Lemon print – pretty printing data frames and tables

How it works

knitr uses a S3 generic function, knit_print, to print objects. Method dispatching in R means that by defining a function, say knit_print.data.frame, calling knit_print(df) will call our function if df was a data frame.

To have knitr use our function when outputting a data frame, we define the function knit_print.data.frame (and similar for data frames passed through dplyr functions, i.e tbl_df and grouped_df).

Disabling the function is by usual chunk options, render = normal_print. check ?knit_print.

Disabling the functions.

```{r normal_print,render=normal_print}
head(USArrests)
```

##            Murder Assault UrbanPop Rape
## Alabama      13.2     236       58 21.2
## Alaska       10.0     263       48 44.5
## Arizona       8.1     294       80 31.0
## Arkansas      8.8     190       50 19.5
## California    9.0     276       91 40.6
## Colorado      7.9     204       78 38.7

You can still ask for kable.

```{r results='asis',render=normal_print}
kable(head(USArrests),caption='Normal `kable` usage.')
```

Why does it matter

The second code block, when edited in RStudio, would look like this: And, as demonstrated, the rendered document has the data frame printed nicely.

If we do not wish this behaviour, but still use kable,

kable(head(USArrests))

RStudio would look like this:

Other objects covered

Summaries are nicely formatted, but these require setting the S3 method for table.

```{r kable.opts=list(caption='Summary tables are printed with some default options.')}
knit_print.table <- lemon_print
summary(USArrests)
```

Cross-tabulations are however not nicely formatted.

head(mtcars)

with(mtcars, table(cyl, gear))

##    gear
## cyl  3 4 5
##   4  1 8 2
##   6  2 4 1
##   8 12 0 2

```{r dplyr,kable.opts=list(caption='Also works on `dplyr` objects.')}
library(dplyr)
mtcars %>% group_by(cyl) %>% summarise(mean(disp))
```