Skip to contents

sasr

Code Coverage

This package provides interface to SAS through saspy and reticulate.

Prerequisites

To use sasr, you need to make sure you have the following

  1. An SAS server that is accessible from the machine that you want to run sasr on
  2. The machine that you want to run sasr has Python and Java

Installation

To install sasr, please use the following command

remotes::install_github(repo = 'insightsengineering/sasr')

Reticulate will be installed automatically, but Python package saspy will not.

If you do not have Python, you can use the following code to install Python, or it can be installed automatically after you call some python related stuffs.

library(reticulate)
install_python()

To install saspy, use the following code

library(sasr)
install_saspy()

After the installation completes, you are ready to use sasr package.

Short Tutorial

To use sasr, you need to follow these steps

  1. Configure your SAS server in sascfg_personal.py under your working directory or the home directory. This is the default file that sasr will look at. However, you can still change that through options(sascfg = ), then sasr will try to find any name that is available in your specified option.
    1. If you don’t know how to create this file, use sascfg() to create the file. Required arguments include host and saspath.
      1. sascfg() only creates ssh based SAS session.
      2. Only password-less ssh connection is supported, e.g. ssh via public keys.
      3. host is the hostname of the SAS server.
      4. saspath is the SAS executable path on the SAS server.
      5. Other arguments are added to the configuration file directly.
      6. tunnel and rtunnel are required if you want to transfer datasets between R and SAS. Use integers like tunnel = 9999L in R, or modify sascfg_personal.py to make sure they are integers.
    2. You can create the configuration by yourself and then SAS connection will not be restricted to ssh.
    3. You can have multiple configuration files with different file names
  2. Create the SAS session based on the configuration file
    1. To use the default connection specified in the configuration file, you can run any command like run_sas, df2sd or sd2df.
      1. The session will be created if there is no session available stored in .sasr_env$.sas_session
      2. If .sasr_env$.sas_session is created, this session will be used by default.
      3. Do not create any variable called .sas_session in environment sasr:::.sasr_env
    2. To create the session manually, you can call sas_session_ssh()
      1. SAS_session have one argument sascfg, pointing to the SAS session configuration file.
    3. To use multiple sessions, you need to store the session your_session <- sas_session_ssh(sascfg)
  3. Transfer the datasets from R to SAS using df2sd
    1. Tunneling must be enabled to transfer datasets.
    2. The variable names of the datasets should not contain dots otherwise SAS may not recognize.
    3. The index (row names) will not be transferred to SAS.
  4. Use run_sas to submit SAS code to the SAS server.
    1. The returned value is a named list, LST is the result and LOG is the log file
    2. run_sas has argument results=, it can be either “TEXT” or “HTML”. This argument decides the LST format.
  5. Transfer SAS datasets back to R use sd2df

Short Example 

library(sasr)
df2sd(mtcars, "mt")
result <- run_sas("
  proc freq data = mt;
  run;
")

cat(result$LOG)

cat(result$LST)

FAQ

Q: Why use saspy instead of using ssh tunnels?

A: Although we can use ssh tunnels to transfer data and execute SAS commands, there are many restrictions: it only supports ssh connection. Using saspy, the official Python interface to SAS, we can enable all connection types, without reinventing the wheel, e.g. we can also connect to a local SAS installation with the same syntax, or connect to a remote SAS Viya through http. In addition, SAS sessions in saspy will not end until you terminate it (or encounter net work issues), it will be nice to execute multiple SAS code one by one, not necessarily putting them in one script and execute the whole script at once. Also, with the update of saspy over time, sasr will be easily extensible, to include functionalities other than transferring data and executing SAS code.