In-memory workbooks are not compressed on close()

[theag3nt]

Using XlsxWriter with the in_memory option results in files which are not properly compressed.

The issue is caused by XlsxWriter constructing its own ZipInfo objects and using ZipFile.writestr() to write them without specifying the compression type. It's not documented, but from the Python zipfile module's source I was able to tell that in this case ZipInfo files do not inherit the ZipFile's compression and default to ZIP_STORED. As the ZipFile is instantiated with ZIP_DEFLATED I assume this is not intentional.

I've verified the problem occurs with XlsxWriter 1.1.1 (and the latest development version too) both with Python version 2.7.15 and 3.6.6.

A short script that can be used to reproduce the problem:

from xlsxwriter import Workbook

with open("sample.xlsx", "wb") as f, Workbook(f, {"in_memory": True}) as wb:
    ws = wb.add_worksheet()
    ws.write_number("A1", 0)

I've first noticed the size differences when comparing generated files with ones created by Excel. But it can be verified with the zipinfo command:

$ zipinfo sample.xlsx
Archive:  sample.xlsx
Zip file size: 13503 bytes, number of entries: 9
?rw-------  2.0 unx      516 b- stor 80-Jan-01 00:00 xl/worksheets/sheet1.xml
?rw-------  2.0 unx      550 b- stor 80-Jan-01 00:00 xl/workbook.xml
?rw-------  2.0 unx      784 b- stor 80-Jan-01 00:00 docProps/app.xml
?rw-------  2.0 unx      592 b- stor 80-Jan-01 00:00 docProps/core.xml
?rw-------  2.0 unx     1031 b- stor 80-Jan-01 00:00 [Content_Types].xml
?rw-------  2.0 unx      867 b- stor 80-Jan-01 00:00 xl/styles.xml
?rw-------  2.0 unx     6994 b- stor 80-Jan-01 00:00 xl/theme/theme1.xml
?rw-------  2.0 unx      587 b- stor 80-Jan-01 00:00 _rels/.rels
?rw-------  2.0 unx      556 b- stor 80-Jan-01 00:00 xl/_rels/workbook.xml.rels
9 files, 12477 bytes uncompressed, 12477 bytes compressed:  0.0%

The expected compression level is greater than 0%, also the files are listed with STORE compression.

[jmcnamara]

Thanks for the detailed bug report and investigation. I'll do some extra testing and merge up the PR.

John

[jmcnamara]

Fixed in version 1.1.2.

Thanks.

[theag3nt]

Great, thank you!