Fix approach to extract rois regex to be more robust

Author: michaelhallquist

NAMESPACE

@@ -15,7 +15,7 @@ export(extract_bids_info)
 export(extract_rois)
 export(filtfilt_cpp)
 export(get_fmriprep_outputs)
-export(get_postproc_stream_outputs)
+export(get_postproc_output_files)
 export(get_project_status)
 export(get_step_title)
 export(get_subject_status)

NEWS.md

@@ -8,6 +8,7 @@
 * Jobs now write a manifest of files and times to the job tracking database for more thorough completeness tests
 * Added optional low-pass filtering of motion parameters, matching (Gratton)
 * Gracefully adjust motion filtering parameters if they fall above Nyquist at this TR
+* Modify extract ROIs config to avoid input_regex and always generate it internally from postproc stream
 
 # BrainGnomes 0.7-5
 

R/process_subject.R

@@ -638,11 +638,18 @@ submit_extract_rois <- function(
 
   # pull the requested extraction stream from the broader list
   ex_cfg <- scfg$extract_rois[[ex_stream]]
+  if (!is.null(ex_cfg$input_regex)) {
+    msg <- "extract_rois/input_regex is ignored by run_project; inputs are derived from postprocess streams."
+    if (!is.null(lg)) {
+      to_log(lg, "warn", msg)
+    } else {
+      warning(msg, call. = FALSE)
+    }
+  }
   if (isTRUE(scfg$force)) ex_cfg$overwrite <- TRUE # enable overwrite of ROIs if force=TRUE
 
-  # Every extract_rois stream can pull for 1+ postprocess streams. Based on postprocess input stream(s), generate regular expressions
-  # need to find outputs of postproc stream. A little tricky given that desc may not be in input_regex. This is handled inside extract_cli.R,
-  # which runs once the job fires (and any expected files are now available from earlier processing stages)
+  # Every extract_rois stream can pull for 1+ postprocess streams. Pass through the input spec
+  # and bids_desc for each stream so extract_cli.R can target postprocessed outputs directly.
   ex_cfg$input_regex <- sapply(ex_cfg$input_streams, function(ss) scfg$postprocess[[ss]]$input_regex, USE.NAMES = FALSE)
 
   # the bids_desc of the postprocess stream is used to update the matched files (to get the outputs of postprocessing)

R/setup_postprocess.R

@@ -1428,51 +1428,43 @@ setup_apply_aroma <- function(ppcfg = list(), fields = NULL) {
 
 
 
-#' Determine expected output files for a postprocessing stream
+#' List postprocessed output files for a stream based on its input spec
 #'
-#' Given a directory of candidate input files and a specification for
-#' matching those inputs, this function returns the full paths to the
-#' corresponding postprocessed NIfTI files. The output filenames are
-#' derived by replacing the `desc` entity in each input file with the
-#' stream's `bids_desc` value.
+#' Converts a postprocess input specification into a pattern that targets
+#' postprocessed outputs by ensuring the `desc` entity matches `bids_desc`.
 #'
-#' @param input_dir Directory containing the input NIfTI files to be
-#'   postprocessed.
-#' @param input_regex Specification used to match the input files. This may be
-#'   a space-separated set of BIDS entities (e.g., "desc:preproc suffix:bold")
-#'   or a regular expression prefixed with "regex:".
-#' @param bids_desc The `desc` value to use for the output filenames.
+#' @param input_dir Directory containing postprocessed outputs.
+#' @param input_regex Specification used to match the input files for the stream.
+#'   May be a space-separated set of BIDS entities or a regex prefixed with "regex:".
+#' @param bids_desc The `desc` value used for postprocessed outputs.
 #'
-#' @return A character vector of full paths to the expected postprocessed
-#'   NIfTI files. The vector is named with the corresponding input file.
+#' @return A character vector of full paths to matching postprocessed outputs.
 #' @export
-#' @examples
-#' \dontrun{
-#' get_postproc_stream_outputs(
-#'   input_dir = "/path/to/subject",
-#'   input_regex = "desc:preproc suffix:bold",
-#'   bids_desc = "clean"
-#' )
-#' }
-#' @importFrom checkmate assert_directory_exists assert_string
-get_postproc_stream_outputs <- function(input_dir, input_regex, bids_desc) {
+get_postproc_output_files <- function(input_dir, input_regex, bids_desc) {
   checkmate::assert_directory_exists(input_dir)
   if (is.null(input_regex)) input_regex <- "desc:preproc suffix:bold"
-  checkmate::assert_string(input_regex)
+  checkmate::assert_character(input_regex, min.len = 1L)
   checkmate::assert_string(bids_desc)
 
-  pattern <- construct_bids_regex(input_regex)
-  in_files <- list.files(path = input_dir, pattern = pattern, recursive = TRUE, full.names = TRUE)
+  resolve_spec <- function(spec) {
+    spec <- trimws(spec)
+    if (grepl("^regex:", spec)) {
+      return(trimws(sub("^regex\\s*:", "", spec)))
+    }
 
-  if (length(in_files) == 0L) return(character(0))
+    if (grepl("\\bdesc:", spec)) {
+      spec <- sub("\\bdesc:[^[:space:]]+", paste0("desc:", bids_desc), spec)
+    } else {
+      spec <- paste(spec, paste0("desc:", bids_desc))
+    }
+
+    construct_bids_regex(spec)
+  }
 
-  out_files <- vapply(in_files, function(f) {
-    info <- as.list(extract_bids_info(f))
-    info$description <- bids_desc
-    construct_bids_filename(info, full.names = TRUE)
-  }, character(1))
+  patterns <- vapply(input_regex, resolve_spec, character(1))
+  files <- unlist(lapply(patterns, function(pat) {
+    list.files(path = input_dir, pattern = pat, recursive = TRUE, full.names = TRUE)
+  }), use.names = FALSE)
 
-  # leave as unnamed for now (don't need to remember input file)
-  # names(out_files) <- in_files
-  out_files
+  unique(files)
 }

inst/extract_cli.R

@@ -69,7 +69,7 @@ if (!checkmate::test_directory(cfg$input)) {
   stop("No valid directory provided as --input")
 }
 
-input_files <- get_postproc_stream_outputs(cfg$input, cfg$input_regex, cfg$bids_desc)
+input_files <- get_postproc_output_files(cfg$input, cfg$input_regex, cfg$bids_desc)
 
 if (length(input_files) == 0L) {
   stop("Cannot find files for ROI extraction with --input: ", cfg$input)

man/get_all_nodes.Rd

@@ -0,0 +1,25 @@
+test_that("get_postproc_output_files targets postprocessed outputs", {
+  tmp_dir <- tempfile("bg_postproc_")
+  dir.create(tmp_dir, recursive = TRUE)
+
+  postproc <- file.path(
+    tmp_dir,
+    "sub-01_task-impressions_space-MNI152NLin2009cAsym_desc-postproc_clean_bold.nii.gz"
+  )
+  preproc <- file.path(
+    tmp_dir,
+    "sub-01_task-impressions_space-MNI152NLin2009cAsym_desc-preproc_bold.nii.gz"
+  )
+  file.create(postproc)
+  file.create(preproc)
+
+  rx_with_desc <- "desc:preproc task:impressions space:MNI152NLin2009cAsym suffix:bold"
+  res_with_desc <- get_postproc_output_files(tmp_dir, rx_with_desc, "postproc_clean")
+  expect_true(postproc %in% res_with_desc)
+  expect_false(preproc %in% res_with_desc)
+
+  rx_no_desc <- "task:impressions space:MNI152NLin2009cAsym suffix:bold"
+  res_no_desc <- get_postproc_output_files(tmp_dir, rx_no_desc, "postproc_clean")
+  expect_true(postproc %in% res_no_desc)
+  expect_false(preproc %in% res_no_desc)
+})