Merge pull request #125 from sasjs/makedata

`mp_makedata()` improvements

all.sas

@@ -1532,6 +1532,45 @@ Usage:
   %else %do;1%end;
 
 %mend mf_isint;/**
+  @file
+  @brief Checks whether a string follows correct library.dataset format
+  @details Many macros in the core library accept a library.dataset parameter
+  referred to as 'libds'.  This macro validates the structure of that parameter,
+  eg:
+
+    @li 8 character libref?
+    @li 32 character dataset?
+    @li contains a period?
+
+  It does NOT check whether the dataset exists, or if the library is assigned.
+
+  Usage:
+
+      %put %mf_islibds(work.something)=1;
+      %put %mf_islibds(nolib)=0;
+      %put %mf_islibds(badlibref.ds)=0;
+      %put %mf_islibds(w.t.f)=0;
+
+  @param [in] libds The string to be checked
+
+  @return output Returns 1 if libds is valid, 0 if it is not
+
+  <h4> Related Macros </h4>
+  @li mf_islibds.test.sas
+  @li mp_validatecol.sas
+
+  @version 9.2
+**/
+
+%macro mf_islibds(libds
+)/*/STORE SOURCE*/;
+
+%local regex;
+%let regex=%sysfunc(prxparse(%str(/^[_a-z]\w{0,7}\.[_a-z]\w{0,31}$/i)));
+
+%sysfunc(prxmatch(&regex,&libds))
+
+%mend mf_islibds;/**
   @file
   @brief Returns physical location of various SAS items
   @details Returns location of the PlatformObjectFramework tools
@@ -1805,6 +1844,60 @@ Usage:
   %exit_success:
 
 %mend mf_verifymacvars;
+/**
+  @file
+  @brief Returns words that are in both string 1 and string 2
+  @details  Compares two space separated strings and returns the words that are
+  in both.
+  Usage:
+
+      %put %mf_wordsInStr1andStr2(
+        Str1=blah sss blaaah brah bram boo
+        ,Str2=   blah blaaah brah ssss
+      );
+
+  returns:
+  > blah blaaah brah
+
+  @param str1= string containing words to extract
+  @param str2= used to compare with the extract string
+
+  @warning CASE SENSITIVE!
+
+  @version 9.2
+  @author Allan Bowe
+
+**/
+
+%macro mf_wordsInStr1andStr2(
+  Str1= /* string containing words to extract */
+  ,Str2= /* used to compare with the extract string */
+)/*/STORE SOURCE*/;
+
+%local count_base count_extr i i2 extr_word base_word match outvar;
+%if %length(&str1)=0 or %length(&str2)=0 %then %do;
+  %put %str(WARN)ING: empty string provided!;
+  %put base string (str1)= &str1;
+  %put compare string (str2) = &str2;
+  %return;
+%end;
+%let count_base=%sysfunc(countw(&Str2));
+%let count_extr=%sysfunc(countw(&Str1));
+
+%do i=1 %to &count_extr;
+  %let extr_word=%scan(&Str1,&i,%str( ));
+  %let match=0;
+  %do i2=1 %to &count_base;
+    %let base_word=%scan(&Str2,&i2,%str( ));
+    %if &extr_word=&base_word %then %let match=1;
+  %end;
+  %if &match=1 %then %let outvar=&outvar &extr_word;
+%end;
+
+  &outvar
+
+%mend mf_wordsInStr1andStr2;
+
 /**
   @file
   @brief Returns words that are in string 1 but not in string 2
@@ -7586,8 +7679,6 @@ lock &libds clear;
     according to the variable types and formats.
 
     TODO:
-      @li Respect PKs
-      @li Respect NOT NULLs
       @li Consider dates, datetimes, times, integers etc
 
   Usage:
@@ -7603,16 +7694,23 @@ lock &libds clear;
       );
       %mp_makedata(work.example)
 
-  @param [in] libds The empty table in which to create data
-  @param [out] obs= (500) The number of records to create.
+  @param [in] libds The empty table (libref.dataset) in which to create data
+  @param [out] obs= (500) The maximum number of records to create.  The table
+    is sorted with nodup on the primary key, so the actual number of records may
+    be lower than this.
 
   <h4> SAS Macros </h4>
   @li mf_getuniquename.sas
   @li mf_getvarlen.sas
+  @li mf_getvarlist.sas
+  @li mf_islibds.sas
   @li mf_nobs.sas
   @li mp_getcols.sas
   @li mp_getpk.sas
 
+  <h4> Related Macros </h4>
+  @li mp_makedata.test.sas
+
   @version 9.2
   @author Allan Bowe
 
@@ -7620,45 +7718,59 @@ lock &libds clear;
 
 %macro mp_makedata(libds
   ,obs=500
+  ,seed=1
 )/*/STORE SOURCE*/;
 
-%local ds1 c1 n1 i col charvars numvars;
+%local ds1 ds2 lib ds pk_fields i col charvars numvars ispk;
 
-%if %mf_nobs(&libds)>0 %then %do;
+%if %mf_islibds(&libds)=0 %then %do;
+  %put &sysmacroname: Invalid libds (&libds) - should be library.dataset format;
+  %return;
+%end;
+%else %if %mf_nobs(&libds)>0 %then %do;
   %put &sysmacroname: &libds has data, it will not be recreated;
   %return;
 %end;
 
-%local ds1 c1 n1;
-%let ds1=%mf_getuniquename(prefix=mp_makedata);
-%let c1=%mf_getuniquename(prefix=mp_makedatacol);
-%let n1=%mf_getuniquename(prefix=mp_makedatacol);
-data &ds1;
+/* set up temporary vars */
+%let ds1=%mf_getuniquename(prefix=mp_makedatads1);
+%let ds2=%mf_getuniquename(prefix=mp_makedatads2);
+%let lib=%scan(&libds,1,.);
+%let ds=%scan(&libds,2,.);
+
+/* grab the primary key vars */
+%mp_getpk(&lib,ds=&ds,outds=&ds1)
+
+proc sql noprint;
+select coalescec(pk_fields,'_all_') into: pk_fields from &ds1;
+
+data &ds2;
   if 0 then set &libds;
   do _n_=1 to &obs;
-    &c1=repeat(uuidgen(),10);
-    &n1=ranuni(1)*5000000;
-    drop &c1 &n1;
     %let charvars=%mf_getvarlist(&libds,typefilter=C);
     %if &charvars ^= %then %do i=1 %to %sysfunc(countw(&charvars));
       %let col=%scan(&charvars,&i);
-      &col=subpad(&c1,1,%mf_getvarlen(&libds,&col));
+      /* create random value based on observation number and colum length */
+      &col=repeat(put(md5(cats(_n_)),$hex32.),%mf_getvarlen(&libds,&col)/32);
     %end;
 
     %let numvars=%mf_getvarlist(&libds,typefilter=N);
     %if &numvars ^= %then %do i=1 %to %sysfunc(countw(&numvars));
       %let col=%scan(&numvars,&i);
-      &col=&n1;
+      &col=_n_;
     %end;
     output;
   end;
 run;
+proc sort data=&ds2 nodupkey;
+  by &pk_fields;
+run;
 
-proc append base=&libds data=&ds1;
+proc append base=&libds data=&ds2;
 run;
 
 proc sql;
-drop table &ds1;
+drop table &ds1, &ds2;
 
 %mend mp_makedata;/**
   @file

base/mf_islibds.sas

@@ -0,0 +1,40 @@
+/**
+  @file
+  @brief Checks whether a string follows correct library.dataset format
+  @details Many macros in the core library accept a library.dataset parameter
+  referred to as 'libds'.  This macro validates the structure of that parameter,
+  eg:
+
+    @li 8 character libref?
+    @li 32 character dataset?
+    @li contains a period?
+
+  It does NOT check whether the dataset exists, or if the library is assigned.
+
+  Usage:
+
+      %put %mf_islibds(work.something)=1;
+      %put %mf_islibds(nolib)=0;
+      %put %mf_islibds(badlibref.ds)=0;
+      %put %mf_islibds(w.t.f)=0;
+
+  @param [in] libds The string to be checked
+
+  @return output Returns 1 if libds is valid, 0 if it is not
+
+  <h4> Related Macros </h4>
+  @li mf_islibds.test.sas
+  @li mp_validatecol.sas
+
+  @version 9.2
+**/
+
+%macro mf_islibds(libds
+)/*/STORE SOURCE*/;
+
+%local regex;
+%let regex=%sysfunc(prxparse(%str(/^[_a-z]\w{0,7}\.[_a-z]\w{0,31}$/i)));
+
+%sysfunc(prxmatch(&regex,&libds))
+
+%mend mf_islibds;

base/mf_wordsinstr1andstr2.sas

@@ -0,0 +1,54 @@
+/**
+  @file
+  @brief Returns words that are in both string 1 and string 2
+  @details  Compares two space separated strings and returns the words that are
+  in both.
+  Usage:
+
+      %put %mf_wordsInStr1andStr2(
+        Str1=blah sss blaaah brah bram boo
+        ,Str2=   blah blaaah brah ssss
+      );
+
+  returns:
+  > blah blaaah brah
+
+  @param str1= string containing words to extract
+  @param str2= used to compare with the extract string
+
+  @warning CASE SENSITIVE!
+
+  @version 9.2
+  @author Allan Bowe
+
+**/
+
+%macro mf_wordsInStr1andStr2(
+  Str1= /* string containing words to extract */
+  ,Str2= /* used to compare with the extract string */
+)/*/STORE SOURCE*/;
+
+%local count_base count_extr i i2 extr_word base_word match outvar;
+%if %length(&str1)=0 or %length(&str2)=0 %then %do;
+  %put %str(WARN)ING: empty string provided!;
+  %put base string (str1)= &str1;
+  %put compare string (str2) = &str2;
+  %return;
+%end;
+%let count_base=%sysfunc(countw(&Str2));
+%let count_extr=%sysfunc(countw(&Str1));
+
+%do i=1 %to &count_extr;
+  %let extr_word=%scan(&Str1,&i,%str( ));
+  %let match=0;
+  %do i2=1 %to &count_base;
+    %let base_word=%scan(&Str2,&i2,%str( ));
+    %if &extr_word=&base_word %then %let match=1;
+  %end;
+  %if &match=1 %then %let outvar=&outvar &extr_word;
+%end;
+
+  &outvar
+
+%mend mf_wordsInStr1andStr2;
+

base/mp_makedata.sas

@@ -10,8 +10,6 @@
     according to the variable types and formats.
 
     TODO:
-      @li Respect PKs
-      @li Respect NOT NULLs
       @li Consider dates, datetimes, times, integers etc
 
   Usage:
@@ -27,16 +25,23 @@
       );
       %mp_makedata(work.example)
 
-  @param [in] libds The empty table in which to create data
-  @param [out] obs= (500) The number of records to create.
+  @param [in] libds The empty table (libref.dataset) in which to create data
+  @param [out] obs= (500) The maximum number of records to create.  The table
+    is sorted with nodup on the primary key, so the actual number of records may
+    be lower than this.
 
   <h4> SAS Macros </h4>
   @li mf_getuniquename.sas
   @li mf_getvarlen.sas
+  @li mf_getvarlist.sas
+  @li mf_islibds.sas
   @li mf_nobs.sas
   @li mp_getcols.sas
   @li mp_getpk.sas
 
+  <h4> Related Macros </h4>
+  @li mp_makedata.test.sas
+
   @version 9.2
   @author Allan Bowe
 
@@ -44,44 +49,58 @@
 
 %macro mp_makedata(libds
   ,obs=500
+  ,seed=1
 )/*/STORE SOURCE*/;
 
-%local ds1 c1 n1 i col charvars numvars;
+%local ds1 ds2 lib ds pk_fields i col charvars numvars ispk;
 
-%if %mf_nobs(&libds)>0 %then %do;
+%if %mf_islibds(&libds)=0 %then %do;
+  %put &sysmacroname: Invalid libds (&libds) - should be library.dataset format;
+  %return;
+%end;
+%else %if %mf_nobs(&libds)>0 %then %do;
   %put &sysmacroname: &libds has data, it will not be recreated;
   %return;
 %end;
 
-%local ds1 c1 n1;
-%let ds1=%mf_getuniquename(prefix=mp_makedata);
-%let c1=%mf_getuniquename(prefix=mp_makedatacol);
-%let n1=%mf_getuniquename(prefix=mp_makedatacol);
-data &ds1;
+/* set up temporary vars */
+%let ds1=%mf_getuniquename(prefix=mp_makedatads1);
+%let ds2=%mf_getuniquename(prefix=mp_makedatads2);
+%let lib=%scan(&libds,1,.);
+%let ds=%scan(&libds,2,.);
+
+/* grab the primary key vars */
+%mp_getpk(&lib,ds=&ds,outds=&ds1)
+
+proc sql noprint;
+select coalescec(pk_fields,'_all_') into: pk_fields from &ds1;
+
+data &ds2;
   if 0 then set &libds;
   do _n_=1 to &obs;
-    &c1=repeat(uuidgen(),10);
-    &n1=ranuni(1)*5000000;
-    drop &c1 &n1;
     %let charvars=%mf_getvarlist(&libds,typefilter=C);
     %if &charvars ^= %then %do i=1 %to %sysfunc(countw(&charvars));
       %let col=%scan(&charvars,&i);
-      &col=subpad(&c1,1,%mf_getvarlen(&libds,&col));
+      /* create random value based on observation number and colum length */
+      &col=repeat(put(md5(cats(_n_)),$hex32.),%mf_getvarlen(&libds,&col)/32);
     %end;
 
     %let numvars=%mf_getvarlist(&libds,typefilter=N);
     %if &numvars ^= %then %do i=1 %to %sysfunc(countw(&numvars));
       %let col=%scan(&numvars,&i);
-      &col=&n1;
+      &col=_n_;
     %end;
     output;
   end;
 run;
+proc sort data=&ds2 nodupkey;
+  by &pk_fields;
+run;
 
-proc append base=&libds data=&ds1;
+proc append base=&libds data=&ds2;
 run;
 
 proc sql;
-drop table &ds1;
+drop table &ds1, &ds2;
 
 %mend mp_makedata;

tests/crossplatform/mf_islibds.test.sas

@@ -0,0 +1,46 @@
+/**
+  @file
+  @brief Testing mf_islibds macro
+
+      %put %mf_islibds(work.something)=1;
+      %put %mf_islibds(nolib)=0;
+      %put %mf_islibds(badlibref.ds)=0;
+      %put %mf_islibds(w.t.f)=0;
+
+  <h4> SAS Macros </h4>
+  @li mf_islibds.sas
+  @li mp_assert.sas
+
+**/
+
+%mp_assert(
+  iftrue=(
+    %mf_islibds(work.something)=1
+  ),
+  desc=%str(Checking mf_islibds(work.something)=1),
+  outds=work.test_results
+)
+
+%mp_assert(
+  iftrue=(
+    %mf_islibds(nolib)=0
+  ),
+  desc=%str(Checking mf_islibds(nolib)=0),
+  outds=work.test_results
+)
+
+%mp_assert(
+  iftrue=(
+    %mf_islibds(badlibref.ds)=0
+  ),
+  desc=%str(Checking mf_islibds(badlibref.ds)=0),
+  outds=work.test_results
+)
+
+%mp_assert(
+  iftrue=(
+    %mf_islibds(w.t.f)=0
+  ),
+  desc=%str(Checking mf_islibds(w.t.f)=0),
+  outds=work.test_results
+)

tests/crossplatform/mf_wordsinstr1andstr2.test.sas

@@ -0,0 +1,20 @@
+/**
+  @file
+  @brief Testing mf_wordsinstr1andstr2 macro
+
+  <h4> SAS Macros </h4>
+  @li mf_wordsinstr1andstr2.sas
+  @li mp_assert.sas
+
+**/
+
+%let x=%mf_wordsinstr1andstr2(str1=xx DOLLAR x $CHAR xxx W MONNAME
+  ,str2=DOLLAR $CHAR W MONNAME xxxxxx
+);
+%mp_assert(
+  iftrue=(
+    "&x"="DOLLAR $CHAR W MONNAME"
+  ),
+  desc=Checking basic string,
+  outds=work.test_results
+)

tests/crossplatform/mf_wordsinstr1butnotstr2.test.sas

@@ -0,0 +1,20 @@
+/**
+  @file
+  @brief Testing mf_wordsinstr1butnotstr2 macro
+
+  <h4> SAS Macros </h4>
+  @li mf_wordsinstr1butnotstr2.sas
+  @li mp_assert.sas
+
+**/
+
+%let x=%mf_wordsinstr1butnotstr2(str1=xx DOLLAR x $CHAR xxx W MONNAME
+  ,str2=ff xx x xxx xxxxxx
+);
+%mp_assert(
+  iftrue=(
+    "&x"="DOLLAR $CHAR W MONNAME"
+  ),
+  desc=Checking basic string,
+  outds=work.test_results
+)

tests/crossplatform/mp_makedata.test.sas

@@ -0,0 +1,59 @@
+/**
+  @file
+  @brief Testing mp_makedata.sas macro
+
+  <h4> SAS Macros </h4>
+  @li mf_nobs.sas
+  @li mp_makedata.sas
+  @li mp_assert.sas
+
+**/
+
+/**
+  * Test 1 - Regular makedata call
+  */
+
+proc sql;
+create table work.example(
+  TX_FROM float format=datetime19.,
+  DD_TYPE char(16),
+  DD_SOURCE char(2048),
+  DD_SHORTDESC char(256),
+  constraint pk primary key(tx_from, dd_type,dd_source),
+  constraint nnn not null(DD_SHORTDESC)
+);
+%mp_makedata(work.example,obs=500)
+
+%mp_assert(
+  iftrue=("%mf_nobs(work.example)"="500"),
+  desc=Check that 500 rows were created,
+  outds=work.test_results
+)
+
+data _null_;
+  set work.example;
+  call symputx('lenvar',length(dd_source));
+  stop;
+run;
+%mp_assert(
+  iftrue=("&lenvar"="2048"),
+  desc=Check that entire length of variable is populated,
+  outds=work.test_results
+)
+
+
+proc sql;
+create table work.example2(
+  TX_FROM float format=datetime19.,
+  DD_TYPE char(16),
+  DD_SOURCE char(2048),
+  DD_SHORTDESC char(256),
+  some_num num
+);
+%mp_makedata(work.example2)
+
+%mp_assert(
+  iftrue=(&syscc=0),
+  desc=Ensure tables without keys still generate,
+  outds=work.test_results
+)