| Summary: | Protected Workbook Read Error | ||
|---|---|---|---|
| Product: | POI | Reporter: | chang.zhao |
| Component: | HSSF | Assignee: | POI Developers List <dev> |
| Status: | RESOLVED WONTFIX | ||
| Severity: | major | ||
| Priority: | P3 | ||
| Version: | 2.5-FINAL | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows XP | ||
We're not planning support for protected workbooks at this stage (it's in the FAQ I believe). It is unclear to us from a legal point of view whether we're allowed to tackle decrypting protected workbooks so we're playing it safe. If you wish to independently tackle this feel free but unless we can get a clear ruling on this it's probably not going to make it into POI proper. |
Create a new Excel Document, select tools->protect workbook, and type in a password. This will cause POIFSFileSystem fs = new POIFSFileSystem(i); HSSFWorkbook wb = new HSSFWorkbook(fs); <= to fail. Upon investigation of this bug, it seems that a protected workbook (but not a protected sheet) will cause the system to report length inaccurately: i.e. in FontRecord fillFields(byte [] data, short size, int offset) ... field_10_font_name_len = data[ 14 + offset ]; ... on an unprotected workbook would give a sane value like 5. However, on a protected workbook it would report a wierd value like 49. I believe the previous bugs with ArrayOutofBoundsException s might be related to this incorrect length reporting. This is not limited to length, other fields are going insane as well. Further, further investigation turns out that protected workbooks have an extra record of type 47, length 56 right between BOF and InterfaceHeader, and it looks like every record after that record (including the interfaceheader) got everything scambled. On a related note, as I was looking at FontRecord.java, in fillFields(byte [] data, short size, int offset) if (data[ 15 ] == 0) { // is compressed unicode field_11_font_name = StringUtil.getFromCompressedUnicode(data, 16, LittleEndian.ubyteToInt(field_10_font_name_len)); } should perhaps be if (data[ 15 + offset] == 0) { field_11_font_name = StringUtil.getFromCompressedUnicode(data, 16 + offset, LittleEndian.ubyteToInt(field_10_font_name_len)); } ? Or is there any reason that offset shouldn't be added? Thanks, CZ