Newer
Older
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
# due to unequal inner_dim ('time') length across file_dim ('sdate').
remove_additional_na_from_merge <- function(inner_dims_across_files, final_dims, across_inner_dim,
length_inner_across_dim, data_array) {
across_file_dim <- names(inner_dims_across_files) #TODO: more than one?
# Get the length of these two dimensions in final_dims
length_inner_across_store_dims <- final_dims[across_inner_dim]
length_file_across_store_dims <- final_dims[across_file_dim]
# Create a logical array for merge_across_dims
logi_array <- array(rep(FALSE,
length_file_across_store_dims * length_inner_across_store_dims),
dim = c(length_inner_across_store_dims, length_file_across_store_dims))
for (i in 1:length_file_across_store_dims) { #1:4
logi_array[1:length_inner_across_dim[[i]], i] <- TRUE
}
# First, get the data array with final_dims dimension
data_array_final_dims <- array(bigmemory::as.matrix(data_array), dim = final_dims)
# Change the NA derived from additional spaces to -9999, then remove these -9999
func_remove_blank <- function(data_array, logi_array) {
# dim(data_array) = [time, file_date]
# dim(logi_array) = [time, file_date]
# Change the blank spaces from NA to -9999
data_array[which(!logi_array)] <- -9999
return(data_array)
}
data_array_final_dims <- multiApply::Apply(data_array_final_dims,
target_dims = c(across_inner_dim, across_file_dim), #c('time', 'file_date')
output_dims = c(across_inner_dim, across_file_dim),
fun = func_remove_blank,
logi_array = logi_array)$output1
## reorder back to the correct dim
tmp <- match(names(final_dims), names(dim(data_array_final_dims)))
data_array_final_dims <- .aperm2(data_array_final_dims, tmp)
data_array_tmp <- data_array_final_dims[data_array_final_dims != -9999] # become a vector
return(data_array_tmp)
}
# When merge_across_dims = TRUE and split_multiselected_dims = TRUE, rearrange the chunks
# (i.e., work_piece) is necessary if one file contains values for discrete dimensions
rebuild_array_merge_split <- function(data_array_tmp, indices_chunk, all_split_dims,
final_dims_fake, across_inner_dim, length_inner_across_dim) {
# generate the correct order list from indices_chunk
final_order_list <- list()
i <- 1
j <- 1
a <- indices_chunk[i]
while (i <= length(indices_chunk)) {
while (indices_chunk[i+1] == indices_chunk[i] & i < length(indices_chunk)) {
a <- c(a, indices_chunk[i+1])
i <- i + 1
}
final_order_list[[j]] <- a
a <- indices_chunk[i+1]
i <- i + 1
j <- j + 1
}
names(final_order_list) <- sapply(final_order_list, '[[', 1)
final_order_list <- lapply(final_order_list, length)
if (!all(diff(as.numeric(names(final_order_list))) > 0)) {
# shape the vector into the array without split_dims
split_dims_pos <- match(all_split_dims[[1]], final_dims_fake)
new_dims <- c()
if (split_dims_pos[1] > 1) {
new_dims <- c(new_dims, final_dims_fake[1:(split_dims_pos[1] - 1)])
}
new_dims <- c(new_dims, prod(all_split_dims[[1]]))
names(new_dims)[split_dims_pos[1]] <- across_inner_dim
if (split_dims_pos[length(split_dims_pos)] < length(final_dims_fake)) {
new_dims <- c(new_dims, final_dims_fake[(split_dims_pos[length(split_dims_pos)] + 1):length(final_dims_fake)])
}
data_array_no_split <- array(data_array_tmp, dim = new_dims)
# seperate 'time' dim into each work_piece length
data_array_seperate <- list()
tmp <- cumsum(unlist(length_inner_across_dim))
tmp <- c(0, tmp)
for (i in 1:length(length_inner_across_dim)) {
data_array_seperate[[i]] <- Subset(data_array_no_split, across_inner_dim,
(tmp[i] + 1):tmp[i + 1])
}
# re-build the array: chunk
which_chunk <- as.numeric(names(final_order_list))
sort_which_chunk <- sort(unique(which_chunk))
which_chunk <- sapply(lapply(which_chunk, '==', sort_which_chunk), which)
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
how_many_indices <- unlist(final_order_list)
array_piece <- list()
ind_in_array_seperate <- as.list(rep(1, length(data_array_seperate)))
for (i in 1:length(final_order_list)) {
array_piece[[i]] <- Subset(data_array_seperate[[which_chunk[i]]],
across_inner_dim,
ind_in_array_seperate[[which_chunk[i]]]:(ind_in_array_seperate[[which_chunk[i]]] + how_many_indices[i] - 1))
ind_in_array_seperate[[which_chunk[i]]] <- ind_in_array_seperate[[which_chunk[i]]] + how_many_indices[i]
}
# re-build the array: paste
data_array_tmp <- array_piece[[1]]
along_pos <- which(names(dim(data_array_tmp)) == across_inner_dim)
if (length(array_piece) > 1) {
for (i in 2:length(array_piece)) {
data_array_tmp <- abind::abind(data_array_tmp, array_piece[[i]],
along = along_pos)
}
}
}
return(data_array_tmp)
}
# Create a list of metadata of the variable (e.g., tas)
create_metadata_list <- function(array_of_metadata_flags, metadata_dims, pattern_dims,
loaded_metadata_files, loaded_metadata, dat_names,
dataset_has_files) {
#NOTE: Here, metadata can be saved in one of two ways: one for $common and the other for $dat
# for $common, it is a list of metadata length. For $dat, it is a list of dat length,
# and each sublist has the metadata for each dat.
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
dim_of_metadata <- dim(array_of_metadata_flags)[metadata_dims]
if (!any(names(dim_of_metadata) == pattern_dims) |
(any(names(dim_of_metadata) == pattern_dims) &
dim_of_metadata[pattern_dims] == 1)) { # put under $common; old code
return_metadata <- vector('list',
length = prod(dim_of_metadata))
return_metadata[as.numeric(loaded_metadata_files)] <- loaded_metadata
dim(return_metadata) <- dim_of_metadata
} else { # put under $dat. metadata_dims has 'dat' and dat length > 1
return_metadata <- vector('list',
length = dim_of_metadata[pattern_dims])
names(return_metadata) <- dat_names
for (kk in 1:length(return_metadata)) {
return_metadata[[kk]] <- vector('list', length = prod(dim_of_metadata[-1])) # 1 is dat
}
loaded_metadata_count <- 1
for (kk in 1:length(return_metadata)) {
for (jj in 1:length(return_metadata[[kk]])) {
if (dataset_has_files[kk]) {
if (loaded_metadata_count %in% loaded_metadata_files) {
return_metadata[[kk]][jj] <- loaded_metadata[[which(loaded_metadata_files == loaded_metadata_count)]]
names(return_metadata[[kk]])[jj] <- names(loaded_metadata[[which(loaded_metadata_files == loaded_metadata_count)]])
} else {
return_metadata[[kk]][jj] <- NULL
loaded_metadata_count <- loaded_metadata_count + 1
} else {
return_metadata[[kk]][jj] <- NULL
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
return(return_metadata)
}
# This function adds the metadata of the variable (e.g., tas) into the list of picked_vars or
# picked_common_vars. The metadata is only retrieved when 'retrieve = TRUE'.
combine_metadata_picked_vars <- function(return_metadata, picked_vars, picked_common_vars,
metadata_dims, pattern_dims, length_dat) {
#NOTE: The metadata of variables can be saved in one of the two different structures.
# (1) metadata_dims != 'dat', or (metadata_dims == 'dat' & length(dat) == 1):
# put under $common
# (2) (metadata_dims == 'dat' & length(dat) > 1):
# put under $dat1, $dat2, .... Put it in picked_vars list
#TODO: The current (2) uses the inefficient method. Should define the list structure first
# then fill the list, rather than expand it in the for loop.
if (any(metadata_dims == pattern_dims) & length_dat > 1) { # (2)
for (kk in 1:length(return_metadata)) {
sublist_names <- lapply(return_metadata, names)[[kk]]
if (!is.null(sublist_names)) {
for (jj in 1:length(sublist_names)) {
picked_vars[[kk]][[sublist_names[jj]]] <- return_metadata[[kk]][[jj]]
}
}
}
Variables_list <- c(list(common = picked_common_vars), picked_vars)
} else { #(1)
len <- unlist(lapply(return_metadata, length))
len <- sum(len) + length(which(len == 0)) #0 means NULL
name_list <- lapply(return_metadata, names)
new_list <- vector('list', length = len)
count <- 1
for (kk in 1:length(return_metadata)) {
if (length(return_metadata[[kk]]) == 0) { #NULL
count <- count + 1
} else {
for (jj in 1:length(return_metadata[[kk]])) {
new_list[[count]] <- return_metadata[[kk]][[jj]]
names(new_list)[count] <- name_list[[kk]][jj]
count <- count + 1
}
}
}
Variables_list <- c(list(common = c(picked_common_vars, new_list)), picked_vars)
}
return(Variables_list)
}
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
# This function generates a list of 3, containing picked(_common)_vars,
# picked(_common)_vars_ordered, and picked(_common)_vars_unorder_indices for the 'var_to_read'
# of this dataset (i) and file (j).
generate_picked_var_of_read <- function(var_to_read, var_to_check, array_of_files_to_load,
var_dims, array_of_var_files, file_var_reader,
file_object, synonims, associated_dim_name,
dim_reorder_params, aiat, current_indices, var_params,
either_picked_vars,
either_picked_vars_ordered,
either_picked_vars_unorder_indices) {
var_file_dims <- NULL
if (any(names(dim(array_of_files_to_load)) %in% var_to_check)) {
var_file_dims <- dim(array_of_files_to_load)[which(names(dim(array_of_files_to_load)) %in%
var_to_check)]
}
if (is.null(either_picked_vars)) {
if (any(names(var_file_dims) %in% names(var_dims))) {
stop("Found a requested var in 'return_var' requested for a ",
"file dimension which also appears in the dimensions of ",
"the variable inside the file.\n", array_of_var_files)
}
first_sample <- file_var_reader(NULL, file_object, NULL,
var_to_read, synonims)
if (any(class(first_sample) %in% names(time_special_types()))) {
array_size <- prod(c(var_file_dims, var_dims))
new_array <- rep(time_special_types()[[class(first_sample)[1]]](NA), array_size)
dim(new_array) <- c(var_file_dims, var_dims)
} else {
new_array <- array(dim = c(var_file_dims, var_dims))
}
attr(new_array, 'variables') <- attr(first_sample, 'variables')
either_picked_vars <- new_array
pick_ordered <- FALSE
if (var_to_read %in% unlist(var_params)) {
if (associated_dim_name %in% names(dim_reorder_params) && !aiat) {
either_picked_vars_ordered <- new_array
pick_ordered <- TRUE
}
}
if (!pick_ordered) {
either_picked_vars_ordered <- NULL
}
} else {
array_var_dims <- dim(either_picked_vars)
full_array_var_dims <- array_var_dims
if (any(names(array_var_dims) %in% names(var_file_dims))) {
array_var_dims <- array_var_dims[-which(names(array_var_dims) %in% names(var_file_dims))]
}
if (names(array_var_dims) != names(var_dims)) {
stop("Error while reading the variable '", var_to_read, "' from ",
"the file. Dimensions do not match.\nExpected ",
paste(paste0("'", names(array_var_dims), "'"), collapse = ', '),
" but found ",
paste(paste0("'", names(var_dims), "'"), collapse = ', '),
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
}
if (any(var_dims > array_var_dims)) {
longer_dims <- which(var_dims > array_var_dims)
if (length(longer_dims) == 1) {
longer_dims_in_full_array <- longer_dims
if (any(names(full_array_var_dims) %in% names(var_file_dims))) {
candidates <- (1:length(full_array_var_dims))[-which(names(full_array_var_dims) %in% names(var_file_dims))]
longer_dims_in_full_array <- candidates[longer_dims]
}
padding_dims <- full_array_var_dims
padding_dims[longer_dims_in_full_array] <-
var_dims[longer_dims] - array_var_dims[longer_dims]
var_class <- class(either_picked_vars)
if (any(var_class %in% names(time_special_types()))) {
padding_size <- prod(padding_dims)
padding <- rep(time_special_types()[[var_class[1]]](NA), padding_size)
dim(padding) <- padding_dims
} else {
padding <- array(dim = padding_dims)
}
either_picked_vars <- .abind2(either_picked_vars, padding,
names(full_array_var_dims)[longer_dims_in_full_array])
} else {
stop("Error while reading the variable '", var_to_read, "' from ",
"the file. Found size (", paste(var_dims, collapse = ' x '),
") is greater than expected maximum size (", array_var_dims, ").")
}
}
}
var_store_indices <- c(as.list(current_indices[names(var_file_dims)]),
lapply(var_dims, function(x) 1:x))
var_values <- file_var_reader(NULL, file_object, NULL, var_to_read, synonims)
if (var_to_read %in% unlist(var_params)) {
if ((associated_dim_name %in% names(dim_reorder_params)) && !aiat) {
## Is this check really needed?
if (length(dim(var_values)) > 1) {
stop("Requested a '", associated_dim_name, "_reorder' for a dimension ",
"whose coordinate variable that has more than 1 dimension. This is ",
"not supported.")
}
ordered_var_values <- dim_reorder_params[[associated_dim_name]](var_values)
attr(ordered_var_values$x, 'variables') <- attr(var_values, 'variables')
if (!all(c('x', 'ix') %in% names(ordered_var_values))) {
stop("All the dimension reorder functions must return a list with the components 'x' and 'ix'.")
}
# Save the indices to reorder the ordered variable values back to original order.
# 'unorder' refers to the indices of 'ordered_var_values' if it is unordered.
# This will be used to define the first round indices.
unorder <- sort(ordered_var_values$ix, index.return = TRUE)$ix
either_picked_vars_ordered <- do.call('[<-',
c(list(x = either_picked_vars_ordered),
var_store_indices,
list(value = ordered_var_values$x)))
either_picked_vars_unorder_indices <- do.call('[<-',
c(list(x = either_picked_vars_unorder_indices),
var_store_indices,
list(value = unorder)))
}
}
either_picked_vars <- do.call('[<-',
c(list(x = either_picked_vars),
var_store_indices,
list(value = var_values)))
# Turn time zone back to UTC if this var_to_read is 'time'
if (all(class(either_picked_vars) == names(time_special_types))) {
attr(either_picked_vars, "tzone") <- 'UTC'
}
return(list(either_picked_vars = either_picked_vars,
either_picked_vars_ordered = either_picked_vars_ordered,
either_picked_vars_unorder_indices = either_picked_vars_unorder_indices))
}
# Trnasforms a vector of indices v expressed in a world of
# length N from 1 to N, into a world of length M, from
# 1 to M. Repeated adjacent indices are collapsed.
transform_indices <- function(v, n, m) {
#unique2 turns e.g. 1 1 2 2 2 3 3 1 1 1 into 1 2 3 1
unique2 <- function(v) {
if (length(v) < 2) {
v
} else {
v[c(1, v[2:length(v)] - v[1:(length(v) - 1)]) != 0]
}
}
unique2(round(((v - 1) / (n - 1)) * (m - 1))) + 1 # this rounding may generate 0s. what then?
}
replace_character_with_indices <- function(selectors, data_dims, chunk_amount) {
} else if (selectors == 'first') {
selectors <- indices(1)
} else if (selectors == 'last') {