englandr@phoenix.Princeton.EDU (Scott Englander) (07/16/89)
I'd like to delete duplicate records in a FoxBase file. I tried setting
the index to unique and then copying the database to a new file, but the
duplicates remained. I ultimately did it with a simple and SLOW program
that checked for duplicates and deleted them (as i'm typing, it's
running in the background under Multifinder -- there are 4000 records!).
But there must be a better way to do this using the unique index, since
only the first record with a given value of the index field is indexed.
--
- Scottbote@csense.UUCP (John Boteler) (07/17/89)
From article <9426@phoenix.Princeton.EDU>, by englandr@phoenix.Princeton.EDU (Scott Englander): > I'd like to delete duplicate records in a FoxBase file. I tried setting > the index to unique and then copying the database to a new file, but the > duplicates remained. I ultimately did it with a simple and SLOW program > that checked for duplicates and deleted them (as i'm typing, it's > running in the background under Multifinder -- there are 4000 records!). > But there must be a better way to do this using the unique index, since > only the first record with a given value of the index field is indexed. Assuming that a 'duplicate record' in your case is one which has the same key as another, and the other fields do not enter into the decision, here is one suggestion: set unique on use &tablename index &indexname go top do while .not. eof() delete && marks this record to be copied later skip && advances to next unique record in index enddo copy to &temp for deleted() && copy only those marked !mv tablename.dbf tablename.old !mv temp.dbf tablename.dbf go home && relax! Now, before you remove the old file, are you sure that those records which did not make it into the new file are truly bogus? rm is forever. -- Bote uunet!cyclops!csense!bote {mimsy,sundc}!{prometheus,hqda-ai}!media!cyclops!csense!bote
timk@xenitec.uucp (Tim Kuehn) (07/17/89)
englandr@phoenix.Princeton.EDU (Scott Englander) writes: >I'd like to delete duplicate records in a FoxBase file. ... Previous attempt deleted ... >But there must be a better way to do this using the unique index, since >only the first record with a given value of the index field is indexed. >-- > > - Scott try this: Definintions: ------------- dborig: the original database dbunique: the database with duplicate records removed ndxuniq: the index file on the key with only unique values of the key <key>: the key value you are indexing on Program: -------- use dborig copy structure to dbuniq select 0 use dbunique select dborig index on <key> to ndxuniq unique set index to ndxuniq go top do while .not. eof() select dbuniq append blank replace dbuniq->field1 with dborig->field1,; dbuniq->field2 with dborig->field2,; dbuniq->field3 with dborig->field3,; dbuniq->field4 with dborig->field4 <...etc...> select dborig skip enddo hope this helps! +-----------------------------------------------------------------------------+ |Timothy D. Kuehn timk@xenitec | |TDK Consulting Services !watmath!xenitec!timk | |871 Victoria St. North, Suite 217A | |Kitchener, Ontario, Canada N2B 3S4 (519)-741-3623 | |DOS/Xenix - SW/HW. uC, uP, DBMS. Satisfaction Gauranteed| +-----------------------------------------------------------------------------+
roger@batcomputer.tn.cornell.edu (Roger Boissonnas) (07/18/89)
In article <1989Jul17.165249.26554@xenitec.uucp> timk@egvideo.UUCP (Tim Kuehn) writes: >englandr@phoenix.Princeton.EDU (Scott Englander) writes: >>I'd like to delete duplicate records in a FoxBase file. > >... Previous attempt deleted ... > >Program: >-------- >use dborig >copy structure to dbuniq >select 0 >use dbunique >select dborig >index on <key> to ndxuniq unique >set index to ndxuniq >go top >do while .not. eof() > select dbuniq > append blank > replace dbuniq->field1 with dborig->field1,; > dbuniq->field2 with dborig->field2,; > dbuniq->field3 with dborig->field3,; > dbuniq->field4 with dborig->field4 > > <...etc...> > select dborig > skip >enddo You can write a program to do the same thing in only three lines: USE dborig INDEX ON <key> TO ndxuniq UNIQUE COPY TO dbuniq Roger Boissonnas roger@tcgould.tn.cornell.edu Cornell Theory Center (607) 255-8306 Roger Boissonnas roger@tcgould.tn.cornell.edu Cornell Theory Center (607) 255-8306
timk@xenitec.uucp (Tim Kuehn) (07/20/89)
roger@tcgould.tn.cornell.edu (Roger Boissonnas) writes: >In article <1989Jul17.165249.26554@xenitec.uucp> timk@egvideo.UUCP (Tim Kuehn) writes: >> >>Program: >>-------- ... source code deleted ... >You can do the same thing in only three lines: >USE dborig >INDEX ON <key> TO ndxuniq UNIQUE >COPY TO dbuniq > > >Roger Boissonnas roger@tcgould.tn.cornell.edu >Cornell Theory Center (607) 255-8306 You know what - you're right! I always thought the copy command used the entire database, irrespective of index keys. Learn something new every day I guess.... +-----------------------------------------------------------------------------+ |Timothy D. Kuehn timk@egvideo (soon to be timk@xenitec) | |TDK Consulting Services !watmath!xenitec!timk | |871 Victoria St. North, Suite 217A | |Kitchener, Ontario, Canada N2B 3S4 (519)-741-3623 | |DOS/Xenix - SW/HW. uC, uP, DBMS. Satisfaction Gauranteed| +-----------------------------------------------------------------------------+