feat: adding mcf_length to mp_getmaxvarlengths

BREAKING CHANGE: mp_getmaxvarlengths now returns 0 for non-special missings, and will use numeric length (as opposed to cast-to-character length) by default

base/mp_getmaxvarlengths.sas

@@ -1,28 +1,46 @@
 /**
-  @file mp_getmaxvarlengths.sas
+  @file
   @brief Scans a dataset to find the max length of the variable values
   @details
   This macro will scan a base dataset and produce an output dataset with two
   columns:
 
   - NAME    Name of the base dataset column
-  - MAXLEN Maximum length of the data contained therein.
+  - MAXLEN  Maximum length of the data contained therein.
 
-  Character fields may be allocated very large widths (eg 32000) of which the
-  maximum  value is likely to be much narrower.  This macro was designed to
-  enable a HTML table to be appropriately sized however this could be used as
-  part of a data audit to ensure we aren't over-sizing our tables in relation to
-  the data therein.
+  Character fields are often allocated very large widths (eg 32000) of which the
+  maximum  value is likely to be much narrower.  Identifying such cases can be
+  helpful in the following scenarios:
+
+  @li Enabling a HTML table to be appropriately sized (`num2char=YES`)
+  @li Reducing the size of a dataset to save on storage (mp_ds2squeeze.sas)
+  @li Identifying columns containing nothing but missing values (`MAXLEN=0` in
+    the output table)
+
+  If the entire column is made up of (non-special) missing values then a value
+  of 0 is returned.
 
-  Numeric fields are converted using the relevant format to determine the width.
   Usage:
 
       %mp_getmaxvarlengths(sashelp.class,outds=work.myds)
 
-  @param libds Two part dataset (or view) reference.
-  @param outds= The output dataset to create
+  @param [in] libds Two part dataset (or view) reference.
+  @param [in] num2char= (NO) When set to NO, numeric fields are sized according
+    to the number of bytes used (or set to zero in the case of non-special
+    missings). When YES, the numeric field is converted to character (using the
+    format, if available), and that is sized instead, using `lengthn()`.
+  @param [out] outds= The output dataset to create, eg:
+  |NAME:$8.|MAXLEN:best.|
+  |---|---|
+  |`Name `|`7 `|
+  |`Sex `|`1 `|
+  |`Age `|`3 `|
+  |`Height `|`8 `|
+  |`Weight `|`3 `|
 
   <h4> SAS Macros </h4>
+  @li mcf_length.sas
+  @li mf_getuniquename.sas
   @li mf_getvarlist.sas
   @li mf_getvartype.sas
   @li mf_getvarformat.sas
@@ -30,20 +48,32 @@
   @version 9.2
   @author Allan Bowe
 
+  <h4> Related Macros </h4>
+  @li mp_ds2squeeze.sas
+  @li mp_getmaxvarlengths.test.sas
+
 **/
 
 %macro mp_getmaxvarlengths(
-    libds      /* libref.dataset to analyse */
-    ,outds=work.mp_getmaxvarlengths /* name of output dataset to create */
+  libds
+  ,num2char=NO
+  ,outds=work.mp_getmaxvarlengths
 )/*/STORE SOURCE*/;
 
-%local vars x var fmt;
+%local vars prefix x var fmt;
 %let vars=%mf_getvarlist(libds=&libds);
+%let prefix=%substr(%mf_getuniquename(),1,25);
+%let num2char=%upcase(&num2char);
+
+%if &num2char=NO %then %do;
+  /* compile length function for numeric fields */
+  %mcf_length(wrap=YES, insert_cmplib=YES)
+%end;
 
 proc sql;
 create table &outds (rename=(
     %do x=1 %to %sysfunc(countw(&vars,%str( )));
-      ________&x=%scan(&vars,&x)
+      &prefix.&x=%scan(&vars,&x)
     %end;
     ))
   as select
@@ -51,18 +81,21 @@ create table &outds (rename=(
       %let var=%scan(&vars,&x);
       %if &x>1 %then ,;
       %if %mf_getvartype(&libds,&var)=C %then %do;
-        max(length(&var)) as ________&x
+        max(lengthn(&var)) as &prefix.&x
       %end;
-      %else %do;
+      %else %if &num2char=YES %then %do;
         %let fmt=%mf_getvarformat(&libds,&var);
         %put fmt=&fmt;
         %if %str(&fmt)=%str() %then %do;
-          max(length(cats(&var))) as ________&x
+          max(lengthn(cats(&var))) as &prefix.&x
         %end;
         %else %do;
-          max(length(put(&var,&fmt))) as ________&x
+          max(lengthn(put(&var,&fmt))) as &prefix.&x
         %end;
       %end;
+      %else %do;
+        max(mcf_length(&var)) as &prefix.&x
+      %end;
     %end;
   from &libds;
 

base/mp_init.sas

@@ -33,37 +33,39 @@
 %macro mp_init(prefix=SASJS
 )/*/STORE SOURCE*/;
 
-  %global
-    &prefix._INIT_NUM   /* initialisation time as numeric                   */
-    &prefix._INIT_DTTM  /* initialisation time in E8601DT26.6 format        */
-    &prefix.WORK        /* avoid typing %sysfunc(pathname(work)) every time */
-  ;
-  %if %eval(&&&prefix._INIT_NUM>0) %then %return;  /* only run once */
+%global
+  SASJS_PREFIX       /* the ONLY hard-coded global macro variable in SASjs    */
+  &prefix._INIT_NUM  /* initialisation time as numeric                        */
+  &prefix._INIT_DTTM /* initialisation time in E8601DT26.6 format             */
+  &prefix.WORK       /* avoid typing %sysfunc(pathname(work)) every time      */
+;
+%if %length(&sasjs_prefix>0) %then %return;  /* only run once */
+%let sasjs_prefix=&prefix;
 
-  data _null_;
-    dttm=datetime();
-    call symputx("&prefix._init_num",dttm,'g');
-    call symputx("&prefix._init_dttm",put(dttm,E8601DT26.6),'g');
-    call symputx("&prefix.work",pathname('WORK'),'g');
-  run;
+data _null_;
+  dttm=datetime();
+  call symputx("&sasjs_prefix._init_num",dttm,'g');
+  call symputx("&sasjs_prefix._init_dttm",put(dttm,E8601DT26.6),'g');
+  call symputx("&sasjs_prefix.work",pathname('WORK'),'g');
+run;
 
-  options
-    noautocorrect           /* disallow misspelled procedure names            */
-    compress=CHAR           /* default is none so ensure we have something!   */
-    datastmtchk=ALLKEYWORDS /* protection from overwriting input datasets     */
-    dsoptions=note2err      /* undocumented - convert bad NOTEs to ERRs       */
-    %str(err)orcheck=STRICT /* catch errs in libname/filename statements      */
-    fmterr                  /* ensure err when a format cannot be found       */
-    mergenoby=%str(ERR)OR   /* throw err when a merge has no BY variables     */
-    missing=.               /* changing this can cause hard to detect errs    */
-    noquotelenmax           /* avoid warnings for long strings                */
-    noreplace               /* avoid overwriting permanent datasets           */
-    ps=max                  /* reduce log size slightly                       */
-    ls=max                  /* reduce log even more and avoid word truncation */
-    validmemname=COMPATIBLE /* avoid special characters etc in table names    */
-    validvarname=V7         /* avoid special characters etc in variable names */
-    varinitchk=%str(ERR)OR  /* avoid data mistakes from variable name typos   */
-    varlenchk=%str(ERR)OR   /* fail hard if truncation (data loss) can result */
-  ;
+options
+  noautocorrect           /* disallow misspelled procedure names            */
+  compress=CHAR           /* default is none so ensure we have something!   */
+  datastmtchk=ALLKEYWORDS /* protection from overwriting input datasets     */
+  dsoptions=note2err      /* undocumented - convert bad NOTEs to ERRs       */
+  %str(err)orcheck=STRICT /* catch errs in libname/filename statements      */
+  fmterr                  /* ensure err when a format cannot be found       */
+  mergenoby=%str(ERR)OR   /* throw err when a merge has no BY variables     */
+  missing=.               /* changing this can cause hard to detect errs    */
+  noquotelenmax           /* avoid warnings for long strings                */
+  noreplace               /* avoid overwriting permanent datasets           */
+  ps=max                  /* reduce log size slightly                       */
+  ls=max                  /* reduce log even more and avoid word truncation */
+  validmemname=COMPATIBLE /* avoid special characters etc in table names    */
+  validvarname=V7         /* avoid special characters etc in variable names */
+  varinitchk=%str(ERR)OR  /* avoid data mistakes from variable name typos   */
+  varlenchk=%str(ERR)OR   /* fail hard if truncation (data loss) can result */
+;
 
 %mend mp_init;