src/jalview/io/FeaturesFile.java

   1 /*\r
   2 * Jalview - A Sequence Alignment Editor and Viewer\r
   3 * Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle\r
   4 *\r
   5 * This program is free software; you can redistribute it and/or\r
   6 * modify it under the terms of the GNU General Public License\r
   7 * as published by the Free Software Foundation; either version 2\r
   8 * of the License, or (at your option) any later version.\r
   9 *\r
  10 * This program is distributed in the hope that it will be useful,\r
  11 * but WITHOUT ANY WARRANTY; without even the implied warranty of\r
  12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\r
  13 * GNU General Public License for more details.\r
  14 *\r
  15 * You should have received a copy of the GNU General Public License\r
  16 * along with this program; if not, write to the Free Software\r
  17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA\r
  18 */\r
  19 package jalview.io;\r
  20 \r
  21 import jalview.datamodel.*;\r
  22 \r
  23 import java.io.*;\r
  24 import java.util.*;\r
  25 import jalview.schemes.UserColourScheme;\r
  26 \r
  27 \r
  28 /**\r
  29  * DOCUMENT ME!\r
  30  *\r
  31  * @author $author$\r
  32  * @version $Revision$\r
  33  */\r
  34 public class FeaturesFile extends AlignFile\r
  35 {\r
  36     /**\r
  37      * Creates a new FeaturesFile object.\r
  38      */\r
  39     public FeaturesFile()\r
  40     {\r
  41     }\r
  42 \r
  43 \r
  44     /**\r
  45      * Creates a new FeaturesFile object.\r
  46      *\r
  47      * @param inFile DOCUMENT ME!\r
  48      * @param type DOCUMENT ME!\r
  49      *\r
  50      * @throws IOException DOCUMENT ME!\r
  51      */\r
  52     public FeaturesFile(String inFile, String type)\r
  53        throws IOException\r
  54     {\r
  55         super(inFile, type);\r
  56     }\r
  57 \r
  58     /**\r
  59      * The Application can render HTML, but the applet will\r
  60      * remove HTML tags and replace links with %LINK%\r
  61      * Both need to read links in HTML however\r
  62      *\r
  63      * @throws IOException DOCUMENT ME!\r
  64      */\r
  65     public boolean parse(AlignmentI align,\r
  66                          Hashtable colours,\r
  67                          boolean removeHTML)\r
  68     {\r
  69       return parse(align, colours, null, removeHTML);\r
  70     }\r
  71     /**\r
  72      * The Application can render HTML, but the applet will\r
  73      * remove HTML tags and replace links with %LINK%\r
  74      * Both need to read links in HTML however\r
  75      *\r
  76      * @throws IOException DOCUMENT ME!\r
  77      */\r
  78     public boolean parse(AlignmentI align,\r
  79                          Hashtable colours,\r
  80                          Hashtable featureLink,\r
  81                          boolean removeHTML)\r
  82     {\r
  83       String line = null;\r
  84       try\r
  85       {\r
  86         SequenceI seq = null;\r
  87         String type, desc, token=null;\r
  88 \r
  89         int index, start, end;\r
  90         float score;\r
  91         StringTokenizer st;\r
  92         SequenceFeature sf;\r
  93         String featureGroup = null, groupLink = null;\r
  94         Hashtable typeLink = new Hashtable();\r
  95 \r
  96         boolean GFFFile = true;\r
  97 \r
  98         while ( (line = nextLine()) != null)\r
  99         {\r
 100           if(line.startsWith("#"))\r
 101             continue;\r
 102 \r
 103           st = new StringTokenizer(line, "\t");\r
 104           if (st.countTokens()>1 && st.countTokens() < 4 )\r
 105           {\r
 106             GFFFile = false;\r
 107             type = st.nextToken();\r
 108             if (type.equalsIgnoreCase("startgroup"))\r
 109             {\r
 110               featureGroup = st.nextToken();\r
 111               if (st.hasMoreElements())\r
 112               {\r
 113                 groupLink = st.nextToken();\r
 114                 featureLink.put(featureGroup, groupLink);\r
 115               }\r
 116             }\r
 117             else if (type.equalsIgnoreCase("endgroup"))\r
 118             {\r
 119               //We should check whether this is the current group,\r
 120               //but at present theres no way of showing more than 1 group\r
 121               st.nextToken();\r
 122               featureGroup = null;\r
 123               groupLink = null;\r
 124             }\r
 125             else\r
 126             {\r
 127               UserColourScheme ucs = new UserColourScheme(st.nextToken());\r
 128               colours.put(type, ucs.findColour("A"));\r
 129               if (st.hasMoreElements())\r
 130               {\r
 131                 String link = st.nextToken();\r
 132                 typeLink.put(type, link);\r
 133                 if(featureLink==null)\r
 134                   featureLink = new Hashtable();\r
 135                 featureLink.put(type, link);\r
 136               }\r
 137 \r
 138             }\r
 139             continue;\r
 140           }\r
 141 \r
 142           while (st.hasMoreElements())\r
 143           {\r
 144 \r
 145             if(GFFFile)\r
 146             {\r
 147               // Still possible this is an old Jalview file,\r
 148               // which does not have type colours at the beginning\r
 149               token = st.nextToken();\r
 150               seq = align.findName(token);\r
 151               if(seq != null)\r
 152               {\r
 153                 desc = st.nextToken();\r
 154                 type = st.nextToken();\r
 155                 start = Integer.parseInt(st.nextToken());\r
 156                 end = Integer.parseInt(st.nextToken());\r
 157                 try\r
 158                 {\r
 159                   score =  new Float(st.nextToken()).floatValue();\r
 160                 }\r
 161                 catch (NumberFormatException ex)\r
 162                 {\r
 163                   score = 0;\r
 164                 }\r
 165 \r
 166                 sf = new SequenceFeature(type, desc, start, end, score, null);\r
 167 \r
 168                 try\r
 169                 {\r
 170                   sf.setValue("STRAND", st.nextToken());\r
 171                   sf.setValue("FRAME", st.nextToken());\r
 172                 }\r
 173                 catch (Exception ex)\r
 174                 {}\r
 175 \r
 176                 seq.getDatasetSequence().addSequenceFeature(sf);\r
 177 \r
 178                 break;\r
 179               }\r
 180             }\r
 181 \r
 182             if(GFFFile && seq==null)\r
 183             {\r
 184               desc = token;\r
 185             }\r
 186             else\r
 187               desc = st.nextToken();\r
 188 \r
 189 \r
 190             token = st.nextToken();\r
 191             if (!token.equals("ID_NOT_SPECIFIED"))\r
 192             {\r
 193               seq = align.findName(token);\r
 194               st.nextToken();\r
 195             }\r
 196             else\r
 197             {\r
 198               try{\r
 199                 index = Integer.parseInt(st.nextToken());\r
 200                 seq = align.getSequenceAt(index);\r
 201               }\r
 202               catch(NumberFormatException ex)\r
 203               {\r
 204                 seq = null;\r
 205               }\r
 206             }\r
 207 \r
 208             if(seq==null)\r
 209             {\r
 210               System.out.println("Sequence not found: "+line);\r
 211               break;\r
 212             }\r
 213 \r
 214             start = Integer.parseInt(st.nextToken());\r
 215             end = Integer.parseInt(st.nextToken());\r
 216 \r
 217             type = st.nextToken();\r
 218 \r
 219             if (!colours.containsKey(type))\r
 220             {\r
 221               // Probably the old style groups file\r
 222               UserColourScheme ucs = new UserColourScheme(type);\r
 223               colours.put(type, ucs.findColour("A"));\r
 224             }\r
 225 \r
 226             sf = new SequenceFeature(type, desc, "", start, end, featureGroup);\r
 227 \r
 228             seq.addSequenceFeature(sf);\r
 229 \r
 230             if(groupLink!=null && removeHTML)\r
 231             {\r
 232               sf.addLink(groupLink);\r
 233               sf.description += "%LINK%";\r
 234             }\r
 235             if(typeLink.containsKey(type) && removeHTML)\r
 236             {\r
 237               sf.addLink(typeLink.get(type).toString());\r
 238               sf.description += "%LINK%";\r
 239             }\r
 240 \r
 241             parseDescriptionHTML(sf, removeHTML);\r
 242 \r
 243             //If we got here, its not a GFFFile\r
 244             GFFFile = false;\r
 245           }\r
 246         }\r
 247       }\r
 248       catch (Exception ex)\r
 249       {\r
 250         System.out.println(line);\r
 251         ex.printStackTrace();\r
 252         System.out.println("Error parsing feature file: " + ex +"\n"+line);\r
 253         return false;\r
 254       }\r
 255 \r
 256       return true;\r
 257     }\r
 258 \r
 259     void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)\r
 260     {\r
 261       StringBuffer sb = new StringBuffer();\r
 262       StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");\r
 263       String token,  link;\r
 264       while(st.hasMoreElements())\r
 265       {\r
 266         token = st.nextToken("<>");\r
 267         if(token.equalsIgnoreCase("html") || token.startsWith("/"))\r
 268           continue;\r
 269 \r
 270         if(token.startsWith("a href="))\r
 271         {\r
 272           link = token.substring(token.indexOf("\"")+1, token.length()-1);\r
 273           String label = st.nextToken("<>");\r
 274           sf.addLink(label+"|"+link);\r
 275           sb.append(label+"%LINK%");\r
 276         }\r
 277         else if(token.equalsIgnoreCase("br"))\r
 278           sb.append("\n");\r
 279         else\r
 280           sb.append(token);\r
 281       }\r
 282 \r
 283       if(removeHTML)\r
 284         sf.description = sb.toString();\r
 285   }\r
 286 \r
 287     /**\r
 288      * DOCUMENT ME!\r
 289      *\r
 290      * @param s DOCUMENT ME!\r
 291      * @param len DOCUMENT ME!\r
 292      * @param gaps DOCUMENT ME!\r
 293      * @param displayId DOCUMENT ME!\r
 294      *\r
 295      * @return DOCUMENT ME!\r
 296      */\r
 297     public String printJalviewFormat(SequenceI [] seqs,\r
 298                                      Hashtable visible)\r
 299     {\r
 300         StringBuffer out = new StringBuffer();\r
 301         SequenceFeature [] next;\r
 302 \r
 303         if(visible==null || visible.size()<1)\r
 304           return "No Features Visible";\r
 305 \r
 306         Enumeration en = visible.keys();\r
 307         String type;\r
 308         int color;\r
 309         while( en.hasMoreElements() )\r
 310         {\r
 311           type = en.nextElement().toString();\r
 312           color = Integer.parseInt( visible.get(type).toString() );\r
 313           out.append(type + "\t"\r
 314                      + jalview.util.Format.getHexString(\r
 315                          new java.awt.Color(color)  )\r
 316                      +"\n");\r
 317         }\r
 318 \r
 319         //Work out which groups are both present and visible\r
 320         Vector groups = new Vector();\r
 321         int groupIndex = 0;\r
 322 \r
 323         for(int i=0; i<seqs.length; i++)\r
 324         {\r
 325           next = seqs[i].getSequenceFeatures();\r
 326           if(next!=null)\r
 327           {\r
 328             for(int j=0; j<next.length; j++)\r
 329             {\r
 330               if (!visible.containsKey(next[j].type))\r
 331                 continue;\r
 332 \r
 333               if (    next[j].featureGroup != null\r
 334                   && !groups.contains(next[j].featureGroup))\r
 335                 groups.addElement(next[j].featureGroup);\r
 336             }\r
 337           }\r
 338         }\r
 339 \r
 340         String group = null;\r
 341 \r
 342         do\r
 343         {\r
 344 \r
 345 \r
 346           if (groups.size() > 0 && groupIndex < groups.size())\r
 347           {\r
 348             group = groups.elementAt(groupIndex).toString();\r
 349             out.append("\nSTARTGROUP\t" + group + "\n");\r
 350           }\r
 351           else\r
 352             group = null;\r
 353 \r
 354           for (int i = 0; i < seqs.length; i++)\r
 355           {\r
 356             next = seqs[i].getSequenceFeatures();\r
 357             if (next != null)\r
 358             {\r
 359               for (int j = 0; j < next.length; j++)\r
 360               {\r
 361                 if (!visible.containsKey(next[j].type))\r
 362                   continue;\r
 363 \r
 364                 if (group != null\r
 365                     && (next[j].featureGroup==null\r
 366                         || !next[j].featureGroup.equals(group))\r
 367                   )\r
 368                   continue;\r
 369 \r
 370                 if(group==null && next[j].featureGroup!=null)\r
 371                   continue;\r
 372 \r
 373                 if(next[j].description==null || next[j].description.equals(""))\r
 374                   out.append(next[j].type+"\t");\r
 375                 else\r
 376                   out.append(next[j].description + "\t");\r
 377 \r
 378                 out.append(  seqs[i].getName() + "\t-1\t"\r
 379                            + next[j].begin + "\t"\r
 380                            + next[j].end + "\t"\r
 381                            + next[j].type + "\n"\r
 382                     );\r
 383               }\r
 384             }\r
 385           }\r
 386 \r
 387           if(group!=null)\r
 388           {\r
 389             out.append("ENDGROUP\t"+group+"\n");\r
 390             groupIndex++;\r
 391           }\r
 392           else\r
 393             break;\r
 394 \r
 395         }\r
 396         while(groupIndex < groups.size()+1);\r
 397 \r
 398 \r
 399       return out.toString();\r
 400     }\r
 401 \r
 402     public String printGFFFormat(SequenceI [] seqs, Hashtable visible)\r
 403     {\r
 404       StringBuffer out = new StringBuffer();\r
 405       SequenceFeature [] next;\r
 406 \r
 407       for(int i=0; i<seqs.length; i++)\r
 408       {\r
 409         if(seqs[i].getSequenceFeatures()!=null)\r
 410         {\r
 411           next = seqs[i].getSequenceFeatures();\r
 412           for(int j=0; j<next.length; j++)\r
 413           {\r
 414             if(!visible.containsKey(next[j].type))\r
 415               continue;\r
 416 \r
 417             out.append(seqs[i].getName() + "\t"\r
 418                        + next[j].description + "\t"\r
 419                        + next[j].type  + "\t"\r
 420                        + next[j].begin + "\t"\r
 421                        + next[j].end   + "\t"\r
 422                        + next[j].score + "\t"\r
 423                       );\r
 424 \r
 425             if(next[j].getValue("STRAND")!=null)\r
 426               out.append(next[j].getValue("STRAND")+"\t");\r
 427             else\r
 428               out.append(".\t");\r
 429             if(next[j].getValue("FRAME")!=null)\r
 430               out.append(next[j].getValue("FRAME")+"\n");\r
 431             else\r
 432               out.append(".\n");\r
 433 \r
 434           }\r
 435         }\r
 436       }\r
 437 \r
 438       return out.toString();\r
 439     }\r
 440 \r
 441     public void parse()\r
 442     {\r
 443       //IGNORED\r
 444     }\r
 445 \r
 446     /**\r
 447      * DOCUMENT ME!\r
 448      *\r
 449      * @return DOCUMENT ME!\r
 450      */\r
 451     public String print()\r
 452     {\r
 453         return "USE printGFFFormat() or printJalviewFormat()";\r
 454     }\r
 455 }\r