The Data

The data is stored in my github in each of the three formats requested. The structure of the data is slightly different in each format due to the differing formatting requirements but is fundamentally the same information.

xml_address = "https://raw.githubusercontent.com/Tillmawitz/data_607/refs/heads/main/assignment_7/books.xml"
html_address = "https://raw.githubusercontent.com/Tillmawitz/data_607/refs/heads/main/assignment_7/books.html"
json_address = "https://raw.githubusercontent.com/Tillmawitz/data_607/refs/heads/main/assignment_7/books.json"

Parsing the Data

First we will parse the xml. This was by far the trickiest format because of the way the interpreter handled the initial formating. It was impossible for the unnest_wider function to simplify each element of the original “list” as it treated each “book” as another list, so each element was actually a list of lists. This required an additional mutation accross all columns that applies list_simplify twice on each element.

xml_data <- as_list(read_xml(xml_address)) |>
  as_tibble() |>
  unnest_wider(books, names_repair = "universal", simplify = TRUE)
## New names:
## • `author` -> `author...2`
## • `author` -> `author...3`
xml_data |>
  mutate(across(everything(), ~ list_simplify(list_simplify(.x))))
## # A tibble: 3 × 5
##   title                                  author...2 author...3 publisher edition
##   <chr>                                  <chr>      <chr>      <chr>     <chr>  
## 1 Forecasting: Principles and Practice   Rob J Hyn… George At… OTexts    3      
## 2 Applied Predictive Modeling            Max Kuhn   Kjell Joh… Springer  5      
## 3 Perfect Puppy in 7 Days: How to Start… Sophia Yin Sophia Yin CattleDo… 1

The html format was by far the simplist to handle, as the html page consisted of only a single table which the html_table function was easily able to interpret.

html_data <- read_html(html_address)
html_data |> html_table()
## [[1]]
## # A tibble: 3 × 4
##   Title                                                Authors Publisher Edition
##   <chr>                                                <chr>   <chr>       <int>
## 1 Forecasting: Principles and Practice                 Rob J … OTexts          3
## 2 Applied Predictive Modeling                          Max Ku… Springer        5
## 3 Perfect Puppy in 7 Days: How to Start Your Puppy Of… Sophia… CattleDo…       1

Finally, the JSON format was relatively easy to interpret as well due to the unnest function which was able to nicely pull the elements from the response.

json_data <- fromJSON(json_address)
json_data |> 
  as_tibble() |>
  unnest(Books)
## # A tibble: 3 × 4
##   Title                                                Authors Publisher Edition
##   <chr>                                                <list>  <chr>       <int>
## 1 Forecasting: Principles and Practice                 <chr>   OTexts          3
## 2 Applied Predictive Modeling                          <chr>   Springer        5
## 3 Perfect Puppy in 7 Days: How to Start Your Puppy Of… <chr>   CattleDo…       1

Concluding Thoughts

It is important to note that each dataframe had different ways of handling multiple authors. This is partly due to how the data is formatted in each document and partly due to how each format is parsed. The xml version has uniquely named author columns, the html version has a single string with author names separated by commas, and the JSON version has a list in the Authors column. Ultimately it would be easy to coerce these columns into any of the other formats depending on what was most convenient for the analysis being performed so they were left in their original formats to illustrate the differences in the parsing methods.