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=3)

table = table[, c(1, 2, 8, 3:6)]

names(table) <- c("Source", "Date", "Size", "DC", "DP", "RC", "RP")

df = table[which(table$DC=="Bernie Sanders" & table$RC=="Donald Trump"), c(1:3, 5, 7)]

names(df)[4:5] <- c("Sanders", "Trump")

for (i in 4:5) {
  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$Size = as.numeric(sub(",", "", df$Size))

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

names(mdata)[5:6] = c("Candidate", "Support")

colors = c("#3333FF", "#FF3333")
labels = c("Sanders", "Trump")

results = mdata

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", "Sanders vs. Trump"))) + 
  scale_size_area(max_size=3,
                  labels=function(x) x^2,
                  name="Sample Size") + 
  scale_y_continuous(breaks=seq(0,1,0.05), minor_breaks=seq(0,1,0.01), labels=percent,
                     limits=c(0.35, 0.6)) + 
  scale_x_date(labels=date_format("%b %d"),
               breaks=sort(c(seq(as.Date("2016/1/1"), as.Date("2016/10/1"), "month"),
                             as.Date("2016/11/8"))),
               limits=c(as.Date("2016/1/1"), as.Date("2016/11/8"))) +
  theme(panel.grid.minor=element_line(size=0.2),
        panel.grid.major=element_line(size=0.6))

#save plot as "st.svg"
svg(filename="st.svg", 
    width=9, 
    height=4, 
    pointsize=12,
    bg="transparent")
d
dev.off()