Set up

I recommend creating an RStudio Project with three files important to this task:

1. Your document template prepared in R Markdown (file type .Rmd)
2. Your style-reference document (.docx) that we’ll put in a style/ subfolder
3. A script (.R) file for reading your data and looping through each element

Aside from your .Rproj file, your folder should also contain:

• a subfolder to hold the output files (e.g. output/)
• any additional subfolders you need for input files (e.g. data/ and images/)
• an optional README.md file to explain the purpose of your work

A simple folder setup for your working directory

Below is a more detailed explanation using the Dawson’s Creek example. Click to jump to each section:

• the R Markdown template
• the Word document used as a style reference
• the R script for iterating over the template

1. The template

---
output:
  word_document:
    reference_docx: style/dawson_style.docx
---

`r paste0(episode$production_code, ": '", episode$title, "'")`

episode %>% 
  dplyr::select(  # choose these columns for display
    Season = season,
    `Number in season` = number_in_season,
    `Number in series` = number_in_series,
    `Original air date` = original_air_date
  ) %>% 
  knitr::kable()  # print to output table

2. The style

The default output when rendering R Markdown to Word is a bit boring. Instead, you can create a separate dummy Word document that contains modified styles to suit your theme.

You can read more about creating a style document in Richard Layton’s ‘happy collaboration with Rmd to docx’ on the R Markdown site. Basically you create a dummy document in which you’ve set the formatting for each style element (title, third-level header, paragraph text, etc), store it in your working repository and refer to it from the YAML of your R Markdown report (as mentioned in the section above).

The style document

You can download the example style file used in this demo from the GitHub repo.

3. Iterate

So we have the template (.Rmd) and style (.docx) files in place. Now we need an R script (.R) to orchestrate the iteration of our outputs.

This script (i) reads the data and (ii) executes a loop that applies episode data to the template, saving to a specified subfolder with a unique file name. You’ll get one document for each unique element; in this demo we’ll get 13 because there are 13 rows (episodes) in our data set.

You can see the script file for this demo in the source code on GitHub.

After reading in the data with something like data <- readr::read_csv(file = "data/dawsons-creek-season-1.csv") you can write a loop that looks something like this:

for (i in data$production_code){  # for each unique episode...
  
  # Isolate that episode from the dataset
  
  episode <- data[data$production_code == i, ]  
  
  # The one-row dataframe object 'episode' we just created will be used as the
  # source for filling the details in on the template file
  
  # Now render ('knit') the R Markdown file to Word format, name it and save it
  
  rmarkdown::render(
    input = "01_template.Rmd",  # path to the template
    output_file = paste0("episode_", i, ".docx"),  # name the output
    output_dir = "output"  # folder in which to put the output file
  )

}  # end of loop

So this code:

• loops through episodes given unique values in the production_code column
• subsets the data object (a data frame of all episodes) to isolate the production code for each episode in turn
• renders the template document after filling it with the episode data
• gives the file a name that incorporates the production code
• saves this file to the output/ subfolder in our project directory