Templating with knit_expand()
In knitr: A General-Purpose Package for Dynamic Report Generation in R

A few simple examples:

library(knitr)
knit_expand(text = 'The value of pi is {{pi}}.')
knit_expand(text = 'The value of a is {{a}}, so a + 1 is {{a+1}}.', a = rnorm(1))
knit_expand(text = 'The area of a circle with radius {{r}} is {{pi*r^2}}', r = 5)

Any number of variables:

knit_expand(text = 'a is {{a}} and b is {{b}}, with my own pi being {{pi}} instead of {{base::pi}}', a=1, b=2, pi=3)

Custom delimiter <% %>:

knit_expand(text = 'I do not like curly braces, so use % with <> instead: a is <% a %>.', a = 8, delim = c("<%", "%>"))

The pyexpander delimiter:

knit_expand(text = 'hello $(LETTERS[24]) and $(pi)!', delim = c("$(", ")"))

Arbitrary R code:

knit_expand(text = 'you cannot see the value of x {{x=rnorm(1)}}but it is indeed created: x = {{x}}')
res = knit_expand(text = c(' x | x^2', '{{x=1:5;paste(sprintf("%2d | %3d", x, x^2), collapse = "\n")}}'))
cat(res)

The m4 example: https://en.wikipedia.org/wiki/M4_(computer_language)

res = knit_expand(text = c('{{i=0;h2=function(x){i<<-i+1;sprintf("<h2>%d. %s</h2>", i, x)} }}<html>',
'{{h2("First Section")}}', '{{h2("Second Section")}}', '{{h2("Conclusion")}}', '</html>'))
cat(res)

Build regression models based on a template; loop through all variables in mtcars:

src = lapply(names(mtcars)[-1], function(i) {
knit_expand(text=c("# Regression on {{i}}", '```r}}', 'lm(mpg~{{i}}, data=mtcars)', '```'))
})
# knit the source
res = knit_child(text = unlist(src))
res = paste('<pre><code>', gsub('^\\s*|\\s*$', '', res), '</code></pre>', sep = '')