Buddy Cop Movies - Part 1

library(rvest)

page <- read_html("https://en.wikipedia.org/wiki/Buddy_cop_film")
amFilms <- page %>% html_nodes("ul:nth-child(13) i a")
movTitle <- amFilms %>% html_text # Western Films
wikiLink <- paste0("https://en.wikipedia.org",amFilms %>% html_attr("href"))
df.mov <- data.frame(title=movTitle, wiki.link=wikiLink, stringsAsFactors=F)

# Need to remove sequels - though buddies may change in sequels (Die Hard), I'm going to say that they dont
df.mov <- df.mov[-which(df.mov$title == "Bad Boys II"),]
df.mov <- df.mov[-which(df.mov$title == "22 Jump Street"),]
df.mov <- df.mov[-which(df.mov$title == "Another 48 Hrs."),]
df.mov <- df.mov[-which(df.mov$title == "Ride Along 2"),]
df.mov <- df.mov[-which(df.mov$title == "Die Hard with a Vengeance"),]
df.mov <- df.mov[-which(df.mov$title == "Live Free or Die Hard"),]

# put in buddies
df.mov$buddy.1 <- NA
df.mov$buddy.2 <- NA
base.node <- "#mw-content-text > table.infobox.vevent > tr"
for (k in 1:nrow(df.mov)) {
  page <- read_html(df.mov$wiki.link[k])
  star.node <- grep("Starring\n", page %>% html_nodes(base.node) %>% html_text) # Get node to create further
  buddies <- page %>% html_node(paste0(base.node, ":nth-child(", star.node, ")")) %>% html_nodes("a") %>% html_text # Extract buddies
  df.mov$buddy.1[k] <- buddies[1]
  df.mov$buddy.2[k] <- buddies[2]
  Sys.sleep(0.5)
}

# Fix a few movies
df.mov$buddy.2[df.mov$title == "The Other Guys"] <- "Will Ferrel"
df.mov$buddy.2[df.mov$title == "Skiptrace"] <- "Jimmy Fallon"