|
1
31 package org.pdfbox.cmapparser;
32
33 import java.io.FileInputStream  ;
34 import java.io.InputStream ;
35 import java.io.IOException ;
36 import java.io.RandomAccessFile ;
37
38 import java.util.List ;
39
40 import org.pdfbox.cmaptypes.CMap;
41 import org.pdfbox.cmaptypes.CodespaceRange;
42
43 import org.pdfbox.cos.COSArray;
44 import org.pdfbox.cos.COSName;
45 import org.pdfbox.cos.COSNumber;
46 import org.pdfbox.cos.COSString;
47
48 import org.pdfbox.pdfparser.PDFStreamParser;
49
50 import org.pdfbox.util.PDFOperator;
51
52
58 public class CMapParser
59 {
60 private static final String BEGIN_CODESPACE_RANGE = "begincodespacerange";
61 private static final String BEGIN_BASE_FONT_CHAR = "beginbfchar";
62 private static final String BEGIN_BASE_FONT_RANGE = "beginbfrange";
63
64 private InputStream input;
65 private CMap result;
66 private RandomAccessFile file;
67
68
74 public CMapParser( InputStream in, RandomAccessFile raf )
75 {
76 input = in;
77 file = raf;
78 }
79
80
85 public CMap getResult()
86 {
87 return result;
88 }
89
90
95 public void parse() throws IOException
96 {
97 result = new CMap();
98 PDFStreamParser parser = new PDFStreamParser( input, file );
99 parser.parse();
100 List tokens = parser.getTokens();
101 for( int i=0; i<tokens.size(); i++ )
102 {
103 Object token = tokens.get( i );
104 if( token instanceof PDFOperator )
105 {
106 PDFOperator op = (PDFOperator)token;
107 if( op.getOperation().equals( BEGIN_CODESPACE_RANGE ) )
108 {
109 COSNumber cosCount = (COSNumber)tokens.get( i-1 );
110 for( int j=0; j<cosCount.intValue(); j++ )
111 {
112 i++;
113 COSString startRange = (COSString)tokens.get( i );
114 i++;
115 COSString endRange = (COSString)tokens.get( i );
116 CodespaceRange range = new CodespaceRange();
117 range.setStart( startRange.getBytes() );
118 range.setEnd( endRange.getBytes() );
119 result.addCodespaceRange( range );
120 }
121 }
122 else if( op.getOperation().equals( BEGIN_BASE_FONT_CHAR ) )
123 {
124 COSNumber cosCount = (COSNumber)tokens.get( i-1 );
125 for( int j=0; j<cosCount.intValue(); j++ )
126 {
127 i++;
128 COSString inputCode = (COSString)tokens.get( i );
129 i++;
130 Object nextToken = tokens.get( i );
131 if( nextToken instanceof COSString )
132 {
133 byte[] bytes = ((COSString)nextToken).getBytes();
134 String value = createStringFromBytes( bytes );
135 result.addMapping( inputCode.getBytes(), value );
136 }
137 else if( nextToken instanceof COSName )
138 {
139 result.addMapping( inputCode.getBytes(), ((COSName)nextToken).getName() );
140 }
141 else
142 {
143 throw new IOException ( "Error parsing CMap beginbfchar, expected{COSString " +
144 "or COSName} and not " + nextToken );
145 }
146 }
147 }
148 else if( op.getOperation().equals( BEGIN_BASE_FONT_RANGE ) )
149 {
150 COSNumber cosCount = (COSNumber)tokens.get( i-1 );
151
152 for( int j=0; j<cosCount.intValue(); j++ )
153 {
154 i++;
155 COSString startCode = (COSString)tokens.get( i );
156 i++;
157 COSString endCode = (COSString)tokens.get( i );
158 i++;
159 Object nextToken = tokens.get( i );
160 COSArray array = null;
161 if( nextToken instanceof COSArray )
162 {
163 array = (COSArray)nextToken;
164 }
165
166 byte[] startBytes = startCode.getBytes();
167 byte[] endBytes = endCode.getBytes();
168 byte[] tokenBytes = null;
169 if( array == null )
170 {
171 tokenBytes = ((COSString)nextToken).getBytes();
172 }
173 else
174 {
175 tokenBytes = ((COSString)array.getObject( 0 )).getBytes();
176 }
177
178 String value = null;
179
180 int arrayIndex = 0;
181 boolean done = false;
182 while( !done )
183 {
184 if( compare( startBytes, endBytes ) >= 0 )
185 {
186 done = true;
187 }
188 value = createStringFromBytes( tokenBytes );
189 result.addMapping( startBytes, value );
190 increment( startBytes );
191
192 if( array == null )
193 {
194 increment( tokenBytes );
195 }
196 else
197 {
198 if( arrayIndex < array.size() )
199 {
200 tokenBytes = ((COSString)array.getObject( arrayIndex++ )).getBytes();
201 }
202 }
203 }
204 }
205 }
206 }
207 }
208 }
209
210 private void increment( byte[] data )
211 {
212 increment( data, data.length-1 );
213 }
214
215 private void increment( byte[] data, int position )
216 {
217 if( position > 0 && (data[position]+256)%256 == 255 )
218 {
219 data[position]=0;
220 increment( data, position-1);
221 }
222 else
223 {
224 data[position] = (byte)(data[position]+1);
225 }
226 }
227
228 private String createStringFromBytes( byte[] bytes ) throws IOException
229 {
230 String retval = null;
231 if( bytes.length == 1 )
232 {
233 retval = new String ( bytes );
234 }
235 else
236 {
237 retval = new String ( bytes, "UTF-16BE" );
238 }
239 return retval;
240 }
241
242 private int compare( byte[] first, byte[] second )
243 {
244 int retval = 1;
245 boolean done = false;
246 for( int i=0; i<first.length && !done; i++ )
247 {
248 if( first[i] == second[i] )
249 {
250 }
252 else if( ((first[i]+256)%256) < ((second[i]+256)%256) )
253 {
254 done = true;
255 retval = -1;
256 }
257 else
258 {
259 done = true;
260 retval = 1;
261 }
262 }
263 return retval;
264 }
265
266
273 public static void main( String [] args ) throws Exception
274 {
275 if( args.length != 1 )
276 {
277 System.err.println( "usage: java org.pdfbox.cmapparser.CMapParser <CMAP File>" );
278 System.exit( -1 );
279 }
280 CMapParser parser = new CMapParser( new FileInputStream ( args[0] ), null );
281 parser.parse();
282 CMap result = parser.getResult();
283 System.out.println( "Result:" + result );
284 }
285 } |
Popular Tags |
Java API By Example, From Geeks To Geeks.