From c9d8df0a4814f291eea94f28ad2e402d140c3daf Mon Sep 17 00:00:00 2001
From: munja <allnbwe@gmail.com>
Date: Wed, 15 Dec 2021 10:16:24 +0000
Subject: [PATCH] fix: updates from testing mp_storediffs (impact on some
 related macros)

---
 all.sas                  | 241 ++++++++++++++++++++++++++++++++++++++-
 base/mf_getvarlist.sas   |   2 +-
 base/mp_include.sas      |   1 +
 base/mp_lockanytable.sas |   2 +-
 base/mp_storediffs.sas   |  45 ++++++--
 5 files changed, 279 insertions(+), 12 deletions(-)

diff --git a/all.sas b/all.sas
index b6be7ca..bc64229 100644
--- a/all.sas
+++ b/all.sas
@@ -1128,7 +1128,7 @@ https://github.com/yabwon/SAS_PACKAGES/blob/main/packages/baseplus.md#functionex
     %put &sysmacroname: SYSMSG= %sysfunc(sysmsg());
     %let rc=%sysfunc(close(&dsid));
   %end;
-  &outvar
+  %do;%unquote(&outvar)%end;
 %mend mf_getvarlist;/**
   @file
   @brief Returns the position of a variable in dataset (varnum attribute).
@@ -6359,6 +6359,7 @@ run;
 /* prepare the errds */
 data &errds;
   length msg mac $1000;
+  call missing(msg,mac);
   iftrue='1=0';
 run;
 
@@ -7004,7 +7005,7 @@ run;
     %else %do;
       data _null_;
         putlog 'NOTE-' / 'NOTE-';
-        putlog "NOTE- &sysmacroname: Table &lib..&ds locked at "@
+        putlog "NOTE- &sysmacroname: Table &lib..&ds locked at "@;
         putlog " %sysfunc(datetime(),datetime19.) (iteration &x)"@;
         putlog 'NOTE-' / 'NOTE-';
       run;
@@ -8085,6 +8086,242 @@ run;
 
 
 %mend mp_sortinplace;/**
+  @file
+  @brief Converts deletes/changes/appends into a single audit table.
+  @details When tracking changes to data over time, it can be helpful to have
+    a single base table to track ALL modifications - enabling audit trail,
+    data recovery, and change re-application.  This macro is one of many
+    data management utilities used in [Data Controller for SAS](
+    https:datacontroller.io) - a comprehensive data ingestion solution, which
+    works on any SAS platform (Viya, SAS 9, Foundation) and is free for up to 5
+    users.
+
+    NOTE - this macro does not validate the inputs. It is assumed that the
+    datasets containing the new / changed / deleted rows are CORRECT, contain
+    no additional (or missing columns), and that the originals dataset contains
+    all relevant base records (and no additionals).
+
+    Usage:
+
+        data work.orig work.deleted work.changed work.appended;
+          set sashelp.class;
+          if _n_=1 then do;
+            output work.orig work.deleted;
+          end;
+          else if _n_=2 then do;
+            output work.orig;
+            age=99;
+            output work.changed;
+          end;
+          else do;
+            name='Newbie';
+            output work.appended;
+            stop;
+          end;
+        run;
+
+        %mp_storediffs(sashelp.class,work.orig,NAME
+          ,delds=work.deleted
+          ,modds=work.changed
+          ,appds=work.appended
+          ,outds=work.final
+          ,mdebug=1
+        )
+
+  @param [in] libds Target table against which the changes were applied
+  @param [in] origds Dataset with original (unchanged) records.  Can be empty if
+    only appending.
+  @param [in] key Space seperated list of key variables
+  @param [in] delds= (0) Dataset with deleted records
+  @param [in] appds= (0) Dataset with appended records
+  @param [in] modds= (0) Dataset with modified records
+  @param [out] outds= (work.mp_storediffs) Output table containing stored data.
+    Has the following format:
+
+        proc sql;
+        create table &outds(
+          load_ref char(36) label='unique load reference',
+          processed_dttm num format=E8601DT26.6 label='Processed at timestamp',
+          libref char(8) label='Library Reference (8 chars)',
+          dsn char(32) label='Dataset Name (32 chars)',
+          key_hash char(32) label=
+            'MD5 Hash of primary key values (pipe seperated)',
+          move_type char(1) label='Either (A)ppended, (D)eleted or (M)odified',
+          is_pk num label='Is Primary Key Field? (1/0)',
+          is_diff num label=
+            'Did value change? (1/0/-1).  Always -1 for appends and deletes.',
+          tgtvar_type char(1) label='Either (C)haracter or (N)umeric',
+          tgtvar_nm char(32) label='Target variable name (32 chars)',
+          oldval_num num label='Old (numeric) value',
+          newval_num num label='New (numeric) value',
+          oldval_char char(32767) label='Old (character) value',
+          newval_char char(32767) label='New (character) value',
+          constraint pk_mpe_audit
+            primary key(load_ref,libref,dsn,key_hash,tgtvar_nm)
+        );
+
+    @param [in] processed_dttm= (0) Provide a datetime constant in relation to
+      the actual load time.  If not provided, current timestamp is used.
+    @param [in] mdebug= set to 1 to enable DEBUG messages and preserve outputs
+    @param [out] loadref= (0) Provide a unique key to reference the load,
+      otherwise a UUID will be generated.
+
+  <h4> SAS Macros </h4>
+  @li mf_getquotedstr.sas
+  @li mf_getuniquename.sas
+  @li mf_getvarlist.sas
+
+  @version 9.2
+  @author Allan Bowe
+**/
+/** @cond */
+
+%macro mp_storediffs(libds
+  ,origds
+  ,key
+  ,delds=0
+  ,appds=0
+  ,modds=0
+  ,outds=work.mp_storediffs
+  ,loadref=0
+  ,processed_dttm=0
+  ,mdebug=0
+)/*/STORE SOURCE*/;
+%local dbg;
+%if &mdebug=1 %then %do;
+  %put &sysmacroname entry vars:;
+  %put _local_;
+%end;
+%else %let dbg=*;
+
+/* set up unique and temporary vars */
+%local ds1 ds2 ds3 ds4 hashkey inds_auto inds_keep dslist;
+%let ds1=%upcase(work.%mf_getuniquename(prefix=mpsd_ds1));
+%let ds2=%upcase(work.%mf_getuniquename(prefix=mpsd_ds2));
+%let ds3=%upcase(work.%mf_getuniquename(prefix=mpsd_ds3));
+%let ds4=%upcase(work.%mf_getuniquename(prefix=mpsd_ds4));
+%let hashkey=%upcase(%mf_getuniquename(prefix=mpsd_hashkey));
+%let inds_auto=%upcase(%mf_getuniquename(prefix=mpsd_inds_auto));
+%let inds_keep=%upcase(%mf_getuniquename(prefix=mpsd_inds_keep));
+
+%let dslist=&origds;
+%if &delds ne 0 %then %do;
+  %let delds=%upcase(&delds);
+  %if %scan(&delds,-1,.)=&delds %then %let delds=WORK.&delds;
+  %let dslist=&dslist &delds;
+%end;
+%if &appds ne 0 %then %do;
+  %let appds=%upcase(&appds);
+  %if %scan(&appds,-1,.)=&appds %then %let appds=WORK.&appds;
+  %let dslist=&dslist &appds;
+%end;
+%if &modds ne 0 %then %do;
+  %let modds=%upcase(&modds);
+  %if %scan(&modds,-1,.)=&modds %then %let modds=WORK.&modds;
+  %let dslist=&dslist &modds;
+%end;
+
+%let origds=%upcase(&origds);
+%if %scan(&origds,-1,.)=&origds %then %let origds=WORK.&origds;
+
+%let key=%upcase(&key);
+
+/* hash the key and append all the tables (marking the source) */
+data &ds1;
+  set &dslist indsname=&inds_auto;
+  &hashkey=put(md5(catx('|',%mf_getquotedstr(&key,quote=N))),$hex32.);
+  &inds_keep=&inds_auto;
+proc sort;
+  by &inds_keep &hashkey;
+run;
+
+/* transpose numeric & char vars */
+proc transpose data=&ds1
+    out=&ds2(rename=(&hashkey=key_hash _name_=tgtvar_nm col1=newval_num));
+  by &inds_keep &hashkey;
+  var _numeric_;
+run;
+proc transpose data=&ds1
+    out=&ds3(
+      rename=(&hashkey=key_hash _name_=tgtvar_nm col1=newval_char)
+      where=(tgtvar_nm not in ("&hashkey","&inds_keep"))
+    );
+  by &inds_keep &hashkey;
+  var _character_;
+run;
+data &ds4;
+  length &inds_keep $41 tgtvar_nm $32;
+  set &ds2 &ds3 indsname=&inds_auto;
+
+  tgtvar_nm=upcase(tgtvar_nm);
+  if tgtvar_nm in (%upcase(%mf_getvarlist(&libds,dlm=%str(,),quote=DOUBLE)));
+
+  if &inds_auto="&ds2" then tgtvar_type='N';
+  else if &inds_auto="&ds3" then tgtvar_type='C';
+  else do;
+    putlog "%str(ERR)OR: unidentified vartype input!" &inds_auto;
+    call symputx('syscc',98);
+  end;
+
+  if &inds_keep="&appds" then move_type='A';
+  else if &inds_keep="&delds" then move_type='D';
+  else if &inds_keep="&modds" then move_type='M';
+  else if &inds_keep="&origds" then move_type='O';
+  else do;
+    putlog "%str(ERR)OR: unidentified movetype input!" &inds_keep;
+    call symputx('syscc',99);
+  end;
+  tgtvar_nm=upcase(tgtvar_nm);
+  if tgtvar_nm in (%mf_getquotedstr(&key)) then is_pk=1;
+  else is_pk=0;
+  drop &inds_keep;
+run;
+
+%if "&loadref"="0" %then %let loadref=%sysfunc(uuidgen());
+%if &processed_dttm=0 %then %let processed_dttm=%sysfunc(datetime());
+%let libds=%upcase(&libds);
+
+/* join orig vals for modified & deleted */
+proc sql;
+create table &outds as
+  select "&loadref" as load_ref length=36
+    ,&processed_dttm as processed_dttm format=E8601DT26.6
+    ,"%scan(&libds,1,.)" as libref length=8
+    ,"%scan(&libds,2,.)" as dsn length=32
+    ,b.key_hash length=32
+    ,b.move_type length=1
+    ,b.tgtvar_nm length=32
+    ,b.is_pk
+    ,case when b.move_type ne 'M' then -1
+      when a.newval_num=b.newval_num and a.newval_char=b.newval_char then 0
+      else 1
+      end as is_diff
+    ,b.tgtvar_type length=1
+    ,case when b.move_type='D' then b.newval_num
+      else a.newval_num
+      end as oldval_num
+    ,case when b.move_type='D' then .
+      else b.newval_num
+      end as newval_num
+    ,case when b.move_type='D' then b.newval_char
+      else a.newval_char
+      end as oldval_char length=32767
+    ,case when b.move_type='D' then ''
+      else b.newval_char
+      end as newval_char length=32767
+  from &ds4(where=(move_type='O')) as a
+  full join &ds4(where=(move_type ne 'O')) as b
+  on a.tgtvar_nm=b.tgtvar_nm
+  and a.key_hash=b.key_hash
+  order by move_type, key_hash,is_pk desc, tgtvar_nm;
+
+%if &mdebug=0 %then %do;
+  proc sql;
+  drop table &ds1, &ds2, &ds3, &ds4;
+%end;
+
+%mend mp_storediffs;
+/** @endcond *//**
   @file
   @brief Capture session start / finish times and request details
   @details For details, see
diff --git a/base/mf_getvarlist.sas b/base/mf_getvarlist.sas
index 731042e..b9625da 100755
--- a/base/mf_getvarlist.sas
+++ b/base/mf_getvarlist.sas
@@ -70,5 +70,5 @@
     %put &sysmacroname: SYSMSG= %sysfunc(sysmsg());
     %let rc=%sysfunc(close(&dsid));
   %end;
-  &outvar
+  %do;%unquote(&outvar)%end;
 %mend mf_getvarlist;
\ No newline at end of file
diff --git a/base/mp_include.sas b/base/mp_include.sas
index 678d9e8..4669ac4 100644
--- a/base/mp_include.sas
+++ b/base/mp_include.sas
@@ -88,6 +88,7 @@ run;
 /* prepare the errds */
 data &errds;
   length msg mac $1000;
+  call missing(msg,mac);
   iftrue='1=0';
 run;
 
diff --git a/base/mp_lockanytable.sas b/base/mp_lockanytable.sas
index ea91ca7..d25be87 100644
--- a/base/mp_lockanytable.sas
+++ b/base/mp_lockanytable.sas
@@ -180,7 +180,7 @@ run;
     %else %do;
       data _null_;
         putlog 'NOTE-' / 'NOTE-';
-        putlog "NOTE- &sysmacroname: Table &lib..&ds locked at "@
+        putlog "NOTE- &sysmacroname: Table &lib..&ds locked at "@;
         putlog " %sysfunc(datetime(),datetime19.) (iteration &x)"@;
         putlog 'NOTE-' / 'NOTE-';
       run;
diff --git a/base/mp_storediffs.sas b/base/mp_storediffs.sas
index 1971c82..2920876 100644
--- a/base/mp_storediffs.sas
+++ b/base/mp_storediffs.sas
@@ -9,6 +9,11 @@
     works on any SAS platform (Viya, SAS 9, Foundation) and is free for up to 5
     users.
 
+    NOTE - this macro does not validate the inputs. It is assumed that the
+    datasets containing the new / changed / deleted rows are CORRECT, contain
+    no additional (or missing columns), and that the originals dataset contains
+    all relevant base records (and no additionals).
+
     Usage:
 
         data work.orig work.deleted work.changed work.appended;
@@ -33,6 +38,7 @@
           ,modds=work.changed
           ,appds=work.appended
           ,outds=work.final
+          ,mdebug=1
         )
 
   @param [in] libds Target table against which the changes were applied
@@ -44,10 +50,11 @@
   @param [in] modds= (0) Dataset with modified records
   @param [out] outds= (work.mp_storediffs) Output table containing stored data.
     Has the following format:
+
         proc sql;
         create table &outds(
           load_ref char(36) label='unique load reference',
-          processed_dttm num format=E8601DT26.6, label='Processed at timestamp'
+          processed_dttm num format=E8601DT26.6 label='Processed at timestamp',
           libref char(8) label='Library Reference (8 chars)',
           dsn char(32) label='Dataset Name (32 chars)',
           key_hash char(32) label=
@@ -60,17 +67,22 @@
           tgtvar_nm char(32) label='Target variable name (32 chars)',
           oldval_num num label='Old (numeric) value',
           newval_num num label='New (numeric) value',
-          oldval_char char(32767) label='Old (character) value',,
+          oldval_char char(32767) label='Old (character) value',
           newval_char char(32767) label='New (character) value',
+          constraint pk_mpe_audit
+            primary key(load_ref,libref,dsn,key_hash,tgtvar_nm)
         );
+
     @param [in] processed_dttm= (0) Provide a datetime constant in relation to
       the actual load time.  If not provided, current timestamp is used.
+    @param [in] mdebug= set to 1 to enable DEBUG messages and preserve outputs
     @param [out] loadref= (0) Provide a unique key to reference the load,
       otherwise a UUID will be generated.
 
   <h4> SAS Macros </h4>
   @li mf_getquotedstr.sas
   @li mf_getuniquename.sas
+  @li mf_getvarlist.sas
 
   @version 9.2
   @author Allan Bowe
@@ -86,7 +98,14 @@
   ,outds=work.mp_storediffs
   ,loadref=0
   ,processed_dttm=0
+  ,mdebug=0
 )/*/STORE SOURCE*/;
+%local dbg;
+%if &mdebug=1 %then %do;
+  %put &sysmacroname entry vars:;
+  %put _local_;
+%end;
+%else %let dbg=*;
 
 /* set up unique and temporary vars */
 %local ds1 ds2 ds3 ds4 hashkey inds_auto inds_keep dslist;
@@ -95,7 +114,7 @@
 %let ds3=%upcase(work.%mf_getuniquename(prefix=mpsd_ds3));
 %let ds4=%upcase(work.%mf_getuniquename(prefix=mpsd_ds4));
 %let hashkey=%upcase(%mf_getuniquename(prefix=mpsd_hashkey));
-%let inds_auto=%upcase(%mf_getuniquename(prefix=mpsd_inds_auto);
+%let inds_auto=%upcase(%mf_getuniquename(prefix=mpsd_inds_auto));
 %let inds_keep=%upcase(%mf_getuniquename(prefix=mpsd_inds_keep));
 
 %let dslist=&origds;
@@ -123,7 +142,7 @@
 /* hash the key and append all the tables (marking the source) */
 data &ds1;
   set &dslist indsname=&inds_auto;
-  &hashkey=put(md5(cats(%mf_getquotedstr(&key,quote=N))),$hex32.);
+  &hashkey=put(md5(catx('|',%mf_getquotedstr(&key,quote=N))),$hex32.);
   &inds_keep=&inds_auto;
 proc sort;
   by &inds_keep &hashkey;
@@ -146,6 +165,10 @@ run;
 data &ds4;
   length &inds_keep $41 tgtvar_nm $32;
   set &ds2 &ds3 indsname=&inds_auto;
+
+  tgtvar_nm=upcase(tgtvar_nm);
+  if tgtvar_nm in (%upcase(%mf_getvarlist(&libds,dlm=%str(,),quote=DOUBLE)));
+
   if &inds_auto="&ds2" then tgtvar_type='N';
   else if &inds_auto="&ds3" then tgtvar_type='C';
   else do;
@@ -187,11 +210,15 @@ create table &outds as
       else 1
       end as is_diff
     ,b.tgtvar_type length=1
-    ,a.newval_num as oldval_num
+    ,case when b.move_type='D' then b.newval_num
+      else a.newval_num
+      end as oldval_num
     ,case when b.move_type='D' then .
       else b.newval_num
       end as newval_num
-    ,a.newval_char as oldval_char length=32767
+    ,case when b.move_type='D' then b.newval_char
+      else a.newval_char
+      end as oldval_char length=32767
     ,case when b.move_type='D' then ''
       else b.newval_char
       end as newval_char length=32767
@@ -201,8 +228,10 @@ create table &outds as
   and a.key_hash=b.key_hash
   order by move_type, key_hash,is_pk desc, tgtvar_nm;
 
-%mp_dropmembers(&ds1 &ds2 &ds3 &ds4)
+%if &mdebug=0 %then %do;
+  proc sql;
+  drop table &ds1, &ds2, &ds3, &ds4;
+%end;
 
 %mend mp_storediffs;
-
 /** @endcond */
\ No newline at end of file