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
28 * Parse and create Jalview Features files
\r
29 * Detects GFF format features files and parses.
\r
30 * Does not implement standard print() - call specific printFeatures or printGFF.
\r
31 * Uses AlignmentI.findSequence(String id) to find the sequence object for the features annotation - this normally works on an exact match.
\r
33 * @version $Revision$
\r
35 public class FeaturesFile
\r
39 * Creates a new FeaturesFile object.
\r
41 public FeaturesFile()
\r
46 * Creates a new FeaturesFile object.
\r
48 * @param inFile DOCUMENT ME!
\r
49 * @param type DOCUMENT ME!
\r
51 * @throws IOException DOCUMENT ME!
\r
53 public FeaturesFile(String inFile, String type)
\r
56 super(inFile, type);
\r
60 * The Application can render HTML, but the applet will
\r
61 * remove HTML tags and replace links with %LINK%
\r
62 * Both need to read links in HTML however
\r
64 * @throws IOException DOCUMENT ME!
\r
66 public boolean parse(AlignmentI align,
\r
70 return parse(align, colours, null, removeHTML);
\r
74 * The Application can render HTML, but the applet will
\r
75 * remove HTML tags and replace links with %LINK%
\r
76 * Both need to read links in HTML however
\r
78 * @throws IOException DOCUMENT ME!
\r
80 public boolean parse(AlignmentI align,
\r
82 Hashtable featureLink,
\r
88 SequenceI seq = null;
\r
89 String type, desc, token = null;
\r
91 int index, start, end;
\r
95 String featureGroup = null, groupLink = null;
\r
96 Hashtable typeLink = new Hashtable();
\r
98 boolean GFFFile = true;
\r
100 while ( (line = nextLine()) != null)
\r
102 if (line.startsWith("#"))
\r
107 st = new StringTokenizer(line, "\t");
\r
108 if (st.countTokens() > 1 && st.countTokens() < 4)
\r
111 type = st.nextToken();
\r
112 if (type.equalsIgnoreCase("startgroup"))
\r
114 featureGroup = st.nextToken();
\r
115 if (st.hasMoreElements())
\r
117 groupLink = st.nextToken();
\r
118 featureLink.put(featureGroup, groupLink);
\r
121 else if (type.equalsIgnoreCase("endgroup"))
\r
123 //We should check whether this is the current group,
\r
124 //but at present theres no way of showing more than 1 group
\r
126 featureGroup = null;
\r
131 UserColourScheme ucs = new UserColourScheme(st.nextToken());
\r
132 colours.put(type, ucs.findColour('A'));
\r
133 if (st.hasMoreElements())
\r
135 String link = st.nextToken();
\r
136 typeLink.put(type, link);
\r
137 if (featureLink == null)
\r
139 featureLink = new Hashtable();
\r
141 featureLink.put(type, link);
\r
148 while (st.hasMoreElements())
\r
153 // Still possible this is an old Jalview file,
\r
154 // which does not have type colours at the beginning
\r
155 token = st.nextToken();
\r
156 seq = align.findName(token);
\r
159 desc = st.nextToken();
\r
160 type = st.nextToken();
\r
162 start = Integer.parseInt(st.nextToken());
\r
163 } catch (NumberFormatException ex)
\r
168 end = Integer.parseInt(st.nextToken());
\r
170 catch (NumberFormatException ex)
\r
176 score = new Float(st.nextToken()).floatValue();
\r
178 catch (NumberFormatException ex)
\r
183 sf = new SequenceFeature(type, desc, start, end, score, null);
\r
187 sf.setValue("STRAND", st.nextToken());
\r
188 sf.setValue("FRAME", st.nextToken());
\r
190 catch (Exception ex)
\r
193 if (st.hasMoreTokens())
\r
195 StringBuffer attributes = new StringBuffer();
\r
196 while (st.hasMoreTokens())
\r
198 attributes.append("\t" + st.nextElement());
\r
200 sf.setValue("ATTRIBUTES", attributes.toString());
\r
203 seq.addSequenceFeature(sf);
\r
209 if (GFFFile && seq == null)
\r
215 desc = st.nextToken();
\r
218 token = st.nextToken();
\r
219 if (!token.equals("ID_NOT_SPECIFIED"))
\r
221 seq = align.findName(token);
\r
228 index = Integer.parseInt(st.nextToken());
\r
229 seq = align.getSequenceAt(index);
\r
231 catch (NumberFormatException ex)
\r
239 System.out.println("Sequence not found: " + line);
\r
243 start = Integer.parseInt(st.nextToken());
\r
244 end = Integer.parseInt(st.nextToken());
\r
246 type = st.nextToken();
\r
248 if (!colours.containsKey(type))
\r
250 // Probably the old style groups file
\r
251 UserColourScheme ucs = new UserColourScheme(type);
\r
252 colours.put(type, ucs.findColour('A'));
\r
255 sf = new SequenceFeature(type, desc, "", start, end, featureGroup);
\r
257 seq.addSequenceFeature(sf);
\r
259 if (groupLink != null && removeHTML)
\r
261 sf.addLink(groupLink);
\r
262 sf.description += "%LINK%";
\r
264 if (typeLink.containsKey(type) && removeHTML)
\r
266 sf.addLink(typeLink.get(type).toString());
\r
267 sf.description += "%LINK%";
\r
270 parseDescriptionHTML(sf, removeHTML);
\r
272 //If we got here, its not a GFFFile
\r
277 catch (Exception ex)
\r
279 System.out.println(line);
\r
280 System.out.println("Error parsing feature file: " + ex + "\n" + line);
\r
281 ex.printStackTrace(System.err);
\r
288 public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
\r
290 if (sf.getDescription() == null)
\r
295 if (removeHTML && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
\r
297 removeHTML = false;
\r
300 StringBuffer sb = new StringBuffer();
\r
301 StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
\r
302 String token, link;
\r
305 while (st.hasMoreElements())
\r
307 token = st.nextToken("&>");
\r
308 if (token.equalsIgnoreCase("html") || token.startsWith("/"))
\r
314 startTag = token.indexOf("<");
\r
318 tag = token.substring(startTag + 1);
\r
319 token = token.substring(0, startTag);
\r
322 if (tag != null && tag.toUpperCase().startsWith("A HREF="))
\r
324 if (token.length() > 0)
\r
328 link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
\r
329 String label = st.nextToken("<>");
\r
330 sf.addLink(label + "|" + link);
\r
331 sb.append(label + "%LINK%");
\r
333 else if (tag != null && tag.equalsIgnoreCase("br"))
\r
337 else if (token.startsWith("lt;"))
\r
339 sb.append("<" + token.substring(3));
\r
341 else if (token.startsWith("gt;"))
\r
343 sb.append(">" + token.substring(3));
\r
345 else if (token.startsWith("amp;"))
\r
347 sb.append("&" + token.substring(4));
\r
357 sf.description = sb.toString();
\r
365 * @param s DOCUMENT ME!
\r
366 * @param len DOCUMENT ME!
\r
367 * @param gaps DOCUMENT ME!
\r
368 * @param displayId DOCUMENT ME!
\r
370 * @return DOCUMENT ME!
\r
372 public String printJalviewFormat(SequenceI[] seqs,
\r
375 StringBuffer out = new StringBuffer();
\r
376 SequenceFeature[] next;
\r
378 if (visible == null || visible.size() < 1)
\r
380 return "No Features Visible";
\r
383 Enumeration en = visible.keys();
\r
386 while (en.hasMoreElements())
\r
388 type = en.nextElement().toString();
\r
389 color = Integer.parseInt(visible.get(type).toString());
\r
390 out.append(type + "\t"
\r
391 + jalview.util.Format.getHexString(
\r
392 new java.awt.Color(color))
\r
396 //Work out which groups are both present and visible
\r
397 Vector groups = new Vector();
\r
398 int groupIndex = 0;
\r
400 for (int i = 0; i < seqs.length; i++)
\r
402 next = seqs[i].getSequenceFeatures();
\r
405 for (int j = 0; j < next.length; j++)
\r
407 if (!visible.containsKey(next[j].type))
\r
412 if (next[j].featureGroup != null
\r
413 && !groups.contains(next[j].featureGroup))
\r
415 groups.addElement(next[j].featureGroup);
\r
421 String group = null;
\r
426 if (groups.size() > 0 && groupIndex < groups.size())
\r
428 group = groups.elementAt(groupIndex).toString();
\r
429 out.append("\nSTARTGROUP\t" + group + "\n");
\r
436 for (int i = 0; i < seqs.length; i++)
\r
438 next = seqs[i].getSequenceFeatures();
\r
441 for (int j = 0; j < next.length; j++)
\r
443 if (!visible.containsKey(next[j].type))
\r
449 && (next[j].featureGroup == null
\r
450 || !next[j].featureGroup.equals(group))
\r
456 if (group == null && next[j].featureGroup != null)
\r
461 if (next[j].description == null || next[j].description.equals(""))
\r
463 out.append(next[j].type + "\t");
\r
467 if (next[j].links != null
\r
468 && next[j].getDescription().indexOf("<html>") == -1)
\r
470 out.append("<html>");
\r
473 out.append(next[j].description + " ");
\r
474 if (next[j].links != null)
\r
476 for (int l = 0; l < next[j].links.size(); l++)
\r
478 String label = next[j].links.elementAt(l).toString();
\r
479 String href = label.substring(label.indexOf("|") + 1);
\r
480 label = label.substring(0, label.indexOf("|"));
\r
482 if (next[j].description.indexOf(href) == -1)
\r
484 out.append("<a href=\""
\r
492 if (next[j].getDescription().indexOf("</html>") == -1)
\r
494 out.append("</html>");
\r
501 out.append(seqs[i].getName() + "\t-1\t"
\r
502 + next[j].begin + "\t"
\r
503 + next[j].end + "\t"
\r
504 + next[j].type + "\n"
\r
512 out.append("ENDGROUP\t" + group + "\n");
\r
521 while (groupIndex < groups.size() + 1);
\r
523 return out.toString();
\r
526 public String printGFFFormat(SequenceI[] seqs, Hashtable visible)
\r
528 StringBuffer out = new StringBuffer();
\r
529 SequenceFeature[] next;
\r
532 for (int i = 0; i < seqs.length; i++)
\r
534 if (seqs[i].getSequenceFeatures() != null)
\r
536 next = seqs[i].getSequenceFeatures();
\r
537 for (int j = 0; j < next.length; j++)
\r
539 if (!visible.containsKey(next[j].type))
\r
544 source = next[j].featureGroup;
\r
545 if (source == null)
\r
547 source = next[j].getDescription();
\r
550 out.append(seqs[i].getName() + "\t"
\r
552 + next[j].type + "\t"
\r
553 + next[j].begin + "\t"
\r
554 + next[j].end + "\t"
\r
555 + next[j].score + "\t"
\r
558 if (next[j].getValue("STRAND") != null)
\r
560 out.append(next[j].getValue("STRAND") + "\t");
\r
567 if (next[j].getValue("FRAME") != null)
\r
569 out.append(next[j].getValue("FRAME"));
\r
576 if (next[j].getValue("ATTRIBUTES") != null)
\r
578 out.append(next[j].getValue("ATTRIBUTES"));
\r
587 return out.toString();
\r
590 public void parse()
\r
598 * @return DOCUMENT ME!
\r
600 public String print()
\r
602 return "USE printGFFFormat() or printJalviewFormat()";
\r