Programmatically creating Markdown tables in R with KnitR

Question

I am just starting to learn about KnitR and the use of Markdown in generating R documents and reports. This looks to be perfect for a lot of the day to day reporting that I

Answer 1

Just wanted to update this with what I settled on doing. I am using the hwriter package right now to print out tables, and using the row.* and col.* features to put CSS classes on to different elements. Then, I wrote custom CSS to make my display as I wanted it. So, here's an example in case anyone else is dealing with something similar.

First, create a file that will do the knitting and change the Markdown into HTML:

FILE: file_knit.r
#!/usr/bin/env Rscript

library(knitr)
library(markdown)

knit("file.Rmd")
markdownToHTML("file.md","file.html",stylesheet="~/custom.css")

Next, create the actual Markdown file:

FILE: file.Rmd
Report of Fruit vs. Animal Choices
==================================

This is a report of fruit vs. animal choices.

```{r echo=FALSE,results='asis'}
library(hwriter)
set.seed(9850104)
my.df <- data.frame(Var1=sample(x=c("Apple","Orange","Banana"),size=40,replace=TRUE),
                    Var2=sample(x=c("Dog","Cat","Bunny"),size=40,replace=TRUE))

tbl1 <- table(my.df$Var1,my.df$Var2)

tbl1 <- cbind(tbl1,rowSums(tbl1))
tbl1 <- rbind(tbl1,colSums(tbl1))

colnames(tbl1)[4] <- "TOTAL"
rownames(tbl1)[4] <- "TOTAL"

# Because I used results='asis' for this chunk, I can just use cat() and hwrite() to 
# write out the table in HTML. Using hwrite()'s row.* function, I can assign classes
# to the various table elements.
cat(hwrite(tbl1,
           border=NA,
           table.class="t1",
           row.class=list(c("header col_first","header col","header col","header col", "header col_last"),
                          c("col_first","col","col","col","col_last"),
                          c("col_first","col","col","col","col_last"),
                          c("col_first","col","col","col","col_last"),
                          c("footer col_first","footer col","footer col","footer col","footer col_last"))))
```

Finally, just create a custom CSS file.

FILE: custom.css
body {
  font-family: sans-serif;
  background-color: white;
  font-size: 12px;
  margin: 20px;
}

h1 {font-size:1.5em;}

table {
  border: solid;
  border-color: black;
  border-width: 2px;
  border-collapse: collapse;
  margin-bottom: 20px;
  text-align: center;
  padding: 0px;
}

.t1 .header {
  color: white;
  background-color: black;
  border-bottom: solid;
  border-color: black;
  border-width: 2px;
  font-weight: bold;
}

.t1 .footer {
  border-top: solid;
  border-color: black;
  border-width: 2px;
}

.t1 .col_first {
  border-right: solid;
  border-color: black;
  border-width: 2px;
  text-align: left;
  font-weight: bold;
  width: 75px;
}

.t1 .col {
  width: 50px;
}

.t1 .col_last {
  width: 50px;
  border-left: solid;
  border-color: black;
  border-width: 2px;
}

Executing ./file_knit.r gives me file.html, which looks like this:

So, hopefully this might be helpful to others who want a bit more formatting in Markdown output!

Answer 2

It is not very hard to make your own customized function. Here is a very simple proof of concept to generate an rmarkdown table of a data.frame:

   rmarkdownTable <- function(df){
      cat(paste(names(df), collapse = "|"))
      cat("\n")
      cat(paste(rep("-", ncol(df)), collapse = "|"))
      cat("\n")

      for(i in 1:nrow(df)){
        cat(paste(df[i,], collapse = "|"))
        cat("\n")
        }
    invisible(NULL)
    }

In .Rmd document you would then use the function with results = 'asis':

```{r, results = 'asis'}
rmarkdownTable <- function(df){
  cat(paste(names(df), collapse = "|"))
  cat("\n")
  cat(paste(rep("-", ncol(df)), collapse = "|"))
  cat("\n")

  for(i in 1:nrow(df)){
    cat(paste(df[i,], collapse = "|"))
    cat("\n")
    }
invisible(NULL)
}

rmarkdownTable(head(iris))
```

The code above would give you the following figure (in the example this is the pdf output, but since the table is in markdwon, you could knit into html or word too).

From here - and reading other people´s code - you can figure out how to manipulate the text to generate the table you want and create more personalized functions.