/**
* Returns the contents of the cell, with no formatting applied
*
* @return
*/
String unformattedContents() {
switch (currentCell.getType()) {
case "s": // string stored in shared table
int idx = Integer.parseInt(lastContents);
return new XSSFRichTextString(sst.getEntryAt(idx)).toString();
case "inlineStr": // inline string (not in sst)
return new XSSFRichTextString(lastContents).toString();
default:
return lastContents;
}
}
/**
* Read the numeric format string out of the styles table for this cell. Stores the result in the
* Cell.
*
* @param startElement
* @param cell
*/
void setFormatString(StartElement startElement, StreamingCell cell) {
Attribute cellStyle = startElement.getAttributeByName(new QName("s"));
String cellStyleString = (cellStyle != null) ? cellStyle.getValue() : null;
XSSFCellStyle style = null;
if (cellStyleString != null) {
style = stylesTable.getStyleAt(Integer.parseInt(cellStyleString));
} else if (stylesTable.getNumCellStyles() > 0) {
style = stylesTable.getStyleAt(0);
}
if (style != null) {
cell.setNumericFormatIndex(style.getDataFormat());
String formatString = style.getDataFormatString();
if (formatString != null) {
cell.setNumericFormat(formatString);
} else {
cell.setNumericFormat(BuiltinFormats.getBuiltinFormat(cell.getNumericFormatIndex()));
}
} else {
cell.setNumericFormatIndex(null);
cell.setNumericFormat(null);
}
}
/**
* Tries to format the contents of the last contents appropriately based on the type of cell and
* the discovered numeric format.
*
* @return
*/
String formattedContents() {
switch (currentCell.getType()) {
case "s": // string stored in shared table
int idx = Integer.parseInt(lastContents);
return new XSSFRichTextString(sst.getEntryAt(idx)).toString();
case "inlineStr": // inline string (not in sst)
return new XSSFRichTextString(lastContents).toString();
case "str": //
return lastContents;
case "e": // error type
return StringUtils.EMPTY; // "ERROR: " + lastContents;
case "n": // numeric type
if (currentCell.getNumericFormat() != null && lastContents.length() > 0) {
return dataFormatter.formatRawCellContents(
Double.parseDouble(lastContents),
currentCell.getNumericFormatIndex(),
currentCell.getNumericFormat());
} else {
return lastContents;
}
default:
return lastContents;
}
}
/**
* Handles a Stream event.
*
* @param event
* @throws SAXException
*/
private void handleEvent(XMLEvent event) throws SAXException {
if (event.getEventType() == XMLStreamConstants.CHARACTERS) {
Characters c = event.asCharacters();
lastContents += c.getData();
} else if (event.getEventType() == XMLStreamConstants.START_ELEMENT) {
StartElement startElement = event.asStartElement();
String tagLocalName = startElement.getName().getLocalPart();
if ("row".equals(tagLocalName)) {
Attribute rowIndex = startElement.getAttributeByName(new QName("r"));
if (firstRowIndex == -1) {
firstRowIndex = Integer.parseInt(rowIndex.getValue());
}
currentRow = new StreamingRow(Integer.parseInt(rowIndex.getValue()) - 1);
} else if ("cols".equals(tagLocalName)) {
parsingCols = true;
} else if ("col".equals(tagLocalName) && parsingCols) {
colNumber = colNumber + 1;
} else if ("c".equals(tagLocalName)) {
Attribute ref = startElement.getAttributeByName(new QName("r"));
String[] coord = ref.getValue().split("(?<=\\D)(?=\\d)|(?<=\\d)(?=\\D)");
currentCell =
new StreamingCell(
CellReference.convertColStringToIndex(coord[0]), Integer.parseInt(coord[1]) - 1);
setFormatString(startElement, currentCell);
Attribute type = startElement.getAttributeByName(new QName("t"));
if (type != null) {
currentCell.setType(type.getValue());
} else {
currentCell.setType("n");
}
Attribute style = startElement.getAttributeByName(new QName("s"));
if (style != null) {
String indexStr = style.getValue();
try {
int index = Integer.parseInt(indexStr);
currentCell.setCellStyle(stylesTable.getStyleAt(index));
} catch (NumberFormatException nfe) {
LOGGER.warn("Ignoring invalid style index {}", indexStr);
}
}
// we store the dimension as well to revert with this method when cols not found
// can happen see xlsx attached here https://jira.talendforge.org/browse/TDP-1957
// <dimension ref="A1:B60"/>
} else if ("dimension".equals(tagLocalName)) {
Attribute attribute = startElement.getAttributeByName(new QName("ref"));
if (attribute != null) {
this.dimension = attribute.getValue();
}
}
// Clear contents cache
lastContents = "";
} else if (event.getEventType() == XMLStreamConstants.END_ELEMENT) {
EndElement endElement = event.asEndElement();
String tagLocalName = endElement.getName().getLocalPart();
if ("v".equals(tagLocalName) || "t".equals(tagLocalName)) {
currentCell.setRawContents(unformattedContents());
currentCell.setContents(formattedContents());
} else if ("row".equals(tagLocalName) && currentRow != null) {
rowCache.add(currentRow);
} else if ("c".equals(tagLocalName)) {
currentRow.getCellMap().put(currentCell.getColumnIndex(), currentCell);
} else if ("cols".equals(tagLocalName)) {
parsingCols = false;
}
}
}