engine xlsxwriter fails for np.inf

[arthurgerigk-rocket]

this works for openpyxl but it fails for xlsxwriter

TypeError("NAN/INF not supported in write_number()")

[jreback]

do nans raise the same? or just inf?

[jreback]

cc @jmcnamara

[arthurgerigk-rocket]

nans are fine

On Thu, Apr 3, 2014 at 7:19 PM, jreback notifications@github.com wrote:

cc @jmcnamara https://github.com/jmcnamara

Reply to this email directly or view it on GitHubhttps://github.com//issues/6782#issuecomment-39478741
.

Arthur Gerigk
Global CMO

Rocket Internet GmbH | Johannisstraße 20 | 10117 Berlin | Deutschland
Mobile: +49 162 290 46 39 | skype: arthur.gerigk | mail:
arthur.gerigk@rocket-internet.de
www.rocket-internet.de

Geschäftsführer: Dr. Johannes Bruder, Arnt Jeschke, Alexander Kudlich
Eingetragen beim Amtsgericht Berlin, HRB 109262 USt-ID DE256469659

[jmcnamara]

I don't think NAN/INF are supported by Excel so I am curious to see what OpenPyXL is doing in this case.

Can you post a small example that demonstrates the issue and I'll look into it.

[arthurgerigk-rocket]

to_excel has a parameter "na_rep" which defaults to the empty string.
actually there are NA values in excel
http://office.microsoft.com/en-001/excel-help/na-HP005209188.aspx

On Thu, Apr 3, 2014 at 7:39 PM, John McNamara notifications@github.comwrote:

I don't think NAN/INF are supported by Excel so I am curious to see what
OpenPyXL is doing in this case.

Can you post a small example that demonstrates the issue and I'll look
into it.

Reply to this email directly or view it on GitHubhttps://github.com//issues/6782#issuecomment-39481271
.

Arthur Gerigk
Global CMO

Rocket Internet GmbH | Johannisstraße 20 | 10117 Berlin | Deutschland
Mobile: +49 162 290 46 39 | skype: arthur.gerigk | mail:
arthur.gerigk@rocket-internet.de
www.rocket-internet.de

Geschäftsführer: Dr. Johannes Bruder, Arnt Jeschke, Alexander Kudlich
Eingetragen beim Amtsgericht Berlin, HRB 109262 USt-ID DE256469659

[jmcnamara]

NA is only supported in Excel as a formula or as the result of a formula. Not as a value.

If you post an example I'll have a look at it.

[arthurgerigk-rocket]

import numpy as np
import pandas as pd
df = pd.DataFrame([(1, np.nan), (2, 3)])
df.to_excel('test.xlsx') # works

df = pd.DataFrame([(1, np.inf), (2, 3)])
df.to_excel('test2.xlsx') # fails

since there is a parameter np_rep I think it has to work ;-)

On Thu, Apr 3, 2014 at 8:14 PM, John McNamara notifications@github.comwrote:

NA is only supported in Excel as a formula or as the result of a formula.
Not as a value.

If you post an example I'll have a look at it.

Reply to this email directly or view it on GitHubhttps://github.com//issues/6782#issuecomment-39485412
.

Arthur Gerigk
Global CMO

Rocket Internet GmbH | Johannisstraße 20 | 10117 Berlin | Deutschland
Mobile: +49 162 290 46 39 | skype: arthur.gerigk | mail:
arthur.gerigk@rocket-internet.de
www.rocket-internet.de

Geschäftsführer: Dr. Johannes Bruder, Arnt Jeschke, Alexander Kudlich
Eingetragen beim Amtsgericht Berlin, HRB 109262 USt-ID DE256469659

[jtratner]

Let's follow the model of to_csv/read_csv and output np.inf as the string 'inf', which will be able to be round-tripped appropriately.

Side note - I can't open up the openpyxl file after it's written - so I'm not sure if openpyxl is doing the right thing.