## DATA SOURCE: the file named e41m.txt was downloaded 
## 2011-05-04 from http://enr.elections.ca/DownloadResults.aspx 
## (after clicking the 'latest results' link near the bottom of
## the left-hand column).  Some trailing blank lines were removed.
e <- read.csv("e41m.txt", sep="\t", encoding="latin1")
## shorten column names
colNames <- c("distNum", "distName", "distNameFR","type","typeFR","lname","mname","fname",
              "affil", "affilFR","votes","percentVotes","rejected","total")
names(e) <- colNames
## remove columns that we won't use
e <- e[, names(e) != "distNameFR"]
e <- e[, names(e) != "typeFR"]
e <- e[, names(e) != "affilFR"]
e <- e[, names(e) != "mname"]

## shorten relevant party names
levels(e$affil)[levels(e$affil) == "Bloc Québécois"] <- "Bloc"
levels(e$affil)[levels(e$affil) == "Conservative"] <- "Con"
levels(e$affil)[levels(e$affil) == "Green Party"] <- "Green"
levels(e$affil)[levels(e$affil) == "Liberal"] <- "Lib"
levels(e$affil)[levels(e$affil) == "NDP-New Democratic Party"] <- "NDP"

## categorize by district
f <- factor(e$distNum)
ee <- split(e, f)
district <- NULL
bloc <- NULL
green <- NULL
con <- NULL
lib <- NULL
ndp <- NULL
conWon <- NULL
conWouldWin <- NULL
nd <- length(ee)
for (i in 1:nd) {
    d <- ee[i][[1]]
    ## restrict to validated data, if we have them
    if (any(d$type == "validated"))
        d <- d[d$type == "validated",]
    affil <- as.character(d$affil)
    percentVotes <- d$percentVotes
    district <- c(district, as.character(d$distName[1]))
    thisbloc <- if (any(affil=="Bloc")) percentVotes[affil=="Bloc"] else 0
    thiscon <- if (any(affil=="Con")) percentVotes[affil=="Con"] else 0
    thisgreen <- if (any(affil=="Green")) percentVotes[affil=="Green"] else 0
    thislib <- if (any(affil=="Lib")) percentVotes[affil=="Lib"] else 0
    thisndp <- if (any(affil=="NDP")) percentVotes[affil=="NDP"] else 0
    con <- c(con, thiscon)
    ndp <- c(ndp, thisndp)
    lib <- c(lib, thislib)
    bloc <- c(bloc, thisbloc)
    green <- c(green, thisgreen)
    conWon <- c(conWon, thiscon > max(thisndp, thislib, thisbloc, thisgreen))
    conWouldWin <- c(conWouldWin, thiscon > sum(thisndp, thislib, thisbloc, thisgreen))
}
results <- data.frame(district, con, ndp, lib, bloc, green, conWon, conWouldWin)
results <- results[order(results$district), ]

png("split.png", width=8.5, height=11, unit="in", res=200, pointsize=7)
par(mfrow=c(1,1), mar=c(0,0,0,0),mgp=c(0,0,0))
plot(c(0,1), c(0,1), xlab="", ylab="", type='n')
jconwon <- 1
jconwonnot <- 1
jconwouldwin <- 1
jconwouldwinnot <- 1
yy <- 0.0051
cex <- 0.45
text(x=0.1, y=1.01, "Actual results", adj=0)
text(x=0.7, y=1.01, "If 'left' not split", adj=0)
text(x=0, y=1.005, "Conservatives won", adj=0, cex=2/3)
text(x=0.2, y=1.005, "Conservatives lost", adj=0, cex=2/3)
text(x=0.6, y=1.005, "Conservatives would win", adj=0, cex=2/3)
text(x=0.8, y=1.005, "Conservatives would lose", adj=0, cex=2/3)
for (i in 1:nd) {
    district <- iconv(as.character(results$district[i]), to="UTF8")
    if (results$conWon[i]) {
        text(x=0, y=1-yy*jconwon, district, cex=cex, adj=0)
        jconwon <- jconwon + 1
    }
    if (!results$conWon[i]) {
        text(x=0.2, y=1-yy*jconwonnot, district, cex=cex, adj=0)
        jconwonnot <- jconwonnot + 1
    }
    if (results$conWouldWin[i]) {
        text(x=0.6, y=1-yy*jconwouldwin, district, cex=cex, adj=0)
        jconwouldwin <- jconwouldwin + 1
    }
    if (!results$conWouldWin[i]) {
        text(x=0.8, y=1-yy*jconwouldwinnot, district, cex=cex, adj=0)
        jconwouldwinnot <- jconwouldwinnot + 1
    }
}
dev.off()