Обсуждение: FW: Case Statement
Hi Stephan, I have a lot of fields, so I'm not sure if a function or case is the way to go. Basically, I have, say 50 boolean fields that are being evaluated, and I want to have a column which is the sum of the number of "TRUE" values of those 50 columns, and then ORDER BY that column. So, for example of the 50 fields to be evaluated, the SELECT statement should return records ordered by those that match the most number of criteria. Does that make sense? Your example below may actually work for me, on the other hand: "case when field1 is true then 1 else 0 end +case when field2 is true then 2 else 0 end +case when field3 is true then 4else 0 end Let's suppose field1 is true and field2, field3 are not, would it return 1, and if all three are true it would return 7 for the case? If so, I think that's the way to go, as all this is being dynamically built in any case. Thanks, Tom -----Original Message----- From: pgsql-sql-owner@postgresql.org [mailto:pgsql-sql-owner@postgresql.org]On Behalf Of Stephan Szabo Sent: Wednesday, July 31, 2002 4:17 PM To: Tom Haddon Cc: pgsql-sql@postgresql.org Subject: Re: [SQL] Case Statement On Wed, 31 Jul 2002, Tom Haddon wrote: > Hi Folks, > > Pretty basic one here. I'm trying to do a SELECT statement that includes a > CASE structure that will be used in the ORDER BY clause, and I want the CASE > structure to evaluate a bunch of criteria and in each case increment its > value by a certain amount: > > SELECT id, description CASE > field1 IS TRUE THEN CASE=CASE+1 > field2 IS TRUE THEN CASE=CASE+2 > END > FROM table > ORDER BY CASE, description; > > Am I misunderstanding what CASE can do here, and if so, how else do I do > what I'm trying to do? I know that that isn't right - any pointers > appreciated. Are you trying to get a 0-3 based on field1 and field2 or some kind of aggregate? The first you can do that with something likecase when field1 is true then 1 else 0 end +case when field2 istrue then 2 else 0 end I think. (Extend for more columns) Or if you've got like 7-8 columns maybe just make a function that does the work for you for clarity. You'll also need a comma after description and you'll want to alias the case expression and use that in order by or use the position number. ---------------------------(end of broadcast)--------------------------- TIP 6: Have you searched our list archives? http://archives.postgresql.org
Just 'cuz you can do something with a language feature does not necessarily mean that it is a good way to do it. If this is a common query and you have a large dataset, then this is going to be computationally intensive. Why not create a field as part of the table and create a trigger to update the field with the sum of your true bool fields when the record is created or updated. This spreads out this computationally intensive query and then it becomes a simple query... Just some random thoughts... .. Otto > -----Original Message----- > From: pgsql-sql-owner@postgresql.org > [mailto:pgsql-sql-owner@postgresql.org]On Behalf Of Tom Haddon > Sent: Wednesday, July 31, 2002 4:33 PM > To: pgsql-sql@postgresql.org > Subject: FW: Case Statement > > > Hi Stephan, > > I have a lot of fields, so I'm not sure if a function or case > is the way to > go. Basically, I have, say 50 boolean fields that are being > evaluated, and I > want to have a column which is the sum of the number of > "TRUE" values of > those 50 columns, and then ORDER BY that column. So, for > example of the 50 > fields to be evaluated, the SELECT statement should return > records ordered > by those that match the most number of criteria. Does that > make sense? Your > example below may actually work for me, on the other hand: > > "case when field1 is true then 1 else 0 end + > case when field2 is true then 2 else 0 end + > case when field3 is true then 4 else 0 end > > Let's suppose field1 is true and field2, field3 are not, > would it return 1, > and if all three are true it would return 7 for the case? If > so, I think > that's the way to go, as all this is being dynamically built > in any case. > > Thanks, Tom > > > > -----Original Message----- > From: pgsql-sql-owner@postgresql.org > [mailto:pgsql-sql-owner@postgresql.org]On Behalf Of Stephan Szabo > Sent: Wednesday, July 31, 2002 4:17 PM > To: Tom Haddon > Cc: pgsql-sql@postgresql.org > Subject: Re: [SQL] Case Statement > > > On Wed, 31 Jul 2002, Tom Haddon wrote: > > > Hi Folks, > > > > Pretty basic one here. I'm trying to do a SELECT statement > that includes a > > CASE structure that will be used in the ORDER BY clause, > and I want the > CASE > > structure to evaluate a bunch of criteria and in each case > increment its > > value by a certain amount: > > > > SELECT id, description CASE > > field1 IS TRUE THEN CASE=CASE+1 > > field2 IS TRUE THEN CASE=CASE+2 > > END > > FROM table > > ORDER BY CASE, description; > > > > Am I misunderstanding what CASE can do here, and if so, how > else do I do > > what I'm trying to do? I know that that isn't right - any pointers > > appreciated. > > Are you trying to get a 0-3 based on field1 and field2 or some kind of > aggregate? The first you can do that with something like > case when field1 is true then 1 else 0 end + > case when field2 is true then 2 else 0 end > I think. (Extend for more columns) > > Or if you've got like 7-8 columns maybe just make a function that does > the work for you for clarity. > > You'll also need a comma after description and you'll want to > alias the > case expression and use that in order by or use the position number. > > > > ---------------------------(end of > broadcast)--------------------------- > TIP 6: Have you searched our list archives? > > http://archives.postgresql.org > > > ---------------------------(end of > broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to > majordomo@postgresql.org > >
On Wed, 31 Jul 2002, Tom Haddon wrote: > Hi Stephan, > > I have a lot of fields, so I'm not sure if a function or case is the way to > go. Basically, I have, say 50 boolean fields that are being evaluated, and I > want to have a column which is the sum of the number of "TRUE" values of > those 50 columns, and then ORDER BY that column. So, for example of the 50 > fields to be evaluated, the SELECT statement should return records ordered > by those that match the most number of criteria. Does that make sense? Your > example below may actually work for me, on the other hand: > > "case when field1 is true then 1 else 0 end + > case when field2 is true then 2 else 0 end + > case when field3 is true then 4 else 0 end > > Let's suppose field1 is true and field2, field3 are not, would it return 1, > and if all three are true it would return 7 for the case? If so, I think > that's the way to go, as all this is being dynamically built in any case. Yes, that's what it should do (I did the 2 variable case and all 4 combinations to try getting all 0-3). If you're doing many more selects than update/inserts to the table where you want to do this (and the fields you care about is a constant set) then a trigger on insert/update is the way to go and just store the value.