Assignment 7

Introduction

In this file, three data frames (tables) are created. Each one shows the same data, but each one has data from different types of sources. The first one is from a json file, the second is from a html file, and the third is from an xml file. All the tables are at the end of this document.

library(rvest)
library(rjson)
library(jsonlite)

## 
## Attaching package: 'jsonlite'

## The following objects are masked from 'package:rjson':
## 
##     fromJSON, toJSON

jsonData <- fromJSON("https://raw.githubusercontent.com/juliaDataScience-22/cuny-fall-23/Assignment-7/books.json")

library(RCurl)
library(XML)
url <- getURL("https://github.com/juliaDataScience-22/cuny-fall-23/blob/Assignment-7/books.html", ssl.verifyPeer=FALSE)
htmlData1 <- read_html(url)
tables <- html_nodes(htmlData1, "table")
htmlData <- html_table(tables)
htmlData <- as.data.frame(htmlData)


library(xml2)
newXml <- read_xml("https://raw.githubusercontent.com/juliaDataScience-22/cuny-fall-23/edefcf444d4ef72ead919f974d8836d0f62e58ab/books.xml")
parsed <- xmlParse(newXml)
xmlData <- xmlToDataFrame(parsed)

library(stringr)
jsonData$books$book[c('Author_One', 'Author_Two')] <- str_split_fixed(jsonData$books$book$Author, ',', 2)
jsonData$books$book <- jsonData$books$book[,-2]

htmlData[c('Author_One', 'Author_Two')] <- str_split_fixed(htmlData$Author, ',', 2)
htmlData <- htmlData[,-2]

xmlData[c('Author_One', 'Author_Two')] <- str_split_fixed(xmlData$Author, ',', 2)
xmlData <- xmlData[,-2]

library(gt)
gt(jsonData$books$book) |>
  tab_header(
    title = "Table 1",
    subtitle = "JSON Data"
  )

Table 1
JSON Data
Title	Pages	Publisher	Year	Author_One	Author_Two
Illuminae	608	Knopf Books for Young Readers	2015	Amie Kaufman	Jay Kristoff
The Martian	387	Broadway Books	2014	Andy Weir
Stringers	428	Angry Robot	2022	Chris Panatier

gt(htmlData) |>
  tab_header(
    title = "Table 2",
    subtitle = "HTML Data"
  )

Table 2
HTML Data
Title	Pages	Publisher	Year	Author_One	Author_Two
Illuminae	608	Knopf Books for Young Readers	2015	Amie Kaufman	Jay Kristoff
The Martian	387	Broadway Books	2014	Andy Weir
Stringers	428	Angry Robot	2022	Chris Panatier

gt(xmlData) |>
  tab_header(
    title = "Table 3",
    subtitle = "XML Data"
  )

Table 3
XML Data
Title	Pages	Publisher	Year	Author_One	Author_Two
Illuminae	608	Knopf Books for Young Readers	2015	Amie Kaufman	Jay Kristoff
The Martian	387	Broadway Books	2014	Andy Weir
Stringers	428	Angry Robot	2022	Chris Panatier