In a possibly completely warped approach if all of the files are 1) SAS data sets and 2) in the same folder I would be tempted to assign a library name to the folder and then use the SAS dictionary.members table to get the names. ... View more

> Peter, > Thanks for the help.When I use option mixed =yes, > the date filed is giving me a funky output. The > first two values in the data field as shown below > are displayed as > 39980 (instead of 06/16/2009) > 39993 (instead of 06/29/2009) > > The rest of the date values are displayed correctly! Apply a SAS date FORMAT to the variable, MMDDYY10. would work, to display as human useable date. Those values are the SAS day offset values it uses for numerical purposes. Notice that they differ by 13 matching the date difference. ... View more

This looks like multiple sessions of SAS are trying to access the same physical WORK library. If you test your batch job with an interactive session running I would expect this error. ... View more

It appears that the single 0 is being used like a digit selector in a PICTURE statement for some reason. Try assigning another format to the range below 0 such as below. value eq low - <0 = [best8.] 1-high = '0'; run; ... View more

> Why don't you request[pre]class classVar txtVAr1 > ;[/pre]to get the 6 values? If you want only the cross product of the class variables but not the idividual summaries of each class variable use the NWAY option on the proc statement with this approach. ... View more

While Proc Import can be very useful it does have limitations when the data isn't especially well formed. For example many types of ID fields such as product 'numbers' or ZIP codes may have mostly values that may be interpretted as numerics but have exceptions. The program will look at a number of lines to try to guess whether a variable should be string or numeric but may not look far enough if you have large data sets. I would recommend running proc import and then looking at the generated code in the log. If any of the variables are numeric, informat is $xx., then copy the code out of the log into the editor and change the informat, format and input statements for that variable. ... View more

One thought, if the sole purpose of the second data step is to get a count of the unique variables it has one problem in that the data should be sorted by the variable you are trying to count. Another approach would be to use some thing like: Proc freq data=INPUT1 order=freq; table mname /nopercent missing; run; Your example actually didn't have spaces but were null strings (empty). From the online help for input statement, column: Both leading and trailing blanks within the field are ignored. Therefore, if numeric values contain blanks that represent zeros or if you want to retain leading and trailing blanks in character values, read the value with an informat. See INPUT Statement, Formatted. Missing Values Missing data do not require a place-holder. The INPUT statement interprets a blank field as missing and reads other values correctly. If a numeric or character field contains a single period, the variable value is set to missing. ... View more

I forgot to mention that you will need to use a style that has PROTECTSPECIALCHARACTERS=OFF set. I believe this is mentioned in http://support.sas.com/resources/papers/sgf09/318-2009.pdf ... View more

I think the way to generate your TOC from separately created RTF would be by using the RTF field codes using RTF TEXT. Place this code in document portion and position that you want the TOC to appear: ODS RTF TEXT="{\field{\*\fldinst {\\TOC \\f \\h} } }"; to insert the field code for a table of contents that will also create hyperlinks to the entries. Before each item in your report insert at table of contents entry code using: ods rtf text= " {\tc \tcl }"; The is used to create indent levels when desired, use 2 or greater such as \tcl2. If you just have \tc then all items will be at the main level of the TOC. Style elements can be used to modify the text by placing the style descriptors before the {. The text will also appear in the body of your document using the styles but the TOC settings for the document will control appearance in the TOC. After all of the document sections are merged, I assume using Word, then find the TOC field and do the Update all entries to display the table. I found it a bit difficult to find the field as there isn't anything visible but if you place it on an empty page near the start of the document you should be okay. Note that this does not use any of the SAS procedure TOC entry behaviors so individual tables to not get entries, just what you supply. DO NOT use the highlight all and menu item to insert a table of contents as you will create a separate TOC entry. I hope this helps. ... View more

If you haven't solved the problem I'd try: if index(page, "?" ) > 0 then page = substr( page, 1, (index( page, "?")-1)); The extra parantheses should force an evaluation of the number. ... View more

If I am understanding your problem I believe that proc tabulate may have a good shot at what you need. A skeleton program would look like: Proc tabulate data= ; class /missing; /*the missing is important as otherwise the default behavior for tabulate will only count the records that all have nonmissing values for the class variables. */ /*one table per variable*/ table n, ; table n, ; . . . /* or one row with all of the variables across the page */ table n, ; run; Depending on the number of categories for each variable I would look at how many columns you want across a page and select the number of variables per row. It may be that you want to use tabulate with a noprint option to create an output dataset and then proc transpose to arrange data for proc print. ... View more

Taking advantage of SAS boolean behavior of returning 1 for true and 0 for false: ags =( cpt1 = 'D9110' ) + ( cpt2 = 'D9110' ) + ( cpt3 = 'D9110' ) + ( cpt4 = 'D9110') + ( cpt5 = 'D9110'); By extension a total count could be generated by including all of the pairs of variable and value if desired. This is not intuitively obvious and should be documented when using this trick for those that haven't seen it. This sometimes runs a bit faster than lots of explicit if-then statements as well. ... View more

Without examples of the data or the Access import and/or table specifications this is hard to diagnose. With that said, from headaches with Access in the past I would look for delimiters in a text variable. If that is the case your data needs to be cleaned, a different delimiter needs to be used or ensure that string variables are quote qualified. ... View more

> This is highly restricted data and code. No > examples > > I saw a note that this maybe fixed in verison 9.2. > I wonder if it was. I'm on 9.1.3. Probably not. I have code that generated the same error message but the cause was not the data. I was writing a message to the log where I had a condition of interest about variables that were named like var1 through var50. The put statement looked something like this: put "Condition was raised by var"i ; Which would place a note in the log that looked like what I wanted: Condition was raised by var23 instead of the 'Condition was raised by var 23' that would be generated by put "Condition was raised by var" i ; The lack of a space between the var" and the i was the source of the message. In my case the solution was to create a string variable by concatenating "var" and i and modifying the code to look like: put "Condition was raised by " string; Warning message goes away. I would look for any PUT or similar instructions combining fixed text and a variable. All you may need to do is place a space between the end of the quoted string and the variable. ... View more

Having been involved in this situation I would first say it is not a SAS programming problem but a contract problem. Your contract with the vendor should have specified the data layout and they should use that specification. If you don't start early in a project you'll be fighting re-ordered columns, differently named columns and likely different formats for years. (I was the 'vendor' in this case with the client sending us data files that changed as often as twice weekly, but the principal was the same.) ... View more