Search before asking
Fesod version
2.1.0-incubating (main branch, commit 4e3f3aa)
JDK version
Temurin 1.8.0_472
Operating system
Windows 10 (amd64)
Steps To Reproduce
Read a file into a model with a ReadCellData field, or read with readDefaultReturn(ReadDefaultReturnEnum.READ_CELL_DATA).
Model:
public class CellDataReadData {
@DateTimeFormat("yyyy年MM月dd日")
private ReadCellData<String> date;
private ReadCellData<Integer> integer1;
private Integer integer2;
private ReadCellData<?> formulaValue;
}Code:
FesodSheet.write(file, CellDataWriteData.class).sheet().doWrite(data);
CollectingReadListener<CellDataReadData> listener = new CollectingReadListener<>();
FesodSheet.read(file, CellDataReadData.class, listener).sheet().doRead();
CellDataReadData row = listener.getFirstRow();
row.getDate().getRowIndex(); // null, the cell is A2, so row 1
row.getDate().getColumnIndex(); // null, the cell is A2, so column 0
Reproduced on xlsx, xls and csv, so it does not depend on the file format.
Current Behavior
Every ReadCellData that reaches a listener has getRowIndex() == null and getColumnIndex() == null, for a model field and for READ_CELL_DATA alike. WriteCellData on the write side does carry them.
Parsers do set the coordinates on the cell data: CellTagHandler (lines 150-151) for xlsx, FormulaRecordHandler and BoolErrRecordHandler for xls, CsvExcelReadExecutor (lines 237-239) for csv.
They are lost in ConverterUtils.convertToJavaObject: when the target type is CellData or ReadCellData it calls cellData.clone() (line 161) and ReadCellData.clone() copies the value, the number, the boolean, the data format and the formula, but not rowIndex/columnIndex. The method even receives rowIndex and columnIndex as arguments (lines 151-152) at the moment it drops them.
Measured before the fix, xlsx / xls / csv:
date(row=null,col=null) integer1(row=null,col=null) formula(row=null,col=null)
So a listener cannot tell which cell a value came from, which is the main reason to read a cell as ReadCellData instead of as a plain value.
Expected Behavior
A cloned cell describes the same cell, so it keeps the coordinates:
date(row=1,col=0) integer1(row=1,col=1) formula(row=1,col=3)
After the change the values above are read back on xlsx, xls and csv.
Anything else?
The fix copies rowIndex and columnIndex in ReadCellData.clone().
Tests: a round trip assertion in CellDataDataTest (3 formats) and a unit test ReadCellDataTest.cloneKeepsCoordinates. Without the fix 4 of the 5 new assertions fail, with it all of them pass.
PR follows.
Are you willing to submit a PR?
Search before asking
Fesod version
2.1.0-incubating (main branch, commit 4e3f3aa)
JDK version
Temurin 1.8.0_472
Operating system
Windows 10 (amd64)
Steps To Reproduce
Read a file into a model with a
ReadCellDatafield, or read withreadDefaultReturn(ReadDefaultReturnEnum.READ_CELL_DATA).Model:
Code:
Reproduced on xlsx, xls and csv, so it does not depend on the file format.
Current Behavior
Every
ReadCellDatathat reaches a listener hasgetRowIndex() == nullandgetColumnIndex() == null, for a model field and forREAD_CELL_DATAalike.WriteCellDataon the write side does carry them.Parsers do set the coordinates on the cell data:
CellTagHandler(lines 150-151) for xlsx,FormulaRecordHandlerandBoolErrRecordHandlerfor xls,CsvExcelReadExecutor(lines 237-239) for csv.They are lost in
ConverterUtils.convertToJavaObject: when the target type isCellDataorReadCellDatait callscellData.clone()(line 161) andReadCellData.clone()copies the value, the number, the boolean, the data format and the formula, but notrowIndex/columnIndex. The method even receivesrowIndexandcolumnIndexas arguments (lines 151-152) at the moment it drops them.Measured before the fix, xlsx / xls / csv:
So a listener cannot tell which cell a value came from, which is the main reason to read a cell as
ReadCellDatainstead of as a plain value.Expected Behavior
A cloned cell describes the same cell, so it keeps the coordinates:
After the change the values above are read back on xlsx, xls and csv.
Anything else?
The fix copies
rowIndexandcolumnIndexinReadCellData.clone().Tests: a round trip assertion in
CellDataDataTest(3 formats) and a unit testReadCellDataTest.cloneKeepsCoordinates. Without the fix 4 of the 5 new assertions fail, with it all of them pass.PR follows.
Are you willing to submit a PR?