@Override
public RecordSet getRecords() throws MalformedSourceException {
try {
openReader();
} catch (IOException e) {
logger_.info("Problem loading file: " + filename + " (" + e.getMessage() + ")");
throw new MalformedSourceException(
"Problem loading file: " + filename + " (" + e.getMessage() + ")");
} catch (InvalidFormatException e) {
logger_.info("Problem loading file: " + filename + " (" + e.getMessage() + ")");
throw new MalformedSourceException(
"Problem loading file: " + filename + " (" + e.getMessage() + ")");
}
RecordSet ret = new RecordSet();
// Currently we need this flag in order for
// TransformationEngine not to go into an infinite loop.
if (!isRead_) {
logger_.info("Opening file: " + filename);
int nSheets = wb_.getNumberOfSheets();
logger_.info("number of sheets: " + nSheets);
for (int i = 0; i < nSheets; i++) {
Sheet cSheet = wb_.getSheetAt(i);
String cSheetName = cSheet.getSheetName();
for (int j = skipLines_; j <= cSheet.getLastRowNum(); j++) {
if (ignoreLinesAfter_ != 0 && j >= ignoreLinesAfter_) {
break;
}
Row row = cSheet.getRow(j);
MapRecord rec = new MapRecord();
for (int k = 0; k < row.getLastCellNum(); k++) {
if (!fieldMap_.keySet().contains(k)) {
continue;
}
StringValue val;
Cell cCell = row.getCell(k);
if (cCell == null) {
continue;
}
int cellType = cCell.getCellType();
switch (cellType) {
case Cell.CELL_TYPE_STRING:
val = new StringValue(cCell.getStringCellValue());
break;
case Cell.CELL_TYPE_NUMERIC:
if (DateUtil.isCellDateFormatted(cCell)) {
SimpleDateFormat sdf = new SimpleDateFormat("dd/MM/yyyy");
val = new StringValue(sdf.format(cCell.getDateCellValue()));
} else {
val = new StringValue(String.valueOf(cCell.getNumericCellValue()));
}
break;
case Cell.CELL_TYPE_BLANK:
val = new StringValue("");
break;
case Cell.CELL_TYPE_BOOLEAN:
val = new StringValue(String.valueOf(cCell.getBooleanCellValue()));
break;
default:
val = new StringValue("Unsupported cell type");
}
rec.addValue(fieldMap_.get(k), val);
}
// TODO remove the hardcoded value
rec.addValue("ExcelSheetName", new StringValue(cSheetName));
ret.addRecord(rec);
}
}
isRead_ = true;
}
return ret;
}
@Override
public RecordSet getRecords() throws MalformedSourceException {
if (reader_ == null) {
throw new EmptySourceException("Input file is not open");
}
RecordSet records = new RecordSet();
try {
String line;
boolean in_record = false;
int line_cnt = 0;
MapRecord rec = null;
String ris_tag = null;
while ((line = reader_.readLine()) != null) {
line_cnt++;
// Ignore empty lines
if (line.isEmpty() || line.equals("") || line.matches("^\\s*$")) {
continue;
}
Pattern ris_pattern = Pattern.compile("^([A-Z][A-Z0-9]) - (.*)$");
Matcher ris_matcher = ris_pattern.matcher(line);
Value val;
if (ris_matcher.matches()) {
ris_tag = ris_matcher.group(1);
if (!in_record) {
// The first tag of the record should be "TY". If we
// encounter it we should create a new record.
if (ris_tag.equals("TY")) {
in_record = true;
rec = new MapRecord();
} else {
logger_.info(
"Line: " + line_cnt + " in file " + filename + " should contain tag \"TY\"");
throw new MalformedSourceException(
"Line: " + line_cnt + " in file " + filename + " should contain tag \"TY\"");
}
}
// If the tag is the end record tag ("ER") we stop
// being in a record. Add the current record in the
// record set and skip the rest of the processing.
if (ris_tag.equals("ER")) {
in_record = false;
records.addRecord(rec);
rec = null;
continue;
}
// If there is no mapping for the current tag we do not
// know what to do with it, so we ignore it.
if (!field_map_.containsKey(ris_tag)) {
logger_.warn("Tag \"" + ris_tag + "\" is not in the field map. Ignoring");
continue;
}
val = new StringValue(ris_matcher.group(2));
} else {
val = new StringValue(line);
}
String field = field_map_.get(ris_tag);
if (field != null) {
rec.addValue(field, val);
}
}
} catch (IOException e) {
logger_.info("Error while reading from file " + filename);
throw new MalformedSourceException("Error while reading from file " + filename);
}
return records;
}
