
49
50
56 public class PDFToImage
57 {
58
59 private static final String PASSWORD = "-password";
60 private static final String START_PAGE = "-startPage";
61 private static final String END_PAGE = "-endPage";
62 private static final String IMAGE_TYPE = "-imageType";
63 private static final String OUTPUT_PREFIX = "-outputPrefix";
64
65
68 private PDFToImage()
69 {
70 }
72
73
80 public static void main( String [] args ) throws Exception
81 {
82 String password = "";
83 String pdfFile = null;
84 String outputPrefix = null;
85 String imageType = "jpg";
86 int startPage = 1;
87 int endPage = Integer.MAX_VALUE;
88 for( int i=0; i<args.length; i++ )
89 {
90 if( args[i].equals( PASSWORD ) )
91 {
92 i++;
93 if( i >= args.length )
94 {
95 usage();
96 }
97 password = args[i];
98 }
99 else if( args[i].equals( START_PAGE ) )
100 {
101 i++;
102 if( i >= args.length )
103 {
104 usage();
105 }
106 startPage = Integer.parseInt( args[i] );
107 }
108 else if( args[i].equals( END_PAGE ) )
109 {
110 i++;
111 if( i >= args.length )
112 {
113 usage();
114 }
115 endPage = Integer.parseInt( args[i] );
116 }
117 else if( args[i].equals( IMAGE_TYPE ) )
118 {
119 i++;
120 imageType = args[i];
121 }
122 else if( args[i].equals( OUTPUT_PREFIX ) )
123 {
124 i++;
125 outputPrefix = args[i];
126 }
127 else
128 {
129 if( pdfFile == null )
130 {
131 pdfFile = args[i];
132 }
133 }
134 }
135
136 if( pdfFile == null )
137 {
138 usage();
139 }
140 else
141 {
142 if(outputPrefix == null)
143 {
144 outputPrefix = pdfFile.substring( 0, pdfFile.lastIndexOf( '.' ));
145 }
146
147 PDDocument document = null;
148 try
149 {
150 document = PDDocument.load( pdfFile );
151
152
153 if( document.isEncrypted() )
155 {
156 try
157 {
158 document.decrypt( password );
159 }
160 catch( InvalidPasswordException e )
161 {
162 if( args.length == 4 ) {
164 System.err.println( "Error: The supplied password is incorrect." );
165 System.exit( 2 );
166 }
167 else
168 {
169 System.err.println( "Error: The document is encrypted." );
171 usage();
172 }
173 }
174 }
175 List pages = document.getDocumentCatalog().getAllPages();
176 for( int i=startPage-1; i<endPage && i<pages.size(); i++ )
177 {
178 ImageOutputStream output = null;
179 ImageWriter imageWriter = null;
180 try
181 {
182 PDPage page = (PDPage)pages.get( i );
183 BufferedImage image = page.convertToImage();
184 String fileName = outputPrefix + (i+1) + "." + imageType;
185 System.out.println( "Writing:" + fileName );
186 output = ImageIO.createImageOutputStream( new File ( fileName ) );
187
188 boolean foundWriter = false;
189 Iterator writerIter = ImageIO.getImageWritersByFormatName( imageType );
190 while( writerIter.hasNext() && !foundWriter )
191 {
192 try
193 {
194 imageWriter = (ImageWriter )writerIter.next();
195 ImageWriteParam writerParams = imageWriter.getDefaultWriteParam();
196 if(writerParams.canWriteCompressed() )
197 {
198 writerParams.setCompressionMode(ImageWriteParam.MODE_EXPLICIT);
199 writerParams.setCompressionQuality(1.0f);
200 }
201
202
203 imageWriter.setOutput( output );
204 imageWriter.write( null, new IIOImage ( image, null, null), writerParams );
205 foundWriter = true;
206 }
207 catch( IIOException io )
208 {
209 }
211 finally
212 {
213 if( imageWriter != null )
214 {
215 imageWriter.dispose();
216 }
217 }
218 }
219 if( !foundWriter )
220 {
221 throw new RuntimeException ( "Error: no writer found for image type '" + imageType + "'" );
222 }
223 }
224 finally
225 {
226 if( output != null )
227 {
228 output.flush();
229 output.close();
230 }
231 }
232 }
233 }
234 finally
235 {
236 if( document != null )
237 {
238 document.close();
239 }
240 }
241 }
242 }
243
244
247 private static void usage()
248 {
249 System.err.println( "Usage: java org.pdfbox.PDFToImage [OPTIONS] <PDF file>\n" +
250 " -password <password> Password to decrypt document\n" +
251 " -imageType <image type> (" + getImageFormats() + ")\n" +
252 " -outputPrefix <output prefix> Filename prefix for image files\n" +
253 " -startPage <number> The first page to start extraction(1 based)\n" +
254 " -endPage <number> The last page to extract(inclusive)\n" +
255 " <PDF file> The PDF document to use\n"
256 );
257 System.exit( 1 );
258 }
259
260 private static String getImageFormats()
261 {
262 StringBuffer retval = new StringBuffer ();
263 String [] formats = ImageIO.getReaderFormatNames();
264 for( int i=0; i<formats.length; i++ )
265 {
266 retval.append( formats[i] );
267 if( i+1<formats.length )
268 {
269 retval.append( "," );
270 }
271 }
272 return retval.toString();
273 }
274 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.