pandas to_csv output quoting issue

Question

I'm having trouble getting the pandas dataframe.to_csv(...) output quoting strings right.

import pandas as pd

text = 'this is "out text"'
df = pd.DataFrame(index=['1'],columns=['1','2'])
df.loc['1','1']=123
df.loc['1','2']=text
df.to_csv('foo.txt',index=False,header=False)

The output is:

123,"this is ""out text"""

But I would like:

123,this is "out text"

Does anyone know how to get this right?

Solution 1

You could pass quoting=csv.QUOTE_NONE, for example:

>>> df.to_csv('foo.txt',index=False,header=False)
>>> !cat foo.txt
123,"this is ""out text"""
>>> import csv
>>> df.to_csv('foo.txt',index=False,header=False, quoting=csv.QUOTE_NONE)
>>> !cat foo.txt
123,this is "out text"

but in my experience it's better to quote more, rather than less.

Solution 2

Note: there is currently a small error in the Pandas to_string documentation. It says:

• quoting : int, Controls whether quotes should be recognized. Values are taken from csv.QUOTE_* values. Acceptable values are 0, 1, 2, and 3 for QUOTE_MINIMAL, QUOTE_ALL, QUOTE_NONE, and QUOTE_NONNUMERIC,
  respectively.

But this reverses how csv defines the QUOTE_NONE and QUOTE_NONNUMERIC variables.

In [13]: import csv
In [14]: csv.QUOTE_NONE
Out[14]: 3

Solution 3

To use quoting=csv.QUOTE_NONE, you need to set the escapechar, e.g.

# Create a tab-separated file with quotes
$ echo abc$'\t'defg$'\t'$'"xyz"' > in.tsv
$ cat in.tsv
abc defg    "xyz"

# Gotcha the quotes disappears in `"..."`
$ python3
>>> import pandas as pd
>>> import csv
>>> df = pd.read("in.tsv", sep="\t")
>>> df = pd.read_csv("in.tsv", sep="\t")
>>> df
Empty DataFrame
Columns: [abc, defg, xyz]
Index: []


# When reading in pandas, to read the `"..."` quotes,
# you have to explicitly say there's no `quotechar`
>>> df = pd.read_csv("in.tsv", sep="\t", quotechar='\0')
>>> df
Empty DataFrame
Columns: [abc, defg, "xyz"]
Index: []

# To print out without the quotes.
>> df.to_csv("out.tsv", , sep="\t", quoting=csv.QUOTE_NONE, quotechar="",  escapechar="\\")

Solution 4

To use without escapechar:

Replace comma char , (Unicode:U+002C) in your df with an single low-9 quotation mark character ‚ (Unicode: U+201A)

After this, you can simply use:

import csv df.to_csv('foo.txt', index=False, header=False, quoting=csv.QUOTE_NONE)

View more solutions

Comments

[user3199761]

I tried this but got "Error: need to escape, but no escapechar set". I also tried to add escapechar='\\' or escapechar=None, but does not seem to work...

[DSM]

@user3199761: that's because your real data -- unlike the example you pasted -- has something that you need to escape, such as a string with a comma. If you output a,"b,c",d you can parse the results, but what are the columns if you write out a,b,c,d? You can't tell.

[user3199761]

I did try your code with the same example above, but got the error. I am using pandas 0.11.0 and python 2.7.6, not sure if this helps.

[DSM]

@user3199761: it might, though I'd be a little surprised. I'm using (pre-release) 0.13.0.

[ericmjl]

@DSM, I also tried your code, and it returned the exact same error.

[user3199761]

@DSM: both pandas 0.11.0 and 0.12.0 does not work, but upgrading to pandas 0.13.0 solved the problem!

[user5359531]

so does this mean that in order to disable quoting, you need to load another library and call a function from it within the function for pandas?

[DSM]

@user5359531: no, it means you import the built-in csv library just to get the csv.QUOTE_NONE constant (which happens to be the number 3, but you've got no guarantee that's always going to be true).

[Leukonoe]

Got error "Error: need to escape, but no escapechar set". I'm aware, that there might be issues with columns, but need the solution anyway.

[DamDam]

This solution works for me, even with commas into some strings. I just use a different separator : df.to_csv(f"./export_{today}-{current_time}.tsv", sep="\t", index=False, quoting=csv.QUOTE_NONE)