import jalview.datamodel.SequenceI;
import jalview.util.Comparison;
-import java.util.ArrayList;
-import java.util.List;
import java.util.Vector;
import com.stevesoft.pat.Regex;
+/**
+ * Implements the search algorithm for the Find dialog
+ */
public class Finder
{
- /**
- * Implements the search algorithms for the Find dialog box.
+ /*
+ * match residue locations
*/
- SearchResultsI searchResults;
+ private SearchResultsI searchResults;
- AlignmentI alignment;
+ /*
+ * sequences matched by id or description
+ */
+ private Vector<SequenceI> idMatch;
- SequenceGroup selection = null;
+ /*
+ * the alignment to search over
+ */
+ private AlignmentI alignment;
- Vector<SequenceI> idMatch = null;
+ /*
+ * (optional) selection to restrict search to
+ */
+ private SequenceGroup selection;
- boolean caseSensitive = false;
+ /*
+ * set true for case-sensitive search (default is false)
+ */
+ private boolean caseSensitive;
- private boolean includeDescription = false;
+ /*
+ * set true to search sequence description (default is false)
+ */
+ private boolean includeDescription;
- boolean findAll = false;
+ /*
+ * set true to return all matches (default is next match only)
+ */
+ private boolean findAll;
- Regex regex = null;
+ /*
+ * sequence index in alignment to search from
+ */
+ private int seqIndex;
- /**
- * holds last-searched position between calls to find(false)
+ /*
+ * residue position in sequence to search from, base 1
+ * (position of last match for a repeat search)
*/
- int seqIndex = 0, resIndex = -1;
+ private int resIndex;
- public Finder(AlignmentI alignment, SequenceGroup selection)
+ /**
+ * Constructor to start searching an alignment, optionally restricting results
+ * to a selection
+ *
+ * @param al
+ * @param sel
+ */
+ public Finder(AlignmentI al, SequenceGroup sel)
{
- this.alignment = alignment;
- this.selection = selection;
+ this(al, sel, 0, -1);
}
/**
- * restart search at given sequence and residue on alignment and (optionally)
- * contained in selection
+ * Constructor to resume search at given sequence and residue on alignment and
+ * (optionally) restricted to a selection
*
- * @param alignment
- * @param selectionGroup
- * @param seqIndex
- * @param resIndex
+ * @param al
+ * @param sel
+ * @param seqindex
+ * @param resindex
*/
- public Finder(AlignmentI alignment, SequenceGroup selectionGroup,
- int seqIndex, int resIndex)
+ public Finder(AlignmentI al, SequenceGroup sel, int seqindex,
+ int resindex)
{
- this(alignment, selectionGroup);
- this.seqIndex = seqIndex;
- this.resIndex = resIndex;
+ this.alignment = al;
+ this.selection = sel;
+ this.seqIndex = seqindex;
+ this.resIndex = resindex;
}
- public boolean find(String searchString)
+ /**
+ * Performs a find for the given search string. By default the next match is
+ * found, but if setFindAll(true) has been called, then all matches are found.
+ * Sequences matched by id or description can be retrieved by getIdMatch(),
+ * and matched residue patterns by getSearchResults().
+ *
+ * @param theSearchString
+ * @return
+ */
+ public void find(String theSearchString)
{
- boolean hasResults = false;
- if (!caseSensitive)
- {
- searchString = searchString.toUpperCase();
- }
- regex = new Regex(searchString);
+ String searchString = caseSensitive ? theSearchString.toUpperCase()
+ : theSearchString;
+ Regex regex = new Regex(searchString);
regex.setIgnoreCase(!caseSensitive);
searchResults = new SearchResults();
- idMatch = new Vector<SequenceI>();
- String item = null;
- boolean found = false;
- int end = alignment.getHeight();
-
- // /////////////////////////////////////////////
+ idMatch = new Vector<>();
- if (selection != null)
+ if (selection != null && selection.getSize() < 1)
{
- if ((selection.getSize() < 1)
- || ((selection.getEndRes() - selection.getStartRes()) < 2))
- {
- selection = null;
- }
+ selection = null; // ? ignore column-only selection
}
- SearchResultMatchI lastm = null;
- while (!found && (seqIndex < end))
+ boolean finished = false;
+ int end = alignment.getHeight();
+
+ while (!finished && (seqIndex < end))
{
SequenceI seq = alignment.getSequenceAt(seqIndex);
- if ((selection != null && selection.getSize() > 0)
- && !selection.getSequences(null).contains(seq))
+ if ((selection != null) && !selection.contains(seq))
{
+ // this sequence is not in the selection - skip to next sequence
seqIndex++;
resIndex = -1;
-
continue;
}
+
if (resIndex < 0)
{
+ /*
+ * at start of sequence; try find by residue number, in sequence id,
+ * or (optionally) in sequence description
+ */
resIndex = 0;
- // test for one off matches - sequence position and sequence ID
- // //// is the searchString a residue number?
- try
- {
- int res = Integer.parseInt(searchString);
- // possibly a residue number - check if valid for seq
- if (seq.getEnd() >= res)
- {
- searchResults.addResult(seq, res, res);
- hasResults = true;
- // resIndex=seq.getLength();
- // seqIndex++;
- if (!findAll)
- {
- found = true;
- break;
- }
- }
- } catch (NumberFormatException ex)
+ if (doNonMotifSearches(seq, searchString, regex))
{
+ return;
}
+ }
- if (regex.search(seq.getName()) && !idMatch.contains(seq))
- {
- idMatch.addElement(seq);
- hasResults = true;
- if (!findAll)
- {
- // stop and return the match
- found = true;
- break;
- }
- }
+ finished = searchSequenceString(seq, regex) && !findAll;
- if (isIncludeDescription() && seq.getDescription() != null
- && regex.search(seq.getDescription())
- && !idMatch.contains(seq))
- {
- idMatch.addElement(seq);
- hasResults = true;
- if (!findAll)
- {
- // stop and return the match
- found = true;
- break;
- }
- }
+ if (!finished)
+ {
+ seqIndex++;
+ resIndex = -1;
}
- item = seq.getSequenceAsString();
+ }
+ }
- if ((selection != null)
- && (selection.getEndRes() < alignment.getWidth() - 1))
+ /**
+ * Searches the sequence, starting from <code>resIndex</code> (base 1), and
+ * adds matches to <code>searchResults</code>. The search is restricted to the
+ * <code>selection</code> region if there is one. Answers true if any match is
+ * added, else false.
+ *
+ * @param seq
+ * @param regex
+ * @return
+ */
+ protected boolean searchSequenceString(SequenceI seq, Regex regex)
+ {
+ /*
+ * Restrict search to selected region if there is one
+ */
+ int seqColStart = 0;
+ int seqColEnd = seq.getLength() - 1;
+ int residueOffset = 0;
+ if (selection != null)
+ {
+ int selColEnd = selection.getEndRes();
+ int selColStart = selection.getStartRes();
+ if (selColStart > seqColEnd)
{
- item = item.substring(0, selection.getEndRes() + 1);
+ return false; // sequence doesn't reach selection region
}
+ seqColStart = selColStart;
+ seqColEnd = Math.min(seqColEnd, selColEnd);
+ residueOffset = seq.findPosition(selection.getStartRes())
+ - seq.getStart();
+ }
+ String seqString = seq.getSequenceAsString(seqColStart, seqColEnd + 1);
- // /Shall we ignore gaps???? - JBPNote: Add Flag for forcing this or not
- StringBuilder noGapsSB = new StringBuilder();
- int insertCount = 0;
- List<Integer> spaces = new ArrayList<Integer>();
+ String noGaps = AlignSeq.extractGaps(Comparison.GapChars, seqString);
- for (int j = 0; j < item.length(); j++)
- {
- if (!Comparison.isGap(item.charAt(j)))
- {
- noGapsSB.append(item.charAt(j));
- spaces.add(Integer.valueOf(insertCount));
- }
- else
- {
- insertCount++;
- }
- }
+ SearchResultMatchI lastMatch = null;
+ boolean found = false;
- String noGaps = noGapsSB.toString();
- for (int r = resIndex; r < noGaps.length(); r++)
+ for (int r = resIndex; r < noGaps.length(); r++)
+ {
+ /*
+ * searchFrom position is base 0, r is base 1,
+ * so search is from the position after the r'th residue
+ */
+ if (regex.searchFrom(noGaps, r))
{
-
- if (regex.searchFrom(noGaps, r))
+ resIndex = regex.matchedFrom();
+ resIndex += residueOffset; // add back #residues before selection region
+ int matchStartPosition = resIndex + seq.getStart();
+ int matchEndPosition = matchStartPosition + regex.charsMatched()
+ - 1;
+ if (lastMatch == null || !lastMatch.contains(seq,
+ matchStartPosition, matchEndPosition))
{
- resIndex = regex.matchedFrom();
-
- if ((selection != null && selection.getSize() > 0) && (resIndex
- + spaces.get(resIndex) < selection.getStartRes()))
- {
- continue;
- }
- // if invalid string used, then regex has no matched to/from
- int sres = seq.findPosition(resIndex + spaces.get(resIndex));
- int eres = seq.findPosition(regex.matchedTo() - 1
- + (spaces.get(regex.matchedTo() - 1)));
- // only add result if not contained in previous result
- if (lastm == null || (lastm.getSequence() != seq
- || (!(lastm.getStart() <= sres
- && lastm.getEnd() >= eres))))
- {
- lastm = searchResults.addResult(seq, sres, eres);
- }
- hasResults = true;
- if (!findAll)
- {
- // thats enough, break and display the result
- found = true;
- resIndex++;
-
- break;
- }
-
- r = resIndex;
+ lastMatch = searchResults.addResult(seq, matchStartPosition,
+ matchEndPosition);
+ found = true;
}
- else
+ if (!findAll)
{
- break;
+ resIndex++;
+ return true;
}
+ r = resIndex;
}
-
- if (!found)
+ else
{
- seqIndex++;
- resIndex = -1;
+ break;
}
}
-
- /**
- * We now search the Id string in the main search loop. for (int id = 0; id
- * < alignment.getHeight(); id++) { if
- * (regex.search(alignment.getSequenceAt(id).getName())) {
- * idMatch.addElement(alignment.getSequenceAt(id)); hasResults = true; } }
- */
- return hasResults;
- }
-
- /**
- * @return the alignment
- */
- public AlignmentI getAlignment()
- {
- return alignment;
- }
-
- /**
- * @param alignment
- * the alignment to set
- */
- public void setAlignment(AlignmentI alignment)
- {
- this.alignment = alignment;
+ return found;
}
/**
- * @return the caseSensitive
+ * Does searches other than for residue patterns. Currently this includes
+ * <ul>
+ * <li>find residue by position (if search string is a number)</li>
+ * <li>match search string to sequence id</li>
+ * <li>match search string to sequence description (optional)</li>
+ * </ul>
+ * Answers true if a match is found and we are not doing 'find all' (so this
+ * search action is complete), else false.
+ *
+ * @param seq
+ * @param searchString
+ * @param regex
+ * @return
*/
- public boolean isCaseSensitive()
+ protected boolean doNonMotifSearches(SequenceI seq, String searchString,
+ Regex regex)
{
- return caseSensitive;
+ if (searchForResidueNumber(seq, searchString) && !findAll)
+ {
+ return true;
+ }
+ if (searchSequenceName(seq, regex) && !findAll)
+ {
+ return true;
+ }
+ if (searchSequenceDescription(seq, regex) && !findAll)
+ {
+ return true;
+ }
+ return false;
}
/**
- * @param caseSensitive
- * the caseSensitive to set
+ * Searches for a match with the sequence description, if that option was
+ * requested, and if found, adds the sequence to the list of match ids (but
+ * not as a duplicate). Answers true if a match was added, else false.
+ *
+ * @param seq
+ * @param regex
+ * @return
*/
- public void setCaseSensitive(boolean caseSensitive)
+ protected boolean searchSequenceDescription(SequenceI seq, Regex regex)
{
- this.caseSensitive = caseSensitive;
+ if (!includeDescription)
+ {
+ return false;
+ }
+ String desc = seq.getDescription();
+ if (desc != null && regex.search(desc) && !idMatch.contains(seq))
+ {
+ idMatch.addElement(seq);
+ return true;
+ }
+ return false;
}
/**
- * @return the findAll
+ * Searches for a match with the sequence name, and if found, adds the
+ * sequence to the list of match ids (but not as a duplicate). Answers true if
+ * a match was added, else false.
+ *
+ * @param seq
+ * @param regex
+ * @return
*/
- public boolean isFindAll()
+ protected boolean searchSequenceName(SequenceI seq, Regex regex)
{
- return findAll;
+ if (regex.search(seq.getName()) && !idMatch.contains(seq))
+ {
+ idMatch.addElement(seq);
+ return true;
+ }
+ return false;
}
/**
- * @param findAll
- * the findAll to set
+ * Tries to interpret the search string as a residue position, and if valid,
+ * adds the position to the search results
*/
- public void setFindAll(boolean findAll)
+ protected boolean searchForResidueNumber(SequenceI seq, String searchString)
{
- this.findAll = findAll;
+ try
+ {
+ int res = Integer.parseInt(searchString);
+ if (seq.getStart() <= res && seq.getEnd() >= res)
+ {
+ searchResults.addResult(seq, res, res);
+ return true;
+ }
+ } catch (NumberFormatException ex)
+ {
+ }
+ return false;
}
/**
- * @return the selection
+ * Sets whether the search is case sensitive (default is no)
+ *
+ * @param value
*/
- public jalview.datamodel.SequenceGroup getSelection()
+ public void setCaseSensitive(boolean value)
{
- return selection;
+ this.caseSensitive = value;
}
/**
- * @param selection
- * the selection to set
+ * Sets whether search returns all matches. Default is to return the next
+ * match only.
+ *
+ * @param value
*/
- public void setSelection(jalview.datamodel.SequenceGroup selection)
+ public void setFindAll(boolean value)
{
- this.selection = selection;
+ this.findAll = value;
}
/**
}
/**
- * @return the regex
- */
- public com.stevesoft.pat.Regex getRegex()
- {
- return regex;
- }
-
- /**
* @return the searchResults
*/
public SearchResultsI getSearchResults()
}
/**
- * @param resIndex
- * the resIndex to set
- */
- public void setResIndex(int resIndex)
- {
- this.resIndex = resIndex;
- }
-
- /**
* @return the seqIndex
*/
public int getSeqIndex()
}
/**
- * @param seqIndex
- * the seqIndex to set
+ * Sets whether search also searches in sequence description text (default is
+ * no)
+ *
+ * @param value
*/
- public void setSeqIndex(int seqIndex)
- {
- this.seqIndex = seqIndex;
- }
-
- public boolean isIncludeDescription()
- {
- return includeDescription;
- }
-
- public void setIncludeDescription(boolean includeDescription)
+ public void setIncludeDescription(boolean value)
{
- this.includeDescription = includeDescription;
+ this.includeDescription = value;
}
}
git://source.jalview.org / jalview.git / blobdiff