X-Git-Url: http://source.jalview.org/gitweb/?a=blobdiff_plain;f=src%2Fjalview%2Fanalysis%2FFinder.java;h=fc6537910140ac4c8bfb378f34c1c1c283fc1589;hb=b7014049e421026d0f2d57179af9688472cd78f2;hp=adf7ff64bfa49820f5f6000828ccff0593da021c;hpb=85b18307da5f85a9cb5c13bb6be97eaf2c7f7965;p=jalview.git diff --git a/src/jalview/analysis/Finder.java b/src/jalview/analysis/Finder.java index adf7ff6..fc65379 100644 --- a/src/jalview/analysis/Finder.java +++ b/src/jalview/analysis/Finder.java @@ -20,25 +20,28 @@ */ package jalview.analysis; +import jalview.api.AlignViewportI; +import jalview.api.FinderI; import jalview.datamodel.AlignmentI; -import jalview.datamodel.Range; import jalview.datamodel.SearchResultMatchI; import jalview.datamodel.SearchResults; import jalview.datamodel.SearchResultsI; import jalview.datamodel.SequenceGroup; import jalview.datamodel.SequenceI; -import jalview.datamodel.VisibleContigsIterator; import jalview.util.Comparison; +import jalview.util.MapList; +import java.util.ArrayList; +import java.util.Arrays; +import java.util.Iterator; import java.util.List; -import java.util.Vector; import com.stevesoft.pat.Regex; /** * Implements the search algorithm for the Find dialog */ -public class Finder +public class Finder implements FinderI { /* * matched residue locations @@ -48,203 +51,312 @@ public class Finder /* * sequences matched by id or description */ - private Vector idMatch; + private List idMatches; /* - * the alignment to search over + * the viewport to search over */ - private AlignmentI alignment; + private AlignViewportI viewport; /* - * (optional) selection to restrict search to + * sequence index in alignment to search from */ - private SequenceGroup selection; + private int sequenceIndex; /* - * set true for case-sensitive search (default is false) + * position offset in sequence to search from, base 0 + * (position after start of last match for a 'find next') */ - private boolean caseSensitive; + private int residueIndex; /* - * set true to search sequence description (default is false) + * the true sequence position of the start of the + * last sequence searched (when 'ignore hidden regions' does not apply) */ - private boolean includeDescription; + private int searchedSequenceStartPosition; /* - * set true to return all matches (default is next match only) + * when 'ignore hidden regions' applies, this holds the mapping from + * the visible sequence positions (1, 2, ...) to true sequence positions */ - private boolean findAll; + private MapList searchedSequenceMap; - /* - * sequence index in alignment to search from - */ - private int sequenceIndex; - - /* - * column position in sequence to search from, base 0 - * - absolute column number including any hidden columns - * (position of last match for a repeat search) - */ - private int columnIndex; + private String seqToSearch; /** - * Constructor to start searching an alignment, optionally restricting results - * to a selection + * Constructor for searching a viewport * - * @param al - * @param sel + * @param av */ - public Finder(AlignmentI al, SequenceGroup sel) + public Finder(AlignViewportI av) { - this(al, sel, 0, -1); + this.viewport = av; + this.sequenceIndex = 0; + this.residueIndex = -1; } - /** - * Constructor to resume search at given sequence and residue on alignment and - * (optionally) restricted to a selection - * - * @param al - * @param sel - * @param seqindex - * @param colindex - */ - public Finder(AlignmentI al, SequenceGroup sel, int seqindex, - int colindex) + @Override + public void findAll(String theSearchString, boolean matchCase, + boolean searchDescription, boolean ignoreHidden) + { + /* + * search from the start + */ + sequenceIndex = 0; + residueIndex = -1; + + doFind(theSearchString, matchCase, searchDescription, true, + ignoreHidden); + + /* + * reset to start for next search + */ + sequenceIndex = 0; + residueIndex = -1; + } + + @Override + public void findNext(String theSearchString, boolean matchCase, + boolean searchDescription, boolean ignoreHidden) { - this.alignment = al; - this.selection = sel; - this.sequenceIndex = seqindex; - this.columnIndex = colindex; + doFind(theSearchString, matchCase, searchDescription, false, + ignoreHidden); + + if (searchResults.isEmpty() && idMatches.isEmpty()) + { + /* + * search failed - reset to start for next search + */ + sequenceIndex = 0; + residueIndex = -1; + } } /** - * 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(). + * Performs a 'find next' or 'find all' * * @param theSearchString - * @return + * @param matchCase + * @param searchDescription + * @param findAll + * @param ignoreHidden */ - public void find(String theSearchString) + protected void doFind(String theSearchString, boolean matchCase, + boolean searchDescription, boolean findAll, boolean ignoreHidden) { - if (findAll) - { - sequenceIndex = 0; - columnIndex = -1; - } + searchResults = new SearchResults(); + idMatches = new ArrayList<>(); - String searchString = caseSensitive ? theSearchString + String searchString = matchCase ? theSearchString : theSearchString.toUpperCase(); Regex searchPattern = new Regex(searchString); - searchPattern.setIgnoreCase(!caseSensitive); - searchResults = new SearchResults(); - idMatch = new Vector<>(); + searchPattern.setIgnoreCase(!matchCase); + SequenceGroup selection = viewport.getSelectionGroup(); if (selection != null && selection.getSize() < 1) { selection = null; // ? ignore column-only selection } + AlignmentI alignment = viewport.getAlignment(); int end = alignment.getHeight(); - while (sequenceIndex < end) + getSequence(ignoreHidden); + + boolean found = false; + while (!found || findAll) { - SequenceI seq = alignment.getSequenceAt(sequenceIndex); - boolean found = findNext(seq, searchString, searchPattern); - if (found && !findAll) + found = findNextMatch(searchString, searchPattern, searchDescription, + ignoreHidden); + if (sequenceIndex >= end) { - return; - } - if (!found) - { - sequenceIndex++; - columnIndex = -1; + break; } } } /** - * Answers the start-end column range of the visible region of - * sequence starting at or after the given column. - * If there are no hidden columns, this just returns the remaining width of - * the sequence. The range is restricted to the current selection - * if there is one. Answers null if there are no visible columns at or after - * column. + * Calculates and saves the sequence string to search. The string is restricted + * to the current selection region if there is one, and is saved with all gaps + * removed. + *

+ * If there are hidden columns, and option {@ignoreHidden} is selected, then + * only visible positions of the sequence are included, and a mapping is also + * constructed from the returned string positions to the true sequence + * positions. + *

+ * Note we have to do this each time {@code findNext} or {@code findAll} is + * called, in case the alignment, selection group or hidden columns have + * changed. In particular, if the sequence at offset {@code sequenceIndex} in + * the alignment is (no longer) in the selection group, search is advanced to + * the next sequence that is. + *

+ * Sets sequence string to the empty string if there are no more sequences (in + * selection group if any) at or after {@code sequenceIndex}. + *

+ * Returns true if a sequence could be found, false if end of alignment was + * reached + * + * @param ignoreHidden + * @return */ - protected Range getNextVisibleSequenceRegion(SequenceI sequence, - int column) + private boolean getSequence(boolean ignoreHidden) { - int seqColStart = column; - int seqColEnd = sequence.getLength() - 1; + AlignmentI alignment = viewport.getAlignment(); + if (sequenceIndex >= alignment.getHeight()) + { + seqToSearch = ""; + return false; + } + SequenceI seq = alignment.getSequenceAt(sequenceIndex); + SequenceGroup selection = viewport.getSelectionGroup(); + if (selection != null && !selection.contains(seq)) + { + if (!nextSequence(ignoreHidden)) + { + return false; + } + seq = alignment.getSequenceAt(sequenceIndex); + } - /* - * restrict search to (next) visible column region, - * in case there are hidden columns - */ - VisibleContigsIterator visibleRegions = alignment.getHiddenColumns() - .getVisContigsIterator(column, alignment.getWidth(), - false); - int[] visible = visibleRegions.hasNext() ? visibleRegions.next() : null; - if (visible == null) + String seqString = null; + if (ignoreHidden) + { + seqString = getVisibleSequence(seq); + } + else { - columnIndex = seqColEnd + 1; - return null; + int startCol = 0; + int endCol = seq.getLength() - 1; + this.searchedSequenceStartPosition = seq.getStart(); + if (selection != null) + { + startCol = selection.getStartRes(); + endCol = Math.min(endCol, selection.getEndRes()); + this.searchedSequenceStartPosition = seq.findPosition(startCol); + } + seqString = seq.getSequenceAsString(startCol, endCol + 1); } - seqColStart = Math.max(seqColStart, visible[0]); - seqColEnd = Math.min(seqColEnd, visible[1]); /* - * restrict search to selected region if there is one + * remove gaps; note that even if this leaves an empty string, we 'search' + * the sequence anyway (for possible match on name or description) */ - if (selection != null) + String ungapped = AlignSeq.extractGaps(Comparison.GapChars, seqString); + this.seqToSearch = ungapped; + + return true; + } + + /** + * Returns a string consisting of only the visible residues of {@code seq} from + * alignment column {@ fromColumn}, restricted to the current selection region + * if there is one. + *

+ * As a side-effect, also computes the mapping from the true sequence positions + * to the positions (1, 2, ...) of the returned sequence. This is to allow + * search matches in the visible sequence to be converted to sequence positions. + * + * @param seq + * @return + */ + private String getVisibleSequence(SequenceI seq) + { + int seqStartCol = seq.findIndex(seq.getStart()); + int seqEndCol = seq.findIndex(seq.getStart() + seq.getLength() - 1); + Iterator visibleColumns = viewport.getViewAsVisibleContigs(true); + StringBuilder visibleSeq = new StringBuilder(seqEndCol - seqStartCol); + List fromRanges = new ArrayList<>(); + + while (visibleColumns.hasNext()) { - int selectionStart = selection.getStartRes(); - int selectionEnd = selection.getEndRes(); - if (selectionStart > seqColEnd || selectionEnd < seqColStart) + int[] range = visibleColumns.next(); + if (range[0] > seqEndCol) + { + // beyond the end of the sequence + break; + } + if (range[1] < seqStartCol) + { + // before the start of the sequence + continue; + } + String subseq = seq.getSequenceAsString(range[0], range[1] + 1); + String ungapped = AlignSeq.extractGaps(Comparison.GapChars, subseq); + visibleSeq.append(ungapped); + if (!ungapped.isEmpty()) { /* - * sequence region doesn't overlap selection region + * visible region includes at least one non-gap character, + * so add the range to the mapping being constructed */ - columnIndex = seqColEnd + 1; - return null; + int seqResFrom = seq.findPosition(range[0]); + int seqResTo = seq.findPosition(range[1]); + fromRanges.add(new int[] { seqResFrom, seqResTo }); } - seqColStart = Math.max(seqColStart, selectionStart); - seqColEnd = Math.min(seqColEnd, selectionEnd); } - return new Range(seqColStart, seqColEnd); + /* + * construct the mapping + * from: visible sequence positions 1..length + * to: true residue positions of the alignment sequence + */ + List toRange = Arrays + .asList(new int[] + { 1, visibleSeq.length() }); + searchedSequenceMap = new MapList(fromRanges, toRange, 1, 1); + + return visibleSeq.toString(); } /** - * Finds the next match in the given sequence, starting at column at - * columnIndex. Answers true if a match is found, else false. If - * a match is found, columnIndex is advanced to the column after - * the start of the matched region, ready for a search from the next position. + * Advances the search to the next sequence in the alignment. Sequences not in + * the current selection group (if there is one) are skipped. The (sub-)sequence + * to be searched is extracted, gaps removed, and saved, or set to null if there + * are no more sequences to search. + *

+ * Returns true if a sequence could be found, false if end of alignment was + * reached * - * @param seq + * @param ignoreHidden + */ + private boolean nextSequence(boolean ignoreHidden) + { + sequenceIndex++; + residueIndex = -1; + + return getSequence(ignoreHidden); + } + + /** + * Finds the next match in the given sequence, starting at offset + * {@code residueIndex}. Answers true if a match is found, else false. + *

+ * If a match is found, {@code residueIndex} is advanced to the position after + * the start of the matched region, ready for the next search. + *

+ * If no match is found, {@code sequenceIndex} is advanced ready to search the + * next sequence. + * + * @param seqToSearch * @param searchString * @param searchPattern + * @param matchDescription + * @param ignoreHidden * @return */ - protected boolean findNext(SequenceI seq, String searchString, - Regex searchPattern) + protected boolean findNextMatch(String searchString, + Regex searchPattern, boolean matchDescription, + boolean ignoreHidden) { - if (selection != null && !selection.contains(seq)) - { - /* - * this sequence is not in the selection - advance to next sequence - */ - return false; - } - - if (columnIndex < 0) + if (residueIndex < 0) { /* * at start of sequence; try find by residue number, in sequence id, * or (optionally) in sequence description */ - if (doNonMotifSearches(seq, searchString, searchPattern)) + if (doNonMotifSearches(searchString, searchPattern, + matchDescription)) { return true; } @@ -253,83 +365,66 @@ public class Finder /* * search for next match in sequence string */ - int end = seq.getLength(); - while (columnIndex < end) + int end = seqToSearch.length(); + while (residueIndex < end) { - if (searchNextVisibleRegion(seq, searchPattern)) + boolean matched = searchPattern.searchFrom(seqToSearch, residueIndex); + if (matched) { - return true; + if (recordMatch(searchPattern, ignoreHidden)) + { + return true; + } + } + else + { + residueIndex = Integer.MAX_VALUE; } - } - return false; - } - - /** - * Searches the sequence, starting from columnIndex, and adds the - * next match (if any) to searchResults. The search is restricted - * to the next visible column region, and to the selection region - * if there is one. Answers true if a match is added, else false. - * - * @param seq - * @param searchPattern - * @return - */ - protected boolean searchNextVisibleRegion(SequenceI seq, Regex searchPattern) - { - Range visible = getNextVisibleSequenceRegion(seq, columnIndex); - if (visible == null) - { - return false; - } - String seqString = seq.getSequenceAsString(visible.start, visible.end + 1); - String noGaps = AlignSeq.extractGaps(Comparison.GapChars, seqString); - - if (searchPattern.search(noGaps)) - { - int sequenceStartPosition = seq.findPosition(visible.start); - recordMatch(seq, searchPattern, sequenceStartPosition); - return true; - } - else - { - /* - * no match - advance columnIndex past this visible region - * so the next visible region (if any) is searched next - */ - columnIndex = visible.end + 1; } + nextSequence(ignoreHidden); return false; } /** * Adds the match held in the searchPattern Regex to the * searchResults, unless it is a subregion of the last match - * recorded. columnIndex is advanced to the position after the + * recorded. residueIndex is advanced to the position after the * start of the matched region, ready for the next search. Answers true if a * match was added, else false. + *

+ * Matches that lie entirely within hidden regions of the alignment are not + * added. * - * @param seq * @param searchPattern - * @param firstResiduePosition + * @param ignoreHidden * @return */ - protected boolean recordMatch(SequenceI seq, Regex searchPattern, - int firstResiduePosition) + protected boolean recordMatch(Regex searchPattern, boolean ignoreHidden) { + SequenceI seq = viewport.getAlignment().getSequenceAt(sequenceIndex); + /* - * get start/end of the match in sequence coordinates + * convert start/end of the match to sequence coordinates */ int offset = searchPattern.matchedFrom(); - int matchStartPosition = firstResiduePosition + offset; + int matchStartPosition = this.searchedSequenceStartPosition + offset; int matchEndPosition = matchStartPosition + searchPattern.charsMatched() - 1; /* - * update columnIndex to next column after the start of the match + * update residueIndex to next position after the start of the match * (findIndex returns a value base 1, columnIndex is held base 0) */ - columnIndex = seq.findIndex(matchStartPosition); + residueIndex += offset + 1; + + /* + * return false if the match is entirely in a hidden region + */ + if (allHidden(seq, matchStartPosition, matchEndPosition)) + { + return false; + } /* * check that this match is not a subset of the previous one (JAL-2302) @@ -341,7 +436,7 @@ public class Finder if (lastMatch == null || !lastMatch.contains(seq, matchStartPosition, matchEndPosition)) { - searchResults.addResult(seq, matchStartPosition, matchEndPosition); + addMatch(seq, matchStartPosition, matchEndPosition, ignoreHidden); return true; } @@ -349,6 +444,63 @@ public class Finder } /** + * Adds one match to the stored list. If hidden residues are being skipped, then + * the match may need to be split into contiguous positions of the sequence (so + * it does not include skipped residues). + * + * @param seq + * @param matchStartPosition + * @param matchEndPosition + * @param ignoreHidden + */ + private void addMatch(SequenceI seq, int matchStartPosition, + int matchEndPosition, boolean ignoreHidden) + { + if (!ignoreHidden) + { + /* + * simple case + */ + searchResults.addResult(seq, matchStartPosition, matchEndPosition); + return; + } + + /* + * get start-end contiguous ranges in underlying sequence + */ + int[] truePositions = searchedSequenceMap + .locateInFrom(matchStartPosition, matchEndPosition); + for (int i = 0; i < truePositions.length - 1; i += 2) + { + searchResults.addResult(seq, truePositions[i], truePositions[i + 1]); + } + } + + /** + * Returns true if all residues are hidden, else false + * + * @param seq + * @param fromPos + * @param toPos + * @return + */ + private boolean allHidden(SequenceI seq, int fromPos, int toPos) + { + if (!viewport.hasHiddenColumns()) + { + return false; + } + for (int res = fromPos; res <= toPos; res++) + { + if (isVisible(seq, res)) + { + return false; + } + } + return true; + } + + /** * Does searches other than for residue patterns. Currently this includes *

* Answers true if a match is found, else false. * - * @param seq * @param searchString * @param searchPattern + * @param includeDescription * @return */ - protected boolean doNonMotifSearches(SequenceI seq, String searchString, - Regex searchPattern) + protected boolean doNonMotifSearches(String searchString, + Regex searchPattern, boolean includeDescription) { + SequenceI seq = viewport.getAlignment().getSequenceAt(sequenceIndex); + /* * position sequence search to start of sequence */ - columnIndex = 0; - - if (searchForResidueNumber(seq, searchString)) + residueIndex = 0; + try { - return true; + int res = Integer.parseInt(searchString); + return searchForResidueNumber(seq, res); + } catch (NumberFormatException ex) + { + // search pattern is not a number } + if (searchSequenceName(seq, searchPattern)) { return true; @@ -397,9 +555,9 @@ public class Finder protected boolean searchSequenceDescription(SequenceI seq, Regex searchPattern) { String desc = seq.getDescription(); - if (desc != null && searchPattern.search(desc) && !idMatch.contains(seq)) + if (desc != null && searchPattern.search(desc) && !idMatches.contains(seq)) { - idMatch.addElement(seq); + idMatches.add(seq); return true; } return false; @@ -416,108 +574,61 @@ public class Finder */ protected boolean searchSequenceName(SequenceI seq, Regex searchPattern) { - if (searchPattern.search(seq.getName()) && !idMatch.contains(seq)) + if (searchPattern.search(seq.getName()) && !idMatches.contains(seq)) { - idMatch.addElement(seq); + idMatches.add(seq); return true; } return false; } /** - * Tries to interpret the search string as a residue position, and if valid, - * adds the position to the search results and returns true, else answers - * false + * If the residue position is valid for the sequence, and in a visible column, + * adds the position to the search results and returns true, else answers false. + * + * @param seq + * @param resNo + * @return */ - protected boolean searchForResidueNumber(SequenceI seq, String searchString) + protected boolean searchForResidueNumber(SequenceI seq, int resNo) { - try + if (seq.getStart() <= resNo && seq.getEnd() >= resNo) { - int res = Integer.parseInt(searchString); - if (seq.getStart() <= res && seq.getEnd() >= res) + if (isVisible(seq, resNo)) { - searchResults.addResult(seq, res, res); + searchResults.addResult(seq, resNo, resNo); return true; } - } catch (NumberFormatException ex) - { } return false; } /** - * Sets whether the search is case sensitive (default is no) + * Returns true if the residue is in a visible column, else false * - * @param value + * @param seq + * @param res + * @return */ - public void setCaseSensitive(boolean value) + private boolean isVisible(SequenceI seq, int res) { - this.caseSensitive = value; + if (!viewport.hasHiddenColumns()) + { + return true; + } + int col = seq.findIndex(res); // base 1 + return viewport.getAlignment().getHiddenColumns().isVisible(col - 1); // base 0 } - /** - * Sets whether search returns all matches. Default is to return the next - * match only. - * - * @param value - */ - public void setFindAll(boolean value) + @Override + public List getIdMatches() { - this.findAll = value; + return idMatches; } - /** - * Returns the (possibly empty) list of sequences matched on sequence name or - * description - * - * @return - */ - public Vector getIdMatch() - { - return idMatch; - } - - /** - * Answers the search results (possibly empty) from the last search - * - * @return - */ + @Override public SearchResultsI getSearchResults() { return searchResults; } - - /** - * Answers the absolute column position (base 0, including any hidden columns) - * of the start of the last sequence motif (residue pattern) match found. A - * 'Find next' will search from the next position. - * - * @return - */ - public int getColumnIndex() - { - return columnIndex; - } - - /** - * Answers the offset in the alignment (0..) of the sequence in which the last - * match was found (if any) - * - * @return - */ - public int getSequenceIndex() - { - return sequenceIndex; - } - - /** - * Sets whether search also searches in sequence description text (default is - * no) - * - * @param value - */ - public void setIncludeDescription(boolean value) - { - this.includeDescription = value; - } }