New analytical SQL sample scripts

Author: Connor McDonald

plsql/error-management/basic-error-logging-package.sql plsql/error-management/basic-error-logging-package.sql

@@ -0,0 +1 @@
+Run these "analytics"-prefixed scripts standalone, or follow them as part of the complete Analytics class at https://tinyurl.com/devgym-classes

sql/analytical-sql-examples

@@ -0,0 +1,41 @@
+REM   Script: 18c approx_count function
+REM   Data analysis applications heavily use aggregate functions. Approximate query processing (available since Oracle Database 12c Release 1) aims to deliver faster results for these queries. The approximate results are not identical to the exact results but they are very close. New approximate SQL functions for rank, sum and count are now available for Top-N style queries.
+
+By making use of approximate query processing, you can instantly improve the performance of existing analytic workloads and enable faster ad-hoc data exploration
+
+drop table t purge;
+
+create table t as  
+select * from all_objects 
+where owner in ('SYS','SYSTEM','PUBLIC','SCOTT','HR','SALES');
+
+select owner, count(*)  
+from t 
+group by owner 
+order by 1;
+
+-- For the new approx_count functions, you must have a matching approx_rank function as part of the GROUP BY definition
+select owner, approx_count(*)  
+from t 
+group by owner 
+order by 1;
+
+select owner, approx_count(*)  
+from t 
+group by owner 
+having approx_rank(partition by owner order by approx_count(*) desc) <= 1 
+order by 1;
+
+select owner, approx_count(*) , approx_rank(partition by owner order by approx_count(*) desc) 
+from t 
+group by owner 
+having approx_rank(partition by owner order by approx_count(*) desc) <= 1 
+order by 1;
+
+-- You can see the benefit of the restriction when you see that the partition by clause can be a subset of the group by aggregation columns.  Hence in this example, we get the top 8 ranked object types *per owner*
+select owner, object_type, approx_count(*) , approx_rank(partition by owner order by approx_count(*) desc) 
+from t 
+group by owner, object_type 
+having approx_rank(partition by owner order by approx_count(*) desc) <= 8 
+order by 1;
+

sql/analytical-sql-examples/README.md

@@ -0,0 +1,32 @@
+REM   Script: 18c character encoding validation
+REM   New routines in UTL_I18N allow for validation of characters within a particular characterset
+
+REM   Script: 18c character encoding validation 
+
+
+REM   New routines in UTL_I18N allow for validation of characters within a particular characterset 
+
+
+select *  from database_properties  
+where property_name in ('NLS_NCHAR_CHARACTERSET','NLS_CHARACTERSET') ;
+
+drop table charset_test;
+
+create table charset_test(col1 varchar2(20), col2 nvarchar2(20)) ;
+
+insert into charset_test   
+values(  
+  unistr('foo\D800bar'),  
+  unistr('foo\D800bar')  
+) ;
+
+commit
+
+
+select * from charset_test ;
+
+select   
+  utl_i18n.validate_character_encoding(col1) invalid_offset_column1,  
+  utl_i18n.validate_character_encoding(col2) invalid_offset_column2  
+from charset_test ;
+

sql/analytical-sql-examples/analytics-18c-approx_count-function.sql

@@ -0,0 +1,51 @@
+REM   Script: Display columns as rows for any SQL query
+REM   If you have a SQL query that produces output that might wrap on a terminal, then this script will let you run it and print columns as rows rather than columns. 
+
+The 'p_query' parameter contains the query to be run. If you want to run this from SQL*Plus or SQLcl, simply change the contents to "&1" to pass the SQL query as a parameter to this script
+
+-- To use this as a utility script for a dynamically provided SQL query, replace "select * from scott.emp" with "&1", and save the script as (say) "printrows.sql". Then the script can be run from SQLcl, SQL*Plus or SQL Developer with a simple: @printrows "select * from scott.emp"
+declare 
+    p_query varchar2(32767) := q'{select * from scott.emp}'; 
+ 
+    l_theCursor     integer default dbms_sql.open_cursor; 
+    l_columnValue   varchar2(4000); 
+    l_status        integer; 
+    l_descTbl       dbms_sql.desc_tab; 
+    l_colCnt        number; 
+    n number := 0; 
+  procedure p(msg varchar2) is 
+    l varchar2(4000) := msg; 
+  begin 
+    while length(l) > 0 loop 
+      dbms_output.put_line(substr(l,1,80)); 
+      l := substr(l,81); 
+    end loop; 
+  end; 
+begin 
+    execute immediate  'alter session set nls_date_format=''dd-MON-yyyy hh24:mi:ss'' '; 
+ 
+    dbms_sql.parse(  l_theCursor,  p_query, dbms_sql.native ); 
+    dbms_sql.describe_columns( l_theCursor, l_colCnt, l_descTbl ); 
+ 
+    for i in 1 .. l_colCnt loop 
+        dbms_sql.define_column(l_theCursor, i, l_columnValue, 4000); 
+    end loop; 
+ 
+    l_status := dbms_sql.execute(l_theCursor); 
+ 
+    while ( dbms_sql.fetch_rows(l_theCursor) > 0 ) loop 
+        for i in 1 .. l_colCnt loop 
+            dbms_sql.column_value( l_theCursor, i, l_columnValue ); 
+            p( rpad( l_descTbl(i).col_name, 30 ) 
+              || ': ' ||  
+              l_columnValue ); 
+        end loop; 
+        dbms_output.put_line( '-----------------' ); 
+        n := n + 1; 
+    end loop; 
+    if n = 0 then 
+      dbms_output.put_line( chr(10)||'No data found '||chr(10) ); 
+    end if; 
+end; 
+/
+

sql/analytical-sql-examples/analytics-18c-character-encoding-validation.sql

@@ -0,0 +1,38 @@
+REM   Script: Analytics - Dynamically sized windows
+REM   SQL from the KISS (Keep It Simply SQL) Analytic video series by Developer Advocate Connor McDonald. This script is using a PL/SQL function to create dynamic window range.
+
+Run this script standalone, or take it as part of the complete Analytics class at https://tinyurl.com/devgym-classes
+
+select time_id,
+       sum(amount_sold) sold_per_day
+from   sh.sales
+group by time_Id
+order by 1;
+
+create or replace
+function LAST_CLOSE(p_close_date date)
+return number is
+begin
+  return
+    case to_char(p_close_date,'DY')
+     when 'SUN' then 2
+     when 'MON' then 3
+     else 1
+    end;
+end;
+/
+
+select
+  time_id,
+  to_char(time_id,'DY') close_day,
+  sum(sold_per_day) 
+    over ( 
+      order by time_id 
+      range between LAST_CLOSE(time_id) preceding and 0 following) as close_off
+from ( 
+  select time_id,
+         sum(amount_sold) sold_per_day
+  from   sh.sales
+  group by time_Id
+  );
+

sql/analytical-sql-examples/analytics-Deleting-Duplicates.sql

@@ -0,0 +1,61 @@
+REM   Script: Analytics - First look at Window functions
+REM   SQL from the KISS (Keep It Simply SQL) Analytic video series by Developer Advocate Connor McDonald. This script introduces the window clause.
+
+Run this script standalone, or take it as part of the complete Analytics class at https://tinyurl.com/devgym-classes
+
+drop table water purge;
+
+create table water ( name varchar2(30) primary key,
+                     type varchar2(10),
+                     square_km int );
+
+insert into water(name,type,square_km) values ('Pacific Ocean','Ocean',155557000);
+
+insert into water(name,type,square_km) values ('Atlantic Ocean','Ocean',76762000);
+
+insert into water(name,type,square_km) values ('Indian Ocean','Ocean',68556000);
+
+insert into water(name,type,square_km) values ('Southern Ocean','Ocean',20327000);
+
+insert into water(name,type,square_km) values ('Arctic Ocean','Ocean',14056000);
+
+insert into water(name,type,square_km) values ('Mediterranean Sea','Sea',2965800);
+
+insert into water(name,type,square_km) values ('Caribbean Sea','Sea',2718200);
+
+insert into water(name,type,square_km) values ('South China Sea','Sea',2319000);
+
+insert into water(name,type,square_km) values ('Bering Sea','Sea',2291900);
+
+insert into water(name,type,square_km) values ('Gulf of Mexico','Gulf',1592800);
+
+insert into water(name,type,square_km) values ('Okhotsk Sea','Sea',1589700);
+
+insert into water(name,type,square_km) values ('East China Sea','Sea',1249200);
+
+insert into water(name,type,square_km) values ('Hudson Bay','Bay',1232300);
+
+insert into water(name,type,square_km) values ('Japan Sea','Sea',1007800);
+
+insert into water(name,type,square_km) values ('Andaman Sea','Sea',797700);
+
+insert into water(name,type,square_km) values ('North Sea','Sea',575200);
+
+insert into water(name,type,square_km) values ('Red Sea','Sea',438000);
+
+insert into water(name,type,square_km) values ('Baltic Sea','Sea',422200);
+
+commit
+
+
+select * 
+from water
+order by 3 desc;
+
+select name, type, square_km,
+       sum(square_km) over ( order by square_km desc 
+          rows between unbounded preceding and current row
+) as tot
+from water
+order by 3 desc;
+

sql/analytical-sql-examples/analytics-Display-columns-as-rows-for-any-SQL-query.sql

@@ -0,0 +1,79 @@
+REM   Script: Analytics - Grouping ranges of data
+REM   SQL from the KISS (Keep It Simply SQL) Analytic video series by Developer Advocate Connor McDonald. This script looks at the classical problem of identifying groups of data within a set of rows.
+
+Run this script standalone, or take it as part of the complete Analytics class at https://tinyurl.com/devgym-classes
+
+drop table LAB_SAMPLES;
+
+drop sequence LAB_SAMPLES_SEQ;
+
+create sequence LAB_SAMPLES_SEQ;
+
+create table LAB_SAMPLES
+ ( sample_id          int default LAB_SAMPLES_SEQ.NEXTVAL,
+   date_taken  date
+ );
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-01');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-02');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-03');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-04');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-07');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-08');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-09');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-10');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-14');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-15');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-16');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-19');
+
+insert into LAB_SAMPLES ( date_taken) values (date '2015-12-20');
+
+select * from LAB_SAMPLES order by 2;
+
+select
+  date_taken,
+  lag(date_taken) over ( order by date_taken) prev
+from lab_samples
+order by 1;
+
+select
+  date_taken,
+  case
+    when nvl(lag(date_taken) over (order by date_taken),date_taken) != date_taken-1
+       then date_taken end loval
+from lab_samples
+order by 1;
+
+select date_taken, max(loval) over (order by date_taken) loval
+from (
+select date_taken,
+ case
+   when nvl(lag(date_taken) over (order by date_taken),date_taken) != 
+    date_taken-1 then date_taken end loval
+from lab_samples )
+order by 1;
+
+select min(date_taken) range_start, max(date_taken) range_end
+from (
+  select date_taken,max(loval) over (order by date_taken) loval
+  from (
+    select date_taken,
+       case
+         when nvl(lag(date_taken) over (order by date_taken),date_taken) != 
+           date_taken-1 then date_taken end loval
+    from lab_samples))
+group by loval
+order by 1;
+