Assignment 3

Question 5

paths_allowed("https://en.wikipedia.org/wiki/2026_in_film")

##  en.wikipedia.org

## [1] TRUE

Question 6

src_html <- read_html("https://en.wikipedia.org/wiki/2026_in_film")
tables_html <- html_elements(src_html, "table")
highest_grossing <- html_table(tables_html[3])
highest_grossing

## [[1]]
## # A tibble: 10 × 4
##     Rank Title                             Distributor         `Worldwide gross`
##    <int> <chr>                             <chr>               <chr>            
##  1     1 Cheburashka 2 †                   Central Partnership $79,500,909      
##  2     2 28 Years Later: The Bone Temple † Sony                $56,641,191      
##  3     3 Send Help †                       20th Century Studi… $53,524,493      
##  4     4 Border 2 †                        AA Films            $51,204,000[3]   
##  5     5 Mercy                             Amazon MGM Studios… $49,508,904      
##  6     6 Primate †                         Paramount Pictures  $39,677,703      
##  7     7 Return to Silent Hill †           Iconic Events Rele… $38,606,428      
##  8     8 Dracula †                         SND (France)        $33,593,404      
##  9     9 Mana Shankara Vara Prasad Garu †  Gold Box Entertain… $32,225,000[4][5]
## 10    10 Prostokvashino  [ru]†             Cinema Atmosphere   $31,934,076

Question 8

library(rvest)
library(purrr)
library(dplyr)

Function to scrape highest-grossing films

years <- 2020:2025
get_highest_grossing <- function(year) {
  
  url <- paste0("https://en.wikipedia.org/wiki/", year, "_in_film")
  page <- read_html(url)
  tables <- html_elements(page, "table")
  hg_table <- html_table(tables[[3]])
  hg_table %>%
    mutate(year = year)
}

hg_films <- map_dfr(years, get_highest_grossing)
hg_films

## # A tibble: 60 × 5
##     Rank Title                               Distributor `Worldwide gross`  year
##    <int> <chr>                               <chr>       <chr>             <int>
##  1     1 Demon Slayer: Kimetsu no Yaiba Mug… Toho / Ani… $507,127,293[4]    2020
##  2     2 The Eight Hundred                   CMC Pictur… $461,421,559       2020
##  3     3 My People, My Homeland              China Lion  $433,241,288[5]    2020
##  4     4 Bad Boys for Life                   Sony        $426,505,244       2020
##  5     5 Tenet                               Warner Bro… $365,309,519       2020
##  6     6 Sonic the Hedgehog                  Paramount   $319,715,683       2020
##  7     7 Dolittle                            Universal   $251,410,631       2020
##  8     8 Jiang Ziya                          Beijing En… $243,883,429       2020
##  9     9 A Little Red Flower                 HG Enterta… $238,600,000[6][…  2020
## 10    10 Shock Wave 2                        Universe F… $226,400,000       2020
## # ℹ 50 more rows