[R-sig-DB] character to factor transform in package RpgSQL

[Xiaobo Gu]

Hi Seth,
You are right, I have change the fetch function to this:


setMethod("fetch", signature(res="pgSQLResult", n="numeric"),
def=function(res, n, stringsAsFactors = TRUE, ...) {
  cols <- .jcall(res using md, "I", "getColumnCount")
  if (cols < 1) return(NULL)
  l <- list()
  for (i in 1:cols) {
    ct <- .jcall(res using md, "I", "getColumnType", i)
    l[[i]] <- if (ct == -5 | ct ==-6 | (ct >= 2 & ct <= 8)) {
           numeric()
       } else if (ct == 91) {
           structure(numeric(), class = "Date")
       } else if (ct == 93) {
           structure(numeric(), class = class(Sys.time()))
       } else character()
    names(l)[i] <- .jcall(res using md, "S", "getColumnName", i)
  }

  j <- 0
  while (.jcall(res using jr, "Z", "next")) {
    j <- j + 1
    for (i in 1:cols) {
      l[[i]][j] <- if (is.numeric(l[[i]])) {
          l[[i]][j] <- .jcall(res using jr, "D", "getDouble", i)
      } else if (inherits(l[[i]], "Date")) {
        tentativeDate <- .jcall(res using jr, "S", "getString", i)
	    if (length(tentativeDate) == 0 || tentativeDate == "0001-01-01
BC") tentativeDate <- NA
        l[[i]][j] <- as.Date(tentativeDate)
      } else {
        a <- .jcall(res using jr, "S", "getString", i)
        l[[i]][j] <- if (is.null(a)) NA else a
      }
    }
    if (n > 0 && j >= n) break
  }
  if (j)
    as.data.frame(l, row.names=1:j, stringsAsFactors = stringsAsFactors )
  else
    as.data.frame(l, stringsAsFactors = stringsAsFactors)
})


On Sun, Nov 14, 2010 at 12:52 PM, Seth Falcon <seth using userprimary.net> wrote:
> Hi,
>
> On Sat, Nov 13, 2010 at 8:28 PM, Xiaobo Gu <guxiaobo1982 using gmail.com> wrote:
>> Hi,
>> I have looked at the source code of package RpgSQL, it seems the fetch
>> function does not transform character column into factors, but when I
>> use the class function to test the result data frame of function
>> fetch, it all tells factor, can you help identify which part of the
>> function doing the transformation, in our usage we want to keep some
>> character column just as characters, bellowing is the source code of
>> function fetch in package RpgSQL.
>
> A guess is that you are seeing the default behavior of data.frame (and
> likely also as.data.frame) in transforming character vectors into
> factors.
>
> So I would suggest experimenting with this last bit:
>>  if (j)
>>    as.data.frame(l, row.names=1:j)
>>  else
>>    as.data.frame(l)
>> })
>
> For example:
>
>    > sapply(as.data.frame(letters[1:3]), class)
>    letters[1:3]
>        "factor"
>    > sapply(as.data.frame(letters[1:3], stringsAsFactors=FALSE), class)
>    letters[1:3]
>     "character"
>
>
> + seth
>
>
> --
> Seth Falcon | @sfalcon | http://userprimary.net/
>