port X

inst/known_issues.yaml

@@ -1,11 +1,12 @@
 known_issues:
   - backend: HDF5AnnData
     slot:
+      - X
+      - layers
       - obsp
       - varp
       - obsm
       - varm
-      - layers
     dtype:
       - integer_csparse
       - integer_rsparse

tests/testthat/test-roundtrip-X.R

@@ -1,109 +1,104 @@
 skip_if_no_anndata()
-skip_if_not_installed("hdf5r")
+skip_if_not_installed("reticulate")
 
-data <- generate_dataset(10L, 20L)
+library(reticulate)
+testthat::skip_if_not(
+  reticulate::py_module_available("dummy_anndata"),
+  message = "Python dummy_anndata module not available for testing"
+)
 
-test_names <- names(data$layers)
+ad <- reticulate::import("anndata", convert = FALSE)
+da <- reticulate::import("dummy_anndata", convert = FALSE)
+bi <- reticulate::import_builtins()
 
-# TODO: Add denseMatrix support to anndata and anndataR
-test_names <- test_names[!grepl("_dense", test_names)]
+known_issues <- read_known_issues()
+
+test_names <- names(da$matrix_generators)
 
 for (name in test_names) {
-  test_that(paste0("roundtrip with X '", name, "'"), {
-    # create anndata
-    ad <- AnnData(
-      X = data$layers[[name]],
-      obs = data$obs[, c(), drop = FALSE],
-      var = data$var[, c(), drop = FALSE]
+  # first generate a python h5ad
+  adata_py <- da$generate_dataset(
+    x_type = name,
+    obs_types = list(),
+    var_types = list(),
+    layer_types = list(),
+    obsm_types = list(),
+    varm_types = list(),
+    obsp_types = list(),
+    varp_types = list(),
+    uns_types = list(),
+    nested_uns_types = list()
+  )
+
+  # create a couple of paths
+  file_py <- withr::local_file(tempfile(paste0("anndata_py_", name), fileext = ".h5ad"))
+  file_r <- withr::local_file(tempfile(paste0("anndata_r_", name), fileext = ".h5ad"))
+
+  # write to file
+  adata_py$write_h5ad(file_py)
+
+  test_that(paste0("Reading an AnnData with X '", name, "' works"), {
+    msg <- message_if_known(
+      backend = "HDF5AnnData",
+      slot = c("X"),
+      dtype = name,
+      process = "read",
+      known_issues = known_issues
     )
+    skip_if(!is.null(msg), message = msg)
 
-    # write to file
-    filename <- withr::local_file(tempfile(fileext = ".h5ad"))
-    write_h5ad(ad, filename)
-
-    # read from file
-    ad_new <- read_h5ad(filename, to = "HDF5AnnData")
-
-    # expect slots are unchanged
+    adata_r <- read_h5ad(file_py, to = "HDF5AnnData")
     expect_equal(
-      ad_new$X,
-      data$layers[[name]],
-      ignore_attr = TRUE,
-      tolerance = 1e-6
+      adata_r$shape(),
+      unlist(reticulate::py_to_r(adata_py$shape))
     )
+
+    # check that the print output is the same
+    str_r <- capture.output(print(adata_r))
+    str_py <- capture.output(print(adata_py))
+    expect_equal(str_r, str_py)
   })
-}
 
-for (name in test_names) {
-  test_that(paste0("reticulate->hdf5 with X '", name, "'"), {
-    # add rownames
-    X <- data$layers[[name]]
-    obs <- data.frame(row.names = rownames(data$obs))
-    var <- data.frame(row.names = rownames(data$var))
-
-    # TODO: remove this?
-    if (is.matrix(X) && any(is.na(X))) {
-      na_indices <- is.na(X)
-      X[na_indices] <- NaN
-    }
-
-    # create anndata
-    ad <- anndata::AnnData(
-      X = X,
-      obs = obs,
-      var = var
+  # maybe this test simply shouldn't be run if there is a known issue with reticulate
+  test_that(paste0("Comparing an anndata with X '", name, "' with reticulate works"), {
+    msg <- message_if_known(
+      backend = "HDF5AnnData",
+      slot = c("X"),
+      dtype = name,
+      process = c("read", "reticulate"),
+      known_issues = known_issues
     )
+    skip_if(!is.null(msg), message = msg)
 
-    # write to file
-    filename <- withr::local_file(tempfile(fileext = ".h5ad"))
-    ad$write_h5ad(filename)
-
-    # read from file
-    ad_new <- HDF5AnnData$new(filename)
+    adata_r <- read_h5ad(file_py, to = "HDF5AnnData")
 
-    # expect slots are unchanged
     expect_equal(
-      ad_new$X,
-      data$layers[[name]],
+      adata_r$X,
+      py_to_r(adata_py$X),
       tolerance = 1e-6
     )
   })
-}
-
-r2py_names <- test_names
-# TODO: re-enable -- rsparse gets converted to csparse by anndata
-r2py_names <- r2py_names[!grepl("rsparse", r2py_names)]
-
-for (name in r2py_names) {
-  test_that(paste0("hdf5->reticulate with X '", name, "'"), {
-    # write to file
-    filename <- withr::local_file(tempfile(fileext = ".h5ad"))
-
 
-    # make anndata
-    ad <- AnnData(
-      X = data$layers[[name]],
-      obs = data$obs[, c(), drop = FALSE],
-      var = data$var[, c(), drop = FALSE]
+  test_that(paste0("Writing an AnnData with X '", name, "' works"), {
+    msg <- message_if_known(
+      backend = "HDF5AnnData",
+      slot = c("X"),
+      dtype = name,
+      process = c("read", "write"),
+      known_issues = known_issues
     )
-    write_h5ad(ad, filename)
+    skip_if(!is.null(msg), message = msg)
+
+    adata_r <- read_h5ad(file_py, to = "InMemoryAnnData")
+    write_h5ad(adata_r, file_r)
 
     # read from file
-    ad_new <- anndata::read_h5ad(filename)
-
-    # expect slots are unchanged
-    layer_ <- ad_new$X
-    # anndata returns these layers as CsparseMatrix
-    if (grepl("rsparse", name)) {
-      layer_ <- as(layer_, "RsparseMatrix")
-    }
-    # strip rownames
-    dimnames(layer_) <- list(NULL, NULL)
-    expect_equal(
-      layer_,
-      data$layers[[name]],
-      ignore_attr = TRUE,
-      tolerance = 1e-6
+    adata_py2 <- ad$read_h5ad(file_r)
+
+    # expect that the objects are the same
+    expect_py_matrix_equal(
+      adata_py2$X,
+      adata_py$X
     )
   })
 }