Apply R/S-Plus functions to remote groups of DBMS rows (experimental)
Applies R/S-Plus functions to groups of remote DBMS rows without bringing an entire result set all at once. The result set is expected to be sorted by the grouping field.
postgresqlDBApply(res, INDEX, FUN = stop("must specify FUN"), begin = NULL, group.begin = NULL, new.record = NULL, end = NULL, batchSize = 100, maxBatch = 1e6, ..., simplify = TRUE)
a result set (see
a character or integer specifying the field name or field number that defines the various groups.
a function to be invoked upon identifying the last row from every group. This function will be passed a data frame holding the records of the current group, a character string with the group label, plus any other arguments passed to
a function of no arguments to be invoked just prior to retrieve the first row from the result set.
a function of no arguments to be invoked just after retrieving the last row from the result set.
a function of one argument (the group label) to be invoked upon identifying a row from a new group
a function to be invoked as each individual record is fetched. The first argument to this function is a one-row data.frame holding the new record.
the default number of rows to bring from the remote result set. If needed, this is automatically extended to hold groups bigger than
the absolute maximum of rows per group that may be extracted from the result set.
any additional arguments to be passed to
Not yet implemented
This function is meant to handle somewhat gracefully(?) large amounts
of data from the DBMS by bringing into R manageable chunks (about
batchSize records at a time, but not more than
the idea is that the data from individual groups can be handled by R, but
not all the groups at the same time.
The PostgreSQL implementation
postgresqlDBApply allows us to register R
functions that get invoked
when certain fetching events occur. These include the ``begin'' event
(no records have been yet fetched), ``begin.group'' (the record just
fetched belongs to a new group), ``new record'' (every fetched record
generates this event), ``group.end'' (the record just fetched was the
last row of the current group), ``end'' (the very last record from the
result set). Awk and perl programmers will find this paradigm very
familiar (although SAP's ABAP language is closer to what we're doing).
A list with as many elements as there were groups in the result set.
This is an experimental version implemented only in R (there are plans, time permitting, to implement it in S-Plus).
The terminology that we're using is closer to SQL than R. In R what we're referring to ``groups'' are the individual levels of a factor (grouping field in our terminology).