res <- read.csv('2001.csv', as.is=TRUE, header=TRUE, na.string="")
res <- res[!is.na(res$Candidate),]

filldown <- function(d) {
  while(TRUE) {
    x <- which(is.na(d))
    if (length(x)==0) break
    dx <- c(2,diff(x))
    d[x[dx>1]] <- d[x[dx>1]-1]
    }
  d
  }

reg <- read.csv('regions.csv', as.is=TRUE, header=TRUE, na.string="")
reg$Region <- filldown(reg$Region)

const <- res[!is.na(res$Constituency),
             c("Constituency","Electorate","Turnout","Status","Declaration")]
const$Turnout <- as.numeric(substr(const$Turnout,1,nchar(const$Turnout)-1))/100
const$Turnout <- round(const$Turnout*const$Electorate)

cand <- res[,c("Constituency","Candidate","Party","Votes")]
cand[] <- lapply(cand, filldown)

cand$Party[cand$Party=="Labour/Co-operative"] <- "Labour"
cand$Party[cand$Party=="Liberal Democrat"] <- "LibDem"

row.names(reg) <- tolower(reg$Constituency)
const$Region <- reg[tolower(const$Constituency),'Region']
const$Region <- factor(const$Region,
                       levels=c('Southeast','Southwest','East Anglia','East Midlands','West Midlands',
                         'Yorkshire and Humberside','North','Northwest','London',
                         'Scotland','N.Ireland','Wales'))

row.names(const) <- const$Constituency

## Restrict to big3

parties.int <- c('Labour','Conservative','LibDem')
candparties <- cand$Party
candparties[!(candparties %in% parties.int)] <- "Other"
cand.int <- aggregate(cand$Votes,
                      by=list(Party=candparties,Constituency=cand$Constituency),
                      FUN=max)
names(cand.int)[3] <- "Votes"
cand.int$Constituency <- as.character(cand.int$Constituency)
const.int <- reshape(cand.int, direction='wide',
                     idvar='Constituency', timevar='Party'
                     )
const.int[] <- lapply(const.int, function(d) ifelse(is.na(d),0,d))
const.int$Region <- const[const.int$Constituency,'Region']

# Martin Baxter's tables

baxterregions <- c('East Scotland'='Scotland', 'West Scotland'='Scotland',
                   'N.Ireland'='N.Ireland',
                   'North'='North', 'Lancashire'='Northwest','Greater Manchester'='Northwest',
                   'Yorkshire'='Yorkshire and Humberside','Humberside'='Yorkshire and Humberside',
                   'West Midlands'='West Midlands','East Midlands'='East Midlands',
                   'Severn'='West Midlands',
                   'Wales'='Wales',
                   'East Anglia'='East Anglia',
                   'Essex'='Southeast','West'='Southwest',
                   'North London'='London', 'South London'='London',
                   'South West'='Southwest','South'='Southeast','South East'='Southeast')



# 2001, actual boundaries
df <- scan('conlist_flatfile01.txt', what=character(0), sep="\n")
df <- matrix(df, byrow=TRUE, ncol=10)
df <- data.frame(df)
names(df) <- c("Constituency","Candidate","Region","Electorate",
               "Votes.Conservative","Votes.Labour","Votes.LibDem","Votes.Nationalist","Votes.Minority","OtherVotes")
df[] <- lapply(df, as.character)
df$Constituency <- substr(df$Constituency, start=nchar("Name:")+1,  stop=nchar(df$Constituency))
df$Candidate <-    substr(df$Candidate,    start=nchar("MP:")+1,    stop=nchar(df$Candidate))
df$Region <-       substr(df$Region,       start=nchar("Region:")+1,stop=nchar(df$Region))
votecols <- c('Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Nationalist','Votes.Minority','OtherVotes')
df[,votecols] <- lapply(df[,votecols], function(x) as.numeric(substr(x,start=nchar("CON:")+1,stop=nchar(x))))
df$Votes.Other <- pmax(df$Votes.Nationalist,df$Votes.Minority)
df <- df[,c('Constituency','Region','Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Other')]
ni <- const.int[const.int$Region=='N.Ireland',]
ni$Region <- 3
df <- rbind(df,ni[,names(df)])
df$Region <- factor(df$Region, levels=1:20, labels=names(baxterregions))
df$Region <- factor(baxterregions[df$Region], levels=levels(const.int$Region))
baxter2001 <- df

# 1997
df <- scan('conlist_flatfile97.txt', what=character(0), sep="\n")
df <- matrix(df, byrow=TRUE, ncol=10)
df <- data.frame(df)
names(df) <- c("Constituency","Candidate","Region","Electorate",
               "Votes.Conservative","Votes.Labour","Votes.LibDem","Votes.Nationalist","Votes.Minority","OtherVotes")
df[] <- lapply(df, as.character)
df$Constituency <- substr(df$Constituency, start=nchar("Name:")+1,  stop=nchar(df$Constituency))
df$Candidate <-    substr(df$Candidate,    start=nchar("MP:")+1,    stop=nchar(df$Candidate))
df$Region <-       substr(df$Region,       start=nchar("Region:")+1,stop=nchar(df$Region))
votecols <- c('Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Nationalist','Votes.Minority','OtherVotes')
df[,votecols] <- lapply(df[,votecols], function(x) as.numeric(substr(x,start=nchar("CON:")+1,stop=nchar(x))))
df$Votes.Other <- pmax(df$Votes.Nationalist,df$Votes.Minority)
df <- df[,c('Constituency','Region','Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Other')]
ni <- const.int[const.int$Region=='N.Ireland',]
ni$Region <- 3
df <- rbind(df,ni[,names(df)])
df$Region <- factor(df$Region, levels=1:20, labels=names(baxterregions))
df$Region <- factor(baxterregions[df$Region], levels=levels(const.int$Region))
baxter1997 <- df

#1992, notional boundaries
df <- scan('conlist_flatfile92.txt', what=character(0), sep="\n")
df <- matrix(df, byrow=TRUE, ncol=10)
df <- data.frame(df)
names(df) <- c("Constituency","Candidate","Region","Electorate",
               "Votes.Conservative","Votes.Labour","Votes.LibDem","Votes.Nationalist","Votes.Minority","OtherVotes")
df[] <- lapply(df, as.character)
df$Constituency <- substr(df$Constituency, start=nchar("Name:")+1,  stop=nchar(df$Constituency))
df$Candidate <-    substr(df$Candidate,    start=nchar("MP:")+1,    stop=nchar(df$Candidate))
df$Region <-       substr(df$Region,       start=nchar("Region:")+1,stop=nchar(df$Region))
votecols <- c('Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Nationalist','Votes.Minority','OtherVotes')
df[,votecols] <- lapply(df[,votecols], function(x) as.numeric(substr(x,start=nchar("CON:")+1,stop=nchar(x))))
df$Votes.Other <- pmax(df$Votes.Nationalist,df$Votes.Minority)
df <- df[,c('Constituency','Region','Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Other')]
ni <- const.int[const.int$Region=='N.Ireland',]
ni$Region <- 3
df <- rbind(df,ni[,names(df)])
df$Region <- factor(df$Region, levels=1:20, labels=names(baxterregions))
df$Region <- factor(baxterregions[df$Region], levels=levels(const.int$Region))
baxter1992nb <- df


## Read BBC data scraped by DJW

regions <- c('South West', 'South East',
             'London',
             'Eastern', 'East Midlands', 'West Midlands',
             'North West','Yorks & Humber','North East',
             'Scotland', 'Wales', 'Northern Ireland')
partytrans = c('Liberal Democrat'='LibDem')

const <- read.csv('data/2005const.csv', header=TRUE, as.is=TRUE)
turnout <- read.csv('data/2005turnout.csv', header=TRUE, as.is=TRUE)
cand <- read.csv('data/2005cand.csv', header=TRUE, as.is=TRUE)
# Some minor renaming/refactoring
const$Region <- factor(const$Region, levels=regions)
partynames <- partytrans[cand$Party]
cand$Party <- ifelse(!is.na(partynames),partynames,cand$Party)
# Reshape to have a wide table, one row per constituency, columns for interesting parties
parties.int <- c('Labour','Conservative','LibDem')
candparties <- cand$Party
candparties[!(candparties %in% parties.int)] <- "Other"
cand.int <- aggregate(cand$Votes,
                      by=list(Party=candparties,Constituency=cand$Constituency),
                      FUN=max)
names(cand.int)[3] <- "Votes"
cand.int$Constituency <- as.character(cand.int$Constituency)
const.int <- reshape(cand.int, direction='wide',
                     idvar='Constituency', timevar='Party'
                     )
const.int[] <- lapply(const.int, function(d) ifelse(is.na(d),0,d))
# Add in columns for region and turnout
const.int <- merge(const.int, turnout, all=TRUE)
const.int <- merge(const.int, const, all=TRUE)
# Save it
bbc2005 <- const.int
bbc2005$Constituency[bbc2005$Constituency=="Glasgow North East (Speaker)"] <- "Glasgow North East"
row.names(bbc2005) <- bbc2005$Constituency

# 2001, notional boundaries
df <- scan('conlist_flatfile01nb.txt', what=character(0), sep="\n")
df <- matrix(df, byrow=TRUE, ncol=10)
df <- data.frame(df)
names(df) <- c("Constituency","Candidate","Region","Electorate",
               "Votes.Conservative","Votes.Labour","Votes.LibDem","Votes.Nationalist","Votes.Minority","OtherVotes")
df[] <- lapply(df, as.character)
df$Constituency <- substr(df$Constituency, start=nchar("Name:")+1,  stop=nchar(df$Constituency))
df$Candidate <-    substr(df$Candidate,    start=nchar("MP:")+1,    stop=nchar(df$Candidate))
df$Region <-       substr(df$Region,       start=nchar("Region:")+1,stop=nchar(df$Region))
votecols <- c('Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Nationalist','Votes.Minority','OtherVotes')
df[,votecols] <- lapply(df[,votecols], function(x) as.numeric(substr(x,start=nchar("CON:")+1,stop=nchar(x))))
df$Votes.Other <- pmax(df$Votes.Nationalist,df$Votes.Minority)
df <- df[,c('Constituency','Region','Votes.Conservative','Votes.Labour','Votes.LibDem','Votes.Other')]
baxter2001nb <- df
row.names(baxter2001nb) <- baxter2001nb$Constituency

#ni <- const.int[const.int$Region=='N.Ireland',]
#ni$Region <- 3
#df <- rbind(df,ni[,names(df)])
#df$Region <- factor(df$Region, levels=1:20, labels=names(baxterregions))
#df$Region <- factor(baxterregions[df$Region], levels=levels(const.int$Region))