public TableSort() throws DiffException {
config = ComparisonConfig.getInstance();
}
public long sort(
String inputFile,
String outputFile,
String tmpdirectory,
long numberLinePerFile,
int maxNumberOfFile,
boolean saveKey,
boolean keepTmp,
String delimiter,
char escape,
KeyGenerator keyGenerator,
Filter filter)
throws DiffException {
if (maxNumberOfFile > MAX_FILE_NUMBER) {
throw new DiffException("Maxium number of files can't exceed " + MAX_FILE_NUMBER);
}
try {
List<File> tmpFileList = new ArrayList<File>();
// first step: read input table, split into sorted temp tables
TxtTable table = new TxtTable(inputFile, keyGenerator, filter, false, delimiter, escape);
long curSplitRowNo = 0;
int fileCount = 0;
table.setPrimaryFields(config.getLeftPrimaryFields());
String tableTile = table.getTitleRow().getContent();
TxtTable.Row row = table.getNextRow();
MemTable tmpTable = new MemTable();
while (row != null && fileCount < maxNumberOfFile) {
tmpTable.addRow(row);
curSplitRowNo++;
row = table.getNextRow();
if (curSplitRowNo >= numberLinePerFile || row == null) {
File newtmpfile = File.createTempFile("sort_tmp", "flatfile", new File(tmpdirectory));
if (!keepTmp) newtmpfile.deleteOnExit();
tmpFileList.add(newtmpfile);
tmpTable.sortAndSave(newtmpfile);
fileCount++;
if (row != null) {
tmpTable = new MemTable();
curSplitRowNo = 0;
}
}
}
if (fileCount == maxNumberOfFile) {
throw new DiffException("File count exceed maximum allowed count.");
}
table.closeFile();
// second step: merge sort
return mergeSortedFiles(
tmpFileList,
new File(outputFile),
tableTile,
new Comparator<TxtTable.Row>() {
public int compare(TxtTable.Row i, TxtTable.Row j) {
return i.compareTo(j);
}
},
delimiter,
escape,
saveKey);
} catch (IOException e) {
throw new DiffException(e);
}
}
public TableSort(String configFile) throws DiffException {
this();
config.init(configFile);
}
