Download CSV from an iPython Notebook

问题

I run an iPython Notebook server, and would like users to be able to download a pandas dataframe as a csv file so that they can use it in their own environment. There's no personal data, so if the solution involves writing the file at the server (which I can do) and then downloading that file, I'd be happy with that.

回答1:

How about using the FileLinks class from IPython? I use this to provide access to data directly from Jupyter notebooks. Assuming your data is in pandas dataframe p_df:

from IPython.display import FileLink, FileLinks

p_df.to_csv('/path/to/data.csv', index=False)
p_df.to_excel('/path/to/data.xlsx', index=False)

FileLinks('/path/to/')

Run this as a notebook cell and the result will be a list of links to files downloadable directly from the notebook. '/path/to' needs to be accessible for the notebook user of course.

回答2:

For not too large tables you can use the following code:

import base64
import pandas as pd
from IPython.display import HTML

def create_download_link( df, title = "Download CSV file", filename = "data.csv"):
    csv = df.to_csv()
    b64 = base64.b64encode(csv.encode())
    payload = b64.decode()
    html = '<a download="{filename}" href="data:text/csv;base64,{payload}" target="_blank">{title}</a>'
    html = html.format(payload=payload,title=title,filename=filename)
    return HTML(html)

df = pd.DataFrame(data = [[1,2],[3,4]], columns=['Col 1', 'Col 2'])
create_download_link(df)

回答3:

If you want to avoid storing CSVs on the server, you can use this Javascript alternative that create the CSV on the client-side:

from IPython.display import Javascript
js_download = """
var csv = '%s';

var filename = 'results.csv';
var blob = new Blob([csv], { type: 'text/csv;charset=utf-8;' });
if (navigator.msSaveBlob) { // IE 10+
    navigator.msSaveBlob(blob, filename);
} else {
    var link = document.createElement("a");
    if (link.download !== undefined) { // feature detection
        // Browsers that support HTML5 download attribute
        var url = URL.createObjectURL(blob);
        link.setAttribute("href", url);
        link.setAttribute("download", filename);
        link.style.visibility = 'hidden';
        document.body.appendChild(link);
        link.click();
        document.body.removeChild(link);
    }
}
""" % data_in_dataframes.to_csv(index=False).replace('\n','\\n').replace("'","\'")

Javascript(js_download)

Basically, it creates a CSV string in python from the pd dataframe and use it in a small js script that creates a CSV file on the client side and open a saving dialog to save it on the user computer. I tested in my iPython env and it works like a charm!

Note that I am escaping the \n. If I don't do so, the js script string will have the CSV variable written on multiple lines.

For example, print "var csv = '%s'" % industries_revenues.to_csv(index=False).replace('\n','\\n') results to this:

var csv = 'Industry,sum_Amount\nBanking,65892584.0\n(...)Finance,20211917.0\n'

Instead of print "var csv = '%s'" % industries_revenues.to_csv(index=False) without the \n escaping that results on a multiple lined and therefore errored javascript:

var csv = 'Industry,sum_Amount
Banking,65892584.0
(...)
Finance,20211917.0
'

I also escape the ' not to break the variable string in javascript.

回答4:

You can use the fact that the notebook can display html for objects, and data urls, to make the content of a csv downloadable:

import urllib

class CSV(object):
    def _repr_html_(self):
        html = []

        html.append("{},{},{}".format(
                "user",
                "age",
                "city"
            )
        )

        html.append("{},{},{}".format(
                "Alice",
                "39",
                "New York"
            )
        )

        html.append("{},{},{}".format(
                "Bob",
                "30",
                "Denver"
            )
        )

        html.append("{},{},{}".format(
                "Carol",
                "27",
                "Tulsa"
            )
        )


        export = '\n'.join(html)
        export = urllib.quote(export.encode("utf-8"))
        csvData = 'data:application/csv;charset=utf-8,' + export
        return "<a download='export.csv' href='{}' target='_blank'>csv file</a>".format(csvData)

CSV()

回答5:

A function that creates a csv download link, based on Coen Jonker's answer.

The function has an optional delete prompt so you can delete the file after download to keep the notebook server clean.

def csv_download_link(df, csv_file_name, delete_prompt=True):
    """Display a download link to load a data frame as csv from within a Jupyter notebook"""
    df.to_csv(csv_file_name, index=False)
    from IPython.display import FileLink, FileLinks
    display(FileLink(csv_file_name))
    if delete_prompt:
        a = input('Press enter to delete the file after you have downloaded it.')
        import os
        os.remove(csv_file_name)
# Diplay download link:
csv_download_link(df, 'file_name.csv')

回答6:

My simple approach to download all the files from the jupyter notebook would be by simply using this wonderful command

!tar cvfz my_compressed_file_name.tar.gz *

This will download all the files of the server including the notebooks.

In case if your server has multiple folders, you might be willing to use the following command. write ../ before the * for every step up the directory.

tar cvfz zipname.tar.gz ../../*

Hope it helps..

来源：https://stackoverflow.com/questions/31893930/download-csv-from-an-ipython-notebook

标签

csv

pandas

ipython-notebook