import jalview.util.Format;
import java.io.IOException;
+import java.util.ArrayList;
import java.util.Hashtable;
+import java.util.List;
import java.util.StringTokenizer;
-import java.util.Vector;
/**
* DOCUMENT ME!
*
* @param inFile
* DOCUMENT ME!
- * @param sourceType
+ * @param type
* DOCUMENT ME!
*
* @throws IOException
* DOCUMENT ME!
*/
- public MSFfile(String inFile, DataSourceType sourceType)
- throws IOException
+ public MSFfile(String inFile, DataSourceType type) throws IOException
{
- super(inFile, sourceType);
+ super(inFile, type);
}
public MSFfile(FileParse source) throws IOException
}
/**
- * DOCUMENT ME!
+ * Read and parse MSF sequence data
*/
@Override
public void parse() throws IOException
{
- int i = 0;
boolean seqFlag = false;
- String key = new String();
- Vector headers = new Vector();
- Hashtable seqhash = new Hashtable();
- String line;
+ List<String> headers = new ArrayList<String>();
+ Hashtable<String, StringBuilder> seqhash = new Hashtable<String, StringBuilder>();
try
{
+ String line;
while ((line = nextLine()) != null)
{
StringTokenizer str = new StringTokenizer(line);
+ String key = null;
while (str.hasMoreTokens())
{
String inStr = str.nextToken();
if (inStr.indexOf("Name:") != -1)
{
key = str.nextToken();
- headers.addElement(key);
+ headers.add(key);
}
- // if line has // set SeqFlag to 1 so we know sequences are coming
+ // if line has // set SeqFlag so we know sequences are coming
if (inStr.indexOf("//") != -1)
{
seqFlag = true;
}
// Process lines as sequence lines if seqFlag is set
- if ((inStr.indexOf("//") == -1) && (seqFlag == true))
+ if ((inStr.indexOf("//") == -1) && seqFlag)
{
- // seqeunce id is the first field
+ // sequence id is the first field
key = inStr;
- StringBuffer tempseq;
+ StringBuilder tempseq;
// Get sequence from hash if it exists
if (seqhash.containsKey(key))
{
- tempseq = (StringBuffer) seqhash.get(key);
+ tempseq = seqhash.get(key);
}
else
{
- tempseq = new StringBuffer();
+ tempseq = new StringBuilder(64);
seqhash.put(key, tempseq);
}
while (str.hasMoreTokens())
{
// append the word to the sequence
- tempseq.append(str.nextToken());
+ String sequenceBlock = str.nextToken();
+ tempseq.append(sequenceBlock);
}
}
}
this.noSeqs = headers.size();
// Add sequences to the hash
- for (i = 0; i < headers.size(); i++)
+ for (int i = 0; i < headers.size(); i++)
{
- if (seqhash.get(headers.elementAt(i)) != null)
+ if (seqhash.get(headers.get(i)) != null)
{
- String head = headers.elementAt(i).toString();
+ String head = headers.get(i);
String seq = seqhash.get(head).toString();
if (maxLength < head.length())
maxLength = head.length();
}
- // Replace ~ with a sensible gap character
- seq = seq.replace('~', '-');
+ /*
+ * replace ~ (leading/trailing positions) with the gap character;
+ * use '.' as this is the internal gap character required by MSF
+ */
+ seq = seq.replace('~', '.');
Sequence newSeq = parseId(head);
else
{
System.err.println("MSFFile Parser: Can't find sequence for "
- + headers.elementAt(i));
+ + headers.get(i));
}
}
}
return check % 10000;
}
+ /**
+ * DOCUMENT ME!
+ *
+ * @param s
+ * DOCUMENT ME!
+ * @param is_NA
+ * DOCUMENT ME!
+ *
+ * @return DOCUMENT ME!
+ */
@Override
- public String print(SequenceI[] sqs, boolean jvsuffix)
+ public String print(SequenceI[] sqs, boolean jvSuffix)
{
boolean is_NA = Comparison.isNucleotide(sqs);
SequenceI[] s = new SequenceI[sqs.length];
- StringBuffer out = new StringBuffer("!!" + (is_NA ? "NA" : "AA")
- + "_MULTIPLE_ALIGNMENT 1.0");
+ StringBuilder out = new StringBuilder(256);
+ out.append("!!").append(is_NA ? "NA" : "AA")
+ .append("_MULTIPLE_ALIGNMENT 1.0");
// TODO: JBPNote : Jalview doesn't remember NA or AA yet.
out.append(newline);
out.append(newline);
while ((i < sqs.length) && (sqs[i] != null))
{
- // Replace all internal gaps with . and external spaces with ~
- s[i] = new Sequence(sqs[i].getName(), sqs[i].getSequenceAsString()
- .replace('-', '.'), sqs[i].getStart(), sqs[i].getEnd());
+ /*
+ * modify to MSF format: uses '.' for internal gaps,
+ * and '~' for leading or trailing gaps
+ */
+ String seqString = sqs[i].getSequenceAsString()
+ .replace('-', '.');
- StringBuffer sb = new StringBuffer();
- sb.append(s[i].getSequence());
+ StringBuilder sb = new StringBuilder(seqString);
for (int ii = 0; ii < sb.length(); ii++)
{
break;
}
}
+ s[i] = new Sequence(sqs[i].getName(), sb.toString(),
+ sqs[i].getStart(), sqs[i].getEnd());
- s[i].setSequence(sb.toString());
-
- if (s[i].getSequence().length > max)
+ if (sb.length() > max)
{
- max = s[i].getSequence().length;
+ max = sb.length();
}
i++;
while ((i < s.length) && (s[i] != null))
{
- nameBlock[i] = new String(" Name: " + printId(s[i], jvsuffix) + " ");
+ nameBlock[i] = new String(" Name: " + printId(s[i], jvSuffix) + " ");
idBlock[i] = new String("Len: "
+ maxLenpad.form(s[i].getSequence().length) + " Check: "
while ((j < s.length) && (s[j] != null))
{
- String name = printId(s[j], jvsuffix);
+ String name = printId(s[j], jvSuffix);
out.append(new Format("%-" + maxid + "s").form(name + " "));
git://source.jalview.org / jalview.git / blobdiff