sasjs/core
1
0
Fork 0
mirror of https://github.com/sasjs/core.git synced 2026-01-16 21:10:05 +00:00
Code Issues Projects Releases Wiki Activity

feat: adding ignore_cols (and mdebug) parameters to mp_guesspk.sas, as well as a code tidy up

Browse Source
This commit is contained in:
munja
2022-01-17 10:45:43 +01:00
parent 887c797e13
commit 724d3b91a0
1 changed files with 235 additions and 213 deletions
Download Patch File Download Diff File Expand all files Collapse all files

200
base/mp_guesspk.sas
View File

@@ -17,16 +17,21 @@
%inc mc; %inc mc;
%mp_guesspk(sashelp.class,outds=classpks) %mp_guesspk(sashelp.class,outds=classpks)
@param baseds The dataset to analyse @param [in] baseds The dataset to analyse
@param outds= The output dataset to contain the possible PKs @param [out] outds= The output dataset to contain the possible PKs
@param max_guesses= (3) The total number of possible primary keys to generate. @param [in] max_guesses= (3) The total number of possible primary keys to
A table may have multiple unlikely PKs, so no need to list them all. generate. A table may have multiple (unlikely) PKs, so no need to list them
@param min_rows= (5) The minimum number of rows a table should have in order all.
to try and guess the PK. @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> <h4> SAS Macros </h4>
@li mf_getvarlist.sas @li mf_getvarlist.sas
@li mf_getuniquename.sas @li mf_getuniquename.sas
@li mf_wordsInstr1butnotstr2.sas
@li mf_nobs.sas @li mf_nobs.sas
<h4> Related Macros </h4> <h4> Related Macros </h4>
@@ -41,73 +46,84 @@
,outds=mp_guesspk ,outds=mp_guesspk
,max_guesses=3 ,max_guesses=3
,min_rows=5 ,min_rows=5
,ignore_cols=0
,mdebug=0
)/*/STORE SOURCE*/; )/*/STORE SOURCE*/;
%local dbg;
%if &mdebug=1 %then %do;
%put &sysmacroname entry vars:;
%put _local_;
%end;
%else %let dbg=*;
/* declare local vars */ /* declare local vars */
%local var vars vcnt i j k l tmpvar tmpds rows posspks ppkcnt; %local var vars vcnt i j k l tmpvar tmpds rows posspks ppkcnt;
%let vars=%mf_getvarlist(&baseds); %let vars=%upcase(%mf_getvarlist(&baseds));
%let vcnt=%sysfunc(countw(&vars)); %put &=vars;
%let vars=%mf_wordsInStr1ButNotStr2(str1=&vars,str2=%upcase(&ignore_cols));
%put &=vars;
%let vcnt=%sysfunc(countw(&vars));
%if &vcnt=0 %then %do; %if &vcnt=0 %then %do;
%put &sysmacroname: &baseds has no variables! Exiting.; %put &sysmacroname: &baseds has no variables! Exiting.;
%return; %return;
%end; %end;
/* get null count and row count */ /* get null count and row count */
%let tmpvar=%mf_getuniquename(); %let tmpvar=%mf_getuniquename();
proc sql noprint; proc sql noprint;
create table _data_ as select create table _data_ as select
count(*) as &tmpvar count(*) as &tmpvar
%do i=1 %to &vcnt; %do i=1 %to &vcnt;
%let var=%scan(&vars,&i); %let var=%scan(&vars,&i);
,sum(case when &var is missing then 1 else 0 end) as &var ,sum(case when &var is missing then 1 else 0 end) as &var
%end; %end;
from &baseds; from &baseds;
/* transpose table and scan for not null cols */ /* transpose table and scan for not null cols */
proc transpose; proc transpose;
data _null_; data _null_;
set &syslast end=last; set &syslast end=last;
length vars $32767; length vars $32767;
retain vars ; retain vars ;
if _name_="&tmpvar" then call symputx('rows',col1,'l'); if _name_="&tmpvar" then call symputx('rows',col1,'l');
else if col1=0 then vars=catx(' ',vars,_name_); else if col1=0 then vars=catx(' ',vars,_name_);
if last then call symputx('posspks',vars,'l'); if last then call symputx('posspks',vars,'l');
run; run;
%let ppkcnt=%sysfunc(countw(&posspks)); %let ppkcnt=%sysfunc(countw(&posspks));
%if &ppkcnt=0 %then %do; %if &ppkcnt=0 %then %do;
%put &sysmacroname: &baseds has no non-missing variables! Exiting.; %put &sysmacroname: &baseds has no non-missing variables! Exiting.;
%return; %return;
%end; %end;
proc sort data=&baseds(keep=&posspks) out=_data_ noduprec; proc sort data=&baseds(keep=&posspks) out=_data_ noduprec;
by _all_; by _all_;
run; run;
%local pkds; %let pkds=&syslast; %local pkds; %let pkds=&syslast;
%if &rows > %mf_nobs(&pkds) %then %do; %if &rows > %mf_nobs(&pkds) %then %do;
%put &sysmacroname: &baseds has no combination of unique records! Exiting.; %put &sysmacroname: &baseds has no combination of unique records! Exiting.;
%return; %return;
%end; %end;
/* now check cardinality */ /* now check cardinality */
proc sql noprint; proc sql noprint;
create table _data_ as select create table _data_ as select
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let var=%scan(&posspks,&i); %let var=%scan(&posspks,&i);
count(distinct &var) as &var count(distinct &var) as &var
%if &i<&ppkcnt %then ,; %if &i<&ppkcnt %then ,;
%end; %end;
from &pkds; from &pkds;
/* transpose and sort by cardinality */ /* transpose and sort by cardinality */
proc transpose; proc transpose;
proc sort; by descending col1; proc sort; by descending col1;
run; run;
/* create initial PK list and re-order posspks list */ /* create initial PK list and re-order posspks list */
data &outds(keep=pkguesses); data &outds(keep=pkguesses);
length pkguesses $5000 vars $5000; length pkguesses $5000 vars $5000;
set &syslast end=last; set &syslast end=last;
retain vars ; retain vars ;
@@ -117,22 +133,22 @@
output; output;
end; end;
if last then call symputx('posspks',vars,'l'); if last then call symputx('posspks',vars,'l');
run; run;
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: %mf_nobs(&outds) possible primary key values found; %put &sysmacroname: %mf_nobs(&outds) possible primary key values found;
%return; %return;
%end; %end;
%if &ppkcnt=1 %then %do; %if &ppkcnt=1 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %return;
%end; %end;
/* begin scanning for uniques on pairs of PKs */ /* begin scanning for uniques on pairs of PKs */
%let tmpds=%mf_getuniquename(); %let tmpds=%mf_getuniquename();
%local lev1 lev2; %local lev1 lev2;
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let lev1=%scan(&posspks,&i); %let lev1=%scan(&posspks,&i);
%do j=2 %to &ppkcnt; %do j=2 %to &ppkcnt;
%let lev2=%scan(&posspks,&j); %let lev2=%scan(&posspks,&j);
@@ -146,21 +162,21 @@
insert into &outds values("&lev1 &lev2"); insert into &outds values("&lev1 &lev2");
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: Max PKs reached at Level 2 for &baseds; %put &sysmacroname: Max PKs reached at Level 2 for &baseds;
%return; %goto exit;
%end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end;
%if &ppkcnt=2 %then %do; %if &ppkcnt=2 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %goto exit;
%end; %end;
/* begin scanning for uniques on PK triplets */ /* begin scanning for uniques on PK triplets */
%local lev3; %local lev3;
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let lev1=%scan(&posspks,&i); %let lev1=%scan(&posspks,&i);
%do j=2 %to &ppkcnt; %do j=2 %to &ppkcnt;
%let lev2=%scan(&posspks,&j); %let lev2=%scan(&posspks,&j);
@@ -176,22 +192,22 @@
insert into &outds values("&lev1 &lev2 &lev3"); insert into &outds values("&lev1 &lev2 &lev3");
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: Max PKs reached at Level 3 for &baseds; %put &sysmacroname: Max PKs reached at Level 3 for &baseds;
%return; %goto exit;
%end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end;
%if &ppkcnt=3 %then %do; %if &ppkcnt=3 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %goto exit;
%end; %end;
/* scan for uniques on up to 4 PK fields */ /* scan for uniques on up to 4 PK fields */
%local lev4; %local lev4;
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let lev1=%scan(&posspks,&i); %let lev1=%scan(&posspks,&i);
%do j=2 %to &ppkcnt; %do j=2 %to &ppkcnt;
%let lev2=%scan(&posspks,&j); %let lev2=%scan(&posspks,&j);
@@ -210,23 +226,23 @@
insert into &outds values("&lev1 &lev2 &lev3 &lev4"); insert into &outds values("&lev1 &lev2 &lev3 &lev4");
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: Max PKs reached at Level 4 for &baseds; %put &sysmacroname: Max PKs reached at Level 4 for &baseds;
%return; %goto exit;
%end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end; %end;
%end;
%if &ppkcnt=4 %then %do; %if &ppkcnt=4 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %goto exit;
%end; %end;
/* scan for uniques on up to 4 PK fields */ /* scan for uniques on up to 4 PK fields */
%local lev5 m; %local lev5 m;
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let lev1=%scan(&posspks,&i); %let lev1=%scan(&posspks,&i);
%do j=2 %to &ppkcnt; %do j=2 %to &ppkcnt;
%let lev2=%scan(&posspks,&j); %let lev2=%scan(&posspks,&j);
@@ -248,8 +264,7 @@
insert into &outds values("&lev1 &lev2 &lev3 &lev4 &lev5"); insert into &outds values("&lev1 &lev2 &lev3 &lev4 &lev5");
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: Max PKs reached at Level 5 for &baseds; %put &sysmacroname: Max PKs reached at Level 5 for &baseds;
%return; %goto exit;
%end;
%end; %end;
%end; %end;
%end; %end;
@@ -257,15 +272,16 @@
%end; %end;
%end; %end;
%end; %end;
%end;
%if &ppkcnt=5 %then %do; %if &ppkcnt=5 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %goto exit;
%end; %end;
/* scan for uniques on up to 4 PK fields */ /* scan for uniques on up to 4 PK fields */
%local lev6 n; %local lev6 n;
%do i=1 %to &ppkcnt; %do i=1 %to &ppkcnt;
%let lev1=%scan(&posspks,&i); %let lev1=%scan(&posspks,&i);
%do j=2 %to &ppkcnt; %do j=2 %to &ppkcnt;
%let lev2=%scan(&posspks,&j); %let lev2=%scan(&posspks,&j);
@@ -276,8 +292,8 @@
%if &lev1 ne &lev4 and &lev2 ne &lev4 and &lev3 ne &lev4 %then %if &lev1 ne &lev4 and &lev2 ne &lev4 and &lev3 ne &lev4 %then
%do m=5 %to &ppkcnt; %do m=5 %to &ppkcnt;
%let lev5=%scan(&posspks,&m); %let lev5=%scan(&posspks,&m);
%if &lev1 ne &lev5 & &lev2 ne &lev5 & &lev3 ne &lev5 & &lev4 ne &lev5 %then %if &lev1 ne &lev5 & &lev2 ne &lev5 & &lev3 ne &lev5 & &lev4 ne &lev5
%do n=6 %to &ppkcnt; %then %do n=6 %to &ppkcnt;
%let lev6=%scan(&posspks,&n); %let lev6=%scan(&posspks,&n);
%if &lev1 ne &lev6 & &lev2 ne &lev6 & &lev3 ne &lev6 %if &lev1 ne &lev6 & &lev2 ne &lev6 & &lev3 ne &lev6
& &lev4 ne &lev6 & &lev5 ne &lev6 %then & &lev4 ne &lev6 & &lev5 ne &lev6 %then
@@ -293,8 +309,7 @@
values("&lev1 &lev2 &lev3 &lev4 &lev5 &lev6"); values("&lev1 &lev2 &lev3 &lev4 &lev5 &lev6");
%if %mf_nobs(&outds) ge &max_guesses %then %do; %if %mf_nobs(&outds) ge &max_guesses %then %do;
%put &sysmacroname: Max PKs reached at Level 6 for &baseds; %put &sysmacroname: Max PKs reached at Level 6 for &baseds;
%return; %goto exit;
%end;
%end; %end;
%end; %end;
%end; %end;
@@ -303,10 +318,17 @@
%end; %end;
%end; %end;
%end; %end;
%end;
%if &ppkcnt=6 %then %do; %if &ppkcnt=6 %then %do;
%put &sysmacroname: No more PK guess possible; %put &sysmacroname: No more PK guess possible;
%return; %goto exit;
%end; %end;
%exit:
%if &mdebug=0 %then %do;
proc sql;
drop table &tmpds;
%end;
%mend mp_guesspk; %mend mp_guesspk;