Referencing External Files

One way to reference external files is with a fileref. A fileref is a logical name associated with an external file. You can assign a fileref with a File Shortcut in the SAS Explorer window, the My Favorite Folders window, the FILENAME statement, the FILENAME function, or you can use a Windows environment variable to point to the file. This section discusses the different ways to assign filerefs and also shows you how to obtain a listing of the active filerefs and clear filerefs during your SAS session.

In an interactive SAS session, you can use the SAS Explorer window or the My Favorite Folders window to create filerefs. The SAS Explorer File Shortcuts folder contains a listing of active filerefs. To create a new fileref from SAS Explorer:

To assign a file shortcut using the My Favorite Folders window:

You can then use these file shortcuts in your SAS programs.

The FILENAME statement provides a means to associate a logical name with an external file or directory.

The simplest syntax of the FILENAME statement is as follows:

For example, if you want to read the file C:\MYDATA\SCORES.DAT, you can issue the following statement to associate the fileref MYDATA with the file C:\MYDATA\SCORES.DAT:

filename mydata "c:\mydata\scores.dat";

Then you can use this fileref in your SAS programs. For example, the following statements create a SAS data set named TEST, using the data stored in the external file referenced by the fileref MYDATA:

data test;
   infile mydata;
   input name $ score;
run;

You can also use the FILENAME, FILE, and INFILE statements to concatenate directories of external files and to concatenate multiple individual external files into one logical external file. These topics are discussed in Assigning a Fileref to Concatenated Directories and Assigning a Fileref to Concatenated Files.

The * and ? wildcards can be used in either the external filename or file extension for matching input filenames. Use * to match one or more characters and the ? to match a single character. Wildcards are supported for input only in the FILENAME and INFILE statements, and in member-name syntax (aggregate syntax). Wildcards are not valid in the FILE statement. The following filename statement reads input from every file in the current directory that begins with the string wild and ends with .dat:

filename wild 'wild*.dat';
data;
   infile wild;
   input;
run;

The following example reads all files in the current working directory:

filename allfiles '*.*';
data;
   infile allfiles;
   input;
run;

The FILENAME statement accepts various options that enable you to associate device names, such as printers, with external files and to control file characteristics, such as record format and length. Some of these options are illustrated in Advanced External I/O Techniques. For the complete syntax of the FILENAME statement, refer to FILENAME Statement: Windows.

Just as you can define an environment variable to serve as a logical name for a SAS library (see Assigning SAS Libraries Using Environment Variables), you can also use an environment variable to refer to an external file. You can choose either to define a SAS environment variable using the SET system option or to define a Windows environment variable using the Windows SET command. Alternatively, you can define environment variables using the System dialog box, accessed from the Control Panel.

The availability of environment variables makes it simple to assign resources to SAS before invocation. However, the environment variables that you define (using the SET system option) for a particular SAS session are not available to other applications.

For example, to define a SAS environment variable that points to the external file C:\MYDATA\TEST.DAT, you can use the following SET option in your SAS configuration file:

-set myvar c:\mydata\test.dat

Then, in your SAS programs, you can use the environment variable MYVAR to refer to the external file:

data mytest;
   infile myvar;
   input name $ score;
run;

It is recommended that you use the SET system option in your SAS configuration file if you invoke SAS using the Windows Start menu.

An alternative to using the SET system option to define an environment variable is to use the Windows SET command. For example, the Windows SET command that equates to the previous example is

SET MYVAR=C:\MYDATA\TEST.BAT

You can also define SET commands by using System Properties dialog box that you access from the Control Panel.

You must issue all the SET commands that define your environment variables before you invoke SAS. If you define an environment variable in an MS-DOS window, and then start SAS from the Start menu, SAS will not recognize the environment variable.

You can assign a fileref to a directory and then access individual files within that directory using member-name syntax (also called aggregate syntax).

For example, if all your regional sales data for January are stored in the directory C:\SAS\MYDATA, you can issue the following FILENAME statement to assign the fileref JAN to this directory:

filename jan "c:\sas\mydata";

Now you can use this fileref with a member name in your SAS programs. In the following example, you reference two files stored in the JAN directory:

data westsale;
   infile jan(west);
   input name $ 1-16 sales 18-25
         comiss 27-34;
run;
data eastsale;
   infile jan(east);
   input name $ 1-16 sales 18-25
         comiss 27-34;
run;

When you use member-name syntax, you do not have to specify the file extension for the file that you are referencing, as long as the file extension is the expected one. For example, in the previous example, the INFILE statement expects a file extension of .DAT. The following table lists the expected file extensions for the various SAS statements and commands:

For example, the following program submits the file C:\PROGRAMS\TESTPGM.SAS to SAS:

filename test "c:\programs";
%include test(testpgm);

SAS searches for a filename TESTPGM.SAS in the directory C:\PROGRAMS.

If your file has a file extension different from the default file extension, you can use the file extension in the filename, as in the following example:

filename test "c:\programs";
%include test(testpgm.xyz);

If your file has no file extension, you must enclose the filename in quotation marks, as in the following example:

filename test "c:\programs";
%include test("testpgm");

To further illustrate the default file extensions SAS uses, here are some more examples using member-name syntax. Assume that the following FILENAME statement has been submitted:

filename test "c:\mysasdir";

The following example opens the file C:\MYSASDIR\PGM1.DAT for output:

file test(pgm1);

The following example opens the file C:\MYSASDIR\PGM1.DAT for input:

infile test(pgm1);

The following example reads and submits the file C:\MYSASDIR\PGM1:

%include test("pgm1");

These examples use SAS statements. SAS commands, such as the FILE and INCLUDE commands, also accept member-name syntax and have the same default file extensions as shown in Default File Extensions for Referencing External Files with Member-Name Syntax.

Another feature of member-name syntax is that it enables you to reference a subdirectory in the working directory without using a fileref. For example, suppose you have a subdirectory named PROGRAMS that is located beneath the working directory. You can use the subdirectory name PROGRAMS when referencing files within this directory. For example, the following statement submits the program stored in working-directory \PROGRAMS\PGM1.SAS:

%include programs(pgm1);

The next example uses the FILE command to save the contents of the active window to working-directory \PROGRAMS\TESTPGM.DAT:

file programs(testpgm);

Member-name syntax is also handy when you use the FILENAME statement to concatenate directories of external files. For example, suppose you issue the following FILENAME statement:

filename progs ("c:\sas\programs",
                "d:\myprogs");

This statement tells SAS that the fileref PROGS refers to all files stored in both the C:\SAS\PROGRAMS and the D:\MYPROGS directories. When you use the fileref PROGS in your SAS program, SAS looks in these directories for the member that you specify. When you use this concatenation feature, you should be aware of the protocol SAS uses, which depends on whether you are accessing the files for read, write, or update. For more information, see Understanding How Concatenated Directories Are Accessed.

SAS resolves an external file reference that uses member-name syntax by using a set of rules. For example, suppose your external file reference in a SAS statement or command is the following:

progs(member1)

SAS uses the following set of rules to resolve this external file reference. This list represents the order of precedence:

The member name must be a valid physical filename. If no extension is given (as in the previous example), SAS uses the appropriate default extension, as given in Default File Extensions for Referencing External Files with Member-Name Syntax. If the extension is given or the member name is quoted, SAS does not assign an extension, and it looks for the filename exactly as it is given.

You can specify concatenations of files when reading external files from within SAS. Concatenated files consist of two or more file specifications (which might contain wildcard characters) separated by blanks or commas. Here are some examples of valid concatenation specifications:

When you use this concatenation feature, you should be aware of the protocol SAS uses, which depends on whether you are accessing the files for read, write, or update. For more information, see Understanding How Concatenated Files Are Accessed.

SAS supports the use of long filenames. (For more information about valid long filenames, see your Windows operating environment documentation.) You can use long filenames whenever you specify a filename as an argument to a dialog box, command, or any aspect of the SAS language.

When specifying external filenames with the SAS language, such as in a statement or function, you should enclose the filename in double quotation marks to reduce ambiguity (since a single quotation mark is a valid character in a long filename). When you need to specify multiple filenames, enclose each filename in double quotation marks and delimit the names with a blank space.

Here are some examples of valid uses of long filenames within SAS:

SAS supports the use of the Universal Naming Convention (UNC) paths. UNC paths let you connect your computer to network devices without having to refer to a network drive letter. SAS supports UNC paths to the extent that Windows and your network software support them. In general, you can refer to a UNC path anywhere in SAS where you would normally refer to a network drive.

UNC paths have the following syntax:

where

For example, the following command includes a file from the network file server ZAPHOD:

include "\\zaphod\universe\galaxy\stars.sas";

If you have assigned several filerefs during a SAS session and need to refresh your memory as to which fileref points where, you can use either the SAS Explorer window or the FILENAME statement to list all the assigned filerefs.

To use the SAS Explorer window to list the active filerefs, double-click on File Shortcuts. The Explorer window lists all the filerefs active for your current SAS session. Any environment variables that you have defined as filerefs are listed, provided you have used them in your SAS session. If you have defined an environment variable as a fileref but have not used it yet in a SAS program, the fileref is not listed in the Explorer window.

You can use the following FILENAME statement to write the active filerefs to the SAS log:

filename _all_ list;

You can clear a fileref by using the following syntax of the FILENAME statement:

If you specify a fileref, only that fileref is cleared. If you specify the keyword _ALL_, all the filerefs that you have assigned during your current SAS session are cleared.

To clear filerefs using the SAS Explorer File Shortcuts:

SAS automatically clears the association between filerefs and their respective files at the end of your job or session. If you want to associate the fileref with a different file during the current session, you do not have to end the session or clear the fileref. SAS automatically reassigns the fileref when you issue a FILENAME statement for the new file.

When you associate a fileref with more than one physical directory, which file is accessed depends on whether it is being accessed for input or output.

If the file is opened for input or update, the first file found that matches the member name is accessed. For example, if you submit the following statements, and the file PHONE.DAT exists in both the C:\SAMPLES and C:\TESTPGMS directories, the one in C:\SAMPLES is read:

filename test ("c:\samples","c:\testpgms");
data sampdat;
   infile test(phone.dat);
   input name $ phonenum $ city $ state $;
run;

When you open a file for output, SAS writes to the file in the first directory listed in the FILENAME statement, even if a file by the same name exists in a later directory. For example, suppose you input the following FILENAME statement:

filename test ("c:\sas","d:\mysasdir");

Then, when you issue the following FILE command, the file SOURCE.PGM is written to the C:\SAS directory, even if a file by the same name exists in the D:\MYSASDIR directory:

file test(source.pgm);

When you associate a fileref with more than one physical file, the behavior of SAS statements and commands depends on whether you are accessing the files for input or output.

If the file is opened for input, data from all files are input. For example, if you issue the following statements, the %INCLUDE statement submits four programs for execution:

filename mydata ("qtr1.sas","qtr2.sas",
                 "qtr3.sas","qtr4.sas");
%include mydata;

If the file is opened for output, data are written to the first file in the concatenation. For example, if you issue the following statements, the PUT statement writes to MYDAT1.DAT:

filename indata "dogdat.dat";
filename outdata ("mydat1.dat","mydat2.dat",
                  "mydat3.dat","mydat4.dat");
data _null_;
   infile indata;
   input name breed color;
   file outdata;
   put name= breed= color=;
run;

Instead of using a fileref to refer to external files, you can use a quoted Windows filename. For example, if the file C:\MYDIR\ORANGES.SAS contains a SAS program that you want to invoke, you can issue the following statement:

%include "c:\mydir\oranges.sas";

When you use a quoted Windows filename in a SAS statement, you can omit the drive and directory specifications if the file that you want to reference is located in the working directory. For example, if in the previous example the working directory is C:\MYDIR, you can submit this statement:

%include "oranges.sas";

You can use several reserved names as quoted physical filenames. Reserved operating system physical names enable you to do a variety of things, such as read data directly from the communications port (such as COM1). The following table lists these physical names and their corresponding device-type keywords:

You can specify operating system physical names with or without a colon. For example, you can specify either COM1: or COM1. For additional information, see your Windows documentation.

The following example demonstrates how to capture data from an external device or application that is transmitting data via a serial (RS-232C port).

options noxwait xsync;
data _null_;
   if symget("sysscpl") = "XP_PRO" then
          rc = system("mode COM1:9600,n,8,1,xon=on");
      stop;
run;

filename commdata commport "COM1:";

data fruit;
      keep num type;
      infile commdata unbuffered;
      file commdata;
      put "ready";
      input totrecs records $;
      if totrecs = . or records ne "RECORDS" then
        do;
          file log;
          put "ERROR: Unable to determine
                number of records to read.";
          stop;
        end;
      do i = 1 to totrecs;
        input num type $;
        output;
        put "NEXT";
      end;
      stop;
run;

Note the use of the device-type keyword COMMPORT in the FILENAME statement in this example. Because the access protocols for devices are slightly different from the access protocols for files, you should always use the appropriate device-type keyword in combination with the reserved physical name in the FILENAME statement. If you do not use a device-type keyword, SAS defaults to using the access protocols for files, not for devices.

For more information about available device-type keywords in the FILENAME statement, see SAS Statements under Windows. Reading Data from the Communications Port discusses the access protocols for using a communications port device.