Packet allocation

Question

Hello,
Using Oracle 11.2.0.1
Using Windows XP Professional
create table approver
approver_id number(4),
working_place number(2),
hq_id number(2),
packet_no1 number(4),
packet_no2 number(4)   --- if approver got 2 packets due to scarcity of approvers.
create table contents
packet_no number(4),
book_id number(5),
book_region number(2),
approver_id number(4)   -- means this packet has been allotted to this approver.
create table app_rule
app_region number(2),
avoid1 number(2),
avoid2 number(2),
avoid3 number(2),
avoid4 number(2),
avoid5 number(2),
avoid6 number(2),
avoid7 number(2),
avoid8 number(2),
avoid9 number(2)
insert into approver values (34,4,1,0,0);
insert into approver values (35,5,1,0,0);
insert into approver values (36,7,1,0,0);
insert into approver values (37,9,1,0,0);
insert into approver values (38,4,1,0,0);
insert into approver values (39,2,1,0,0);
insert into approver values (40,9,1,0,0);
insert into approver values (41,11,1,0,0);
insert into approver values (42,22,1,0,0);
insert into approver values (43,14,18,0,0);
insert into approver values (44,25,18,0,0);
insert into approver values (45,17,18,0,0);
insert into approver values (46,6,18,0,0);
insert into approver values (47,22,18,0,0);
insert into approver values (48,30,18,0,0);
insert into approver values (49,7,18,0,0);
insert into approver values (50,17,18,0,0);
insert into approver values (51,25,18,0,0);
insert into approver values (52,3,11,0,0);
insert into approver values (53,2,11,0,0);
insert into approver values (54,1,11,0,0);
insert into approver values (55,5,11,0,0);
insert into approver values (56,19,11,0,0);
insert into approver values (57,26,11,0,0);
insert into approver values (58,27,11,0,0);
insert into approver values (59,31,11,0,0);
insert into approver values (60,16,11,0,0);
insert into contents values (1240,5170,5,0);
insert into contents values (1240,5178,5,0);
insert into contents values (1141,1517,1,0);
insert into contents values (1141,1507,1,0);
insert into contents values (1214,4173,4,0);
insert into contents values (1214,4070,4,0);
insert into contents values (1214,7181,7,0);
insert into contents values (1214,7178,7,0);
insert into contents values (1214,8078,8,0);
insert into contents values (1214,6678,6,0);
insert into contents values (1141,19517,19,0);
insert into contents values (1141,19507,19,0);
---If there are more than 9 different book_region in a packet then that packet
---has to manually entertained/dispatch.  After allotment those packets rows should be in a seperate table.
insert into app_rule values (1,1,6,3,10,0,0,0,0,0);
insert into app_rule values (2,2,11,32,4,6,9,19,0,0);
insert into app_rule values (3,3,19,2,14,16,15,0,0,0);
insert into app_rule values (4,4,2,1,0,0,0,0,0,0);
insert into app_rule values (5,5,22,1,18,8,0,0,0,0);
insert into app_rule values (6,6,8,7,5,0,0,0,0,0);
insert into app_rule values (7,7,1,2,0,0,0,0,0,0);
insert into app_rule values (8,8,9,10,11,12,13,14,15,16);
insert into app_rule values (9,9,17,18,19,20,21,0,0,0);
insert into app_rule values (10,10,6,7,8,9,26,0,0,0);
insert into app_rule values (11,11,22,23,24,0,0,0,0,0);
insert into app_rule values (12,12,25,26,27,28,29,30,31,32);
---Upto 33 regions, means person who is belonging to app_region, will not receive any book from avoid columns.
---plus HQ ID regions.Business Rule :
Complexity Level 0 :
-Packet of books sent to approver for their reading and any correction as per rule.
Example :
First Packet No. 1240 which belongs to region 5, because all books are from region 5. First approver No. 34 whose working region is 4 and HQ is 1. Now, we will scan app_rule for app_region=4 which have 4,2,1 regions for avoid and 5 is not in 4,2,1 so can we allot this packet to this approver ? No, we can not not, because approver's HQ (1) is also in avoid list, which should not.
Ok, for 35, no it can not, more over it should not because if it happened, then it will be a blunder because approver's working region id is 5 and books are from region 5, so it should not.
for 36, 7... 7th avoid are 7,1,2.. ok, but 1 is there i.e. approver's HQ.... so it can not.
for 37 9... 9th avoid are 9,17,18,19,20,21. Yes, it is perfect, because neither 5 is in the list nor 1 (Approver's HQ).
Complexity Level 1 :
What about packet no. 1214, the complicated packet because it contains books from 4 different regions i.e. 4,7,8,6; means we will look those approver who is not working in :
for 4 : 4,2,1
for 7 : 7,1,2
for 8 : 8,9,10,11,12,13,14,15,16
for 6 : 6,8,7,5
unique : 1,2,4,5,6,7,8,9,10,11,12,13,14,15,16 regions + their HQ ids. Who is from above... ie. first approver id 44 because neither 25 is in this list nor 18.
Complexity Level 2 :
Suppose there are 10 packets and we have only say 7 approvers who have sent their consent for this year. So, allotment should be that much smart and flexible who can interchange the packet internally for best fitting the allotment and for rest 3 packets it should double-allot the packet following above rule. Something like suppose packet no. 1000 has not been allotted due to less number of approvers and if packet no. 2000 has been allotted to 100th number approver, but this 2000 packet can be allotted to 1002 too, so allotment should be smart enough like packet number 1000 to 100 and packet no. 2000 to 1002 (internally interchange the packet i.e. best fitting of allotment).
Please tell me I am clear in my question or I have to provide more clearity on business rule or I have to provide more inserts for the best checking.
Yes, you can ask me, show us your effort in this regard, this is your job/work, why WE should do your task ?
My reply will be : Basically, till now I remained busy in the admin part or you can say that I have not got more PL/SQL exposure (zero in PL/SQL) and I don't think that I can do this task without your esteem help in this forum, truly speaking I can not get the work done at my own alone. I am really very sorry, If am "giving" you my task, but its not really giving,
I am just looking how this task will be done, if you please help me. We are at the designing stage, so if you can suggests any major changes in the design or there is flaw in the design. Rather than asking the question's part, I thought that it would be better to write the question as a whole.
Regards
Girish Sharma

Answer 1

>
I have just posted the column's definition
>
Thanks. Now it would be good if you do not change these again since we are using them as our baseline. Also the denormalized structure and data (e.g. for app_rule) makes it easier to see the data.
I am using a normalized version of the rulel table I name app_rule1. Here is the DDL and INSERTs for region 4 which I am using to work thru your example for complexity level 0 packet 1240.
create table app_rule1
app_region number(2),
avoid number(2)
insert into app_rule1 values (4,4);
insert into app_rule1 values (4,2);
insert into app_rule1 values (4,1);I'm having trouble with your approver data for this; I think it violates the business rules but perhaps you could explain.
Approver four data is
insert into approver values (34,4,1,0,0);approver_id - 34
working_place - 4 -- should have 'region' in the name because if I understand you correctly this MUST correspond to a region number - please confirm
hq_id - 1 -- should have 'region' in the name for same reason as above - please confirm
So in your explanation as to why you cannot use approver 34 for packet 1240 you said
>
No, we can not not, because approver's HQ (1) is also in avoid list, which should not.
>
My question - since approver 34 is in the avoid list for region 4 why does this approver have '4' for working_place? This '4' is the only thing that points to the rule table but it will always point to a rule record that we cannot use. This means approve 34 as defined by your data could never be used at all.
If this is true then it should be illegal to attempt to put this approver record into the table.
Is that analysis correct? My test code blew up with the first effort to follow your packet 1240 example because it tries to link data that shouldn't be linked.
Related to that - is it your intention that an approver can only point to one 'rule' record? That seems rather limited.
This code does not work properly because I need to change it based on your comments of the above but I will explain my initial effort
-- test1
-- query regions for one packet (could use DISTINCT but may not need to)
with packet_regions as (
select book_region from contents
where packet_no = 1240
-- query regions to avoid for one approver (including the approvers hq_id)
avoid_regions as (
select approver_id, working_place, hq_id avoid_region from approver
union
select approver_id, working_place, avoid avoid_region
from approver a, app_rule1 r
where a.working_place = r.app_region
-- query possible approvers
select approver_id from avoid_regions ar
where avoid_region not in (select book_region from packet_regions)1. The first query is to query regions represented for one packet
2. Second query is to select regions to avoid for each approver
3. Third query is to use the first two to select the list of legal approvers - that is, approvers whose 'avoid' list (including their HQ) is not represented in the regions for the packet.
Perhaps you can see how much simpler it is to have the app_rule data normalized rather than column-oriented (which would require parsing of some sort).
I can continue testing when I understand your response to my questions above.