Can someone please explain how to map the information from a cURL into something usable in PROC HTTP?   I went to a website (sorry for being vague) and manually selected values and used the Chrome developer tools to capture the cURL.    All the cURL values have this general form-- a url followed by several header paramters, followed by some parmeters in a --data list, followed by --compressed:     curl 'url' [http://www.website.net] [header parameters e.g. -H 'Connection: keep-alive' -H 'Cache-Control: max-age=0' etc etc -- data 'PARAM1=value1&PARAM2=value2&PARAM3=&PARAM4=value4' --compressed   I've played around a bit with the cURL in a Mac OS terminal window, so I know that it's valid and returns what I'm looking for.   I can't find anything that lays out how to use this information in PROC HTTP, though. Most examples online either use the "GET" method, or are referencing JSON, or are using curl in an X command in SAS. The documentation shows a very bare bones POST example, as well as examples using a proxy server, but nothing has really addressed what I'm looking for.   I tried using FILE and PUT like this:   data _null_; file test ; if _n_ eq 1 then do ; put 'param1=value1' @; put 'param2=value2' @ ; put 'param3=value3' @ ; /* etc */ end ; run ; and then:   proc http url="https://www.website.net/default.aspx" in='test' out=siteout ; run; But that just gives me back the URL referenced in PROC HTTP, and not the site that would be created by submitting the parameters specified (which I got from the cURL).    Maybe I'm missing something really fundamental and basic, but I've read everything I could find, and nothing seems to relate to this directly. Also, lots of the examples leap quickly into being inside a macro loop or something, so it's hard to figure out the basics in the midst of the more advanced coding context of macros. Don't get me wrong, I know macros, but I'm just trying to work out the basics of PROC HTTP before I turn to Python for the scraping.      Thanks   Jed       ... View more

I'm a little confused about how FOOTNOTES behave in PROC SQL.   I had sort of assumed that FOOTNOTES behaved similar to TITLES. They seem to have similar behavior in terms of persistence, but I can't really make sense of how this code behaves:   proc sql ; title "query 1: unique values of sex" ; select distinct Sex from sashelp.class ; footnote "query 1: &SQLObs rows" ; %put NOTE: There were &SQLObs rows read from sashelp.class. ; title "query 2: unique combinations of sex and age" ; select distinct Sex, Age from sashelp.class ; %put NOTE: There were &SQLObs rows read from sashelp.class. ; footnote2 "query 2: &SQLObs rows" ; quit ; When I run it, i get the following output:   query 1: unique values of sex Sex F M query 2: unique combinations of sex and age Sex Age F 11 F 12 F 13 F 14 F 15 M 11 M 12 M 13 M 14 M 15 M 16 query 1: 2 rows   Note that the footnote appears only at the end of the PROC (which is consistent with the description of the FOOTNOTE statement in the documentation) but the footnote2 doesn't appear.    I looked around but couldn't see anything that explicitly discusses how footnotes behave in PROC SQL when you are executing multiple queries in a single PROC.    I guess my question is, where should I specify footnotes, and how can I expect them to behave? ... View more

I was able to create multi-level summaries in PROC REPORT using the following code.   With the UNDERLINE in the BREAK AFTER, is there a way to have it extend only for some columns, rather than across the entire report?   For example, could it cut across from Origin through Cumpct, but then not interrupt Gtotpct -- Gcumpct?   Also, I figured this out by adapting the code in this excellent paper: http://www2.sas.com/proceedings/sugi27/p120-27.pdf   but it's still not clear to me why I need to reset the cumulative totals when the _BREAK_ value changes even though I am using COMPUTE BEFORE.    Any tips are appreciated!     proc report data = sashelp.cars nowd out = car_report split = '\' ;      column    origin         /* crosstab var 1 */                make           /* crosstab var 2 */                Horsepower     /* REPORT needs a numeric variable to use N-- this was sort of arbitrary */                               /* it's essentially count within group */                pct            /* percent within group */                cumfreq        /* cumulative count within group */                cumpct         /* cumulative percent within group */                gtotpct        /* global percent */                gcumfreq       /* global cumulative frequency */                gcumpct ;      /* global cumulative percent */        ** define all the vars ;      ** thanks, PROC REPORT ;      ** ORIGIN and MAKE are the variables we're grouping by ;      define    origin         / group ;      define    make           / group ;      ** Horsepower is really just used for the N statistic. to do that, it must be declared an ANALYSIS column ;      define    Horsepower     / analysis     format = comma12. n 'Subgroup\Frequency' ;      ** the rest of the columns are computed ;        ** to be calculated within the groups ;      ** this is the percent column ;      define    pct            / computed     format = percent9.2 'Subgroup\Percent' ;      ** this is the cumulative frequency column ;      define    cumfreq        / computed     format = comma12.0  'Subgroup\Cumulative\Frequency' width = 10 ;      ** this is the cumulative percent column ;      define    cumpct         / computed     format = percent9.2 'Subgroup\Cumulative\Percent' width = 10 ;        ** these are for global/grand totals ;      ** this column is the percent across all observations/combinations ;      define    gtotpct        / computed     format = percent9.2 'Percent' ;      ** this column is the cumulative frequency across all combinations ;      define    gcumfreq       / computed     format = comma12.   'Cumulative\Frequency' width = 10 ;      ** this column is the cumulative percent across all combinations ;      define    gcumpct        / computed     format = percent9.2 'Cumulative\Percent' width = 10 ;        ** PROC REPORT requires that the columns be listed in order on the COLUMN statement ;      ** it makes sense to keep them in order in the DEFINE block ;      ** for these COMPUTE blocks, I'm ordering them in more of a logical order for processing ;        ** before anything happens, initialize gtotal as the sum of Horsepower.n and gcumtot to 0 ;      compute before ;           gtotal + Horsepower.n ;           gcumtot = 0 ;      endcomp ;        ** before each new origin value, re-initialize total and cumtot ;      compute before origin ;           total + Horsepower.n ;           cumtot = 0 ;      endcomp ;        ** pct is the count (Horsepower.n) over the total, very straight forward ;      compute pct ;           pct = Horsepower.n / total ;      endcomp ;        ** the cumulative frequency is the running total ;      ** reset it at the breaks ;          compute cumfreq ;           cumtot + Horsepower.n  ;           cumfreq = cumtot ;             ** the relationship between these resets and the COMPUTE BEFORE isn't totally clear to me ;           ** but it doesn't work otherwise ::JJT::08-24-18:: ;           if _break_ = "Origin" THEN cumfreq = . ;           if _break_ = "Origin" THEN total = 0 ;     endcomp ;        ** compute the percent of all combination (uses gtotal) ;      compute gtotpct ;           gtotpct = Horsepower.n / gtotal ;      endcomp ;        ** create running cumulative percent ;      compute cumpct ;           cumpct = cumfreq / total ;      endcomp ;        ** create global cumulative frequency ;      ** it's similar to the "local" counts but... ;      ** in order to get this to work, i added the IF condition ;      ** it makes more sense if you look at the output report data, but some totals are computed BEFORE ;      ** and obviously I didn't want that to happen because it was inflating my counts ;      compute gcumfreq ;           if not(missing(make)) then gcumtot + Horsepower.n ;           gcumfreq = gcumtot ;             ** yeah, again, this is a little bit of magic to me. ;           ** i might post this on SAS Communities and see if Cynthia Zender will explain it to me ;           IF _break_ = "_RBREAK_" THEN gcumfreq = . ;           if _break_ = "_RBREAK_" THEN gtotal = 0 ;      endcomp ;        ** again, this is a relatively straightforward calculation ;      compute gcumpct ;           gcumpct = gcumfreq / gtotal ;      endcomp ;        ** this is required ;      ** you have to choose SOMETHING for proc report to do here ;      ** you can opt to have it SUPPRESS any subgroup summaries ;      ** i kept the underline (UL) because it only shows up in the listing output and makes it a little more readable ;      ** one drawback is the goes across the entire report field, which can probably be fixed but that's icing ;      break after Origin / ul ; run ;       ... View more