How To Import A CSV File Using Google Sheets API V4

Background I'm developing a Python 2.7 script that analyzes data from an SQL table and at the end, generates a CSV file. Once the file is generated, I'm logging into my google shee

Solution 1:

You have two options for importing g CSV file. You can use the Drive API to create a spreadsheet from a CSV, or you can use the Sheets API to create an empty spreadsheet and then use spreadsheets.batchUpdate with a PasteDataRequest to add CSV data.

Solution 2:

I've spent couple of hours trying to make any of the other answers work. Libraries do not explain the authentication well, and don't work with google-provided way of handling credentials. On the other hand, Sam's answer doesn't elaborate on the details of using the API, which might be confusing at times. So, here is a full recipe of uploading CSVs to gSheets. It uses both Sam's and CapoChino's answers plus some of my own research.

1. Authenticate/Setup. Generally, refer to the docs
   • Big blue button will get you credentials.json with no extra steps
   • quickstart.py can easily be adapted into authenticate.py
   • scopes should contain https://www.googleapis.com/auth/spreadsheets

Hopefully by now you have your credentials stored, so let's move to the actual code

2. Recipe that should work out of the box:

import pickle
from googleapiclient.discovery import build

SPREADSHEET_ID = '1BxiMVs0XRA5nFMdKvBdBZjgmUUqptlbs74OgvE2upms' # Get this one from the link in browser
worksheet_name = 'Sheet2'
path_to_csv = 'New Folder/much_data.csv'
path_to_credentials = 'Credentials/token.pickle'


# convenience routines
def find_sheet_id_by_name(sheet_name):
    # ugly, but works
    sheets_with_properties = API \
        .spreadsheets() \
        .get(spreadsheetId=SPREADSHEET_ID, fields='sheets.properties') \
        .execute() \
        .get('sheets')

    for sheet in sheets_with_properties:
        if 'title' in sheet['properties'].keys():
            if sheet['properties']['title'] == sheet_name:
                return sheet['properties']['sheetId']


def push_csv_to_gsheet(csv_path, sheet_id):
    with open(csv_path, 'r') as csv_file:
        csvContents = csv_file.read()
    body = {
        'requests': [{
            'pasteData': {
                "coordinate": {
                    "sheetId": sheet_id,
                    "rowIndex": "0",  # adapt this if you need different positioning
                    "columnIndex": "0", # adapt this if you need different positioning
                },
                "data": csvContents,
                "type": 'PASTE_NORMAL',
                "delimiter": ',',
            }
        }]
    }
    request = API.spreadsheets().batchUpdate(spreadsheetId=SPREADSHEET_ID, body=body)
    response = request.execute()
    return response


# upload
with open(path_to_credentials, 'rb') as token:
    credentials = pickle.load(token)

API = build('sheets', 'v4', credentials=credentials)

push_csv_to_gsheet(
    csv_path=path_to_csv,
    sheet_id=find_sheet_id_by_name(worksheet_name)
)

Good thing about directly using batchUpdate is that it uploads thousands of rows in a second. On a low level gspread does the same and should be as performant. Also there is gspread-pandas.

p.s. the code is tested with python 3.5, but this thread seemed to be most appropriate to submit it to.

Solution 3:

Another alternative to Sam Berlin's answer. If you're using Python, you can use the Drive API via gspread to import a CSV file. Here's an example:

import gspread

# Check how to get `credentials`:
# https://github.com/burnash/gspread

gc = gspread.authorize(credentials)

# Read CSV file contents
content = open('file_to_import.csv', 'r').read()

gc.import_csv('<SPREADSHEET_ID>', content)

Related question: Upload CSV to Google Sheets using gspread

Solution 4:

I like Burnash's gspread library, but the import_csv function in his answer is limited. It always starts the paste at A1 of the first worksheet (tab) and deletes all other tabs.

I needed to paste starting at a particular tab and cell, so I took Sam Berlin's suggestion to use a PasteDataRequest. Here's my function:

def pasteCsv(csvFile, sheet, cell):
    '''
    csvFile - path to csv file to upload
    sheet - a gspread.Spreadsheet object
    cell - string giving starting cell, optionally including sheet/tab name
      ex: 'A1', 'MySheet!C3', etc.
    '''
    if '!' in cell:
        (tabName, cell) = cell.split('!')
        wks = sheet.worksheet(tabName)
    else:
        wks = sheet.sheet1
    (firstRow, firstColumn) = gspread.utils.a1_to_rowcol(cell)

    with open(csvFile, 'r') as f:
        csvContents = f.read()
    body = {
        'requests': [{
            'pasteData': {
                "coordinate": {
                    "sheetId": wks.id,
                    "rowIndex": firstRow-1,
                    "columnIndex": firstColumn-1,
                },
                "data": csvContents,
                "type": 'PASTE_NORMAL',
                "delimiter": ',',
            }
        }]
    }
    return sheet.batch_update(body)

Note that I used a raw pasteData request rather than the higher-level update_cells method to take advantage of Google's automatic (correct) handling of input data that contains quoted strings, which may contain non-delimeter commas.

Solution 5:

An alternative to Sam Berlin's answer, you can turn your CSV into a list of lists and set that to your POST payload.

Such a function looks something like this:

def preprocess(table):
    table.to_csv('pivoted.csv') # I use Pandas but use whatever you'd like
    _file = open('pivoted.csv')
    contents = _file.read()
    array = contents.split('\n')
    master_array = []
    for row in array:
        master_array.append(row.split(','))
    return master_array

That master array gets thrown into the following:

body = {
      'values': newValues
}

    result2 = service.spreadsheets().values().update(spreadsheetId=spreadsheetId, range=rangeName + str(len(values) + start + 1), valueInputOption="USER_ENTERED", body=body).execute()

It works just fine for me.