# File: calrollcall.R
#
# Purpose: Subset large CA rollcall voting matrices using user defined
#          selection criteria.
#
# Author: Jeff Lewis (jblewis@ucla.edu)
#
# Change log:
#
#   August 2010: Initial version
#

# Function: getRollCallsByIndex(filen,votes)
#
# Purpose: Extract from the full matrix of rollcalls a subset of votes
#          by vote number (id column in the "desc" file
# 
# Arguments:
#       filen = Name of a CA vote matrix dataset (eg "ca07-08votes.dat")
#
#       votes = Vector of vote ids to extract
#
getRollCallsByIndex <- function(filen,votes=1:10) {
   v <- scan(filen,what="character",sep="\n")
   n <- length(v)
   k <- length(votes)
   dat <- matrix(NA,n,k)
   name <- character(n)

   # Loop over members and extract selected votes
   for (i in 1:n) {
     name[i] <- sub("^\\s+","",substr(v[i],1,20),perl=TRUE)
     dat[i,] <- strsplit(v[i],split='')[[1]][20+votes] 
   }   
   dat <- as.data.frame( cbind(name,dat) )   
   names(dat) <- c('member',sprintf("V%i",votes))

   # Remove members who cast no votes in the selected subset of rollcalls
   nonzero <- apply(dat[,-1],1,function(x) sum(x != 0))
   dat[nonzero>0,]
}

# Function: Select votes
# 
# Purpose: Extracts vote from the CA vote description file according to a
#          given list of criteria
#
# Arguments:
#        filen = Name of a CA vote description dataset (eg "ca07-08desc.dat")
#
#        criteria = A list of criteria to be "and"ed together to select
#                   a subset of votes from the full dataset. The keys in the list
#                   identify fields. The values in the list are the criteria to
#                   be applied to the values from the corresponding filed (list key)
#
#        ...      = Other arguements to be passed to "grep" function in applying selection
#                   criteria
selectVotes <- function(filen,criteria = list(),...) {
  cnames <- c("id","bill","author","topic","date","location","motion",
              "yeas","noes","passfail")
  dat <- read.csv(filen,sep="\t",col.names=cnames, as.is=TRUE)
  cat(sprintf("%s has %i records...\n",filen, dim(dat)[1]))
  if (length(criteria)==0) {
    return(dat)
  }
  for (i in 1:length(criteria)) {
    value = criteria[[i]]
    field = names(criteria[i])
    cat(sprintf("\tLimit to \"%s\" in \"%s\"...\n", value, field))
    dat <- dat[ grep(value,dat[,field], ...), ]
    cat(sprintf("\t\t%i records remaining.\n", dim(dat)[1]))
  }
  dat 
}


#
# A little example, requires data from ca07-08.zip to be
# be located in the current working directory
# 
example <- function() {
  # Get all Asm Floor votes
  cat("Get Assembly Floor votes...\n")
  crit <- list("location"="ASM. FLOOR")
  desc <- selectVotes("ca07-08desc.dat",criteria=crit)
  print(table(desc$location))
  dat  <- getRollCallsByIndex("ca07-08votes.dat",desc$id)
  print( dim(dat) )

  # Get all Sen Floor votes from June of 2008
  cat("Get Sen Floor votes from March of 2008...\n")
  crit <- list("location"="SEN\\. FLOOR", date="06/../2008")
  desc <- selectVotes("ca07-08desc.dat",criteria=crit, perl=TRUE)
  print(table(desc$location))
  print(table(desc$date))
  dat  <- getRollCallsByIndex("ca07-08votes.dat",desc$id)
  print( dim(dat) )
  return( list(desc=desc,votedata=dat) )
}

# Uncomment next line to run the example...
# res <- example()