library(RPostgreSQL)
library(RQuantLib)
if(.Platform$OS.type == "unix"){
  root.dir <- "/home/share/CorpCDOs"
}else{
  root.dir <- "//WDSENTINEL/share/CorpCDOs"
}

source(file.path(root.dir, "code", "R", "etdb.R"))
source(file.path(root.dir, "code", "R", "yieldcurve.R"))
source(file.path(root.dir, "code", "R", "cds_utils.R"))
source(file.path(root.dir, "code", "R", "intex_deal_functions.R"))
source(file.path(root.dir, "code", "R", "optimization.R"))

workdate <- as.Date("2013-02-11")

MarkitData <- getMarkitIRData(workdate)
L1m <- buildMarkitYC(MarkitData, dt = 1/12)
L2m <- buildMarkitYC(MarkitData, dt = 1/6)
L3m <-  buildMarkitYC(MarkitData)
L6m <- buildMarkitYC(MarkitData, dt = 1/2)
L12m <- buildMarkitYC(MarkitData, dt = 1)
setEvaluationDate(as.Date(MarkitData$effectiveasof))

sanitize.column <- function(vec){
  vec <- gsub(",", "", vec)
  index <- grep("\\(", vec)
  for(l in index){
    vec[l] <- -as.numeric(substr(vec[l], 2, nchar(vec[l])-1))
  }
  return( as.numeric(vec) )
}

allfiles <- list.files(file.path(root.dir, "Scenarios", paste0("Prices_", workdate)), "*.txt")
allfiles <- unique(sapply(strsplit(allfiles, "-"), function(x) x[1]))
allfiles <- allfiles[!(allfiles=="Total")]
dealnames <- list.files(file.path(root.dir, "Scenarios", paste0("Prices_", workdate)), "*COLLAT_INITIAL-CF-Scen1*")
dealnames <- unique(sapply(strsplit(dealnames, "-"), function(x) x[1]))
cusips <- setdiff(allfiles, dealnames)
dealnames <- tolower(dealnames)

getdealcf <- function(dealnames, workdate=Sys.Date()){
    cfdata <- list()
    tranches <- c("COLLAT_REINVEST", "COLLAT_INITIAL")
    fields <-  c("Cashflow", "Principal", "Interest")
    flag <- FALSE
    n.scenarios <- 100
    for(dealname in dealnames){
        cfdata[[dealname]] <- list()
        r <- matrix(0, n.scenarios, 3)
        colnames(r) <- fields
        sqlstring <- sprintf("select marketvalue from latest_deal_model_numbers where dealname='%s'", dealname)
        mv <- dbGetQuery(dbCon, sqlstring)$marketvalue
        sqlstring <- sprintf("select \"Curr Collat Bal\" from latest_clo_universe where dealname='%s'", dealname)
        currbal <- dbGetQuery(dbCon, sqlstring)$"Curr Collat Bal"
        cfdata[[dealname]]$mv <- mv
        cfdata[[dealname]]$currbal <- currbal
        for(tranche in tranches){
            for(i in 1:n.scenarios){
                filename <- paste0(paste(toupper(dealname), tranche, "CF", paste0("Scen", i), sep="-"), ".txt")
                ## we catch the error if there is an error reading the file
                ## happen if the tranche is missing in intex
                data <- tryCatch(
                        read.table(file.path(root.dir, "Scenarios", paste0("Prices_", workdate), filename),
                                   sep="\t", header=F, skip=3, colClasses="character", comment.char=""),
                        error = function(e) e)
                if(inherits(data, "error")){
                    cat(sprintf("file: %s, tranche: %s can't be loaded", dealname, tranche), "\n")
                    flag <- TRUE
                    break
                }
                data <- data[,1:4]
                colnames(data) <- c("Date", "Cashflow", "Principal", "Interest")
                data$Date <- as.Date(data$Date, "%b %d, %Y")
                if(any(is.na(data$Date))){
                    cat(sprintf("file: %s is messed up", filename), "\n")
                    flag <- TRUE
                    break
                }
                futuredates <- data$Date[data$Date>=L3m$params$tradeDate]
                pastdates <- data$Date[data$Date<L3m$params$tradeDate]
                if(i==1||length(futuredates)>length(DC$times)){
                    DC <- DiscountCurve(L3m$params, L3m$tsQuotes, yearFrac(L3m$params$tradeDate, futuredates))
                }
                pv <- c()
                for(field in fields){
                    cleanfield <- tryCatch(sanitize.column(data[,field]),
                                           warning = function(w) w)
                    if(inherits(cleanfield, "warning")){
                        cat("garbled", dealname, i, "\n")
                        flag <- TRUE
                        break
                    }else{
                        data[,field] <- cleanfield
                    }
                    if(length(futuredates) == 0){
                        df <- rep(1, length(pastdates))
                    }else{
                        df <- c(rep(1, length(pastdates)), DC$discounts[1:length(futuredates)])
                    }
                    if(nrow(data)>0){
                        pv <- c(pv, crossprod(df, data[,field]))
                    }else{
                        pv <- c(pv, 0)
                    }
                }
                if(flag){
                    break
                }else{
                    r[i,] <- pv
                }
            }
            if(flag){
                cfdata[[dealname]] <- NULL
                flag <- FALSE
                break
            }else{
                cfdata[[dealname]][[tranche]]<- r
            }
        }
        if(is.null(cfdata[[dealname]])){
            next
        }
        cf <- cfdata[[dealname]][["COLLAT_REINVEST"]][,"Cashflow"] +
            cfdata[[dealname]][["COLLAT_INITIAL"]][,"Cashflow"]
        cfdata[[dealname]]$wapbasis <- (mean(cf)- cfdata[[dealname]]$mv)/cfdata[[dealname]]$mv
        cat(dealname, "\n")
        program <- KLfit(t(cf)/1e8, rep(1/n.scenarios, n.scenarios),
                         cfdata[[dealname]]$mv/1e8)
        cfdata[[dealname]]$weight <- program$weight
    }
    return( cfdata )
}

getcusipcf <- function(cusips, cfdata, workdate=Sys.Date()){
    flag <- FALSE
    cusipdata <- list()
    dealnames <- dealnamefromcusip(cusips)
    n.scenarios <- 100
    intexfields <-  c("Cashflow", "Principal", "Interest", "Balance")
    fields <-  c("Cashflow", "Principal", "Interest")
    for(i in 1:length(cusips)){
        cusip <- cusips[i]
        dealdata <- getdealdata(dealnames[i])
        schedule <- getdealschedule(dealdata)
        r <- matrix(0, n.scenarios, 5)
        colnames(r) <- c(fields, "wal", "duration")
        sqlstring <- sprintf("select curr_balance, spread from cusip_universe where cusip = '%s'", cusip)
        indicdata <- dbGetQuery(dbCon, sqlstring)
        cusipdata[[cusip]]$currbal <- indicdata$curr_balance
        cusipdata[[cusip]]$spread <- indicdata$spread
        for(j in 1:n.scenarios){
            filename <- sprintf("%s-CF-Scen%s.txt", cusip, j)
            if(!file.exists(file.path(root.dir, "Scenarios", paste0("Prices_", workdate), filename))){
                next
            }
            data <- read.table(file.path(root.dir, "Scenarios", paste0("Prices_", workdate), filename),
                               sep = "\t", header=F, colClasses="character", skip = 3, comment.char="")
            data <- data[, 1:5]
            colnames(data) <- c("Date", intexfields)
            data$Date <- as.Date(data$Date, "%b %d, %Y")
            if(any(is.na(data$Date))){
                cat(sprintf("file: %s is messed up", filename), "\n")
                flag <- TRUE
                break
            }
            futuredates <- data$Date[data$Date >= L3m$params$tradeDate]
            pastdates <- data$Date[data$Date < L3m$params$tradeDate]
            if(i==1||length(futuredates)>length(DC$times)){
                DC <- DiscountCurve(L3m$params, L3m$tsQuotes, yearFrac(L3m$params$tradeDate, futuredates))
            }
            if(length(futuredates) == 0){
                df <- rep(1, length(pastdates))
            }else{
                df <- c(rep(1, length(pastdates)), DC$discounts[1:length(futuredates)])
            }
            pv <- c()
            for(field in fields){
                cleanfield <- tryCatch(sanitize.column(data[,field]),
                                       warning = function(w)w)
                if(inherits(cleanfield, "warning")||any(is.na(cleanfield))){
                    cat(sprintf("file: %s is messed up", filename), "\n")
                    flag <- TRUE
                    break
                }else{
                    data[,field] <- cleanfield
                }
                if(length(futuredates) == 0){
                    df <- rep(1, length(pastdates))
                }else{
                    df <- c(rep(1, length(pastdates)), DC$discounts[1:length(futuredates)])
                }
                if(nrow(data)>0){
                    pv <- c(pv, crossprod(df, data[,field]))
                }else{
                    pv <- c(pv, 0)
                }
            }
            if(flag){
                break
            }else{
                data[,"Balance" ] <- tryCatch(sanitize.column(data[,"Balance"]),
                                              warning = function(w){cat("garbled", dealname, i)})
                wal <- crossprod(yearFrac(workdate, data$Date)[-1], -diff(data[,"Balance"]))/
                    indicdata$curr_balance
                if(pv[1] == 0){
                    duration <- 0
                }else{
                    duration <- crossprod(data[,"Cashflow"], df * yearFrac(workdate, data$Date))/pv[1]
                }
                tryCatch(r[j,] <- c(pv, wal, duration),
                         error =function(e) browser())
            }
        }

        if(flag){
            cusipdata[[cusip]] <- NULL
            flag <- FALSE
        }else{
            cat("done", cusip, "\n")
            cusipdata[[cusip]]$fields <- r
            cusipdata[[cusip]]$price <- 100 * crossprod(cfdata[[dealnames[i]]]$weight,
                                                        cusipdata[[cusip]]$fields[,"Cashflow"])/cusipdata[[cusip]]$currbal
            cusipdata[[cusip]]$wal <- crossprod(cfdata[[dealnames[i]]]$weight,
                                                cusipdata[[cusip]]$fields[,"wal"])
            cusipdata[[cusip]]$duration <- crossprod(cfdata[[dealnames[i]]]$weight,
                                                     cusipdata[[cusip]]$fields[,"duration"])
        }
    }
    return(cusipdata)
}

cfdata <- getdealcf(dealnames, workdate)
availablecusips <- cusips[dealnamefromcusip(cusips) %in% names(cfdata)]
cusipdata <- getcusipcf(availablecusips, cfdata, workdate)
## upload wapbasis
for(dealname in names(cfdata)){
    sqlstring <- sprintf(paste0("UPDATE et_deal_model_numbers SET ",
                   "wapbasis = '%s' WHERE dealname= '%s' AND updatedate = '%s'"),
                         cfdata[[dealname]]$wapbasis*100,
                         dealname,
                         strftime(workdate))
    dbSendQuery(dbCon, sqlstring)
}

## upload model data
for(cusip in names(cusipdata)){
    columns <- c("Cusip", "price", "wal", "duration", "updatedate")
    values <- c(cusip, cusipdata[[cusip]]$price, cusipdata[[cusip]]$wal,
                cusipdata[[cusip]]$duration, strftime(workdate))
    colstring <- paste(columns[!is.na(values)], collapse=",")
    valstring <-  paste(values[!is.na(values)], collapse="','")
    sqlstring <- sprintf(paste0("INSERT INTO et_cusip_model_numbers",
                                "(%s) VALUES('%s')"),
                         colstring, valstring)
    dbSendQuery(dbCon, sqlstring)
}