cat("\n","Enter textfile name","\n") # prompt
y<-readLines(n=1) 

x <- paste(y,".txt", sep="")
d <- read.csv(x)[,c(1,2,3)]

colnames(d) <- c("rating", "day.of.week", "day.in.review")
d$day.in.review <- d$day.in.review == "True"

d$day.of.week <- as.character(d$day.of.week) 
d$day.of.week[d$day.of.week == "0"] <- "Sunday"
d$day.of.week[d$day.of.week == "1"] <- "Monday"
d$day.of.week[d$day.of.week == "2"] <- "Tuesday"
d$day.of.week[d$day.of.week == "3"] <- "Wednesday"
d$day.of.week[d$day.of.week == "4"] <- "Thursday"
d$day.of.week[d$day.of.week == "5"] <- "Friday"
d$day.of.week[d$day.of.week == "6"] <- "Saturday"
d$day.of.week <- factor(d$day.of.week, levels = c("Monday","Tuesday","Wednesday",
                                         "Thursday","Friday","Saturday","Sunday"))

d$day.of.week <- as.factor(d$day.of.week) 

#t <- table(d$day.of.week, as.factor(d$rating))
#t2 <- prop.table(t,2)
chisq.test(d$day.of.week, as.factor(d$rating))

# compute and add a set of weights so that we can graph proportions
# using the same data.frame
w <- data.frame(1 / (table(d$day.of.week)/sum(table(d$day.of.week))))
colnames(w) <- c("day.of.week", "plot.weight")
d <- merge(d, w, by="day.of.week", all.x=TRUE, all.y=FALSE)

# draw two graphs
library(ggplot2)

# generate the unscaled count graph
filename.count <- paste(y,"-count.png",sep="")

png(filename.count, width=8, height=6, unit="in", res=200)
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar", main=y)
p <- p + scale_x_discrete("Day of Week") +
  scale_y_continuous("yelp # star review") +
  scale_fill_discrete("", breaks=as.character(rev(1:5)) )
print(p)
dev.off()

# generate the scaled proportion graph
filename.prop <- paste(y,"-proportion.png",sep="")

max.value <- tapply(d$plot.weight, d$day.of.week, sum)[1]
png(filename.prop, width=8,height=6, unit="in", res=200)
p <- qplot(day.of.week, data=d, fill=as.factor(rating), geom="bar",
           weight=plot.weight, main=y)
p <- p + scale_x_discrete("Day of Week") +
  scale_y_continuous("percentage of reviews",
                     breaks=(0:4 * max.value/4), labels=seq(0,100, 25)) +
  scale_fill_discrete("", breaks=as.character(rev(1:5)))
print(p)
dev.off()