can_decrypt <- gargle::secret_has_key("GOOGLESHEETS4_KEY")
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  error = TRUE,
  purl = can_decrypt,
  eval = can_decrypt
)
message("No token available. Code chunks will not be evaluated.")
# article assume current user has access to a Sheet named "chicken-sheet"
# execute this if it seems to have gone missing
x <- drive_find("chicken")
if (nrow(x) != 1) {
  x <- gs4_example("chicken-sheet")
  y <- drive_cp(x, name = "chicken-sheet")
}

Why use googlesheets4 and googledrive together?

googlesheets4 wraps the Sheets API v4, which lets you read, write, and format data in Sheets. The Sheets API is very focused on spreadsheet-oriented data and metadata, i.e. (work)sheets and cells.

The Sheets API offers practically no support for file-level operations, other than basic spreadsheet creation. There is no way to delete, copy, or rename a Sheet or to place it in a folder or to change its sharing permissions. We must use the Drive API for all of this, which is wrapped by the googledrive package (https://googledrive.tidyverse.org).

Another reason to use the googlesheets4 and googledrive packages together is for ease of file (Sheet) identification. The googlesheets4 package requires you to specify the target Sheet by its ID, not by its name. That's because the underlying APIs only accept file IDs. But the googledrive package offers lots of support for navigating between human-friendly file names and their associated IDs. This support applies to all files on Drive and, specifically, to Sheets.

Therefore, it is common to use googledrive and googlesheets4 together in a script or app.

Coordinating auth

How does auth work if you're using googlesheets4 and googledrive? The path of least resistance is to do nothing and just let each package deal with its own auth. This works fine! But it's a bit clunky and you need to make sure you're using the same Google identity (email) with each package/API.

It can be nicer to be proactive about auth and use the same token for your googledrive and googlesheets4 work. Below we show a couple of ways to do this.

Auth with googledrive first, then googlesheets4

Outline:

First attach both packages.

library(googledrive)
library(googlesheets4)

Do auth first with googledrive. Remember googledrive::drive_auth() accepts additional arguments, e.g. to specify a Google identity via email = or to use a service account via path =. Then direct googlesheets4 to use the same token as googledrive.

drive_auth()
gs4_auth(token = drive_token())
# happens in .onLoad() when IN_PKGDOWN, but need this for local dev/preview
googlesheets4:::gs4_auth_docs(drive = TRUE) 

# attempt to reduce quota exhaustion problems
if (identical(Sys.getenv("IN_PKGDOWN"), "true")) Sys.sleep(30)

Now you can use googledrive functions, like googledrive::drive_find() or googledrive::drive_get(), to list files or find them by name, path, or other property. Then, once you've identified the target file, use googlesheets4 to do spreadsheet-specific tasks.

drive_find("chicken")

ss <- drive_get("chicken-sheet")

gs4_get(ss)

read_sheet(ss)

If you ever want to confirm the currently authenticated user, both packages provide a *_user() function that reveals some info:

drive_user()
gs4_user()

We are using a service account to render this article. But if you've used the default OAuth flow, this should correspond to the email of the Google account you logged in with.

gs4_deauth()
drive_deauth()
detach("package:googledrive", character.only = TRUE)
detach("package:googlesheets4", character.only = TRUE)

Auth with googlesheets4 first, then googledrive

Outline:

First attach both packages.

library(googlesheets4)
library(googledrive)

Do auth first with googlesheets4, specifying a Drive scope. Remember gs4_auth() accepts additional arguments, e.g. to specify a Google identity via email = or to use a service account via path =. Then direct googledrive to use the same token as googlesheets4.

gs4_auth(scope = "https://www.googleapis.com/auth/drive")
drive_auth(token = gs4_token())
googlesheets4:::gs4_auth_docs(drive = TRUE)

Now you can use googledrive functions to list files or find them by name, path, or other property. Then, once you've identified the target file, use googlesheets4 to do spreadsheet-specific tasks.

drive_find("chicken")

ss <- drive_get("chicken-sheet")

gs4_get(ss)

read_sheet(ss)
gs4_deauth()
drive_deauth()
detach("package:googledrive", character.only = TRUE)
detach("package:googlesheets4", character.only = TRUE)

Scope savvy

If you only need "read" access to Drive or Sheets, the conservative thing to do is to specify a read-only scope. This is a great way to limit the damage anyone can do with the token -- you or someone else -- through carelessness or malice. If you are storing a token on a remote or shared location, it is wise to use the most conservative scope that still gets the job done.

Here are various scopes relevant to googledrive and googlesheets4 and what they would allow.

drive scope allows reading and writing with Drive and Sheets APIs. This scope is the most powerful and, therefore, the most dangerous.

PACKAGE_auth(
  ...,
  scopes = "https://www.googleapis.com/auth/drive",
  ...
)

drive.readonly still allows file identification via Drive and can be combined with spreadsheets if you plan to edit, create, or delete Sheets.

PACKAGE_auth(
  ...,
  scopes = c(
    "https://www.googleapis.com/auth/drive.readonly",
    "https://www.googleapis.com/auth/spreadsheets"
  ),
  ...
)

If you are just using Drive to identify Sheets and are only reading from those Sheets, the drive.readonly scope is sufficient and means you can't modify anything by accident.

PACKAGE_auth(
  ...,
  scopes = "https://www.googleapis.com/auth/drive.readonly",
  ...
)

If you are not using Drive at all, i.e. you always identify Sheets by file ID, and you are only reading from those Sheets, you only need googlesheets4 and spreadsheets.readonly is sufficient.

gs4_auth(
  ...,
  scopes = "https://www.googleapis.com/auth/spreadsheets.readonly",
  ...
)


jennybc/googlesheets2 documentation built on Dec. 10, 2023, 12:56 a.m.