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

#get the tables from the url
theurl <- getURL("https://en.wikipedia.org/wiki/Quebec_general_election,_2014", ssl.verifyPeer=FALSE)
table.1 <- htmltab(theurl, which=12)[, c(2, 4:10)]
table.2 <- htmltab(theurl, which=13)[, c(2, 4:10)]
names(table.2) <- names(table.1)
df <- rbind(table.1[-nrow(table.1), ], table.2)
names(df)[1] <- "Date"
party.colours <- c("#9999ff", "#dd0000", "#9900cc", "#ff8800", "#666666")

#format numerical and date data
df[is.na(df)] <- 0
df$Date <- as.Date(substr(df$Date, 9, 18))
for (i in c(2:8)) {
  df[[i]] = as.numeric(df[[i]])/100
}
df$Other <- df$Other + df$ON + df$GPQ
df <- df[, -c(6:7)]

#reshape data to have candidate and support as variable
mdata <- melt(df[-c(1, nrow(df)), ], id=c("Date"))
names(mdata)[2:3] <- c("Party", "Support")

election.2012 <- melt(df[nrow(df), ], id=c("Date"))
names(election.2012)[2:3] <- c("Party", "Support")

election.2014 <- melt(df[1, ], id=c("Date"))
names(election.2014)[2:3] <- c("Party", "Support")

#make plot
d <- ggplot(mdata, aes(x=Date, y=Support, colour=Party))
d <- d + geom_vline(xintercept=as.numeric(as.Date("2014-03-05")), linetype=2, color="#666666", show.legend=F)
d <- d + geom_point(alpha=0.5)
d <- d + geom_smooth(span=0.45, size=0.8, alpha=0.3, se=TRUE, show.legend=FALSE)
d <- d + scale_colour_manual(values=party.colours)
d <- d + labs(title="Opinion polls for the 2014 Quebec general election")
d <- d + scale_y_continuous(breaks=seq(0,1,0.05),
                            minor_breaks=seq(0,1,0.01),
                            labels=percent,
                            limits=c(0, 0.45))
d <- d + scale_x_date(labels=c("2012 Election",
                               format(seq(as.Date("2012-10-01"), as.Date("2014-02-01"), by="month"),
                                      format="%b %Y"),
                               "Campaign",
                               "2014 Election"),
                      breaks=c(as.Date("2012-09-04"),
                               seq(as.Date("2012-10-01"), as.Date("2014-02-01"), by="month"),
                               as.Date("2014-03-05"), as.Date("2014-04-07")),
                      minor_breaks=NULL)

d <- d + geom_point(data=election.2014, shape=5, size=5, show.legend=FALSE)
d <- d + geom_point(data=election.2014, size=2.5, show.legend=TRUE)
d <- d + geom_text(data=election.2014, show.legend=F,
                   aes(label=sprintf("%.1f", election.2014$Support*100)),
                   size=3.5, nudge_x=11, hjust=0, color="#000000")

d <- d + geom_point(data=election.2012, shape=5, size=5, show.legend=FALSE)
d <- d + geom_point(data=election.2012, size=2.5, show.legend=FALSE)
d <- d + geom_text(data=election.2012, show.legend=F,
                   aes(label=sprintf("%.1f", election.2012$Support*100)),
                   size=3.5, nudge_x=11, nudge_y=c(0.005,-0.005,0,0,0), hjust=0, color="#000000")

d <- d + theme(panel.grid.minor=element_line(size=0.2),
               panel.grid.major=element_line(size=0.6))
d <- d + theme(axis.text.x=element_text(angle=45, hjust=1))

#save plot as "qc2014.svg"
svg(filename="qc2014.svg", 
    width=9, 
    height=6, 
    pointsize=12,
    bg="transparent")
d
dev.off()