Allow swapping y and x in st_any_pred

R/geometry_operation.R

@@ -13,9 +13,19 @@
 #' @param pred A geometric binary predicate function, such as
 #'   \code{\link{st_intersects}}. It should return an object of class
 #'   \code{sgbp}, for sparse predicates.
+#' @param yx Whether to do \code{pred(y, x)} instead of \code{pred(x, y)}. For
+#'   symmetric predicates, the results should be the same. When x has a large
+#'   number of geometries and y has few, \code{pred(y, x)} is much faster than
+#'   \code{pred(x, y)} for \code{st_intersects}, \code{st_disjoint}, and
+#'   \code{st_is_within_distance}.
+#' @param sparse If \code{TRUE}, returns numeric indices rather than logical
+#'   vector. Defaults to \code{FALSE} for backward compatibility, though the
+#'   default in \code{st_intersects} is \code{TRUE}.
+#' @param ... Arguments passed to \code{pred}.
 #' @return For \code{st_any_*}, a logical vector indicating whether each
 #'   geometry in \code{x} intersects (or other predicates such as is covered by)
-#'   anything in \code{y}. Simplified from the \code{sgbp} results which
+#'   anything in \code{y} or a numeric vector of indices of \code{TRUE} when
+#'   \code{sparse = TRUE}. Simplified from the \code{sgbp} results which
 #'   indicate which item in \code{y} each item in \code{x} intersects, which
 #'   might not always be relevant. For \code{st_n_*}, an integer vector
 #'   indicating the number of geometries in y returns TRUE for each geometry in
@@ -36,18 +46,27 @@
 #' st_any_intersects(pts, pol)
 #' st_n_pred(pts, pol, pred = st_disjoint)
 #' st_n_intersects(pts, pol)
-st_any_pred <- function(x, y, pred) lengths(pred(x, y)) > 0L
-# TODO: put the item with more geometries in the x position. This way is much faster.
-# If swapping positions, it shouldn't be hard to recover it if the pred is symmetric
+st_any_pred <- function(x, y, pred, yx = FALSE, sparse = FALSE, ...) {
+    if (length(x) > length(y)*10 && yx) {
+        res <- unlist(unclass(pred(y, x, ...)))
+        res <- sort(unique(res))
+        if (!sparse) res <- which(seq_along(x) %in% res)
+    } else {
+        res <- lengths(pred(x, y, ...)) > 0L
+        if (sparse) res <- which(res)
+    }
+    res
+}
 
 #' @rdname st_any_pred
 #' @export
-st_any_intersects <- function(x, y) st_any_pred(x, y, st_intersects)
+st_any_intersects <- function(x, y, yx = FALSE, sparse = FALSE) 
+    st_any_pred(x, y, st_intersects, yx = yx, sparse = sparse)
 
 #' @rdname st_any_pred
 #' @export
-st_n_pred <- function(x, y, pred) lengths(pred(x, y))
-
+st_n_pred <- function(x, y, pred, ...) lengths(pred(x, y, ...))
+# Not doing yx here since this is usually not used when length(y) << length(x)
 #' @rdname st_any_pred
 #' @export
 st_n_intersects <- function(x, y) st_n_pred(x, y, st_intersects)
@@ -145,7 +164,8 @@ st_n_intersects <- function(x, y) st_n_pred(x, y, st_intersects)
 #' (i.e. \code{colGeometry}) and an annotation geometry for each sample. For
 #' example, whether each Visium spot intersects with the tissue boundary in each
 #' sample.
-#'
+#' 
+#' @inheritParams st_any_pred
 #' @param sfe An SFE object.
 #' @param colGeometryName Name of column geometry for the predicate.
 #' @param annotGeometryName Name of annotation geometry for the predicate.
@@ -171,12 +191,13 @@ st_n_intersects <- function(x, y) st_n_pred(x, y, st_intersects)
 #' # How many nuclei are there in each Visium spot
 #' n_nuclei <- annotNPred(sfe, "spotPoly", annotGeometryName = "nuclei")
 annotPred <- function(sfe, colGeometryName = 1L, annotGeometryName = 1L,
-                      sample_id = "all", pred = st_intersects) {
+                      sample_id = "all", pred = st_intersects, yx = FALSE) {
     sample_id <- .check_sample_id(sfe, sample_id, one = FALSE)
     ag <- annotGeometry(sfe, type = annotGeometryName, sample_id = sample_id)
     .annot_fun(sfe, ag,
         colGeometryName = colGeometryName,
-        samples_use = sample_id, fun = st_any_pred, pred = pred
+        samples_use = sample_id, fun = st_any_pred, pred = pred,
+        yx = yx, sparse = FALSE
     )
 }
 
@@ -425,7 +446,7 @@ crop <- function(x, y = NULL, colGeometryName = 1L, sample_id = "all",
     }
     preds <- .annot_fun(x, y, colGeometryName,
         samples_use = samples_use,
-        fun = st_any_pred, pred = pred
+        fun = st_any_pred, pred = pred, yx = TRUE
     )
     # Don't remove anything from other samples
     other_bcs <- setdiff(colnames(x), names(preds))

tests/testthat/test-geometry_operation.R

@@ -7,8 +7,16 @@ pts = st_sfc(st_point(c(.5,.5)), st_point(c(1.5, 1.5)), st_point(c(2.5, 2.5)))
 pol = st_polygon(list(rbind(c(0,0), c(2,0), c(2,2), c(0,2), c(0,0))))
 
 test_that("trivial, st_any_intersects", {
-    o <- st_any_intersects(pts, pol)
+    o <- st_any_intersects(pts, pol, sparse = FALSE)
     expect_equal(o, c(TRUE, TRUE, FALSE))
+    o <- st_any_intersects(pts, pol, sparse = TRUE)
+    expect_equal(o, c(1L, 2L))
+})
+
+test_that("Swap x and y for st_intersects", {
+    o1 <- st_any_intersects(pts, pol, sparse = TRUE)
+    o2 <- st_any_intersects(pts, pol, sparse = TRUE, yx = TRUE)
+    expect_equal(o1, o2)
 })
 
 test_that("trivial, st_n_intersects", {
@@ -48,21 +56,21 @@ rowGeometry(sfe_visium, "points", sample_id = "sample01", withDimnames = FALSE)
 test_that("All spots in the cropped SFE objects indeed are covered by the bbox", {
     sfe_cropped <- SpatialFeatureExperiment::crop(sfe_visium, bbox_use, sample_id = "all")
     cg <- spotPoly(sfe_cropped, "all")
-    expect_true(all(st_any_pred(cg, bbox_use, pred = st_covered_by)))
+    expect_true(all(st_any_pred(cg, bbox_use, pred = st_covered_by, sparse = FALSE)))
     expect_true(st_geometry_type(cg, by_geometry = FALSE) == "POLYGON")
     rg <- rowGeometry(sfe_cropped)
-    expect_true(all(st_any_pred(rg, bbox_use, pred = st_covered_by)))
+    expect_true(all(st_any_pred(rg, bbox_use, pred = st_covered_by, sparse = FALSE)))
 })
 
 test_that("Only crop one sample out of two, with sf", {
     expect_error(SpatialFeatureExperiment::crop(sfe_visium, y = bbox_sf, sample_id = "sample02"),
                  "No bounding boxes for samples specified.")
     sfe_cropped2 <- SpatialFeatureExperiment::crop(sfe_visium, y = bbox_sf, sample_id = "sample01")
     expect_true(all(st_any_pred(spotPoly(sfe_cropped2, "sample01"), bbox_use,
-                                pred = st_covered_by
+                                pred = st_covered_by, sparse = FALSE
     )))
     expect_false(all(st_any_pred(spotPoly(sfe_cropped2, "sample02"), bbox_use,
-                                 pred = st_covered_by
+                                 pred = st_covered_by, sparse = FALSE
     )))
     expect_equal(sum(st_any_intersects(
         spotPoly(sfe_cropped2, "sample02"),
@@ -92,10 +100,10 @@ test_that("Using a bounding box to crop SFE objects, current way, one sample", {
                  "No bounding boxes for samples specified.")
     sfe_cropped <- SpatialFeatureExperiment::crop(sfe_visium, y = m)
     expect_true(all(st_any_pred(spotPoly(sfe_cropped, "sample01"), bbox_use,
-                                pred = st_covered_by
+                                pred = st_covered_by, sparse = FALSE
     )))
     expect_false(all(st_any_pred(spotPoly(sfe_cropped, "sample02"), bbox_use,
-                                 pred = st_covered_by
+                                 pred = st_covered_by, sparse = FALSE
     )))
     expect_equal(sum(st_any_intersects(
         spotPoly(sfe_cropped, "sample02"),
@@ -111,10 +119,10 @@ test_that("Using a bounding box to crop SFE objects, current way, all samples",
                                 c("sample01", "sample02")))
     sfe_cropped <- SpatialFeatureExperiment::crop(sfe_visium, y = m)
     expect_true(all(st_any_pred(spotPoly(sfe_cropped, "sample01"), bbox_use,
-                                pred = st_covered_by
+                                pred = st_covered_by, sparse = FALSE
     )))
     expect_true(all(st_any_pred(spotPoly(sfe_cropped, "sample02"), bbox_use2,
-                                pred = st_covered_by
+                                pred = st_covered_by, sparse = FALSE
     )))
 })
 
@@ -289,7 +297,7 @@ test_that("annotOp", {
         rownames(out),
         colnames(sfe_visium)[colData(sfe_visium)$sample_id == "sample01"]
     )
-    p <- st_any_pred(out, bbox_use, st_covered_by)
+    p <- st_any_pred(out, bbox_use, st_covered_by, sparse = FALSE)
     expect_true(all(p[c(1, 2, 5)]))
     expect_false(any(p[3:4]))
 })