Unix command to get file extension

[alepage]

Hello, 

I have made a listing of all the folders  and all the files from the root (/folder1/...) using two different find command.  But after, reviewing the files listing, I have found that some files have does not have file extension.  

Is there a Unix command that I can use in SAS to get the file extension.

Please provide a sample example.

libname dest1 base "/.../Data_Retention/data";
filename folderls pipe "find /folder1/sasdata/ -type d  ";
data dest1.folderslisting;
length text $2000.;
infile folderls;
input;
text=_infile_;
run;
filename filelist pipe "find /folder1/sasdata/ -type f ";
data dest1.fileslisting;
length text $1000.;
infile filelist;
input;
text=_infile_;
run;

[Kurt_Bremser · Accepted solution]

Do it in SAS:

filename filelist pipe "find /folder1/sasdata/ -type f ";
data dest1.fileslisting;
length text $1000. extension $10;
infile filelist;
input;
text=_infile_;
if countw(scan(text,-1,"/"),".") > 1 then extension = scan(text,-1,".");
run;

On the UNIX side, you would need to apply the cut filter command.

[alepage]

Hello Kurt,

I have used exactly a script similar to yours but I found that some files have very large file extension that I am not famillar with. What's the maximum length for a file extension.

I am scanning files that have been created since 2010, therefore I am getting very large file extension, and some file name does not have a dot. So when I scan for file extension, I am getting the complete file name.  Any idea how to solve that issue.

Thank in advance for your help.

[Kurt_Bremser]

If you don't have a dot, then you don't have an extension by definition.

The max filename length (space reserved in the directory file) is 255 for UNIX/Linux. You can modify my code to have an ELSE branch where the whole string after the last slash is taken.

[Tom]

File extension are just a convention and not a requirement even on Windows.  Certainly since the original DOS filename limitations of 8 characters before the period and 3 after.

And on Unix they have even less meaning/impact.

So you can have filenames with no period in them.  Or filenames with multiple periods in them.

And on Unix if you start the filename with a period it is a "hidden" file.  Hidden in the sense that you have to add an option to the ls command if you want them to appear in the result.

Here is the code I use in SAS to extract the extension from a filename.

if index(filename,'.')>1 then extension=scan(filename,-1,'.');

[Patrick]

The printf action of the find command lets you select what information in what structure you want returned. 

I couldn't test below script but I have used similar code in the past.

With below syntax for each folder or file the find command will return a pipe delimited string with a line feed at the end - which in turn makes it easy to read into SAS.

filename ls pipe "find / -printf %nrstr('%h|%f|%y\n')";
/*filename ls pipe "find / -printf %nrstr('%h|%f|%y\n') 2>/dev/null";*/
data work.listing;
  infile ls lrecl=1024 truncover dlm='|' dsd;
  input path:$900. file_name:$200. file_type:$1.;
run;
filename ls clear

/**
-printf: Possible values for directive %y:
  f: Regular file
  d: Directory
  l: Symbolic link
  p: Named pipe (FIFO)
  c: Character special file
  b: Block special file
  s: Socket
**/

The filename in comment with 2>/dev/null is something you could use for dealing with messages in case of insufficient permissions for listing directory content.

[Sajid01]

Hello @alepage 
Your question is  "Is there a Unix command that I can use in SAS to get the file extension."
Well there is . Assume that the file has an extension then use the following command to get the extension.
Note filename is the variable holding the name of the file.

${filename##*.}

The following example shows the usage

[Tom]

Thanks.  Note that method will have the same trouble as SCAN() does with filenames that do no contain a period.  So you would want to first test that the name actually has a period (and perhaps that it does not start with a period) before using that method to extract the characters after the period.

[yabwon]

You already have a lot of interesting answers but, just for fun, one more.

You could use BasePlus package's %dirsAndFiles() macro.

It's OS independent, provides data in wide or long format, with details on files an directories or not.

EXAMPLE 1. Get list of files and directories:

%dirsAndFiles(C:\SAS_WORK\,ODS=work.result1)

EXAMPLE 2. Get detailed info:

%dirsAndFiles(~/,ODS=work.result2,details=1)

Bart

P.S. To install and use basePlus package do:

• Enable the framework [first time only]:

• filename SPFinit url "https://raw.githubusercontent.com/yabwon/SAS_PACKAGES/main/SPF/SPFinit.sas";
  %include SPFinit; /* enable the framework */

• Install the framework and the package on your machine in the folder you created:

• filename packages "</your/directory/for/packages/>"; 
  %installPackage(SPFinit BasePlus) 

• From now on, in your SAS session just run it like this:

• filename packages "</your/directory/for/packages/>";
  %include packages(SPFinit.sas);
  
  %loadPackage(basePlus)  

Link to details.