2 * Jalview - A Sequence Alignment Editor and Viewer
\r
3 * Copyright (C) 2007 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
\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
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
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
24 import jalview.datamodel.*;
\r
25 import jalview.schemes.*;
\r
31 * @version $Revision$
\r
33 public class FeaturesFile
\r
37 * Creates a new FeaturesFile object.
\r
39 public FeaturesFile()
\r
44 * Creates a new FeaturesFile object.
\r
46 * @param inFile DOCUMENT ME!
\r
47 * @param type DOCUMENT ME!
\r
49 * @throws IOException DOCUMENT ME!
\r
51 public FeaturesFile(String inFile, String type)
\r
54 super(inFile, type);
\r
58 * The Application can render HTML, but the applet will
\r
59 * remove HTML tags and replace links with %LINK%
\r
60 * Both need to read links in HTML however
\r
62 * @throws IOException DOCUMENT ME!
\r
64 public boolean parse(AlignmentI align,
\r
68 return parse(align, colours, null, removeHTML);
\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
76 * @throws IOException DOCUMENT ME!
\r
78 public boolean parse(AlignmentI align,
\r
80 Hashtable featureLink,
\r
86 SequenceI seq = null;
\r
87 String type, desc, token = null;
\r
89 int index, start, end;
\r
93 String featureGroup = null, groupLink = null;
\r
94 Hashtable typeLink = new Hashtable();
\r
96 boolean GFFFile = true;
\r
98 while ( (line = nextLine()) != null)
\r
100 if (line.startsWith("#"))
\r
105 st = new StringTokenizer(line, "\t");
\r
106 if (st.countTokens() > 1 && st.countTokens() < 4)
\r
109 type = st.nextToken();
\r
110 if (type.equalsIgnoreCase("startgroup"))
\r
112 featureGroup = st.nextToken();
\r
113 if (st.hasMoreElements())
\r
115 groupLink = st.nextToken();
\r
116 featureLink.put(featureGroup, groupLink);
\r
119 else if (type.equalsIgnoreCase("endgroup"))
\r
121 //We should check whether this is the current group,
\r
122 //but at present theres no way of showing more than 1 group
\r
124 featureGroup = null;
\r
129 UserColourScheme ucs = new UserColourScheme(st.nextToken());
\r
130 colours.put(type, ucs.findColour('A'));
\r
131 if (st.hasMoreElements())
\r
133 String link = st.nextToken();
\r
134 typeLink.put(type, link);
\r
135 if (featureLink == null)
\r
137 featureLink = new Hashtable();
\r
139 featureLink.put(type, link);
\r
146 while (st.hasMoreElements())
\r
151 // Still possible this is an old Jalview file,
\r
152 // which does not have type colours at the beginning
\r
153 token = st.nextToken();
\r
154 seq = align.findName(token);
\r
157 desc = st.nextToken();
\r
158 type = st.nextToken();
\r
159 start = Integer.parseInt(st.nextToken());
\r
160 end = Integer.parseInt(st.nextToken());
\r
163 score = new Float(st.nextToken()).floatValue();
\r
165 catch (NumberFormatException ex)
\r
170 sf = new SequenceFeature(type, desc, start, end, score, null);
\r
174 sf.setValue("STRAND", st.nextToken());
\r
175 sf.setValue("FRAME", st.nextToken());
\r
177 catch (Exception ex)
\r
180 if (st.hasMoreTokens())
\r
182 StringBuffer attributes = new StringBuffer();
\r
183 while (st.hasMoreTokens())
\r
185 attributes.append("\t" + st.nextElement());
\r
187 sf.setValue("ATTRIBUTES", attributes.toString());
\r
190 seq.addSequenceFeature(sf);
\r
196 if (GFFFile && seq == null)
\r
202 desc = st.nextToken();
\r
205 token = st.nextToken();
\r
206 if (!token.equals("ID_NOT_SPECIFIED"))
\r
208 seq = align.findName(token);
\r
215 index = Integer.parseInt(st.nextToken());
\r
216 seq = align.getSequenceAt(index);
\r
218 catch (NumberFormatException ex)
\r
226 System.out.println("Sequence not found: " + line);
\r
230 start = Integer.parseInt(st.nextToken());
\r
231 end = Integer.parseInt(st.nextToken());
\r
233 type = st.nextToken();
\r
235 if (!colours.containsKey(type))
\r
237 // Probably the old style groups file
\r
238 UserColourScheme ucs = new UserColourScheme(type);
\r
239 colours.put(type, ucs.findColour('A'));
\r
242 sf = new SequenceFeature(type, desc, "", start, end, featureGroup);
\r
244 seq.addSequenceFeature(sf);
\r
246 if (groupLink != null && removeHTML)
\r
248 sf.addLink(groupLink);
\r
249 sf.description += "%LINK%";
\r
251 if (typeLink.containsKey(type) && removeHTML)
\r
253 sf.addLink(typeLink.get(type).toString());
\r
254 sf.description += "%LINK%";
\r
257 parseDescriptionHTML(sf, removeHTML);
\r
259 //If we got here, its not a GFFFile
\r
264 catch (Exception ex)
\r
266 System.out.println(line);
\r
267 System.out.println("Error parsing feature file: " + ex + "\n" + line);
\r
274 public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
\r
276 if (sf.getDescription() == null)
\r
281 if (removeHTML && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
\r
283 removeHTML = false;
\r
286 StringBuffer sb = new StringBuffer();
\r
287 StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
\r
288 String token, link;
\r
291 while (st.hasMoreElements())
\r
293 token = st.nextToken("&>");
\r
294 if (token.equalsIgnoreCase("html") || token.startsWith("/"))
\r
300 startTag = token.indexOf("<");
\r
304 tag = token.substring(startTag + 1);
\r
305 token = token.substring(0, startTag);
\r
308 if (tag != null && tag.toUpperCase().startsWith("A HREF="))
\r
310 if (token.length() > 0)
\r
314 link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
\r
315 String label = st.nextToken("<>");
\r
316 sf.addLink(label + "|" + link);
\r
317 sb.append(label + "%LINK%");
\r
319 else if (tag != null && tag.equalsIgnoreCase("br"))
\r
323 else if (token.startsWith("lt;"))
\r
325 sb.append("<" + token.substring(3));
\r
327 else if (token.startsWith("gt;"))
\r
329 sb.append(">" + token.substring(3));
\r
331 else if (token.startsWith("amp;"))
\r
333 sb.append("&" + token.substring(4));
\r
343 sf.description = sb.toString();
\r
351 * @param s DOCUMENT ME!
\r
352 * @param len DOCUMENT ME!
\r
353 * @param gaps DOCUMENT ME!
\r
354 * @param displayId DOCUMENT ME!
\r
356 * @return DOCUMENT ME!
\r
358 public String printJalviewFormat(SequenceI[] seqs,
\r
361 StringBuffer out = new StringBuffer();
\r
362 SequenceFeature[] next;
\r
364 if (visible == null || visible.size() < 1)
\r
366 return "No Features Visible";
\r
369 Enumeration en = visible.keys();
\r
372 while (en.hasMoreElements())
\r
374 type = en.nextElement().toString();
\r
375 color = Integer.parseInt(visible.get(type).toString());
\r
376 out.append(type + "\t"
\r
377 + jalview.util.Format.getHexString(
\r
378 new java.awt.Color(color))
\r
382 //Work out which groups are both present and visible
\r
383 Vector groups = new Vector();
\r
384 int groupIndex = 0;
\r
386 for (int i = 0; i < seqs.length; i++)
\r
388 next = seqs[i].getSequenceFeatures();
\r
391 for (int j = 0; j < next.length; j++)
\r
393 if (!visible.containsKey(next[j].type))
\r
398 if (next[j].featureGroup != null
\r
399 && !groups.contains(next[j].featureGroup))
\r
401 groups.addElement(next[j].featureGroup);
\r
407 String group = null;
\r
412 if (groups.size() > 0 && groupIndex < groups.size())
\r
414 group = groups.elementAt(groupIndex).toString();
\r
415 out.append("\nSTARTGROUP\t" + group + "\n");
\r
422 for (int i = 0; i < seqs.length; i++)
\r
424 next = seqs[i].getSequenceFeatures();
\r
427 for (int j = 0; j < next.length; j++)
\r
429 if (!visible.containsKey(next[j].type))
\r
435 && (next[j].featureGroup == null
\r
436 || !next[j].featureGroup.equals(group))
\r
442 if (group == null && next[j].featureGroup != null)
\r
447 if (next[j].description == null || next[j].description.equals(""))
\r
449 out.append(next[j].type + "\t");
\r
453 if (next[j].links != null
\r
454 && next[j].getDescription().indexOf("<html>") == -1)
\r
456 out.append("<html>");
\r
459 out.append(next[j].description + " ");
\r
460 if (next[j].links != null)
\r
462 for (int l = 0; l < next[j].links.size(); l++)
\r
464 String label = next[j].links.elementAt(l).toString();
\r
465 String href = label.substring(label.indexOf("|") + 1);
\r
466 label = label.substring(0, label.indexOf("|"));
\r
468 if (next[j].description.indexOf(href) == -1)
\r
470 out.append("<a href=\""
\r
478 if (next[j].getDescription().indexOf("</html>") == -1)
\r
480 out.append("</html>");
\r
487 out.append(seqs[i].getName() + "\t-1\t"
\r
488 + next[j].begin + "\t"
\r
489 + next[j].end + "\t"
\r
490 + next[j].type + "\n"
\r
498 out.append("ENDGROUP\t" + group + "\n");
\r
507 while (groupIndex < groups.size() + 1);
\r
509 return out.toString();
\r
512 public String printGFFFormat(SequenceI[] seqs, Hashtable visible)
\r
514 StringBuffer out = new StringBuffer();
\r
515 SequenceFeature[] next;
\r
518 for (int i = 0; i < seqs.length; i++)
\r
520 if (seqs[i].getSequenceFeatures() != null)
\r
522 next = seqs[i].getSequenceFeatures();
\r
523 for (int j = 0; j < next.length; j++)
\r
525 if (!visible.containsKey(next[j].type))
\r
530 source = next[j].featureGroup;
\r
531 if (source == null)
\r
533 source = next[j].getDescription();
\r
536 out.append(seqs[i].getName() + "\t"
\r
538 + next[j].type + "\t"
\r
539 + next[j].begin + "\t"
\r
540 + next[j].end + "\t"
\r
541 + next[j].score + "\t"
\r
544 if (next[j].getValue("STRAND") != null)
\r
546 out.append(next[j].getValue("STRAND") + "\t");
\r
553 if (next[j].getValue("FRAME") != null)
\r
555 out.append(next[j].getValue("FRAME"));
\r
562 if (next[j].getValue("ATTRIBUTES") != null)
\r
564 out.append(next[j].getValue("ATTRIBUTES"));
\r
573 return out.toString();
\r
576 public void parse()
\r
584 * @return DOCUMENT ME!
\r
586 public String print()
\r
588 return "USE printGFFFormat() or printJalviewFormat()";
\r