[snip]
>
> Interesting approach.
>
> For a possibly limited but cross-platform way (i.e. don't need to be
> on Windows or use pywin32) to do the same conversion from .XLSX
> to .XLS files, it is also possible to use an XML parser, such as a SAX-
> capable parser (to read the content (*) of the .XLSX files, and then
> write the same content to .XLS files using, I guess, the Python xlwt
> library, which is mentioned in other messages in this group.). (I have
> not used xlwt (yet), which is why I said "I guess", though I have used
> its counterpart for reading, xlrd, in my xtopdf toolkit.)
>
> (*) Conditions apply - see below.
>
> This alternative method is possible because .XLSX format files are a
> kind of XML. There is a recipe for how to extract the text-only
> content (i.e. numbers and strings, no formatting or images or charts -
> this is the condition mentioned above) of .XLSX files, using SAX, in
> the Python Cookbook 2nd Edition.
XLSX files were introduced by Excel 2007 (i.e. v12). An XLSX file is a
ZIP file containing a bundle of XML documents. The Python Cookbook 2nd
Edition was published in 2005. The recipe (12.7) to which you refer
relates to the XML files that can be produced by Excel 2003 (v11) and
Excel XP (v10), using the "XML Spreadsheet" option of "Save as". The two
formats are XMLly and Microsofty but otherwise rather dissimilar.
> I had tried out that recipe some time
> ago (it worked fine, though I had to tweak it a bit), and used it to
> convert the (text-only) content of .XLSX files to PDF, as part of my
> xtopdf toolkit. That code is not in the xtopdf release yet, but will
> be after some time. If I can dig up the (standalone) code I wrote for
> that conversion, I'll post a link to it here in a few days. But
> basically, it's really easy to read .XLSX content with Python using
> SAX, since there are clearly defined XML elements for tables, rows and
> cells. In fact, that means you can also read the .XLSX content using
> any language that has a SAX XML parser, not just Python.
Any parser within reason can be used, not just SAX. We have an XLSX
parser (using ElementTree) in the queue to be plugged into xlrd. It
handles the basics i.e. open_workbook(..., formatting_info=0).