Here's how I pull data from a DB, stash it in column based binary J
files, and map them using jmf. After that I can just use the inverted
table mechanisms explained so nicely on the J wiki.
I would assume you can do similar with your CSV problem.
require 'dates dd files jmf printf'
DSN =: 'dsn=something;uid=user;pwd=password'
STMT =: 'SELECT date_field, float_field, int_field FROM a_table'
i2b =: 2 & (3!:4)
writeints =: (fappend~ i2b)~
f2b =: 2 & (3!:5)
writefloats =: (fappend~ f2b)~
getdata =: 4 : 0
stmt =. x
dsn =. y
ch =. ddcon dsn
'Executing SQL:\n%s' printf <stmt
sh =. stmt ddsel ch
ddbind sh
total =. 0
while. 0 <: ddfetch sh do.
rcnt =. 0 { ddrow sh NB. ddrow appears to return an
array of one element!
'Saving %d rows.' printf <rcnt
(tsrep dts_jdd_ rcnt {. dddata sh,1) writefloats 'DATE.ijn'
(rcnt {. , dddata sh,2) writeints 'INT.ijn'
(rcnt {. , dddata sh,3) writefloats 'FLOAT.ijn'
total =. total + rcnt
end.
ddend sh
dddis ch
'Saved a total of %d rows.' printf <total
total
)
mapdata =: 3 : 0
JFL map_jmf_ 'DATE';'DATE.ijn'
JINT map_jmf_ 'INT';'INT.ijn'
JFL map_jmf_ 'FLOAT';'FLOAT.ijn'
)
STMT getdata DSN
mapdata ''
+/ INT
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
