library("testthat")
library("recommenderlab")

set.seed(1234)

## create a matrix with ratings
db <- matrix(
  as.numeric(sample(
    c(NA, 0:5), 100, replace = TRUE,
    prob = c(.7, rep(.3 / 6, 6))
  )),
  nrow = 10,
  ncol = 10,
  dimnames = list(
    users = paste('u', 1:10, sep = ''),
    items = paste('i', 1:10, sep = '')
  )
)


## convert
r <- as(db, "realRatingMatrix")

## do we get the original matrix back?
expect_identical(as(r, "matrix"), db)

## Coersions:
expect_identical(r, as(as(r, "matrix"), "realRatingMatrix"))
expect_identical(r, as(as(r, "dgCMatrix"), "realRatingMatrix"))
expect_identical(r, as(as(r, "dgTMatrix"), "realRatingMatrix"))
expect_identical(getRatings(r[c("u1", "u2"), colnames(r)]),
  getRatings(as(as(r, "data.frame"), "realRatingMatrix")[c("u1", "u2"),
    colnames(r)]))

## subset
expect_identical(db[9:10, 2:5], as(r[9:10, 2:5], "matrix"))

## number of ratings
expect_identical(sum(sapply(as(r, "list"), length)), nrow(as(r, "data.frame")))
expect_identical(nratings(r), sum(sapply(as(r, "list"), length)))

## check ratings
expect_identical(as(r, "dgTMatrix")@x, getRatings(r))

## check subset assignment (preserves 0s)
r[2, 1:5] <- 1:5
db[2, 1:5] <- 1:5
expect_identical(as(r, "matrix"), db)

## check row/colSums and counts
expect_identical(rowSums(r), rowSums(db, na.rm = TRUE))
expect_identical(colSums(r), colSums(db, na.rm = TRUE))

expect_identical(rowCounts(r), apply(db, MARGIN = 1, function(x)
  sum(!is.na(x))))
expect_identical(colCounts(r), apply(db, MARGIN = 2, function(x)
  sum(!is.na(x))))