X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fio%2FFeaturesFile.java;h=6958923c2161212d66f7534fd72ab0e397ff2e44;hb=a7f8fa9c06d82efbbdbc6a0bf1054bb21610b353;hp=cfc08fbdcaebb2e8917e9fe0e0c8d0c27e342806;hpb=153dd62dc91da13ae732600e6ea55ddbe15eab39;p=jalview.git

diff --git a/src/jalview/io/FeaturesFile.java b/src/jalview/io/FeaturesFile.java
index cfc08fb..6958923 100755
--- a/src/jalview/io/FeaturesFile.java
+++ b/src/jalview/io/FeaturesFile.java
@@ -1,6 +1,6 @@
 /*
- * Jalview - A Sequence Alignment Editor and Viewer (Version 2.6)
- * Copyright (C) 2010 J Procter, AM Waterhouse, G Barton, M Clamp, S Searle
+ * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
+ * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, G Barton, M Clamp, S Searle
  * 
  * This file is part of Jalview.
  * 
@@ -20,6 +20,15 @@ package jalview.io;
 import java.io.*;
 import java.util.*;
 
+import javax.xml.parsers.ParserConfigurationException;
+
+import org.xml.sax.SAXException;
+
+import fr.orsay.lri.varna.exceptions.ExceptionFileFormatOrSyntax;
+import fr.orsay.lri.varna.exceptions.ExceptionLoadingFailed;
+import fr.orsay.lri.varna.exceptions.ExceptionPermissionDenied;
+
+import jalview.analysis.SequenceIdMatcher;
 import jalview.datamodel.*;
 import jalview.schemes.*;
 import jalview.util.Format;
@@ -58,40 +67,77 @@ public class FeaturesFile extends AlignFile
    * 
    * @throws IOException
    *           DOCUMENT ME!
+ * @throws SAXException 
+ * @throws ParserConfigurationException 
+ * @throws ExceptionFileFormatOrSyntax 
+ * @throws ExceptionLoadingFailed 
+ * @throws ExceptionPermissionDenied 
    */
-  public FeaturesFile(String inFile, String type) throws IOException
+  public FeaturesFile(String inFile, String type) throws IOException, ExceptionFileFormatOrSyntax, ParserConfigurationException, SAXException, ExceptionPermissionDenied, ExceptionLoadingFailed
   {
     super(inFile, type);
   }
 
-  public FeaturesFile(FileParse source) throws IOException
+  public FeaturesFile(FileParse source) throws IOException, ExceptionFileFormatOrSyntax, ParserConfigurationException, SAXException, ExceptionPermissionDenied, ExceptionLoadingFailed
   {
     super(source);
   }
 
   /**
-   * The Application can render HTML, but the applet will remove HTML tags and
-   * replace links with %LINK% Both need to read links in HTML however
-   * 
-   * @throws IOException
-   *           DOCUMENT ME!
+   * Parse GFF or sequence features file using case-independent matching, discarding URLs
+   * @param align - alignment/dataset containing sequences that are to be annotated
+   * @param colours - hashtable to store feature colour definitions
+   * @param removeHTML - process html strings into plain text
+   * @return true if features were added
    */
   public boolean parse(AlignmentI align, Hashtable colours,
           boolean removeHTML)
   {
-    return parse(align, colours, null, removeHTML);
+    return parse(align, colours, null, removeHTML, false);
   }
 
   /**
-   * The Application can render HTML, but the applet will remove HTML tags and
-   * replace links with %LINK% Both need to read links in HTML however
-   * 
-   * @throws IOException
-   *           DOCUMENT ME!
+   * Parse GFF or sequence features file optionally using case-independent matching, discarding URLs
+   * @param align - alignment/dataset containing sequences that are to be annotated
+   * @param colours - hashtable to store feature colour definitions
+   * @param removeHTML - process html strings into plain text
+   * @param relaxedIdmatching - when true, ID matches to compound sequence IDs are allowed
+   * @return true if features were added
+   */
+  public boolean parse(AlignmentI align, 
+          Hashtable colours, boolean removeHTML, boolean relaxedIdMatching)
+  {
+    return parse(align, colours, null, removeHTML, relaxedIdMatching);
+  }
+
+  /**
+   * Parse GFF or sequence features file optionally using case-independent matching
+   * @param align - alignment/dataset containing sequences that are to be annotated
+   * @param colours - hashtable to store feature colour definitions
+   * @param featureLink - hashtable to store associated URLs 
+   * @param removeHTML - process html strings into plain text
+   * @return true if features were added
    */
   public boolean parse(AlignmentI align, Hashtable colours,
           Hashtable featureLink, boolean removeHTML)
   {
+    return parse(align, colours, featureLink, removeHTML, false);
+  }
+
+  /**
+  /**
+   * Parse GFF or sequence features file 
+   * @param align - alignment/dataset containing sequences that are to be annotated
+   * @param colours - hashtable to store feature colour definitions
+   * @param featureLink - hashtable to store associated URLs 
+   * @param removeHTML - process html strings into plain text
+   * @param relaxedIdmatching - when true, ID matches to compound sequence IDs are allowed
+   * @return true if features were added
+   */
+  public boolean parse(AlignmentI align,
+          Hashtable colours, Hashtable featureLink, boolean removeHTML, boolean relaxedIdmatching)
+  {
+
     String line = null;
     try
     {
@@ -165,24 +211,27 @@ public class FeaturesFile extends AlignFile
               // fifth is either 'above','below', or 'none'.
               // sixth is a float value and only required when fifth is either
               // 'above' or 'below'.
-              StringTokenizer gcol = new StringTokenizer(colscheme, "|", true);
+              StringTokenizer gcol = new StringTokenizer(colscheme, "|",
+                      true);
               // set defaults
               int threshtype = AnnotationColourGradient.NO_THRESHOLD;
               float min = Float.MIN_VALUE, max = Float.MAX_VALUE, threshval = Float.NaN;
               boolean labelCol = false;
               // Parse spec line
               String mincol = gcol.nextToken();
-              if (mincol=="|")
+              if (mincol == "|")
               {
                 System.err
-                .println("Expected either 'label' or a colour specification in the line: "+line );
+                        .println("Expected either 'label' or a colour specification in the line: "
+                                + line);
                 continue;
               }
               String maxcol = null;
               if (mincol.toLowerCase().indexOf("label") == 0)
               {
                 labelCol = true;
-                mincol = (gcol.hasMoreTokens() ? gcol.nextToken() : null); // skip '|'
+                mincol = (gcol.hasMoreTokens() ? gcol.nextToken() : null); // skip
+                                                                           // '|'
                 mincol = (gcol.hasMoreTokens() ? gcol.nextToken() : null);
               }
               String abso = null, minval, maxval;
@@ -191,16 +240,20 @@ public class FeaturesFile extends AlignFile
                 // at least four more tokens
                 if (mincol.equals("|"))
                 {
-                  mincol="";
-                } else {
+                  mincol = "";
+                }
+                else
+                {
                   gcol.nextToken(); // skip next '|'
                 }
                 // continue parsing rest of line
                 maxcol = gcol.nextToken();
                 if (maxcol.equals("|"))
                 {
-                  maxcol="";
-                } else {
+                  maxcol = "";
+                }
+                else
+                {
                   gcol.nextToken(); // skip next '|'
                 }
                 abso = gcol.nextToken();
@@ -216,7 +269,8 @@ public class FeaturesFile extends AlignFile
                   gcol.nextToken(); // skip next '|'
                 }
                 maxval = gcol.nextToken();
-                if (gcol.hasMoreTokens()) {
+                if (gcol.hasMoreTokens())
+                {
                   gcol.nextToken(); // skip next '|'
                 }
                 try
@@ -361,7 +415,7 @@ public class FeaturesFile extends AlignFile
             // Still possible this is an old Jalview file,
             // which does not have type colours at the beginning
             seqId = token = st.nextToken();
-            seq = align.findName(seqId, true);
+            seq = findName(align, seqId, relaxedIdmatching);
             if (seq != null)
             {
               desc = st.nextToken();
@@ -470,7 +524,7 @@ public class FeaturesFile extends AlignFile
 
           if (!token.equals("ID_NOT_SPECIFIED"))
           {
-            seq = align.findName(seqId = token, true);
+            seq = findName(align, seqId = token, relaxedIdmatching);
             st.nextToken();
           }
           else
@@ -540,88 +594,64 @@ public class FeaturesFile extends AlignFile
           GFFFile = false;
         }
       }
+      resetMatcher();
     } catch (Exception ex)
     {
       System.out.println(line);
       System.out.println("Error parsing feature file: " + ex + "\n" + line);
       ex.printStackTrace(System.err);
+      resetMatcher();
       return false;
     }
 
     return true;
   }
 
-  public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
+  private AlignmentI lastmatchedAl = null;
+
+  private SequenceIdMatcher matcher = null;
+
+  /**
+   * clear any temporary handles used to speed up ID matching
+   */
+  private void resetMatcher()
   {
-    if (sf.getDescription() == null)
+    lastmatchedAl = null;
+    matcher = null;
+  }
+
+  private SequenceI findName(AlignmentI align, String seqId,
+          boolean relaxedIdMatching)
+  {
+    SequenceI match = null;
+    if (relaxedIdMatching)
     {
-      return;
+      if (lastmatchedAl != align)
+      {
+        matcher = new SequenceIdMatcher(
+                (lastmatchedAl = align).getSequencesArray());
+      }
+      match = matcher.findIdMatch(seqId);
     }
-
-    if (removeHTML
-            && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
+    else
     {
-      removeHTML = false;
+      match = align.findName(seqId, true);
     }
+    return match;
+  }
 
-    StringBuffer sb = new StringBuffer();
-    StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
-    String token, link;
-    int startTag;
-    String tag = null;
-    while (st.hasMoreElements())
+  public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
+  {
+    if (sf.getDescription() == null)
     {
-      token = st.nextToken("&>");
-      if (token.equalsIgnoreCase("html") || token.startsWith("/"))
-      {
-        continue;
-      }
-
-      tag = null;
-      startTag = token.indexOf("<");
-
-      if (startTag > -1)
-      {
-        tag = token.substring(startTag + 1);
-        token = token.substring(0, startTag);
-      }
-
-      if (tag != null && tag.toUpperCase().startsWith("A HREF="))
-      {
-        if (token.length() > 0)
-        {
-          sb.append(token);
-        }
-        link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
-        String label = st.nextToken("<>");
-        sf.addLink(label + "|" + link);
-        sb.append(label + "%LINK%");
-      }
-      else if (tag != null && tag.equalsIgnoreCase("br"))
-      {
-        sb.append("\n");
-      }
-      else if (token.startsWith("lt;"))
-      {
-        sb.append("<" + token.substring(3));
-      }
-      else if (token.startsWith("gt;"))
-      {
-        sb.append(">" + token.substring(3));
-      }
-      else if (token.startsWith("amp;"))
-      {
-        sb.append("&" + token.substring(4));
-      }
-      else
-      {
-        sb.append(token);
-      }
+      return;
     }
+    jalview.util.ParseHtmlBodyAndLinks parsed = new jalview.util.ParseHtmlBodyAndLinks(sf.getDescription(), removeHTML, newline);
 
-    if (removeHTML)
+    sf.description = (removeHTML) ? parsed.getNonHtmlContent() : sf.description;
+    for (String link:parsed.getLinks())
     {
-      sf.description = sb.toString();
+      sf.addLink(link);
     }
 
   }
@@ -718,7 +748,10 @@ public class FeaturesFile extends AlignFile
           color = Format.getHexString(new java.awt.Color(Integer
                   .parseInt(visible.get(type).toString())));
         }
-        out.append(type + "\t" + color + "\n");
+        out.append(type);
+        out.append("\t");
+        out.append(color);
+        out.append(newline);
       }
     }
     // Work out which groups are both present and visible
@@ -757,7 +790,10 @@ public class FeaturesFile extends AlignFile
       if (groups.size() > 0 && groupIndex < groups.size())
       {
         group = groups.elementAt(groupIndex).toString();
-        out.append("\nSTARTGROUP\t" + group + "\n");
+        out.append(newline);
+        out.append("STARTGROUP\t");
+        out.append(group);
+        out.append(newline);
       }
       else
       {
@@ -830,22 +866,28 @@ public class FeaturesFile extends AlignFile
 
               out.append("\t");
             }
-            out.append(seqs[i].getName()
-                    + "\t-1\t"
-                    + next[j].begin
-                    + "\t"
-                    + next[j].end
-                    + "\t"
-                    + next[j].type
-                    + ((next[j].score != Float.NaN) ? "\t" + next[j].score
-                            + "\n" : "\n"));
+            out.append(seqs[i].getName());
+            out.append("\t-1\t");
+            out.append(next[j].begin);
+            out.append("\t");
+            out.append(next[j].end);
+            out.append("\t");
+            out.append(next[j].type);
+            if (next[j].score != Float.NaN)
+            {
+              out.append("\t");
+              out.append(next[j].score);
+            }
+            out.append(newline);
           }
         }
       }
 
       if (group != null)
       {
-        out.append("ENDGROUP\t" + group + "\n");
+        out.append("ENDGROUP\t");
+        out.append(group);
+        out.append(newline);
         groupIndex++;
       }
       else
@@ -904,13 +946,23 @@ public class FeaturesFile extends AlignFile
             source = next[j].getDescription();
           }
 
-          out.append(seqs[i].getName() + "\t" + source + "\t"
-                  + next[j].type + "\t" + next[j].begin + "\t"
-                  + next[j].end + "\t" + next[j].score + "\t");
+          out.append(seqs[i].getName());
+          out.append("\t");
+          out.append(source);
+          out.append("\t");
+          out.append(next[j].type);
+          out.append("\t");
+          out.append(next[j].begin);
+          out.append("\t");
+          out.append(next[j].end);
+          out.append("\t");
+          out.append(next[j].score);
+          out.append("\t");
 
           if (next[j].getValue("STRAND") != null)
           {
-            out.append(next[j].getValue("STRAND") + "\t");
+            out.append(next[j].getValue("STRAND"));
+            out.append("\t");
           }
           else
           {
@@ -925,13 +977,15 @@ public class FeaturesFile extends AlignFile
           {
             out.append(".");
           }
+          // TODO: verify/check GFF - should there be a /t here before attribute
+          // output ?
 
           if (next[j].getValue("ATTRIBUTES") != null)
           {
             out.append(next[j].getValue("ATTRIBUTES"));
           }
 
-          out.append("\n");
+          out.append(newline);
 
         }
       }
@@ -957,4 +1011,5 @@ public class FeaturesFile extends AlignFile
   {
     return "USE printGFFFormat() or printJalviewFormat()";
   }
+
 }