adls: Operations on an Azure Data Lake Storage Gen2 filesystem

In Hong-Revo/AzureStor: Storage Management in 'Azure'

Operations on an Azure Data Lake Storage Gen2 filesystem

Description

Upload, download, or delete a file; list files in a directory; create or delete directories; check file existence.

Usage

list_adls_files(filesystem, dir = "/", info = c("all", "name"),
  recursive = FALSE)

multiupload_adls_file(filesystem, src, dest, recursive = FALSE,
  blocksize = 2^22, lease = NULL, put_md5 = FALSE, use_azcopy = FALSE,
  max_concurrent_transfers = 10)

upload_adls_file(filesystem, src, dest = basename(src), blocksize = 2^24,
  lease = NULL, put_md5 = FALSE, use_azcopy = FALSE)

multidownload_adls_file(filesystem, src, dest, recursive = FALSE,
  blocksize = 2^24, overwrite = FALSE, check_md5 = FALSE,
  use_azcopy = FALSE, max_concurrent_transfers = 10)

download_adls_file(filesystem, src, dest = basename(src), blocksize = 2^24,
  overwrite = FALSE, check_md5 = FALSE, use_azcopy = FALSE)

delete_adls_file(filesystem, file, confirm = TRUE)

create_adls_dir(filesystem, dir)

delete_adls_dir(filesystem, dir, recursive = FALSE, confirm = TRUE)

adls_file_exists(filesystem, file)

adls_dir_exists(filesystem, dir)

Arguments

filesystem	An ADLSgen2 filesystem object.
dir, file	A string naming a directory or file respectively.
info	Whether to return names only, or all information in a directory listing.
recursive	For the multiupload/download functions, whether to recursively transfer files in subdirectories. For list_adls_files, and delete_adls_dir, whether the operation should recurse through subdirectories. For delete_adls_dir, this must be TRUE to delete a non-empty directory.
src, dest	The source and destination paths/files for uploading and downloading. See 'Details' below.
blocksize	The number of bytes to upload/download per HTTP(S) request.
lease	The lease for a file, if present.
put_md5	For uploading, whether to compute the MD5 hash of the file(s). This will be stored as part of the file's properties.
use_azcopy	Whether to use the AzCopy utility from Microsoft to do the transfer, rather than doing it in R.
max_concurrent_transfers	For multiupload_adls_file and multidownload_adls_file, the maximum number of concurrent file transfers. Each concurrent file transfer requires a separate R process, so limit this if you are low on memory.
overwrite	When downloading, whether to overwrite an existing destination file.
check_md5	For downloading, whether to verify the MD5 hash of the downloaded file(s). This requires that the file's Content-MD5 property is set. If this is TRUE and the Content-MD5 property is missing, a warning is generated.
confirm	Whether to ask for confirmation on deleting a file or directory.

Details

upload_adls_file and download_adls_file are the workhorse file transfer functions for ADLSgen2 storage. They each take as inputs a single filename as the source for uploading/downloading, and a single filename as the destination. Alternatively, for uploading, src can be a textConnection or rawConnection object; and for downloading, dest can be NULL or a rawConnection object. If dest is NULL, the downloaded data is returned as a raw vector, and if a raw connection, it will be placed into the connection. See the examples below.

multiupload_adls_file and multidownload_adls_file are functions for uploading and downloading multiple files at once. They parallelise file transfers by using the background process pool provided by AzureRMR, which can lead to significant efficiency gains when transferring many small files. There are two ways to specify the source and destination for these functions:

• Both src and dest can be vectors naming the individual source and destination pathnames.

• The src argument can be a wildcard pattern expanding to one or more files, with dest naming a destination directory. In this case, if recursive is true, the file transfer will replicate the source directory structure at the destination.

upload_adls_file and download_adls_file can display a progress bar to track the file transfer. You can control whether to display this with options(azure_storage_progress_bar=TRUE|FALSE); the default is TRUE.

adls_file_exists and adls_dir_exists test for the existence of a file and directory, respectively.

AzCopy

upload_azure_file and download_azure_file have the ability to use the AzCopy commandline utility to transfer files, instead of native R code. This can be useful if you want to take advantage of AzCopy's logging and recovery features; it may also be faster in the case of transferring a very large number of small files. To enable this, set the use_azcopy argument to TRUE.

Note that AzCopy only supports SAS and AAD (OAuth) token as authentication methods. AzCopy also expects a single filename or wildcard spec as its source/destination argument, not a vector of filenames or a connection.

Value

For list_adls_files, if info="name", a vector of file/directory names. If info="all", a data frame giving the file size and whether each object is a file or directory.

For download_adls_file, if dest=NULL, the contents of the downloaded file as a raw vector.

For adls_file_exists, either TRUE or FALSE.

See Also

adls_filesystem, az_storage, storage_download, call_azcopy

Examples

## Not run: 

fs <- adls_filesystem("https://mystorage.dfs.core.windows.net/myfilesystem", key="access_key")

list_adls_files(fs, "/")
list_adls_files(fs, "/", recursive=TRUE)

create_adls_dir(fs, "/newdir")

upload_adls_file(fs, "~/bigfile.zip", dest="/newdir/bigfile.zip")
download_adls_file(fs, "/newdir/bigfile.zip", dest="~/bigfile_downloaded.zip")

delete_adls_file(fs, "/newdir/bigfile.zip")
delete_adls_dir(fs, "/newdir")

# uploading/downloading multiple files at once
multiupload_adls_file(fs, "/data/logfiles/*.zip")
multidownload_adls_file(fs, "/monthly/jan*.*", "/data/january")

# you can also pass a vector of file/pathnames as the source and destination
src <- c("file1.csv", "file2.csv", "file3.csv")
dest <- paste0("uploaded_", src)
multiupload_adls_file(share, src, dest)

# uploading serialized R objects via connections
json <- jsonlite::toJSON(iris, pretty=TRUE, auto_unbox=TRUE)
con <- textConnection(json)
upload_adls_file(fs, con, "iris.json")

rds <- serialize(iris, NULL)
con <- rawConnection(rds)
upload_adls_file(fs, con, "iris.rds")

# downloading files into memory: as a raw vector, and via a connection
rawvec <- download_adls_file(fs, "iris.json", NULL)
rawToChar(rawvec)

con <- rawConnection(raw(0), "r+")
download_adls_file(fs, "iris.rds", con)
unserialize(con)


## End(Not run)

Hong-Revo/AzureStor documentation built on Aug. 3, 2022, 3:37 p.m.