@sasjs/utils 2.48.2 → 2.49.1

package/macros/mp_getpk.sas

@@ -0,0 +1,281 @@
+/**
+  @file
+  @brief Extract the primary key fields from a table or library
+  @details Examines the constraints to identify primary key fields - indicated
+  by an explicit PK constraint, or a unique index that is also NOT NULL.
+
+  Can be executed at both table and library level.  Supports both BASE engine
+  libraries and SQL Server.
+
+  Usage:
+
+      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 unq unique(tx_from, dd_type),
+        constraint nnn not null(DD_SHORTDESC)
+      );
+      %mp_getpk(work,ds=example)
+
+  Returns:
+
+|libref:$8.|dsn:$32.|memtype:$8.|dbms_memtype:$32.|typemem:$8.|memlabel:$256.|nvar:best.|compress:$8.|pk_fields:$512.|
+|---|---|---|---|---|---|---|---|---|
+|WORK|EXAMPLE|DATA| |DATA| |4|NO|TX_FROM DD_TYPE DD_SOURCE|
+
+
+  @param [in] lib The libref to examine
+  @param [in] ds= (0) Select the dataset to examine, else use 0 for all tables
+  @param [in] mdebug= (0) Set to 1 to preserve temp tables, print var values etc
+  @param [out] outds= (work.mp_getpk) The name of the output table to create.
+
+  <h4> SAS Macros </h4>
+  @li mf_existfeature.sas
+  @li mf_getengine.sas
+  @li mf_getschema.sas
+  @li mp_dropmembers.sas
+  @li mp_getconstraints.sas
+
+  <h4> Related Macros </h4>
+  @li mp_getpk.test.sas
+  @li mp_guesspk.sas
+
+  @version 9.3
+  @author Macro People Ltd
+**/
+
+%macro mp_getpk(
+  lib,
+  ds=0,
+  outds=work.mp_getpk,
+  mdebug=0
+)/*/STORE SOURCE*/;
+
+
+%local engine schema ds1 ds2 ds3 dsn tabs1 tabs2 sum pk4sure pkdefault finalpks
+  pkfromindex;
+
+%let lib=%upcase(&lib);
+%let ds=%upcase(&ds);
+%let engine=%mf_getengine(&lib);
+%let schema=%mf_getschema(&lib);
+
+%let ds1=%mf_getuniquename(prefix=getpk_ds1);
+%let ds2=%mf_getuniquename(prefix=getpk_ds2);
+%let ds3=%mf_getuniquename(prefix=getpk_ds3);
+%let tabs1=%mf_getuniquename(prefix=getpk_tabs1);
+%let tabs2=%mf_getuniquename(prefix=getpk_tabs2);
+%let sum=%mf_getuniquename(prefix=getpk_sum);
+%let pk4sure=%mf_getuniquename(prefix=getpk_pk4sure);
+%let pkdefault=%mf_getuniquename(prefix=getpk_pkdefault);
+%let pkfromindex=%mf_getuniquename(prefix=getpk_pkfromindex);
+%let finalpks=%mf_getuniquename(prefix=getpk_finalpks);
+
+%local dbg;
+%if &mdebug=1 %then %do;
+  %put &sysmacroname entry vars:;
+  %put _local_;
+%end;
+%else %let dbg=*;
+
+proc sql;
+create table &ds1 as
+  select  libname as libref
+    ,upcase(memname) as dsn
+    ,memtype
+    ,upcase(name) as name
+    ,type
+    ,length
+    ,varnum
+    ,label
+    ,format
+    ,idxusage
+    ,notnull
+  from dictionary.columns
+  where upcase(libname)="&lib"
+%if &ds ne 0 %then %do;
+    and upcase(memname)="&ds"
+%end;
+  ;
+
+
+%if &engine=SQLSVR %then %do;
+  proc sql;
+  connect using &lib;
+  create table work.&ds2 as
+  select * from connection to &lib(
+  select
+      s.name as SchemaName,
+      t.name as memname,
+      tc.name as name,
+      ic.key_ordinal as KeyOrderNr
+  from
+      sys.schemas s
+      inner join sys.tables t   on s.schema_id=t.schema_id
+      inner join sys.indexes i  on t.object_id=i.object_id
+      inner join sys.index_columns ic on i.object_id=ic.object_id
+                                    and i.index_id=ic.index_id
+      inner join sys.columns tc on ic.object_id=tc.object_id
+                              and ic.column_id=tc.column_id
+  where i.is_primary_key=1
+    and s.name=%str(%')&schema%str(%')
+  order by t.name, ic.key_ordinal ;
+  );disconnect from &lib;
+  create table &ds3 as
+    select a.*
+      ,case when b.name is not null then 1 else 0 end as pk_ind
+    from work.&ds1 a
+    left join work.&ds2 b
+    on a.dsn=b.memname
+      and upcase(a.name)=upcase(b.name)
+    order by libref,dsn;
+%end;
+%else %do;
+
+  %if &ds = 0 %then %let dsn=;
+
+  /* get all constraints, in constraint order*/
+  %mp_getconstraints(lib=&lib,ds=&dsn,outds=work.&ds2)
+
+  /* extract cols that are clearly primary keys */
+  proc sql;
+  create table &pk4sure as
+    select libref
+      ,table_name
+      ,constraint_name
+      ,constraint_order
+      ,column_name as name
+    from work.&ds2
+    where constraint_type='PRIMARY'
+    order by 1,2,3,4;
+
+  /* extract unique constraints where every col is also NOT NULL */
+  proc sql;
+  create table &sum as
+    select a.libref
+      ,a.table_name
+      ,a.constraint_name
+      ,count(a.column_name) as unq_cnt
+      ,count(b.column_name) as nul_cnt
+    from work.&ds2(where=(constraint_type ='UNIQUE')) a
+    left join work.&ds2(where=(constraint_type ='NOT NULL')) b
+    on a.libref=b.libref
+      and a.table_name=b.table_name
+      and a.column_name=b.column_name
+    group by 1,2,3
+    having unq_cnt=nul_cnt;
+
+  /* extract cols from the relevant unique constraints */
+  create table &pkdefault as
+    select a.libref
+      ,a.table_name
+      ,a.constraint_name
+      ,b.constraint_order
+      ,b.column_name as name
+    from &sum a
+    left join &ds2(where=(constraint_type ='UNIQUE')) b
+    on a.libref=b.libref
+      and a.table_name=b.table_name
+      and a.constraint_name=b.constraint_name
+    order by 1,2,3,4;
+
+  /* extract cols from the relevant unique INDEXES */
+  create table &pkfromindex as
+    select libname as libref
+      ,memname as table_name
+      ,indxname as constraint_name
+      ,indxpos as constraint_order
+      ,name
+    from dictionary.indexes
+    where nomiss='yes' and unique='yes' and upcase(libname)="&lib"
+  %if &ds ne 0 %then %do;
+      and upcase(memname)="&ds"
+  %end;
+    order by 1,2,3,4;
+
+  /* create one table */
+  data &finalpks;
+    set &pkdefault &pk4sure &pkfromindex;
+    pk_ind=1;
+    /* if there are multiple unique constraints, take the first */
+    by libref table_name constraint_name;
+    retain keepme;
+    if first.table_name then keepme=1;
+    if first.constraint_name and not first.table_name then keepme=0;
+    if keepme=1;
+  run;
+
+  /* join back to starting table */
+  proc sql;
+  create table &ds3 as
+    select a.*
+      ,b.constraint_order
+      ,case when b.pk_ind=1 then 1 else 0 end as pk_ind
+    from work.&ds1 a
+    left join work.&finalpks b
+    on a.libref=b.libref
+      and a.dsn=b.table_name
+      and upcase(a.name)=upcase(b.name)
+    order by libref,dsn,constraint_order;
+%end;
+
+
+/* prepare tables */
+proc sql;
+create table work.&tabs1 as select
+  libname as libref
+  ,upcase(memname) as dsn
+  ,memtype
+%if %mf_existfeature(DBMS_MEMTYPE)=1 %then %do;
+  ,dbms_memtype
+%end;
+%else %do;
+  ,'n/a' as dbms_memtype format=$32.
+%end;
+  ,typemem
+  ,memlabel
+  ,nvar
+  ,compress
+from dictionary.tables
+  where upcase(libname)="&lib"
+%if &ds ne 0 %then %do;
+    and upcase(memname)="&ds"
+%end;
+  ;
+data &tabs2;
+  set &ds3;
+  length pk_fields $512;
+  retain pk_fields;
+  by libref dsn constraint_order;
+  if first.dsn then pk_fields='';
+  if pk_ind=1 then pk_fields=catx(' ',pk_fields,name);
+  if last.dsn then output;
+run;
+
+proc sql;
+create table &outds as
+  select a.libref
+    ,a.dsn
+    ,a.memtype
+    ,a.dbms_memtype
+    ,a.typemem
+    ,a.memlabel
+    ,a.nvar
+    ,a.compress
+    ,b.pk_fields
+  from work.&tabs1 a
+  left join work.&tabs2 b
+  on a.libref=b.libref
+    and a.dsn=b.dsn;
+
+/* tidy up */
+%mp_dropmembers(
+  &ds1 &ds2 &ds3 &dsn &tabs1 &tabs2 &sum &pk4sure &pkdefault &finalpks,
+  iftrue=(&mdebug=0)
+)
+
+%mend mp_getpk;

package/macros/mp_gsubfile.sas

@@ -0,0 +1,58 @@
+/**
+  @file
+  @brief Performs a text substitution on a file
+  @details Makes use of the GSUB function in LUA to perform a text substitution
+  in a file - either in-place, or writing to a new location.  The benefit of
+  using LUA is that the entire file can be loaded into a single variable,
+  thereby side stepping the 32767 character limit in a data step.
+
+  Usage:
+
+      %let file=%sysfunc(pathname(work))/file.txt;
+      %let str=replace/me;
+      %let rep=with/this;
+      data _null_;
+        file "&file";
+        put "&str";
+      run;
+      %mp_gsubfile(file=&file, patternvar=str, replacevar=rep)
+      data _null_;
+        infile "&file";
+        input;
+        list;
+      run;
+
+  @param file= (0) The file to perform the substitution on
+  @param patternvar= A macro variable containing the Lua
+    [pattern](https://www.lua.org/pil/20.2.html) to search for.  Due to the use
+    of special (magic) characters in Lua patterns, it is safer to pass the NAME
+    of the macro variable containing the string, rather than the value itself.
+  @param replacevar= The name of the macro variable containing the replacement
+    _string_.
+  @param outfile= (0) The file to write the output to. If zero, then the file
+    is overwritten in-place.
+
+  <h4> SAS Macros </h4>
+  @li ml_gsubfile.sas
+
+  <h4> Related Macros </h4>
+  @li mp_gsubfile.test.sas
+
+  @version 9.4
+  @author Allan Bowe
+**/
+
+%macro mp_gsubfile(file=0,
+  patternvar=,
+  replacevar=,
+  outfile=0
+)/*/STORE SOURCE*/;
+
+  %if "%substr(&sysver.XX,1,4)"="V.04" %then %do;
+    %put %str(ERR)OR: Viya 4 does not support the IO library in lua;
+    %return;
+  %end;
+
+  %ml_gsubfile()
+
+%mend mp_gsubfile;

package/macros/mp_guesspk.sas

@@ -0,0 +1,332 @@
+/**
+  @file
+  @brief Guess the primary key of a table
+  @details Tries to guess the primary key of a table based on the following
+  logic:
+
+      * Columns with nulls are ignored
+      * Return only column combinations that provide unique results
+      * Start from one column, then move out to composite keys of 2 to 6 columns
+
+  The library of the target should be assigned before using this macro.
+
+  Usage:
+
+      filename mc url
+        "https://raw.githubusercontent.com/sasjs/core/main/all.sas";
+      %inc mc;
+      %mp_guesspk(sashelp.class,outds=classpks)
+
+  @param [in] baseds The dataset to analyse
+  @param [out] outds= The output dataset to contain the possible PKs
+  @param [in] max_guesses= (3) The total number of possible primary keys to
+    generate. A table may have multiple (unlikely) PKs, so no need to list them
+    all.
+  @param [in] min_rows= (5) The minimum number of rows a table should have in
+    order to try and guess the PK.
+  @param [in] ignore_cols (0) Space seperated list of columns which you are
+    sure are not part of the primary key (helps to avoid false positives)
+  @param [in] mdebug= Set to 1 to enable DEBUG messages and preserve outputs
+
+  <h4> SAS Macros </h4>
+  @li mf_getvarlist.sas
+  @li mf_getuniquename.sas
+  @li mf_wordsInstr1butnotstr2.sas
+  @li mf_nobs.sas
+
+  <h4> Related Macros </h4>
+  @li mp_getpk.sas
+
+  @version 9.3
+  @author Allan Bowe
+
+**/
+
+%macro mp_guesspk(baseds
+  ,outds=mp_guesspk
+  ,max_guesses=3
+  ,min_rows=5
+  ,ignore_cols=0
+  ,mdebug=0
+)/*/STORE SOURCE*/;
+%local dbg;
+%if &mdebug=1 %then %do;
+  %put &sysmacroname entry vars:;
+  %put _local_;
+%end;
+%else %let dbg=*;
+
+/* declare local vars */
+%local var vars vcnt i j k l tmpvar tmpds rows posspks ppkcnt;
+%let vars=%upcase(%mf_getvarlist(&baseds));
+%let vars=%mf_wordsInStr1ButNotStr2(str1=&vars,str2=%upcase(&ignore_cols));
+%let vcnt=%sysfunc(countw(&vars));
+
+%if &vcnt=0 %then %do;
+  %put &sysmacroname: &baseds has no variables!  Exiting.;
+  %return;
+%end;
+
+/* get null count and row count */
+%let tmpvar=%mf_getuniquename();
+proc sql noprint;
+create table _data_ as select
+  count(*) as &tmpvar
+%do i=1 %to &vcnt;
+  %let var=%scan(&vars,&i);
+  ,sum(case when &var is missing then 1 else 0 end) as &var
+%end;
+  from &baseds;
+
+/* transpose table and scan for not null cols */
+proc transpose;
+data _null_;
+  set &syslast end=last;
+  length vars $32767;
+  retain vars ;
+  if _name_="&tmpvar" then call symputx('rows',col1,'l');
+  else if col1=0 then vars=catx(' ',vars,_name_);
+  if last then call symputx('posspks',vars,'l');
+run;
+
+%let ppkcnt=%sysfunc(countw(&posspks));
+%if &ppkcnt=0 %then %do;
+  %put &sysmacroname: &baseds has no non-missing variables!  Exiting.;
+  %return;
+%end;
+
+proc sort data=&baseds(keep=&posspks) out=_data_ noduprec;
+  by _all_;
+run;
+%local pkds; %let pkds=&syslast;
+
+%if &rows > %mf_nobs(&pkds) %then %do;
+  %put &sysmacroname: &baseds has no combination of unique records! Exiting.;
+  %return;
+%end;
+
+/* now check cardinality */
+proc sql noprint;
+create table _data_ as select
+%do i=1 %to &ppkcnt;
+  %let var=%scan(&posspks,&i);
+  count(distinct &var) as &var
+  %if &i<&ppkcnt %then ,;
+%end;
+  from &pkds;
+
+/* transpose and sort by cardinality */
+proc transpose;
+proc sort; by descending col1;
+run;
+
+/* create initial PK list and re-order posspks list */
+data &outds(keep=pkguesses);
+  length pkguesses $5000 vars $5000;
+  set &syslast end=last;
+  retain vars ;
+  vars=catx(' ',vars,_name_);
+  if col1=&rows then do;
+    pkguesses=_name_;
+    output;
+  end;
+  if last then call symputx('posspks',vars,'l');
+run;
+
+%if %mf_nobs(&outds) ge &max_guesses %then %do;
+  %put &sysmacroname: %mf_nobs(&outds) possible primary key values found;
+  %return;
+%end;
+
+%if &ppkcnt=1 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %return;
+%end;
+
+/* begin scanning for uniques on pairs of PKs */
+%let tmpds=%mf_getuniquename();
+%local lev1 lev2;
+%do i=1 %to &ppkcnt;
+  %let lev1=%scan(&posspks,&i);
+  %do j=2 %to &ppkcnt;
+    %let lev2=%scan(&posspks,&j);
+    %if &lev1 ne &lev2 %then %do;
+      /* check for two level uniqueness */
+      proc sort data=&pkds(keep=&lev1 &lev2) out=&tmpds noduprec;
+        by _all_;
+      run;
+      %if %mf_nobs(&tmpds)=&rows %then %do;
+        proc sql;
+        insert into &outds values("&lev1 &lev2");
+        %if %mf_nobs(&outds) ge &max_guesses %then %do;
+          %put &sysmacroname: Max PKs reached at Level 2 for &baseds;
+          %goto exit;
+        %end;
+      %end;
+    %end;
+  %end;
+%end;
+
+%if &ppkcnt=2 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %goto exit;
+%end;
+
+/* begin scanning for uniques on PK triplets */
+%local lev3;
+%do i=1 %to &ppkcnt;
+  %let lev1=%scan(&posspks,&i);
+  %do j=2 %to &ppkcnt;
+    %let lev2=%scan(&posspks,&j);
+    %if &lev1 ne &lev2 %then %do k=3 %to &ppkcnt;
+      %let lev3=%scan(&posspks,&k);
+      %if &lev1 ne &lev3 and &lev2 ne &lev3 %then %do;
+        /* check for three level uniqueness */
+        proc sort data=&pkds(keep=&lev1 &lev2 &lev3) out=&tmpds noduprec;
+          by _all_;
+        run;
+        %if %mf_nobs(&tmpds)=&rows %then %do;
+          proc sql;
+          insert into &outds values("&lev1 &lev2 &lev3");
+          %if %mf_nobs(&outds) ge &max_guesses %then %do;
+            %put &sysmacroname: Max PKs reached at Level 3 for &baseds;
+            %goto exit;
+          %end;
+        %end;
+      %end;
+    %end;
+  %end;
+%end;
+
+%if &ppkcnt=3 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %goto exit;
+%end;
+
+/* scan for uniques on up to 4 PK fields */
+%local lev4;
+%do i=1 %to &ppkcnt;
+  %let lev1=%scan(&posspks,&i);
+  %do j=2 %to &ppkcnt;
+    %let lev2=%scan(&posspks,&j);
+    %if &lev1 ne &lev2 %then %do k=3 %to &ppkcnt;
+      %let lev3=%scan(&posspks,&k);
+      %if &lev1 ne &lev3 and &lev2 ne &lev3 %then %do l=4 %to &ppkcnt;
+        %let lev4=%scan(&posspks,&l);
+        %if &lev1 ne &lev4 and &lev2 ne &lev4 and &lev3 ne &lev4 %then %do;
+          /* check for four level uniqueness */
+          proc sort data=&pkds(keep=&lev1 &lev2 &lev3 &lev4)
+              out=&tmpds noduprec;
+            by _all_;
+          run;
+          %if %mf_nobs(&tmpds)=&rows %then %do;
+            proc sql;
+            insert into &outds values("&lev1 &lev2 &lev3 &lev4");
+            %if %mf_nobs(&outds) ge &max_guesses %then %do;
+              %put &sysmacroname: Max PKs reached at Level 4 for &baseds;
+              %goto exit;
+            %end;
+          %end;
+        %end;
+      %end;
+    %end;
+  %end;
+%end;
+
+%if &ppkcnt=4 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %goto exit;
+%end;
+
+/* scan for uniques on up to 4 PK fields */
+%local lev5 m;
+%do i=1 %to &ppkcnt;
+  %let lev1=%scan(&posspks,&i);
+  %do j=2 %to &ppkcnt;
+    %let lev2=%scan(&posspks,&j);
+    %if &lev1 ne &lev2 %then %do k=3 %to &ppkcnt;
+      %let lev3=%scan(&posspks,&k);
+      %if &lev1 ne &lev3 and &lev2 ne &lev3 %then %do l=4 %to &ppkcnt;
+        %let lev4=%scan(&posspks,&l);
+        %if &lev1 ne &lev4 and &lev2 ne &lev4 and &lev3 ne &lev4 %then
+        %do m=5 %to &ppkcnt;
+          %let lev5=%scan(&posspks,&m);
+          %if &lev1 ne &lev5 & &lev2 ne &lev5 & &lev3 ne &lev5 & &lev4 ne &lev5 %then %do;
+            /* check for four level uniqueness */
+            proc sort data=&pkds(keep=&lev1 &lev2 &lev3 &lev4 &lev5)
+                out=&tmpds noduprec;
+              by _all_;
+            run;
+            %if %mf_nobs(&tmpds)=&rows %then %do;
+              proc sql;
+              insert into &outds values("&lev1 &lev2 &lev3 &lev4 &lev5");
+              %if %mf_nobs(&outds) ge &max_guesses %then %do;
+                %put &sysmacroname: Max PKs reached at Level 5 for &baseds;
+                %goto exit;
+              %end;
+            %end;
+          %end;
+        %end;
+      %end;
+    %end;
+  %end;
+%end;
+
+%if &ppkcnt=5 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %goto exit;
+%end;
+
+/* scan for uniques on up to 4 PK fields */
+%local lev6 n;
+%do i=1 %to &ppkcnt;
+  %let lev1=%scan(&posspks,&i);
+  %do j=2 %to &ppkcnt;
+    %let lev2=%scan(&posspks,&j);
+    %if &lev1 ne &lev2 %then %do k=3 %to &ppkcnt;
+      %let lev3=%scan(&posspks,&k);
+      %if &lev1 ne &lev3 and &lev2 ne &lev3 %then %do l=4 %to &ppkcnt;
+        %let lev4=%scan(&posspks,&l);
+        %if &lev1 ne &lev4 and &lev2 ne &lev4 and &lev3 ne &lev4 %then
+        %do m=5 %to &ppkcnt;
+          %let lev5=%scan(&posspks,&m);
+          %if &lev1 ne &lev5 & &lev2 ne &lev5 & &lev3 ne &lev5 & &lev4 ne &lev5
+          %then %do n=6 %to &ppkcnt;
+            %let lev6=%scan(&posspks,&n);
+            %if &lev1 ne &lev6 & &lev2 ne &lev6 & &lev3 ne &lev6
+            & &lev4 ne &lev6 & &lev5 ne &lev6 %then
+            %do;
+              /* check for four level uniqueness */
+              proc sort data=&pkds(keep=&lev1 &lev2 &lev3 &lev4 &lev5 &lev6)
+                out=&tmpds noduprec;
+                by _all_;
+              run;
+              %if %mf_nobs(&tmpds)=&rows %then %do;
+                proc sql;
+                insert into &outds
+                  values("&lev1 &lev2 &lev3 &lev4 &lev5 &lev6");
+                %if %mf_nobs(&outds) ge &max_guesses %then %do;
+                  %put &sysmacroname: Max PKs reached at Level 6 for &baseds;
+                  %goto exit;
+                %end;
+              %end;
+            %end;
+          %end;
+        %end;
+      %end;
+    %end;
+  %end;
+%end;
+
+%if &ppkcnt=6 %then %do;
+  %put &sysmacroname: No more PK guess possible;
+  %goto exit;
+%end;
+
+%exit:
+%if &mdebug=0 %then %do;
+  proc sql;
+  drop table &tmpds;
+%end;
+
+%mend mp_guesspk;