1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
|
#############################################################################
##
## Copyright 2016 Novartis Institutes for BioMedical Research Inc.
## Licensed under the Apache License, Version 2.0 (the "License");
## you may not use this file except in compliance with the License.
## You may obtain a copy of the License at
##
## http://www.apache.org/licenses/LICENSE-2.0
##
## Unless required by applicable law or agreed to in writing, software
## distributed under the License is distributed on an "AS IS" BASIS,
## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
## See the License for the specific language governing permissions and
## limitations under the License.
##
#############################################################################
###################################################
## UI for operating on groups and files
###################################################
##' Get the names of the items in the group or at the \code{/} root of the file
##'
##' Works similar to the regular \code{names} function for a list. The names of the items of either a
##' \code{\link{H5File}} at the root or a \code{\link{H5Group}} are returned as a character vector.
##' The items are then accessed, again similar to a list, using \code{[[}.
##' @title Get the names of the items in the group or at the \code{/} root of the file
##' @param x An object of class \code{\link{H5File}} or \code{\link{H5Group}}
##' @return A character vector with the names of the items in the group/file.
##' @author Holger Hoefling
##' @export
names.H5Group <- function(x) {
link_access_pl <- h5const$H5P_DEFAULT
ginfo <- x$group_info()
nlinks <- ginfo$nlinks
res <- character(nlinks)
for(i in seq_len(nlinks)) {
res[i] <- x$link_name_by_idx(i-1, ".", idx_type=h5const$H5_INDEX_NAME, order=h5const$H5_ITER_INC,
link_access_pl=link_access_pl)
}
return(res)
}
##' @rdname names.H5Group
##' @export
names.H5File <- names.H5Group
##' Retrieve object from a group of file
##'
##' Works similar to retrieving objects in a list. \code{x[["my_name"]]} retrieves object \code{my_name} from the
##' HDF5-File or HDF5-Group \code{x}.
##'
##' One can also assign objects under a not yet existing name. For either a \code{\link{H5Group}} or \code{\link{H5D}},
##' a hard link is created. If it is a datatype, \code{\link{H5T}}, this is committed under the chosen name \code{name}.
##' @title Retrieve object from a group of file
##' @param x An object of class \code{\link{H5File}} or \code{\link{H5Group}}
##' @param name Name of the object to retrieve. Has to be a character vector of length one. No integer values allowed.
##' @param ... Currently ignored
##' @param link_access_pl An object of class \code{\link{H5P_LINK_ACCESS}}.
##' @param dataset_access_pl An object of class \code{\link{H5P_DATASET_ACCESS}}.
##' @param type_access_pl Currently always \code{h5const$H5P_DEFAULT}
##' @param value What to assign. Has to be one of \code{\link{H5Group}}, \code{\link{H5D}} or \code{\link{H5T}}
##' @return A \code{\link{H5Group}}, \code{\link{H5D}} or \code{\link{H5T}}, depending on the object saved in the group under
##' the requested name.
##' @author Holger Hoefling
##' @rdname H5Group_access
##' @name H5Group_access
##' @export
'[[.H5Group' <- function(x, name, ..., link_access_pl=h5const$H5P_DEFAULT, dataset_access_pl=h5const$H5P_DEFAULT, type_access_pl=h5const$H5P_DEFAULT) {
if(x$exists(name)) {
return(x$open(name=name, link_access_pl=link_access_pl, dataset_access_pl=dataset_access_pl, type_access_pl=type_access_pl))
}
else {
stop(paste("An object with name", name, "does not exist in this group"))
}
}
##' @rdname H5Group_access
##' @export
'[[.H5File' <- get("[[.H5Group")
##' @rdname H5Group_access
##' @export
'[[<-.H5Group' <- function(x, name, ..., value) {
## if a group is given, make a link
## if a dataset is given, make a link
## if another r-obj is given, create a dataset for that object and write it
## if the object already exists, just fail
if(x$exists(name)) {
## it is a peculiarity of the way R works that when assigning attributes,
## as it is done by value, the resulting object is written over itself (with the attribute attached)
## so here, we need to check if the object being assigned is itself
if(inherits(value, "H5File") || inherits(value, "H5Group") || inherits(value, "H5D") || inherits(value, "H5T")) {
value_oinfo <- value$obj_info()
x_oinfo <- x[[name]]$obj_info()
if(identical(value_oinfo, x_oinfo)) {
return(invisible(x))
}
}
stop("Cannot assign - already exists. Please use the 'link_delete' to delete the object before assigning a new one")
}
if(inherits(value, "H5Group")) {
x$link_create_hard(value, ".", name)
}
else if(inherits(value, "H5D")) {
x$link(value, name)
}
else if(inherits(value, "H5T")) {
x$commit(name, value)
}
else {
x$create_dataset(name, value)
}
return(invisible(x))
}
##' @rdname H5Group_access
##' @export
'[[<-.H5File' <- get("[[<-.H5Group")
###################################################
## UI for working with arguments
###################################################
##' Interface for HDF5 attributes
##'
##' Implements high-level functions that allows interactions with HDF5-attributes in a way very similar to regular R-object attributes
##' in R are handled.
##' @title Interface for HDF5 attributes
##' @param x The object to which to attach the attribute to or retrieve it from. Can be one of \code{\link{H5Group}}, \code{\link{H5D}},
##' \code{\link{H5T}} or \code{\link{H5File}}
##' @param which The name of the attribute to assign it to
##' @param value The value to assign to the attribute.
##' @return For \code{h5attributes}, a named list with the content of the attributes read out. For \code{h5attr_names},
##' a vector of names of the attributes attached to the object \code{x}. For \code{h5attr}, the content of the attribute and
##' for \code{h5attr<-}, the assignment, the original object to which the attributes are attached (so that chaining is possible).
##' @author Holger Hoefling
##' @export
h5attributes <- function(x) {
## return a list with the names and content of all attributes
## first get number of attributes; used deprecated attr_get_number;
## reason is that H5Oget_info can be slow on large datasets (but should be ok on files)
if(inherits(x, "H5File")) {
num_attrs <- x$obj_info()$num_attrs
}
else {
num_attrs <- x$attr_get_number()
}
attr_data <- vector("list", length=num_attrs)
names(attr_data) <- h5attr_names(x)
for(i in seq_along(attr_data)) {
attr_obj <- x$attr_open_by_idx(i - 1, ".")
attr_data[[i]] <- attr_obj$read()
}
return(attr_data)
}
##' @rdname h5attributes
##' @export
h5attr_names <- function(x) {
## get the number of attributes
## first get number of attributes; used deprecated attr_get_number;
## reason is that H5Oget_info can be slow on large datasets (but should be ok on files)
if(inherits(x, "H5File")) {
num_attrs <- x$obj_info()$num_attrs
}
else {
num_attrs <- x$attr_get_number()
}
attr_names <- character(num_attrs)
for(i in seq_along(attr_names)) {
attr_names[i] <- x$attr_name_by_idx(i - 1, ".")
}
return(attr_names)
}
##' @rdname h5attributes
##' @export
h5attr <- function(x, which) {
if(!x$attr_exists_by_name(which, ".")) {
stop("Attribute does not exist")
}
attr_obj <- x$attr_open_by_name(which, ".")
return(attr_obj$read())
}
##' @rdname h5attributes
##' @export
'h5attr<-' <- function(x, which, value) {
if(x$attr_exists_by_name(which, ".")) {
## if it already exists, delete it first
## to make sure that unused object get closed
gc()
x$attr_delete_by_name(which, ".")
}
attr_obj <- x$create_attr(which, robj=value)
return(invisible(x))
}
###################################################
## UI for working with datasets
###################################################
##' Selecting and assigning subsets of HDF5-Spaces and HDF5-Datasets
##'
##' Used for subsetting HDF5-Datasets or HDF5-Spaces or for assigning data into HDF5-Datasets. There are some differences to
##' consider with R itself.
##'
##' Most importantly HDF5-COMPOUND objects only have a single dimension internally to HDF5 (a vector), but they correspond to R-data.frames,
##' which are 2 dimensional. For an HDF5 COMPOUND object, it is currently not possible to only sub-select a specific column.
##' All columns have to be extracted (using 1-dimensional access with \code{[} and can then be subset in R itself.
##' The same is true for writing a COMPOUND object (\code{\link{H5T_COMPOUND}}). A complete data-frame
##' is needed, not just a subset of the columns.
##'
##' Another important differences is for datasets of HDF5-ARRAY type \code{\link{H5T_ARRAY}}
##' where the access to the object is only for the dimension of the object itself, not including the dimensions of the underlying array type.
##' @title Selecting and assigning subsets of HDF5-Spaces and HDF5-Datasets
##' @param x The \code{\link{H5S}} or \code{\link{H5D}} to subset or assign values to
##' @param d1 First dimension of the object
##' @param ... Used for other dimension of the object
##' @param op Operation to perform on the \code{\link{H5S}}. Look into the HDF5 online help
##' \url{https://portal.hdfgroup.org/display/HDF5/H5S_SELECT_ELEMENTS} and
##' \url{https://portal.hdfgroup.org/display/HDF5/H5S_SELECT_HYPERSLAB}
##' @param dataset_xfer_pl An object of class \code{\link{H5P_DATASET_XFER}}.
##' @param flags Some flags governing edge cases of conversion from HDF5 to R. This is related to how integers are being treated and
##' the issue of R not being able to natively represent 64bit integers and not at all being able to represent unsigned 64bit integers
##' (even using add-on packages). The constants governing this are part of \code{\link{h5const}}. The relevant ones start with the term
##' \code{H5TOR} and are documented there. The default set here returns a regular 32bit integer if it doesn't lead to an overflow
##' and returns a 64bit integer from the \code{bit64} package otherwise. For 64bit unsigned int that are larger than 64bit signed int,
##' it return a \code{double}. This looses precision, however. See also documentation or \code{\link{h5const}}.
##' @param value The value to assign to the dataset
##' @param drop Logical. When reading data, should dimensions of size 1 be dropped.
##' @param envir The environment in which the dimension indices \code{d1, ...} are to be evaluated. Usually the environment from
##' where the function is called
##' @return For \code{x} being a \code{\link{H5S}}, the same object is returned, but with the selection set as requested. For
##' \code{\link{H5D}} it retrieves the subset of data requested or sets the subset of data assigned, as for any n-dimensional array
##' in R.
##' @author Holger Hoefling
##' @name H5S_H5D_subset_assign
NULL
#' @rdname H5S_H5D_subset_assign
#' @export
subset_h5.H5S <- function(x,d1, ..., op=h5const$H5S_SELECT_SET, envir=parent.frame()) {
args <- eval(substitute(alist(d1, ...)))
return(x$subset(args=args, op=op, envir=envir))
}
#' @rdname H5S_H5D_subset_assign
#' @export
'[.H5S' <- subset_h5.H5S
#' @rdname H5S_H5D_subset_assign
#' @export
subset_h5.H5D <- function(x, d1, ..., dataset_xfer_pl=h5const$H5P_DEFAULT,
flags=getOption("hdf5r.h5tor_default"), drop=TRUE, envir=parent.frame()) {
args <- eval(substitute(alist(d1, ...)))
return(x$read(args=args, dataset_xfer_pl=dataset_xfer_pl, flags=flags, drop=drop, envir=envir))
}
#' @rdname H5S_H5D_subset_assign
#' @export
"[.H5D" <- subset_h5.H5D
#' @rdname H5S_H5D_subset_assign
#' @export
subset_assign_h5.H5D <- function(x, d1, ..., dataset_xfer_pl=h5const$H5P_DEFAULT, envir=parent.frame(), value) {
args <- eval(substitute(alist(d1, ...)))
return(x$write(args=args, value=value, dataset_xfer_pl=dataset_xfer_pl, envir=envir))
}
#' @rdname H5S_H5D_subset_assign
#' @export
"[<-.H5D" <- subset_assign_h5.H5D
###############################################################
# Helper functions for evaluating arguments
###############################################################
##' Apply a selection to a space
##'
##' Calls the respective stand-alone functions for point-selection or multiple hyperslab selection. The reason for not
##' calling a method of an R6 object is to make it more efficient and make it useable without creating a full R6 object.
##' @title Apply a selection to a space
##' @param space_id The space_id of the space to which to apply the selection to
##' @param selection The selection object of class \code{point_selection} or \code{hyperslab_selection}
##' @return NULL
##' @author Holger Hoefling
##' @keywords internal
apply_selection <- function(space_id, selection) {
op <- h5const$H5S_SELECT_SET
if(inherits(selection, "point_selection")) {
standalone_H5S_select_elements(id=space_id, coord=selection, op=h5const$H5S_SELECT_SET, byrow=TRUE)
}
else if(inherits(selection, "hyperslab_selection")) {
standalone_H5S_select_multiple_hyperslab(id=space_id, hyperslab_array=selection)
}
else {
stop("Unknown selection type; needs to be of class point_selection of hyperslab_selection")
}
return(invisible(NULL))
}
##' Can arguments be interpreted as a scalar?
##'
##' Check if there is only one argument and if it is either empty
##' of of length 1 with value 1, i.e. can be interpreted as a scalar.
##' @title Can arguments be interpreted as a scalar?
##' @param args The arguments to check
##' @return Logical if the arguments can be interpreted as a scalar
##' @author Holger Hoefling
##' @keywords internal
are_args_scalar <- function(args) {
if(length(args) != 1) {
return(FALSE)
}
if(args[[1]]==bquote(expr=) || args[[1]] == 1) {
return(TRUE)
}
else {
return(FALSE)
}
}
##' Check argument for known functions that encode a hyperslab
##'
##' Checks for the functions \code{:}, \code{seq} and \code{seq_len}
##' @title Check argument for known functions that encode a hyperslab
##' @param cur_arg The argument to check
##' @param envir The environment in which to evaluate the argument
##' @return A vector of length 4 describing start, count, stride and block if appropriate
##' @author Holger Hoefling
##' @keywords internal
check_arg_for_hyperslab_func <- function(x, envir) {
res <- c(NA, NA, NA, NA)
if(is.call(x)) {
if(length(x) > 1) {
for(i in 2:length(x)) {
x[[i]] <- eval(x[[i]], envir=envir)
}
}
if(x[[1]] == ":") {
if(length(x) == 3) {
if(x[[3]] < x[[2]]) {
stop("Retrieving hyperslabs in reverse order not supported")
}
start <- as.numeric(x[[2]])
block <- as.numeric(x[[3]]) - start + 1
res <- c(start, 1, 1, block)
}
}
else if(x[[1]] == "seq_len") {
res <- c(1, 1, 1, as.numeric(x[[2]]))
}
else if(x[[1]] == "seq") {
if(length(x) >= 2 && (is.integer(x[[2]]) || is.numeric(x[[2]]))) {
## in both cases use a matching to seq.default, as the signatures are the same
matched_call <- match.call.withDef(seq.default, x)
if(is.null(matched_call$length.out)) {
length.out <- floor(((as.numeric(matched_call$to) - as.numeric(matched_call$from)) / as.numeric(matched_call$by)) + 1)
}
else {
length.out <- as.numeric(matched_call$length.out)
}
stride <- as.numeric(matched_call$by)
start <- as.numeric(matched_call$from)
if(stride==1) {
## it is one block
res <- c(start, 1, stride, length.out)
}
else {
## it is length.out blocks, each of length 1
res <- c(start, length.out, stride, 1)
}
}
}
}
if(!is.na(res[[1]]) && res[[1]] <= 0) {
res <- c(NA, NA, NA, NA)
}
return(res)
}
##' Evaluate if the arguments are regular for hyperslab use
##'
##' For each argument check if it can be used as a hyperslab, potentially after sorting and making unique.
##' @title Evaluate if the arguments are regular for hyperslab use
##' @param args The arguments input; if it was empty, then set to NULL
##' @param ds_dims The dimensions of the input dataset
##' @param envir The environment in which to evaluate the arguments
##' @param post_read Should the reshuffle be computed for post-read (then \code{TRUE}) or pre-write (then \code{FALSE})
##' @return A list with 2 parts; Evaluated arguments, regularity report and reshuffle indicators. Will be returned as a list
##' with components \code{args_in}, \code{args_point}, \code{is_hyperslab}, \code{hyperslab},
##' \code{needs_reshuffle}, code{reshuffle}, \code{result_dims_pre_shuffle}, \code{result_dims_post_shuffle}, \code{max_dims}
##' @author Holger Hoefling
##' @keywords internal
args_regularity_evaluation <- function(args, ds_dims, envir, post_read=TRUE) {
## check that the number of arguments is the same as the dataset dimension
if(length(args) != length(ds_dims)) {
stop("Number of arguments not equal to number of dimensions: ", length(args), " vs. ", length(ds_dims))
}
## create the skeleton for the regularity report
## has columns: start, count, stride, block
hyperslab <- matrix(rep(NA, length(args) * 4), ncol=4)
colnames(hyperslab) <- c("start", "count", "stride", "block")
is_hyperslab <- rep(FALSE, length(args))
needs_reshuffle <- is_hyperslab
args_in <- args
args_point <- vector("list", length(args))
result_dims_pre_shuffle <- numeric(length(args))
result_dims_post_shuffle <- numeric(length(args))
max_dims <- result_dims_pre_shuffle
reshuffle <- args_point
for(i in seq_along(args)) {
cur_arg <- args[[i]]
if(length(args[[i]]) == 1 && args[[i]]==quote(expr=)) {
## set to maximal hyperslab for this dimension
res_hyper <- c(1, 1, 1, ds_dims[[i]])
}
else if(is.call(cur_arg)) {
res_hyper <- check_arg_for_hyperslab_func(cur_arg, envir)
}
else {
res_hyper <- c(NA, NA, NA, NA)
}
## res_hyper contains the information if the argument can be interpreted as a hyperslab
if(!any(is.na(res_hyper))) { ## is a hyperslab made from a function
is_hyperslab[i] <- TRUE
hyperslab[i,] <- res_hyper
needs_reshuffle[i] <- FALSE
result_dims_pre_shuffle[i] <- res_hyper[2] * res_hyper[4] # count * block
result_dims_post_shuffle[i] <- res_hyper[2] * res_hyper[4] # count * block
max_dims[i] <- res_hyper[1] + (res_hyper[2] - 1) * res_hyper[3] + res_hyper[4] - 1 # start + (count - 1) * stride + block - 1
}
else {
cur_arg <- eval(cur_arg, envir=envir)
if(is.logical(cur_arg)) {
## expand it as necessary and convert to an integer vector
cur_arg <- seq_len(ds_dims[[i]])[cur_arg]
}
if(is.null(cur_arg)) {
## make it into a numeric of length 0
cur_arg <- numeric(0)
}
if(is.numeric(cur_arg) || is.integer(cur_arg) || is.integer64(cur_arg)) {
cur_arg <- as.numeric(cur_arg)
## check if it has length 0 or 1; these are special cases
if(length(cur_arg) == 0) {
is_hyperslab[i] <- FALSE
needs_reshuffle[i] <- FALSE
args_point[[i]] <- cur_arg
result_dims_pre_shuffle[i] <- 0
result_dims_post_shuffle[i] <- 0
max_dims[i] <- -Inf
}
else if(length(cur_arg) == 1) {
is_hyperslab[i] <- TRUE
needs_reshuffle[i] <- FALSE
result_dims_pre_shuffle[i] <- 1
result_dims_post_shuffle[i] <- 1
hyperslab[i, ] <- c(cur_arg, 1, 1, 1)
max_dims[i] <- cur_arg
}
else {
## check it is all positive, negative or a mixture of both
if(all(cur_arg < 0)) {
cur_arg <- seq_len(ds_dims[[i]])[cur_arg]
}
else if(!all(cur_arg > 0)) {
stop("In index ", i, " not all subscripts are either positive or negative")
}
## then check if it can be written as a hyperslab
cur_arg_diff_unique <- unique(diff(cur_arg))
if(all(cur_arg_diff_unique > 0)) { ## strictly increasing
needs_reshuffle[i] <- FALSE
max_dims[i] <- cur_arg[length(cur_arg)]
result_dims_pre_shuffle[i] <- length(cur_arg)
result_dims_post_shuffle[i] <- length(cur_arg)
if(length(cur_arg_diff_unique) == 1) {
is_hyperslab[i] <- TRUE
if(cur_arg_diff_unique == 1) {
hyperslab[i, ] <- c(cur_arg[[1]], 1, 1, length(cur_arg))
}
else {
hyperslab[i, ] <- c(cur_arg[[1]], length(cur_arg), cur_arg_diff_unique, 1)
}
}
else {
## can only be done as points, not as a hyperslab
is_hyperslab[i] <- FALSE
args_point[[i]] <- cur_arg
}
}
else { ## not strictly increasing, so sort and make unique
## see if a sorting and making unique could make it into a hyperslab
sort_arg <- sort(unique.default(cur_arg))
needs_reshuffle[i] <- TRUE
if(post_read) {
reshuffle[[i]] <- match(cur_arg, sort_arg)
result_dims_pre_shuffle[i] <- length(sort_arg)
result_dims_post_shuffle[i] <- length(cur_arg)
}
else {
## need to invalidate the first instance of anything that is duplicated
reshuffle[[i]] <- order(cur_arg)[!duplicated(cur_arg, fromLast=TRUE)]
## length of pre and post are reversed here
result_dims_post_shuffle[i] <- length(sort_arg)
result_dims_pre_shuffle[i] <- length(cur_arg)
}
sort_arg_diff <- diff(sort_arg)
sort_arg_diff_unique <- unique(sort_arg_diff)
## now same as before for the unsorted one
## but as is sorted and unique, know already that it is non-decreasing
max_dims[i] <- sort_arg[length(sort_arg)]
if(length(sort_arg_diff_unique) == 1) {
is_hyperslab[i] <- TRUE
if(sort_arg_diff_unique == 1) {
hyperslab[i, ] <- c(sort_arg[[1]], 1, 1, length(sort_arg))
}
else {
hyperslab[i, ] <- c(sort_arg[[1]], length(sort_arg), sort_arg_diff_unique, 1)
}
}
else {
## can only be done as points, not as a hyperslab
is_hyperslab[i] <- FALSE
args_point[[i]] <- sort_arg
}
}
}
}
else {
stop("Can't evaluate argument ", i)
}
}
}
return(list(args_in=args_in, args_point=args_point, is_hyperslab=is_hyperslab, hyperslab=hyperslab,
result_dims_pre_shuffle=result_dims_pre_shuffle, result_dims_post_shuffle=result_dims_post_shuffle,
max_dims=max_dims, needs_reshuffle=needs_reshuffle, reshuffle=reshuffle))
}
##' Single hyperslab dimension to explicit vector
##'
##' Uses the information of a hyperslab and turns it into an explicit vector.
##' @title Single hyperslab dimension to explicit vector
##' @param hyperslab a length 4 vector describing the start, count, stride and block component of a single dimension of a hyperslab
##' @return An explicit vector describing the points in the hyperslab dimension
##' @author Holger Hoefling
##' @keywords internal
hyperslab_to_points <- function(hyperslab) {
stopifnot(length(hyperslab)==4 && is.numeric(hyperslab))
## c("start", "count", "stride", "block")
if(hyperslab[[2]] == 1) { # only one block
res <- (seq_len(hyperslab[[4]]) - 1) + hyperslab[[1]]
}
else if(hyperslab[[4]] == 1) { # only blocks of size 1
res <- ((seq_len(hyperslab[[2]]) - 1) * hyperslab[[3]]) + hyperslab[[1]]
}
else { # mixed; should not occur for us but will write function to handle this anyway
## doesn't need to be overly efficient
res <- rep(seq_len(hyperslab[[4]]), times=hyperslab[[2]])
## now add it the stride component
res <- res + rep((seq_len(hyperslab[[2]]) - 1) * hyperslab[[3]], each=hyperslab[[4]])
res <- res + hyperslab[[1]] - 1
}
return(res)
}
##' Turn regulation evaluation into a selection for a space object
##'
##' Analyzes the results of the regularity evaluation of each dimension and checks if
##' it needs to be into a hyperslab-selection or a point-selection. A hyperslab selection will be chosen
##' whenever there are significantly less of it than the number of selected points. The ratio is determined
##' by the option \code{hdf5r.point_to_hyperslab_ratio}. If this is 1, then always hyperslabs will be used.
##' @title Turn regulation evaluation into a selection for a space object
##' @param reg_eval_res The result of the \code{args_regularity_evaluation} function
##' @return Returns an object with either the point-matrix or the hyperslab-selection array. The resulting object is
##' of class \code{point_selection} or \code{hyperslab_selection}.
##' @author Holger Hoefling
##' @keywords internal
regularity_eval_to_selection <- function(reg_eval_res) {
## first calculate how many points need to be selected and how many hyperslabs would need to be necesary
num_points <- prod(reg_eval_res$result_dims_pre_shuffle)
num_hyperslabs <- prod(reg_eval_res$result_dims_pre_shuffle[!reg_eval_res$is_hyperslab])
## trivially any selection can always be written as a concatenation of hyperslabs; so should be ever do a pointlist?
## likely pointlist more efficient in selection. Use hyperslabs only if we have X-fold fewer hyperslabs than datapoints
## here choose X as >= 4 for now
if(num_hyperslabs == 0) {
point_hyperslab_ratio <- 0
}
else {
point_hyperslab_ratio <- num_points / num_hyperslabs
}
if(point_hyperslab_ratio >= getOption("hdf5r.point_to_hyperslabs_ratio")) {
## use a hyperslab
sel_type <- "hyperslab_selection"
## make a 3-dimensional array, each 2 dimensional sub-array (in the third dimension) is a complete hyperslab description
if(any(!reg_eval_res$is_hyperslab)) {
## not just 1 hyperslab
point_grid <- expand_point_grid(reg_eval_res$args_point[!reg_eval_res$is_hyperslab])
hyperslab_array <- array(0, dim=c(nrow(reg_eval_res$hyperslab), nrow(point_grid), 4))
for(i in 1:4) {
hyperslab_array[,,i] <- reg_eval_res$hyperslab[,i]
}
## c("start", "count", "stride", "block")
## set the dimensions that are the point grid; those currently have NA in them
hyperslab_array[!reg_eval_res$is_hyperslab, , 1] <- t(point_grid)
hyperslab_array[!reg_eval_res$is_hyperslab, , 2] <- 1
hyperslab_array[!reg_eval_res$is_hyperslab, , 3] <- 1
hyperslab_array[!reg_eval_res$is_hyperslab, , 4] <- 1
}
else {
hyperslab_array <- array(reg_eval_res$hyperslab, dim=c(nrow(reg_eval_res$hyperslab), 1, 4))
}
sel <- hyperslab_array
}
else {
sel_type <- "point_selection"
## if there are any hyperslab dimensions, expand them to a point list
args_point <- reg_eval_res$args_point
if(any(reg_eval_res$is_hyperslab)) {
for(i in which(reg_eval_res$is_hyperslab)) {
args_point[[i]] <- hyperslab_to_points(reg_eval_res$hyperslab[i,])
}
}
sel <- expand_point_grid(args_point)
}
return(structure(.Data=sel, class=sel_type))
}
##' Expand list of points for each dimension into a matrix of all combinations
##'
##' The function is similar to the \code{expand.grid} function
##' @title Expand list of points for each dimension into a matrix of all combinations
##' @param point_list A list of the points in each dimension to include
##' @return A matrix with every combination of points for each dimension
##' @author Holger Hoefling
##' @keywords internal
expand_point_grid <- function(point_list) {
point_list_size <- unlist(lapply(point_list, length))
cum_prod_std <- c(1, cumprod(point_list_size))
cum_prod_rev <- rev(c(1, cumprod(rev(point_list_size))))
num_points <- prod(point_list_size)
point_mat <- matrix(numeric(num_points * length(point_list)), ncol=length(point_list))
for(i in seq_along(point_list)) {
point_mat[, i] <- rep(point_list[[i]], each=cum_prod_std[i], times=cum_prod_rev[i+1])
}
return(point_mat)
}
##' Match arguments in a call to function and add default values
##'
##' Given the definition of a function and a call, it matches the arguments
##' so that they are named and inserts any default argument values where
##' those are missing
##' @title Match arguments in a call to function and add default values
##' @param definition Definition of the function to match against
##' @param call The call that should be matched
##' @return A call with named arguments and default values
##' @author Holger Hoefling
##' @keywords internal
match.call.withDef <- function(definition, call) {
def.formals <- formals(definition)
foo <- try(as.list(match.call(definition = definition, call=call)), silent=TRUE)
if(inherits(foo, "try-error")) {
return(foo)
}
## add the default arguments if they are missing
for(v in names(def.formals)) {
if(!(v %in% names(foo))) {
foo <- c(foo, def.formals[v])
}
}
return(foo)
}
##' Reshuffle the input as needed - see \code{args_regularity_evaluation}
##'
##' When necessary, this function performs the reshuffle as defined by \code{args_regularity_evaluation}.
##' @title Reshuffle the input as needed - see \code{args_regularity_evaluation}
##' @param x The array to reshuffle
##' @param reg_eval_res The result of the regularity evaluation
##' @return The reshuffled input
##' @author Holger Hoefling
##' @keywords internal
do_reshuffle <- function (x, reg_eval_res) {
## for the post read shuffle, we just have to use the order as noted in reg_eval_res
## however, data.frames have an additional dimension the evaluator currently
## does not know about, so have to add it
if(inherits(x, "data.frame")) {
if(length(reg_eval_res$needs_reshuffle) != 1) {
stop("For data.frame, the selection can only have 1 dimension")
}
reorder_params <- rep(list(quote(expr=)), 2)
if(reg_eval_res$needs_reshuffle) {
reorder_params[1] <- reg_eval_res$reshuffle[1]
}
}
else {
reorder_params <- rep(list(quote(expr=)), length(reg_eval_res$reshuffle))
reorder_params[reg_eval_res$needs_reshuffle] <- reg_eval_res$reshuffle[reg_eval_res$needs_reshuffle]
}
res <- do.call("[", c(list(x), reorder_params))
return(res)
}
|