fix toolAggregate edge case #227
Conversation
| mapping <- data.frame(a = c("A2", "B1"), b = c("C", "D")) | ||
| weight <- pm | ||
| getItems(weight, 3) <- c("C", "D") | ||
| expect_silent(toolAggregate(pm, mapping, from = "a", to = "b", dim = 3, weight = weight)) |
There was a problem hiding this comment.
Without this PR the following fails (as it should, because from and to are swapped here):
toolAggregate(pm, mapping, from = "b", to = "a", dim = 3, weight = weight)
With this PR this no longer fails, instead 'from' and 'to' are simply swapped. However, there are other cases where 'from' and 'to' are swapped opportunistically already, the following is TRUE even without this PR (using pm and map from test-toolAggregate.R):
a <- toolAggregate(pm, map, from = "reg", to = "from")
b <- toolAggregate(pm, map, from = "from", to = "reg")
identical(noC(a),noC(b))
While I don't like this at all, it seems be okay as this problem is already present and also somewhat inherent to the ultra flexible interface and "smart" default values if from and to are not set explicitly. I've spent quite a while on finding a solution without this downside, but couldn't find one. toolAggregate is doing so many things, allowing input in so many formats and combinations that I find it has become a serious problem to further work on/fix bugs in toolAggregate.
toolAggregate would fail if it should do a simple renaming (rel is a 1 to 1 mapping) where a weight with output categories (as opposed to input categories) is passed, see test added in this PR. This PR fixes that.