# Show Missouri registered voters with census population.

# Franklin Center for Government & Public Integrity
# Earl F Glynn
# 28 July 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/Missouri/Census-Registration-Comparison/")

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

# data known globally
plot.missouri.registration <- function(county)
{
  county.name <- county
  # Remove "County" from census county name to match registration data,
  # except for St Louis County (there is also a St. Louis city)
  if (county.name != "St. Louis County")  # has a period when invoked here
  {
    county.name <- gsub(" County","", county.name)
  }

  # Remove periods in all "St." and "Ste." names for match to work.
  county.name <- gsub("\\.", "", county.name)

  # Missouri breaks Jackson County into Kansas City and the rest of Jackson County.
  # Since the US Census does not do this, let's aggregate Missouri voter data to
  # compare with Census.
  if (county.name == "Jackson")
  {
    mtext("(Includes Kansas City election district)", line=-1, col="blue")
  }

  # Online data
  points(dates.total,  voters.total[county.name,],  pch=3, cex=1.25, col="red")
  points(dates.active, voters.active[county.name,], pch=4, cex=1.25, col="green")

  # Voter file data
  points(voter.file.dates.total,  voter.file.total[county.name,],  pch=1, col="red")
  points(voter.file.dates.active, voter.file.active[county.name,], pch=2, col="green")

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

  legend("bottom", c("Online Reported Total Registration",
                     "Voter File Total Registration",
                     "Online Reported 'Active' Voters",
                     "Voter File 'Active' Voters"),
       title="Registered Voters",
       pch=c(3, 1, 4, 2),
       text.col=c("red", "red", "green", "green"),
       col=c("red","red","green","green"), box.lty="blank")
}

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

# Data from Missouri Secretary of State's web site
filename <- paste("F:/Voter-Registration/Missouri/",
                  "Census-Registration-Comparison/SOS-Data/",
                  "Missouri-County-Voter-Counts-2000-2010-Stats-Active.csv", sep="")
voters.active <- read.csv(filename, row.names=1)

filename <- paste("F:/Voter-Registration/Missouri/",
                  "Census-Registration-Comparison/SOS-Data/",
                  "Missouri-County-Voter-Counts-2000-2010-Stats-Total.csv", sep="")
voters.total <- read.csv(filename, row.names=1)

# FIXUP to add Kansas City's data to Jackson County
voters.active["Jackson",] <- voters.active["Jackson",] + voters.active["Kansas City",]
voters.total["Jackson",]  <- voters.total["Jackson",]  + voters.total["Kansas City",]

# Change column headings Dyyyy.mm.dd to dates.
dates.active <- as.Date(gsub("\\.","-",gsub("D","", colnames(voters.active))))
dates.total <-  as.Date(gsub("\\.","-",gsub("D","", colnames(voters.total ))))

stopifnot( all(dates.active == dates.total) )

# Voter File Data
filename <- paste("F:/Voter-Registration/Missouri/",
                  "Census-Registration-Comparison/Voter-Files/",
                  "Missouri-County-Voter-Counts-2004-2012-Files-Active.csv", sep="")
voter.file.active <- read.csv(filename, as.is=TRUE, row.names=1)

filename <- paste("F:/Voter-Registration/Missouri/",
                  "Census-Registration-Comparison/Voter-Files/",
                  "Missouri-County-Voter-Counts-2004-2012-Files-Total.csv", sep="")
voter.file.total  <- read.csv(filename, as.is=TRUE, row.names=1)

# FIXUP to add Kansas City's data to Jackson County
voter.file.active["Jackson",] <- voter.file.active["Jackson",] +
                                 voter.file.active["Kansas City",]
voter.file.total["Jackson",]  <- voter.file.total["Jackson",]  +
                                 voter.file.total["Kansas City",]

# Change column headings Dyyyy.mm.dd to dates.
voter.file.dates.active <- as.Date(gsub("\\.","-",gsub("D","", colnames(voter.file.active))))
voter.file.dates.total <-  as.Date(gsub("\\.","-",gsub("D","", colnames(voter.file.total ))))

stopifnot( all(voter.file.dates.active == voter.file.dates.total) )

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

# Missouri FIPS code is "29"
d <- get.voting.age.census.data("29")
cat(d$state, "\n")  # verify we have the correct state

# IGNORE this usual check since voter data has "Kansas City" district that
# is not in usual U.S. Census data for counties.
# Verify County names from census match county names of voter counts
# stopifnot( all( gsub(" County","",d$counties) == rownames(voters.active) ) )

# 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.missouri.registration)
dev.off()