About SASJedi

SASJedi · ‎08-15-2011

And just to round out a complete anwer to the original question, an XML file is just a specialy structured text file, so yes - you could read it in a DATA step using INFILE and INPUT statements. Some PERL regular expressions and the SAS PRX functions would facilitate parsing your data from amidst the XML tags. But that sure would be a lot of extra work! As Cynthia and Art pointed out, the SAS XML Mapper is free, automatically generates the XML map file and even provides sample SAS code for accessing the XML data. With LIBNAME access, reading XML data is just as easy as reading a regular SAS data set. Mark

SASJedi · ‎07-20-2011

>any hints as to code (within a macro) to use to check for the version of SAS being run? Automatic macro variables you might find interesting when writing macro code to run in a multi-version SAS environment include: SYSVER - SAS release version (Examples: 9.2, 9.3) SYSVLONG - SAS release number and maintenance level (Examples: 9.02.02M3P041310, 9.03.01M0P051811) SYSVLONG4 - Everything in SYSVLONG, plus the 4 digit year of the release (Examples: 9.02.02M3P04132010 , 9.03.01M0P05182011)

SASJedi · ‎07-18-2011

Run a PROC PRINT on your data and you will find that the data is just fine. Your SAS Viewer XF column is just not wide enough to show all of the text. You can re-size the column by dragging the right boundary of the X5 column header over to the right, and you'll see the rest of your text.

SASJedi · ‎07-12-2011

The code you originally posted looks like it was designed to deterime the number of records in the text file, not the file size in bytes. You'll have a bit of trouble with this code under windows for a couple of reasons: 1. Your code specifies RECFM=F for the input file. This indicates that the text file has fixed length records, i.e. every line in the text file is exactly the same number of characters long. Windows doesn't produce fixed record files - windows text file are variable records length format (RECFM=V). So unless the file was created on another system (probably a mainframe) you should use RECFM=V instead of RECFM=F. 2. The default input buffer size on Windows is 256 bytes. If any individual record in the ext file is more than 256 bytes long, the record will be truncated in the input buffer. Specifying an input buffer size LARGER than the largest record has no ill effect other than consuming a few bytes more of memory. You can specify the input buffer size (Logical RECord Length) with the LRECL= option. 3. If, during a DATA step, you want to put a value into a macro variable for use in downstream program steps, you need a data step function like CALL SYMPUTX(). Try something like this: %let File=C:\Somefile.ext; filename _in "&file" lrecl=32767; data _null_; infile _in end=last; input; ctr+1; if last then call symputx('ctr',ctr); run; filename _in clear; %PUT NOTE: &file contains &ctr records.;

SASJedi · ‎07-10-2011

There is a limit to the number of values ORACLE will accept on an IN operator (in ORACLE 10g, 1000 is the max). Consider using a join to an in-line view. Something like this might do the trick: proc sql; create table TEST as select s.* from oracle.sample as s, (select unique id from orcl.notice) as list where s.id=list.id ; quit; Another option is to make a temp table out of the results of the subquery, and use that for a join. Otherwise, you might try splitting the list results into segments of 1000 or less and using separate IN operators and logical ORs: "where ID in (1,2,3...,999) OR ID in(1000,1001,1002,...1999)" etc

SASJedi · ‎07-01-2011

You can create an email without immediately sending it - details in the YouTube video SAS Programming Tip: Write an Email in SAS and Delay Sending You'd have to somehow automate the "Save as Draft" in Outlook, though. Regards, SASJedi

SASJedi · ‎06-30-2011

Automatic macro variables are the key. SYSHOSTNAME holds the name of the server running SAS, and SYSSCP and SYSSCPL reveal the operating system under which SAS it executing. Try something like this: data _NULL_; PUTLOG "NOTE: This is execuing on server &SYSHOSTNAME"; PUTLOG "NOTE: SAS is running under &SYSSCPL &SYSSCPL"; run;

SASJedi · ‎06-30-2011

To get some visibility into what's going on, try the SASTRACE option. Using this option puts extra info in your log about what portions (if any) of the query are passed to the DB and what was processed in SAS. Here's a generic program template for you to experiment with: options sastrace=',,,d' sastraceloc=saslog nostsuffix; libname MyOra oracle user=UserID password=MyPW schema=SchemaName; proc sql; select * from MyOra.some_table_name where Date < '26JUN1954'd ; quit;

SASJedi · ‎06-28-2011

/* Create the original dataset for testing */ Data original; input Account:$3. Milestone Period:$2.; datalines; SDI 2011 Q2 run; /* Create a simple dataset with the 5 unique measurement observations */ Data Measures; infile datalines truncover; input Measurement $ 1-25; datalines; Approval Rate Days to Approval Fulfillment Rate Days to Ship Compliance ; run; /* Do the actual work */ /* A cartesian join is perfect for this */ proc sql; create table final as select * from original, measures ; quit;

SASJedi · ‎06-20-2011

I copied the code from my previous post, pasted it into a SAS9.2 editor window running on Windows 7 64bit, then submitted the code. The final output looks like this: xyz 123 02/11/2009 03/11/2009 xyz 123 03/11/2009 04/11/2009 xyz 123 04/11/2009 05/11/2009 xyz 123 05/11/2009 . abc 124 05/11/2009 . def 125 02/11/2009 03/11/2009 def 125 03/11/2009 . To me, this appears to be the output you want. I'm having difficulty understanding why your results are not identical. Did you execute the final sort step? If not, the obs are in reverse order, and the "first"record will be missing date2. Please run a PROC PRINT on your final data set and paste the results here so we can compare and troubleshoot further.

SASJedi · ‎06-18-2011

Why wouldn't the lag function work? For example: data start; input Text:$3. ID:3. Date:mmddyy10.; format date mmddyy10.; datalines; xyz 123 02/11/2009 xyz 123 03/11/2009 xyz 123 04/11/2009 xyz 123 05/11/2009 abc 124 05/11/2009 def 125 02/11/2009 def 125 03/11/2009 ; run; proc sort data=start ; by id descending date; run; data final; set start; by id; date2=lag1(date); if (first.ID and last.ID) or first.id then call missing(date2); format date2 mmddyy10.; run; proc sort data=final; by id date; run;

SASJedi · ‎06-17-2011

If they are both numeric, the difference is only the format used to display the values. Just merge or join by product, and it will work fine. If not, you'll need to specify the data type for the two columns.

SASJedi · ‎06-16-2011

It looks like you want to keep ALL observations from STORE1 along with any data from STORE2 that happens to match. Observations in STORE2 that do not match a row in STORE1 should be discarded. If so, you need an indication to tell you if the STORE1 data set is contributing to the current observation being processed by the MERGE, and the IN= dataset option is your answer. Try something like this: DATA store3; MERGE store1(in=KeepThis) store2; BY product; if KeepThis; RUN;

SASJedi · ‎06-15-2011

To correct the code: 1. Use %NRSTR instead of %STR to mask the actual % characters in your text. %STR does not mask % or & characters. 2. Use an extra % (as an escape character) before literal % characters and also before an unbalanced quote: %let a=%nrstr(store_name like "%%MACY%'S%%" ); Finally, after submitting bad code (especially unbalanced quotes), an interactive SAS system may become unresponsive. To purge the system and restore functionality, first try submitting this code repeatedly: *';*";*);%mend; run cancel; You should eventually see this error message in the SAS log: ERROR: No matching %MACRO statement for this %MEND statement. This is the expected response to the %MEND statement in the above text string, and indicates that SAS has resumed normal operation. If this doesn't work, you'll probably need to restart SAS.

SASJedi · ‎06-02-2011

You will probably need to get this with explicit pass-through SQL. Something like this ought to do it: proc sql; connect to oracle (path=Total.world user=xxxx password=xxxx); select * from connection to oracle (SELECT owner, table_name, num_rows FROM ALL_TABLES WHERE OWNER='USA' ) ; disconnect from oracle; quit;

Online Status	Offline
Date Last Visited	Friday