library(RCurl)
library(reshape)
library(htmltab)
library(ggplot2)
library(stringr)
library(scales)

#get the table from the url
theurl <- getURL("https://en.wikipedia.org/wiki/Nationwide_opinion_polling_for_the_United_States_presidential_election,_2016", ssl.verifyPeer=FALSE)
table <- htmltab(theurl, which=8)

table = table[, c(1:8)]

names(table) <- c("Size", "Date", "DC", "DP", "RC", "RP", "LC", "LP")
table$LC = sub(" \\(.*\\)", "", table$LC)

df = table[which(table$DC=="Hillary Clinton" &
                   table$RC=="Donald Trump" &
                   table$LC=="Gary Johnson"), c(1,2,4,6,8)]

df = df[1:(nrow(df)-1),] #remove oldest poll because it is too far from the others

names(df)[3:5] <- c("Clinton", "Trump", "Johnson")

for (i in 3:5) {
  df[[i]] = as.numeric(sub("%", "", df[[i]]))/100
}

df$Date = sub("[0-9]+\\s*–\\s*([0-9]+)", "\\1", df$Date)
df$Date = sub(".*–", "", df$Date)
df$Date = sub("[0-9]+\\s*-\\s*([0-9]+)", "\\1", df$Date)
df$Date = sub(".*-", "", df$Date)
df$Date = trimws(df$Date)
df$Date = as.Date(df$Date, format="%B %d, %Y")

df$Size = sub(".*Sample size: ([0-9,]+).*", "\\1", df$Size)
df$Size = as.numeric(sub(",", "", df$Size))

df$Error = 1/sqrt(df$Size)
mdata <- melt(df, id=c("Date", "Error", "Size"))

names(mdata)[4:5] = c("Candidate", "Support")

colors = c("#3333FF", "#FF3333", "#FED105")
labels = c("Clinton", "Trump", "Johnson")

results = mdata

#breaks() returns n evenly spaced numbers between x and y
#whose squares are divisible by p
#the function is used for the legend
breaks <- function(x, y, n, p) {
  x = sqrt(ceiling(as.integer(x^2) / p) * p)
  y = sqrt(floor(as.integer(y^2) / p) * p)
  s = seq(x, y, length.out=n)
  for (i in 2:(n-1)) {
    s[i] = sqrt(round(s[i]^2 / p) * p)
  }
  return(unique(s))
}

d = ggplot(results, aes(x=Date, y=Support, colour=Candidate, size=1/Error, weight=1/Error)) +
  geom_point(alpha=0.7) + 
  geom_smooth(span=0.8, show.legend=F) + 
  scale_colour_manual(values = colors) + 
  labs(title=expression(atop("Nationwide opinion polling for the 2016 U.S. presidential election", "Three-way race"))) + 
  scale_size_area(max_size=3,
                  breaks=function(x) breaks(x[1], x[2], 5, 250), #5 numbers divisible by 250
                  labels=function(x) comma_format()(x^2),
                  name="Sample Size") + 
  scale_y_continuous(breaks=seq(0,1,0.1), minor_breaks=seq(0,1,0.02), labels=percent,
                     limits=c(0, 0.5)) + 
  scale_x_date(labels=date_format("%b %d"),
               breaks=sort(c(seq(as.Date("2016/1/1"), as.Date("2016/10/1"), "week"),
                             as.Date("2016/11/8"))),
               minor_breaks=date_breaks("days")) +
  theme(panel.grid.minor=element_line(size=0.2),
        panel.grid.major=element_line(size=0.6))

#save plot as "ct.svg"
svg(filename="us2016-threeway.svg", 
    width=9, 
    height=5, 
    pointsize=12,
    bg="transparent")
d
dev.off()