FUN.TRY.R

############################################ FUNCTION TO EXTRACT DECTED OUTLIER AND FORMAT TRY DATA Georges Kunstler
############################################ 14/06/2013

library(MASS, quietly=TRUE)
library(doParallel, quietly=TRUE)
library(mvoutlier, quietly=TRUE)


######################################################## Build a function that extract the variables
##'Description of the function to extract data from original TRY data
##'
##' based on the data structure of extraction from TRY data base
##' @title fun.extract.try
##' @param ObservationID.t  list of data identifier that we want to extract
##' @param data try data object
##' @param Non.Trait.Data list of names of non traits data that we want to extract
##' @param Trait.Data list of names of traits data that we want to extract
##' @return data.frame with one line per observation id with clumns with ObservationID Species Nontrait data for Traits: OrigValue OrigUnit StdValue 
##' @author Kunstler 
fun.extract.try <- function(ObservationID.t, data, Non.Trait.Data, Trait.Data) {
    data.temp <- data[data$ObservationID == ObservationID.t, ]
    ## Non trait data
    Vec.Non.Trait.Data <- rep(NA, length(Non.Trait.Data))
    names(Vec.Non.Trait.Data) <- Non.Trait.Data
    
    for (i in 1:length(Non.Trait.Data)) {
        if (sum(data.temp$DataName == Non.Trait.Data[i]) == 1) {
            Vec.Non.Trait.Data[i] <- data.temp[data.temp$DataName == Non.Trait.Data[i], 
                "OrigValueStr"]
        }
        if (sum(data.temp$DataName == Non.Trait.Data[i]) > 1) {
            ## if(sum(data.temp$DataName==Non.Trait.Data[i] &
            ## grepl('Mean',data.temp$ValueKindName, fixed=TRUE))!=1){ print('error in
            ## ValueKindName')}
            Vec.Non.Trait.Data[i] <- data.temp[data.temp$DataName == Non.Trait.Data[i], 
                "OrigValueStr"][1]
        }
    }
    
    ## Trait data
    Vec.Trait.Data.OrigValue <- Vec.Trait.Data.OrigUnit <- Vec.Trait.Data.StdValue <- rep(NA, 
        length(Trait.Data))
    names(Vec.Trait.Data.OrigValue) <- paste("OrigValue", Trait.Data)
    names(Vec.Trait.Data.OrigUnit) <- paste("OrigUnitName", Trait.Data)
    names(Vec.Trait.Data.StdValue) <- paste("StdValue", Trait.Data)
    
    for (i in 1:length(Trait.Data)) {
        if (sum(grepl(Trait.Data[i], data.temp$TraitName, fixed = TRUE)) == 1) {
            Vec.Trait.Data.OrigValue[i] <- data.temp[grepl(Trait.Data[i], data.temp$TraitName, 
                fixed = TRUE), "OrigValue"]
            Vec.Trait.Data.OrigUnit[i] <- data.temp[grepl(Trait.Data[i], data.temp$TraitName, 
                fixed = TRUE), "OrigUnitStr"]
            Vec.Trait.Data.StdValue[i] <- data.temp[grepl(Trait.Data[i], data.temp$TraitName, 
                fixed = TRUE), "StdValue"]
        }
        
        if (sum(grepl(Trait.Data[i], data.temp$TraitName, fixed = TRUE)) > 1) {
            if (sum((data.temp$ValueKindName %in% c("Best estimate", "Mean", "Site specific mean") & 
                !is.na(data.temp$ValueKindName))) == 1) {
                Vec.Trait.Data.OrigValue[i] <- mean(data.temp[grepl(Trait.Data[i], 
                  data.temp$TraitName, fixed = TRUE) & (data.temp$ValueKindName %in% 
                  c("Best estimate", "Mean", "Site specific mean") & !is.na(data.temp$ValueKindName)), 
                  "OrigValue"])
                Vec.Trait.Data.OrigUnit[i] <- (data.temp[grepl(Trait.Data[i], data.temp$TraitName, 
                  fixed = TRUE) & (data.temp$ValueKindName %in% c("Best estimate", 
                  "Mean", "Site specific mean") & !is.na(data.temp$ValueKindName)), 
                  "OrigUnitStr"])[1]
                Vec.Trait.Data.StdValue[i] <- mean(data.temp[grepl(Trait.Data[i], 
                  data.temp$TraitName, fixed = TRUE) & (data.temp$ValueKindName %in% 
                  c("Best estimate", "Mean", "Site specific mean") & !is.na(data.temp$ValueKindName)),