# Show Virginia registered voters with census population.

# Franklin Center for Government & Public Integrity
# Earl F Glynn
# 28 July 2012.  Update 20 Aug 2012.

################################################################################
### Setup

# Common functions to read/plot voting age census data by county
source("F:/US-Census/Voting-Age-Population/Census-Voting-Age-Population.R")

setwd("F:/Voter-Registration/Virginia/census-registration-comparison/")

# CapLeading attributed to Christian Hoffmann in R mailing list
CapLeading <- function(string)
{
  fn <- function(x)
  {
    v <- unlist(strsplit(x, split = " "))
    u <- sapply(v, function(x){
            x <- tolower(x)
            substring(x, 1, 1) <- toupper(substring(x, 1, 1))
            x})
    paste(u, collapse = " ")
 }
 sapply(string, fn)
}

################################################################################
### Virginia county plotting callback function to plot voter registration

# data known globally
plot.virginia.registration <- function(county)
{
  county.name <- CapLeading(county)
  if (county.name == "Total") county.name <- "TOTAL"

  # CapLeading messed this one up
  if (county.name == "Isle Of Wight County")  county.name <- "Isle of Wight County"
  if (county.name == "King And Queen County") county.name <- "King and Queen County"

  county.data <- subset(registered, registered$Locality == county.name)

  points(county.data$Date, county.data$Total,  pch=3, cex=1.25, col="red")
  points(county.data$Date, county.data$Active, pch=4, cex=1.25, col="green")

  points(voter.file.date, voter.file[county.name, "Total"], pch=16, col="black")
  points(voter.file.date, voter.file[county.name, 1], pch=16, col="black")

  text(voter.file.date, voter.file[county.name, "Total"] - strheight("X"),
       "voter file",
       adj=0.5, col="black")
  text(voter.file.date, voter.file[county.name, "Active"] - strheight("X"),
       "voter file",
       adj=0.5, col="black")

  mtext("  Sources:  U.S. Census; Virginia Board of Elections",
        BOTTOM<-1, adj=0, line=-1, cex=0.75, col="blue", outer=TRUE)

  mtext("2012-08-23", BOTTOM<-1, adj=0.98, line=-2, cex=0.5,
          col="grey", outer=TRUE)

  legend("bottom", c("Total Registration",
                     "'Active' Voters"),
       title="Registered Voters",
       pch=3:4, cex=1.25,
       text.col=c("red", "green"),
       col=c("red","green"), box.lty="blank")
}

################################################################################
### Read Virginia Registered Voter data
### Remember:  Need to have rowname of total line "TOTAL" for use in callback

# County voter counts from Virginia Secretary of State
registered <- read.csv("Virginia-Voter-Registration-2004-2012.csv", as.is=TRUE)
registered$Date <- as.Date(registered$Date)

# Voter file
voter.file <- read.csv("Virginia-Active-Inactive-2012-04-16.csv", as.is=TRUE,
                       row.names=1)
rownames(voter.file) <- CapLeading(rownames(voter.file))

# Kludge fixes:
# Fix things CapLeading messed up
rownames(voter.file)[ 64] <- "Isle of Wight County"
rownames(voter.file)[135] <- "TOTAL"  # Keep this upper case

# Get rid of ampersand in name
rownames(voter.file)[66] <- "King and Queen County"

voter.file.date <- as.Date("2012-04-16")

################################################################################
### Plot Virginia census and voter registration data by county.

# Virginia FIPS code is "51"
d <- get.voting.age.census.data("51")
cat(d$state, "\n")  # verify we have the correct state

# Verify County names from census match county names of voter counts
#stopifnot( all( gsub(" County","",d$counties) == rownames(registered) ) )   ####

# Write census data file
# Add actual dates instead of only year index before writing file
d$census.data$Date <- d$timeline[d$census.data$Year]

write.csv(d$census.data[,c("State", "County", "Year", "Date", "Population", "VotingAge")],
          paste(gsub(" ","-",d$state),
          "-Voting-Age-Population.csv", sep=""), row.names=FALSE)

# Plot all counties in PDF file
pdf(paste(gsub(" ","-",d$state),  # Change blanks to dashes in state names
          "-Voting-Age-Population-and-Registered-Voters.pdf", sep=""), width=8, height=8)
  plot.county.population.charts(d$state, d$counties, d$census.data, d$timeline,
                                plot.virginia.registration)
dev.off()