extract.R

library(tools)

# General information on station
iStatut = c('0'='inconnu', 
            '1'='station avec signification hydrologique', 
            '2'='station sans signification hydrologique', 
            '3'="station d'essai")

iFinalite = c('0'='inconnue', 
              '1'="hydrométrie générale", 
              '2'='alerte de crue', 
              '3'="hydrométrie générale et alerte de crue",
              '4'="gestion d'ouvrage", 
              '5'='police des eaux', 
              '6'="suivi d'étiage", 
              '7'='bassin expérimental', 
              '8'='drainage')

iType = c('0'='inconnu',
          '1'='une échelle',
          '2'='deux échelles, station mère',
          '3'='deux échelles, station fille',
          '4'='débits mesurés',
          '5'='virtuelle')

iInfluence = c('0'='inconnue',
               '1'='nulle ou faible',
               '2'='en étiage seulement',
               '3'='forte en toute saison')

iDebit = c('0'='reconstitué',
           '1'="réel (prise en compte de l'eau rajoutée ou retirée du bassin selon aménagements)",
           '2'='naturel')

iQBE = c('0'='qualité basses eaux inconnue',
         '1'='qualité basses eaux bonne',
         '2'='qualité basses eaux douteuse')

iQME = c('0'='qualité moyennes eaux inconnue',
         '1'='qualité moyennes eaux bonne',
         '2'='qualité moyennes eaux douteuse')

iQHE = c('0'='qualité hautes eaux inconnue',
         '1'='qualité hautes eaux bonne',
         '2'='qualité hautes eaux douteuse')


# Extraction of information
extract_info = function (data_path, filedir, filename) {

    filename = c(filename)
    
    if (all(filename == 'all') | length(filename) > 1) {
        if (all(filename == 'all')) {
            filelist = c()
            filelist_tmp = list.files(file.path(data_path,
                                                filedir))
            for (f in filelist_tmp) {
                if (file_ext(f) == 'txt') {
                    filelist = c(filelist, f) 
                }
            }
        } else if (length(filename > 1)) {
            filelist = filename
        } 
        df_info = data.frame()
        for (f in filelist) {
            df_info = rbind(df_info,
                            extract_info(data_path, 
                                         filedir, 
                                         f)
                            )
        }
        rownames(df_info) = NULL
        return (df_info)
    }

    filename = filename[1]    

    file_path = file.path(data_path, filedir, filename)
    infotxt = c(readLines(file_path, n=41))
    
    info = list(code=trimws(substr(infotxt[11], 38, nchar(infotxt[11]))),
                nom=trimws(substr(infotxt[12], 39, nchar(infotxt[12]))),
                territoire=trimws(substr(infotxt[13], 39, nchar(infotxt[13]))),
                L93X=as.numeric(substr(infotxt[16], 38, 50)),
                L93Y=as.numeric(substr(infotxt[16], 52, 63)),
                surface=as.numeric(substr(infotxt[19], 38, 50)),
                statut=iStatut[trimws(substr(infotxt[26], 38, 50))],
                finalite=iFinalite[trimws(substr(infotxt[26], 52, 56))],
                type=iType[trimws(substr(infotxt[26], 58, 58))],
                influence=iInfluence[trimws(substr(infotxt[26], 60, 60))],
                debit=iDebit[trimws(substr(infotxt[26], 62, 62))],
                QBE=iQBE[trimws(substr(infotxt[26], 72, 72))],
                QME=iQME[trimws(substr(infotxt[26], 74, 74))],
                QHE=iQHE[trimws(substr(infotxt[26], 76, 76))]
                )
    
    df_info = data.frame(info, file_path=file_path)
    
    return (df_info)
}

# df_info = extract_info(
#     "/home/louis/Documents/bouleau/INRAE/CDD_stationnarite/data",
#     '',
#     c('H5920011_HYDRO_QJM.txt', 'K4470010_HYDRO_QJM.txt'))


# Extraction of data
extract_data = function (data_path, filedir, filename) {
    
    filename = c(filename)

    if (all(filename == 'all') | length(filename) > 1) {
        if (all(filename == 'all')) {
            filelist = c()
            filelist_tmp = list.files(file.path(data_path,
                                                filedir))
            for (f in filelist_tmp) {
                if (file_ext(f) == 'txt') {
                    filelist = c(filelist, f) 
                }
            }
        } else if (length(filename > 1)) {
            filelist = filename
        } 
        df_data = data.frame()
        for (f in filelist) {
            df_data = rbind(df_data,
                            extract_data(data_path, 
                                         filedir, 
                                         f)
                            )
        }
        rownames(df_data) = NULL
        return (df_data)
    }

    filename = filename[1]

    file_path = file.path(data_path, filedir, filename)
    
    df_data = read.table(file_path,
                         header=TRUE,
                         na.strings=c('     -99'),
                         sep=';',
                         skip=41)[,1:2]
    
    df_info = extract_info(data_path, filedir, filename)
    code = df_info$code
    
    df_data = cbind(df_data, code=code)

    return (df_data)
}

# df_data = extract_data(
    # "/home/louis/Documents/bouleau/INRAE/CDD_stationnarite/data",
    # '',
    # c('H5920011_HYDRO_QJM.txt', 'K4470010_HYDRO_QJM.txt'))