txt files I need to pattern match contents on with regular expressions RegEx

[kjohnsonm]

Hello,

I have a need to loop through a set of text files and pattern match with RegEx and based on positive hits, I want a simple output of what files contain at least one match of my expression. I have tested my expression against a few RegEx websites and it seems to pass my quality tests. What I don't know how to do is use a “RegEx” search on a file in SAS. Can anyone give me some starter pointers? I am an experienced programmer but new this last spring to SAS. I am willing to read but to date am not sure what function to even be looking at, …or combination of functions. I guess I am a little lost on what "proc import" to use, the files may or may not be fixed, tab, stanza, bla, bla… I just want to read them a line at a time and check for a match once found can skip out of the file and mark it as match found.   TIA.  -Keith

[ballardw]

There are a number of SAS functions but loot at PRXMATCH.

You may also want to look at some reading tricks such as using _infil_ instead of trying to read into variables, possibly wildcard filenaming and the Filename option on the Infile statement.

[Reeza]

You may want to look at system commands that you can pass to OS via SAS instead, that may be more efficient. This would be via the X or SYSEXEC command.

[Haikuo]

Here is something may get you started (Windows version), tweak it to accommodate your environment.

filename x pipe 'dir \\yourfolder\*.txt /s /b'; /*This is pipe in all of the dir result into fileref:x*/

data want;

infile x truncover;

input fname $100.;/*this is to get the text file name*/

infile dummy filevar=fname end=last;

do while(not last);

input content $100.; /*this is to get the content of each text file*/

texfilename=fname;

/*here is where to do your RegEx match*/

output;

end;

run;

filename x clear;

[kjohnsonm]

I finally was able to get the pipe syntax to at least run, I am not 100% sure but it seems to open the file up for viewing.  In my case I eventually want to process every *.txt file on my Network file system so this does not work for my needs but is interesting.    I was able to get something like this to work for me to a degree:

filename myFile 'D:\The_Directory_path\That_file.txt';
data want;
      infile myFile truncover scanover;
      input line   $4096.;
run;
proc contents data=want;
run;

In my logs I get data like this:

747  filename myFile 'D:\The_Directory_path\That_file.txt';
748  data want;
749        infile myFile truncover scanover;
750        input line   $4096.;
751  run;

NOTE: The infile MYFILE is:
      Filename=D:\The_Directory_path\That_file.txt,
      RECFM=V,LRECL=32767,File Size (bytes)=431,
      Last Modified=18Dec2015:09:52:12,
      Create Time=17Nov2015:10:13:37

NOTE: 18 records were read from the infile MYFILE.
      The minimum record length was 0.
      The maximum record length was 50.
NOTE: The data set WORK.WANT has 18 observations and 1 variables.
NOTE: DATA statement used (Total process time):
      real time           0.05 seconds
      cpu time            0.03 seconds

752  proc contents data=want;
753  run;

NOTE: PROCEDURE CONTENTS used (Total process time):
      real time           0.01 seconds
      cpu time            0.01 seconds

Does anyone know how to trap Max/Min, and the time date created or modified in macros vars?  I could use the metadata  🙂   I of course already have the file name and path in my program that this eventually fits into...  TIA.  -KJ

[FreelanceReinh]

Excellent advice from both, @ballardw and @Reeza! And a nice, practical example from @Haikuo.

If you go the data step route, you could combine @ballardw's tips with @Haikuo's code and, e.g., use the automatic variable _INFILE_ rather than a newly created variable CONTENT.

Example 10 of the INFILE statement documentation could also be interesting. They read the external file names from a text file, which would be preferable to the pipe approach if the .txt files were located in various different folders.

I was just playing around with the code presented there and verified that it is possible to skip the rest of a (potentially huge) file, as soon as a match was found, and then continue with the next file:

do until(prxmatch('/your RegEx expression/', _infile_));
  input;
end;

The above DO loop would replace the do while(^eof); ... end; there.

[kjohnsonm]

Thanks for the replies; I am not on my game today because I cannot seem to make any of these ideas produce except the use

prxparse. I will try again later.  

I was able to get a simple data set to work with the command

data _null_;
	if _N_=1 then
	do;
	retain PerlExpression;
		pattern="/(?!0)(?!9[0-9][0-9])\d{3}[-.]{1}(?!00)\d{2}/";
		PerlExpression=prxparse(pattern);
/*		put PerlExpression;*/
	end;
	array match[26] $ 105;
	input data_line $100.;
	position=prxmatch(PerlExpression, data_line);
	mA='Matched Position: ';
	mS='String: ';
	mL='Whole line:';
	if position ^= 0 then
	do;
		current_match = substr(data_line, position,5 );
		put mA position mS current_match mL data_line;
	end;
datalines;
123456789012345678901234567890123456789012345678901234567890123 123-45                             0
023456789012345678901234567890123456789012345678901234567890123 123.45                             0
12345678901234567890123456789-01                                                                    123456789012345678901234567 89-01                                                                   
000000000000000000000000000000000000000000000000000000000000003 123 45                             0
14:56.456                                                                                           
45:32                                                                                               
;
run;

[kjohnsonm]

PS

I tried this:

proc print data=dictionary.tables;
run;

This errors because I am not using a Libref.

[Haikuo]

/*
You could try this
I have limited to 10 obs, as sometimes it takes long to run if 
you have many tables
*/

proc print data=sashelp.vtable(obs=10);
run;

/*
The reason yours does not work is that except one occasion, 
SAS library name is limited to 8 characters, while 'dictionary'
is 10. The following is that one occasion
*/
proc sql inobs=10;
	select * from dictionary.tables
	;
quit;

[kjohnsonm]

...and in my example this where limits me to just my data set want.  ...and anything i might have in my work libname.  Thanks for the knowledge.  thats helps a lot!   -KJ

proc sql inobs=10;
    select * from dictionary.tables
    where libname='WORK' or memname='WANT';
    ;
quit;


proc print data=sashelp.vtable;
where libname='WORK' or memname='WANT';
run;