These functions are what underpins the ability of certain geoms to work automatically in both directions. See the *Extending ggplot2* for how they are used when implementing `Geom`, `Stat`, and `Position` classes.
has_flipped_aes(
data,
params = list(),
main_is_orthogonal = NA,
range_is_orthogonal = NA,
group_has_equal = FALSE,
ambiguous = FALSE,
main_is_continuous = FALSE
)flip_data(data, flip = NULL)
flipped_names(flip = FALSE)
The layer data
The parameters of the `Stat`/`Geom`. Only the `orientation` parameter will be used.
If only `x` or `y` are present do they correspond to the main orientation or the reverse. E.g. If `TRUE` and `y` is present it is not flipped. If `NA` this check will be ignored.
If `xmin`/`xmax` or `ymin`/`ymax` is present do they correspond to the main orientation or reverse. If `NA` this check will be ignored.
Is it expected that grouped data has either a single `x` or `y` value that will correspond to the orientation.
Is the layer ambiguous in its mapping by nature. If so, it will only be flipped if `params$orientation == "y"`
If there is a discrete and continuous axis, does the continuous one correspond to the main orientation?
Logical. Is the layer flipped.
`has_flipped_aes()` returns `TRUE` if it detects a layer in the other orientation and `FALSE` otherwise. `flip_data()` will return the input unchanged if `flip = FALSE` and the data with flipped aesthetic names if `flip = TRUE`. `flipped_names()` returns a named list of strings. If `flip = FALSE` the name of the element will correspond to the element, e.g. `flipped_names(FALSE)$x == "x"` and if `flip = TRUE` it will correspond to the flipped name, e.g. `flipped_names(FALSE)$x == "y"`
How the layer data should be interpreted depends on its specific features. `has_flipped_aes()` contains a range of flags for defining what certain features in the data correspond to:
- `main_is_orthogonal`: This argument controls how the existence of only a `x` or `y` aesthetic is understood. If `TRUE` then the exisiting aesthetic would be then secondary axis. This behaviour is present in [stat_ydensity()] and [stat_boxplot()]. If `FALSE` then the exisiting aesthetic is the main axis as seen in e.g. [stat_bin()], [geom_count()], and [stat_density()]. - `range_is_orthogonal`: This argument controls whether the existance of range-like aesthetics (e.g. `xmin` and `xmax`) represents the main or secondary axis. If `TRUE` then the range is given for the secondary axis as seen in e.g. [geom_ribbon()] and [geom_linerange()]. `FALSE` is less prevalent but can be seen in [geom_bar()] where it may encode the span of each bar. - `group_has_equal`: This argument controls whether to test for equality of all `x` and `y` values inside each group and set the main axis to the one where all is equal. This test is only performed if `TRUE`, and only after less computationally heavy tests has come up empty handed. Examples are [stat_boxplot()] and [stat_ydensity] - `ambiguous`: This argument tells the function that the layer, while bidirectional, doesn't treat each axis differently. It will circumvent any data based guessing and only take hint from the `orientation` element in `params`. If this is not present it will fall back to `FALSE`. Examples are [geom_line()] and [geom_area()] - `main_is_continuous`: This argument controls how the test for discreteness in the scales should be interpreted. If `TRUE` then the main axis will be the one which is not discrete-like. Conversely, if `FALSE` the main axis will be the discrete-like one. Examples of `TRUE` is [stat_density()] and [stat_bin()], while examples of `FALSE` is [stat_ydensity()] and [stat_boxplot()]
`has_flipped_aes()` is used to sniff out the orientation of the layer from the data. It has a range of arguments that can be used to finetune the sniffing based on what the data should look like. `flip_data()` will switch the column names of the data so that it looks like x-oriented data. `flipped_names()` provides a named list of aesthetic names that corresponds to the orientation of the layer.