I assumed that you could just look for non-whitespace, but of course your SQL expression may look like e.g. "a.col1=dsn.col4&a.col3...", without whitespace. So you may want to use "\w" (word character) instead of "\S" (non whitespace). You could also check for the first character being non-numeric, but is that really necessary, if your input is valid SQL code? ... View more

The points raised by @RW9 will need to be addressed, but as the question is presented the following might help.   /* set up data */ data have; infile datalines truncover; input have $char70.; input replace $char30.; datalines4; Benchmark indices registered recovery from the day's low point. (old="low", new="high") Completes inspection at Factory (old="at", new="in") Patient; status; cardiac arrest (old=";",new=',') That"s ok (old='"',new="'") This 'dataline' uses quotes (old="'",new='"') ;;;; data want; set have; length find_str repl_str $ 8; /* get 'old' string and remove quotes */ find_str = strip(tranwrd(scanq(replace,1,','),'(old=','')); find_str = dequote(find_str); /* get 'new' string and remove quotes */ repl_str = strip(tranwrd(scanq(replace,2,','),'new=','')); repl_str = substr(repl_str,1,length(repl_str) - 1); repl_str = dequote(repl_str); /* perform replace */ want = tranwrd(have,strip(find_str),strip(repl_str)); run; ... View more

@keen_sas wrote: The observation consists of text along with the variable names, that variable while concatenating will be converted into value and forms a complete observation. The variable names are embedded  in square brackets. While concatenating they should be stripped to avoid unwanted spaces in the final observation.The number of variables are not fixed , the text may also contain some special characters as well. How to make this functional. Original is the text and output is what is expected,both are displayed below at the end.   Tried with the below code.,but something is missing the do loop while splitting the text based on delimiter ']'   data a; text="It was found that the customer transacted an average of [AVERAGE] transaction per month in [START_YEAR] and closed account in [END_YR]." ; call symputx('txt',countw(text,"]")); run ;   data b; set a ; array txt{&txt} $200. ; do i = 1 to countw(text,"]"); if i =1 then txt{i}=substr(text,1,find(text,"]")); else txt{i}=substr(text,find(text,"]")+1,find(text,"]")); end; run ;   Original  Output It was found that the customer transacted an average of [AVERAGE] transaction per month in [START_YEAR] and closed account in [END_YR]. ; "It was found that the customer transacted an average of" || strip(AVERAGE) || "transaction per month in" || strip(START_YEAR) || "and closed account in" || strip(END_YR).";   Why are you requiring the use of the || operator? And how is the "output" variable used? I ... View more

To do what you are really asking you would need to create a SAS language parser.  Probably not something that you want try to figure out.   Why not just include the prefix in the source code to begin with?   So your snippets of code will look like: XXX.age > 10 or XXX.gender = 'F' Then your metadata (Excel is NOT a good system to use to store data by the way.  It is a spreadsheet, not a database.) might have something that says for alias XXX you want to use the actual dataset (aka Table) named MyData.  So then your final generated code might look like: select * from MyData XXX where XXX.age > 10 or XXX.gender = 'F' ; No editing of the code snippets is needed at all.   ... View more

There are two possible solutions to this. What do you want if you have a string like 'a==b=c'?   Do you want the result to be '=b=c' (which is not a SCAN solution, as consecutive delimiters are treated as one by SCAN), then the INDEXC solution suggested by @novinosrin is the way to go, if you want the result 'b=c' then take a look at the CALL SCAN routine, as suggested by @r_behata. ... View more

Thanks @Reeza sort was great suggestion. below thing should work now and i have tried with 6 dates with it and it worked data bde(drop = i new:); set abc; newvar1=var1; newvar2=var2; newvar3=var3; newvar4=var4; newvar5=var5; newvar6=var6; array vare(*) var1-var6; array new_vare(*) newvar1-newvar6; call sortn(of new_vare[*]); do i = 1 to 6 until(standard<new_vare{i}); if standard>new_vare{i} then maxval = .; else maxval = new_vare{i}; end; format maxval date9.; run; ... View more

tdef only creates the template. You need a TREPLAY to use the defined template.   You may need to supply a lot more information about the name of the graph file you are replaying from gslide and such.   Or possibly wander into the area of PDF layout and SGgraphics instead of device based graphics. ... View more

REFLINE takes a variable (or value) so ensure that variable has the correct values for each panel and it will automatically be different in each panel.    TEXT will allow you to add custom text at a specific location, similar to an annotation. Again, you'll need the data in your dataset to drive the correct placement and text.    Regarding custom titles - either look at BY Titles customization or another TEXT statement.    http://support.sas.com/documentation/cdl/en/grstatproc/69716/HTML/default/viewer.htm#n1r8egn3tnghz2n1vkokrrtoxkk4.htm ... View more

Run Proc Contents on the resulting data set to see the format. Likely  it will be BEST8. The values are likely therr, barring more than 15 decimals just not displaying. If you really care to see that many decimals change the formats to something you preferr like BEST32. of F24.16 or such. ... View more

Hi: When I test code similar to yours, I get this error because you have spaces in the text string: ERROR: The anchor value "Header one " is not valid for the        "RTF" destination; anchors cannot contain white space,        control characters, or any of the following:        "()<>[]{}/%&\". Anchor value will be ignored. Also, are you going to generate a contents page or not? You don't have TOC_DATA turned on as an option. I don't understand how your code is working at all. You should get an error message on the space for BOOKMARK=. cynthia ... View more

As far as displaying an uppercase delta in proc report will the following provide what you want?: ^{unicode 0394}; ... View more

Hi: DATA step output is controlled by the style template much better when you use FILE PRINT ODS. Regular old-fashioned FILE PRINT is considered "batch" level output and really has no formatting, as ballardw explains. I think that unless you want to get into the world of style templates and FILE PRINT ODS, that you will want to investigate PROC PRINT, PROC REPORT or PROC TABULATE for generating your report.   Depending on the type of code you're writing with FILE PRINT and your PUT statements, you may or may not be able to replicate that for ODS RTF. It really depends on what you are trying to create. And since you didn't show data or output, your best bet for help might be to work with Tech Support on this question, because they can look at ALL your code and ALL your data and you can show them what you want to get and then they can make a recommendation. cynthia ... View more

Looks like data_null has answered your questions about suppressing the automatic by-title, and then creating a custom title using the value of the by-variable. There are a couple of ways to control the size of the text: One is to use the goptions:    goptions htitle=16pt htext=10pt; Another is to hard-code the size on the title statement:   title1 h=16pt "Title One";   title2 h=10pt "Title Two"; ... View more