607-wk7-hw7

HTML

url <- "https://raw.githubusercontent.com/curdferguson/607-wk7-hw7/main/recommended_reads_cheese.html"

books_parsed <- read_html(url)

# read the information from the HTML table
books_html <- books_parsed %>% 
  html_table(fill = TRUE) %>% 
  .[[1]] %>% 
  as_tibble() %>% 
  rename(`Link Text` = `Buy Online`)

# create a column for the link url using nodes and list apply functions
x <- nrow(books_html) + 1
y <- c(2:x)

xpaths <- paste("/html/body/table/tr[", y, "]/td[4]/a", sep="")
nodes <- lapply(xpaths, function(z){ 
  html_node(books_parsed, xpath = z)})
link_urls = lapply(nodes, function(i){
  str_match(i, "(?<=(<a href=\")).*(?=(\">[A-Z]))")[,1]})

books_html$`Link Url` <- as_vector(link_urls)

books_html

## # A tibble: 4 x 7
##       ISBN Title        Author  `Link Text` Price Comment       `Link Url`      
##      <dbl> <chr>        <chr>   <chr>       <chr> <chr>         <chr>           
## 1  9.78e12 Composing t~ Brian ~ Amazon      $22.~ "These recip~ https://www.ama~
## 2  9.78e12 HOT CHEESE:~ Polina~ Book Larder $19.~ "From classi~ https://www.boo~
## 3  9.78e12 Cheese and ~ Paul K~ Strand Boo~ $24.~ "A fascinati~ https://www.str~
## 4  9.78e12 The Book of~ Liz Th~ Amazon      $40.~ "With its en~ https://www.ama~

JSON

books_json <- jsonlite::fromJSON("https://raw.githubusercontent.com/curdferguson/607-wk7-hw7/main/recommended_reads_cheese.json")

books_json_tibble <- bind_rows(lapply(books_json[[1]], function(x) {
  tibble(
    ISBN = as.numeric(x$ISBN),
    Title = x$Title,
    Author = x$Author,
    url = x$`Buy Online`[[1]],
    `Link Text` = x$`Buy Online`[[2]],
    Price = x$Price,
    Comment = x$Comment)
  }))

books_json_tibble

## # A tibble: 5 x 7
##       ISBN Title         Author   url           `Link Text` Price Comment       
##      <dbl> <chr>         <chr>    <chr>         <chr>       <chr> <chr>         
## 1  9.78e12 Composing th~ Brian K~ https://www.~ Amazon      $22.~ These recipes~
## 2  9.78e12 Composing th~ Leigh F~ https://www.~ Amazon      $22.~ These recipes~
## 3  9.78e12 HOT CHEESE: ~ Polina ~ https://www.~ Book Larder $19.~ From classics~
## 4  9.78e12 Cheese and C~ Paul Ki~ https://www.~ Strand Boo~ $24.~ A fascinating~
## 5  9.78e12 The Book of ~ Liz Tho~ https://www.~ Amazon      $40.~ With its enga~