X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fdatamodel%2FSequence.java;h=96747e4fba3a1c36dd4b227c7fe67eeb25a7b83b;hb=ff450fad8709ae81919af7a15ea382af7292794c;hp=0055d0e605b71e889530ed3fee929e1e8b867cf5;hpb=4c6cd08af580ff73cfce39bec34ae4658a4b4b34;p=jalview.git

diff --git a/src/jalview/datamodel/Sequence.java b/src/jalview/datamodel/Sequence.java
index 0055d0e..96747e4 100755
--- a/src/jalview/datamodel/Sequence.java
+++ b/src/jalview/datamodel/Sequence.java
@@ -31,6 +31,7 @@ import jalview.util.StringUtils;
 
 import java.util.ArrayList;
 import java.util.Arrays;
+import java.util.BitSet;
 import java.util.Collections;
 import java.util.Enumeration;
 import java.util.List;
@@ -90,6 +91,21 @@ public class Sequence extends ASequence implements SequenceI
 
   private SequenceFeatures sequenceFeatureStore;
 
+  /*
+   * A cursor holding the approximate current view position to the sequence,
+   * as determined by findIndex or findPosition or findPositions.
+   * Using a cursor as a hint allows these methods to be more performant for
+   * large sequences.
+   */
+  private SequenceCursor cursor;
+
+  /*
+   * A number that should be incremented whenever the sequence is edited.
+   * If the value matches the cursor token, then we can trust the cursor,
+   * if not then it should be recomputed. 
+   */
+  private int changeCount;
+
   /**
    * Creates a new Sequence object.
    * 
@@ -244,9 +260,8 @@ public class Sequence extends ASequence implements SequenceI
   protected void initSeqFrom(SequenceI seq,
           AlignmentAnnotation[] alAnnotation)
   {
-    char[] oseq = seq.getSequence();
-    initSeqAndName(seq.getName(), Arrays.copyOf(oseq, oseq.length),
-            seq.getStart(), seq.getEnd());
+    char[] oseq = seq.getSequence(); // returns a copy of the array
+    initSeqAndName(seq.getName(), oseq, seq.getStart(), seq.getEnd());
 
     description = seq.getDescription();
     if (seq != datasetSequence)
@@ -529,6 +544,7 @@ public class Sequence extends ASequence implements SequenceI
   {
     this.sequence = seq.toCharArray();
     checkValidRange();
+    sequenceChanged();
   }
 
   @Override
@@ -546,7 +562,9 @@ public class Sequence extends ASequence implements SequenceI
   @Override
   public char[] getSequence()
   {
-    return sequence;
+    // return sequence;
+    return sequence == null ? null : Arrays.copyOf(sequence,
+            sequence.length);
   }
 
   /*
@@ -649,58 +667,320 @@ public class Sequence extends ASequence implements SequenceI
     return this.description;
   }
 
-  /*
-   * (non-Javadoc)
-   * 
-   * @see jalview.datamodel.SequenceI#findIndex(int)
+  /**
+   * {@inheritDoc}
    */
   @Override
   public int findIndex(int pos)
   {
-    // returns the alignment position for a residue
+    /*
+     * use a valid, hopefully nearby, cursor if available
+     */
+    if (isValidCursor(cursor))
+    {
+      return findIndex(pos, cursor);
+    }
+
     int j = start;
     int i = 0;
-    // Rely on end being at least as long as the length of the sequence.
+    int startColumn = 0;
+
+    /*
+     * traverse sequence from the start counting gaps; make a note of
+     * the column of the first residue to save in the cursor
+     */
     while ((i < sequence.length) && (j <= end) && (j <= pos))
     {
       if (!Comparison.isGap(sequence[i]))
       {
+        if (j == start)
+        {
+          startColumn = i;
+        }
         j++;
       }
-
       i++;
     }
 
-    if ((j == end) && (j < pos))
+    if (j == end && j < pos)
     {
       return end + 1;
     }
-    else
+
+    updateCursor(pos, i, startColumn);
+    return i;
+  }
+
+  /**
+   * Updates the cursor to the latest found residue and column position
+   * 
+   * @param residuePos
+   *          (start..)
+   * @param column
+   *          (1..)
+   * @param startColumn
+   *          column position of the first sequence residue
+   */
+  protected void updateCursor(int residuePos, int column, int startColumn)
+  {
+    int endColumn = cursor == null ? 0 : cursor.lastColumnPosition;
+    if (residuePos == this.end)
     {
-      return i;
+      endColumn = column;
     }
+
+    cursor = new SequenceCursor(this, residuePos, column, startColumn,
+            endColumn, this.changeCount);
   }
 
+  /**
+   * Answers the aligned column position (1..) for the given residue position
+   * (start..) given a 'hint' of a residue/column location in the neighbourhood.
+   * The hint may be left of, at, or to the right of the required position.
+   * 
+   * @param pos
+   * @param curs
+   * @return
+   */
+  protected int findIndex(int pos, SequenceCursor curs)
+  {
+    if (!isValidCursor(curs))
+    {
+      /*
+       * wrong or invalidated cursor, compute de novo
+       */
+      return findIndex(pos);
+    }
+
+    if (curs.residuePosition == pos)
+    {
+      return curs.columnPosition;
+    }
+
+    /*
+     * move left or right to find pos from hint.position
+     */
+    int col = curs.columnPosition - 1; // convert from base 1 to 0-based array
+                                       // index
+    int newPos = curs.residuePosition;
+    int delta = newPos > pos ? -1 : 1;
+
+    while (newPos != pos)
+    {
+      col += delta; // shift one column left or right
+      if (col < 0 || col == sequence.length)
+      {
+        break;
+      }
+      if (!Comparison.isGap(sequence[col]))
+      {
+        newPos += delta;
+      }
+    }
+
+    col++; // convert back to base 1
+    updateCursor(pos, col, curs.firstColumnPosition);
+
+    return col;
+  }
+
+  /**
+   * {@inheritDoc}
+   */
   @Override
-  public int findPosition(int i)
+  public int findPosition(final int column)
   {
+    /*
+     * use a valid, hopefully nearby, cursor if available
+     */
+    if (isValidCursor(cursor))
+    {
+      return findPosition(column + 1, cursor);
+    }
+    
+    // TODO recode this more naturally i.e. count residues only
+    // as they are found, not 'in anticipation'
+
+    /*
+     * traverse the sequence counting gaps; note the column position
+     * of the first residue, to save in the cursor
+     */
+    int firstResidueColumn = 0;
+    int lastPosFound = 0;
+    int lastPosFoundColumn = 0;
+    int seqlen = sequence.length;
+
+    if (seqlen > 0 && !Comparison.isGap(sequence[0]))
+    {
+      lastPosFound = start;
+      lastPosFoundColumn = 0;
+    }
+
     int j = 0;
     int pos = start;
-    int seqlen = sequence.length;
-    while ((j < i) && (j < seqlen))
+
+    while (j < column && j < seqlen)
     {
       if (!Comparison.isGap(sequence[j]))
       {
+        lastPosFound = pos;
+        lastPosFoundColumn = j;
+        if (pos == this.start)
+        {
+          firstResidueColumn = j;
+        }
         pos++;
       }
-
       j++;
     }
+    if (j < seqlen && !Comparison.isGap(sequence[j]))
+    {
+      lastPosFound = pos;
+      lastPosFoundColumn = j;
+      if (pos == this.start)
+      {
+        firstResidueColumn = j;
+      }
+    }
+
+    /*
+     * update the cursor to the last residue position found (if any)
+     * (converting column position to base 1)
+     */
+    if (lastPosFound != 0)
+    {
+      updateCursor(lastPosFound, lastPosFoundColumn + 1,
+              firstResidueColumn + 1);
+    }
 
     return pos;
   }
 
   /**
+   * Answers true if the given cursor is not null, is for this sequence object,
+   * and has a token value that matches this object's changeCount, else false.
+   * This allows us to ignore a cursor as 'stale' if the sequence has been
+   * modified since the cursor was created.
+   * 
+   * @param curs
+   * @return
+   */
+  protected boolean isValidCursor(SequenceCursor curs)
+  {
+    if (curs == null || curs.sequence != this || curs.token != changeCount)
+    {
+      return false;
+    }
+    /*
+     * sanity check against range
+     */
+    if (curs.columnPosition < 0 || curs.columnPosition > sequence.length)
+    {
+      return false;
+    }
+    if (curs.residuePosition < start || curs.residuePosition > end)
+    {
+      return false;
+    }
+    return true;
+  }
+
+  /**
+   * Answers the sequence position (start..) for the given aligned column
+   * position (1..), given a hint of a cursor in the neighbourhood. The cursor
+   * may lie left of, at, or to the right of the column position.
+   * 
+   * @param col
+   * @param curs
+   * @return
+   */
+  protected int findPosition(final int col, SequenceCursor curs)
+  {
+    if (!isValidCursor(curs))
+    {
+      /*
+       * wrong or invalidated cursor, compute de novo
+       */
+      return findPosition(col - 1);// ugh back to base 0
+    }
+
+    if (curs.columnPosition == col)
+    {
+      cursor = curs; // in case this method becomes public
+      return curs.residuePosition; // easy case :-)
+    }
+
+    if (curs.lastColumnPosition > 0 && curs.lastColumnPosition < col)
+    {
+      /*
+       * sequence lies entirely to the left of col
+       * - return last residue + 1
+       */
+      return end + 1;
+    }
+
+    if (curs.firstColumnPosition > 0 && curs.firstColumnPosition > col)
+    {
+      /*
+       * sequence lies entirely to the right of col
+       * - return first residue
+       */
+      return start;
+    }
+
+    // todo could choose closest to col out of column,
+    // firstColumnPosition, lastColumnPosition as a start point
+
+    /*
+     * move left or right to find pos from cursor position
+     */
+    int firstResidueColumn = curs.firstColumnPosition;
+    int column = curs.columnPosition - 1; // to base 0
+    int newPos = curs.residuePosition;
+    int delta = curs.columnPosition > col ? -1 : 1;
+    boolean gapped = false;
+    int lastFoundPosition = curs.residuePosition;
+    int lastFoundPositionColumn = curs.columnPosition;
+
+    while (column != col - 1)
+    {
+      column += delta; // shift one column left or right
+      if (column < 0 || column == sequence.length)
+      {
+        break;
+      }
+      gapped = Comparison.isGap(sequence[column]);
+      if (!gapped)
+      {
+        newPos += delta;
+        lastFoundPosition = newPos;
+        lastFoundPositionColumn = column + 1;
+        if (lastFoundPosition == this.start)
+        {
+          firstResidueColumn = column + 1;
+        }
+      }
+    }
+
+    if (cursor == null || lastFoundPosition != cursor.residuePosition)
+    {
+      updateCursor(lastFoundPosition, lastFoundPositionColumn,
+              firstResidueColumn);
+    }
+
+    /*
+     * hack to give position to the right if on a gap
+     * or beyond the length of the sequence (see JAL-2562)
+     */
+    if (delta > 0 && (gapped || column >= sequence.length))
+    {
+      newPos++;
+    }
+
+    return newPos;
+  }
+
+  /**
    * Returns an int array where indices correspond to each residue in the
    * sequence and the element value gives its position in the alignment
    * 
@@ -784,6 +1064,40 @@ public class Sequence extends ASequence implements SequenceI
   }
 
   @Override
+  public BitSet getInsertionsAsBits()
+  {
+    BitSet map = new BitSet();
+    int lastj = -1, j = 0;
+    int pos = start;
+    int seqlen = sequence.length;
+    while ((j < seqlen))
+    {
+      if (jalview.util.Comparison.isGap(sequence[j]))
+      {
+        if (lastj == -1)
+        {
+          lastj = j;
+        }
+      }
+      else
+      {
+        if (lastj != -1)
+        {
+          map.set(lastj, j);
+          lastj = -1;
+        }
+      }
+      j++;
+    }
+    if (lastj != -1)
+    {
+      map.set(lastj, j);
+      lastj = -1;
+    }
+    return map;
+  }
+
+  @Override
   public void deleteChars(int i, int j)
   {
     int newstart = start, newend = end;
@@ -856,6 +1170,7 @@ public class Sequence extends ASequence implements SequenceI
     start = newstart;
     end = newend;
     sequence = tmp;
+    sequenceChanged();
   }
 
   @Override
@@ -886,6 +1201,7 @@ public class Sequence extends ASequence implements SequenceI
     }
 
     sequence = tmp;
+    sequenceChanged();
   }
 
   @Override
@@ -1080,7 +1396,7 @@ public class Sequence extends ASequence implements SequenceI
 
   private boolean _isNa;
 
-  private long _seqhash = 0;
+  private int _seqhash = 0;
 
   /**
    * Answers false if the sequence is more than 85% nucleotide (ACGTU), else
@@ -1449,7 +1765,12 @@ public class Sequence extends ASequence implements SequenceI
   {
     int startPos = findPosition(fromColumn - 1); // convert base 1 to base 0
     int endPos = findPosition(toColumn - 1);
-
+    // to trace / debug behaviour:
+    // System.out
+    // .println(String
+    // .format("%s.findFeatures columns [%d-%d] positions [%d-%d] leaves cursor %s",
+    // getName(), fromColumn, toColumn, startPos,
+    // endPos, cursor));
     List<SequenceFeature> result = new ArrayList<>();
     if (datasetSequence != null)
     {
@@ -1495,4 +1816,44 @@ public class Sequence extends ASequence implements SequenceI
 
     return result;
   }
+
+  /**
+   * Invalidates any stale cursors (forcing recalculation) by incrementing the
+   * token that has to match the one presented by the cursor
+   */
+  @Override
+  public void sequenceChanged()
+  {
+    changeCount++;
+  }
+
+  /**
+   * {@inheritDoc}
+   */
+  @Override
+  public int replace(char c1, char c2)
+  {
+    if (c1 == c2)
+    {
+      return 0;
+    }
+    int count = 0;
+    synchronized (sequence)
+    {
+      for (int c = 0; c < sequence.length; c++)
+      {
+        if (sequence[c] == c1)
+        {
+          sequence[c] = c2;
+          count++;
+        }
+      }
+    }
+    if (count > 0)
+    {
+      sequenceChanged();
+    }
+
+    return count;
+  }
 }