pandas.errors.ParserError: ',' expected after '"'

Question

I am trying to read this dataset from Kaggle: Amazon sales rank data for print and kindle books

The file amazon_com_extras.csv has a column named "Title" that sometimes contains a comma ',' so all the fields in this .csv are enclosed by quotation marks:

"ASIN","GROUP","FORMAT","TITLE","AUTHOR","PUBLISHER"
"022640014X","book","hardcover","The Diversity Bargain: And Other Dilemmas of Race, Admissions, and Meritocracy at Elite Universities","Natasha K. Warikoo","University Of Chicago Press"

I have read other questions related to this problem but none of them solve it. For example, I have tried:

df = pd.read_csv("amazon_com_extras.csv",engine="python",sep=',')
df = pd.read_csv("amazon_com_extras.csv",engine="python",sep=',',quotechar='"')

But nothing seems to work. I am using Python 3.7.2 and pandas 0.24.1.

Answer 1

This is happening to you because there are fields inside the document that contain unescaped quotes inside the quoted text.

I am not aware of a way to instruct the csv parser to handle that without preprocessing.

If you don't care about those columns, you can use

pd.read_csv("amazon_com_extras.csv", engine="python", sep=',', quotechar='"', error_bad_lines=False)

That will disable the Exception from being raised, but it will remove the affected lines (you will see that in the console).

An example of such a line:

"1405246510","book","hardcover",""Hannah Montana" Annual 2010","Unknown","Egmont Books Ltd"

Notice the quotes.

Instead, a more standard dialect of csv would have rendered:

1405246510,"book","hardcover","""Hannah Montana"" Annual 2010","Unknown","Egmont Books Ltd"

You can, for example, load the file with Libreoffice and re-save it as CSV again to get a working CSV dialect or use other preprocessing techniques.

Answer 2

The problem is that pandas treats the char " for queting, and expects " after every " in a cell, which doesn't happen in this csv.

To make pandas not treat it as a quoting mark, pass the parameter quoting=3 inside the pd.read_csv function.