Increase code coverage (#116)

DESCRIPTION

@@ -26,13 +26,14 @@ Suggests:
     fuzzyjoin,
     igraph,
     knitr,
+    microbenchmark,
+    profmem,
+    purrr,
     rmarkdown,
     stringdist,
     testthat (>= 3.0.0),
     tidyverse,
-    purrr,
-    microbenchmark,
-    profmem
+    vdiffr
 Config/testthat/edition: 3
 URL: https://beniamino.org/zoomerjoin/
 BugReports: https://github.com/beniaminogreen/zoomerjoin/issues/

R/on_load.R

@@ -1,3 +1,4 @@
+# nocov start
 .onAttach <- function(libname, pkgname) {
   if (Sys.getenv("_R_CHECK_LIMIT_CORES_") != "") {
     if (as.logical(Sys.getenv("_R_CHECK_LIMIT_CORES_"))) {
@@ -6,3 +7,4 @@
     }
   }
 }
+# nocov end

R/string_group.R

@@ -52,7 +52,7 @@
 #' @importFrom stats runif
 #' @importFrom utils installed.packages packageVersion
 jaccard_string_group <- function(string, n_gram_width = 2, n_bands = 45, band_width = 8, threshold = .7, progress = FALSE) {
-  if (system.file(package = "igraph") == "") {
+  if (!requireNamespace("igraph")) {
     stop("library 'igraph' must be installed to run this function")
   }
 

tests/testthat/test-curves.R

@@ -0,0 +1,21 @@
+test_that("euclidean curve", {
+  vdiffr::expect_doppelganger(
+    "Basic Euclidean curve",
+    euclidean_curve(20, 5, r = 0.5)
+  )
+})
+
+test_that("jaccard curve", {
+  expect_error(
+    jaccard_curve(c(20, 10), 1),
+    "single integer"
+  )
+  expect_error(
+    jaccard_curve(1, c(20, 10)),
+    "single integer"
+  )
+  vdiffr::expect_doppelganger(
+    "Basic Jaccard curve",
+    jaccard_curve(20, 5)
+  )
+})

tests/testthat/test-test_logical_lsh_join.R

@@ -50,42 +50,50 @@ test_that("jaccard_inner_join works on tiny dataset", {
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2))
   expect_identical(sort(test$id_2), c(1, 2))
 })
+
 test_that("jaccard_full_join works on tiny dataset", {
   capture_messages(
     test <- jaccard_full_join(dataset_1, dataset_2, threshold = .6, n_bands = 300)
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2, 3))
   expect_identical(sort(test$id_2), c(1, 2, 3))
 })
+
 test_that("jaccard_left_join works on tiny dataset", {
   capture_messages(
     test <- jaccard_left_join(dataset_1, dataset_2, threshold = .6, n_bands = 300)
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2, 3))
   expect_identical(sort(test$id_2), c(1, 2))
 })
+
 test_that("jaccard_right_join works on tiny dataset", {
   capture_messages(
     test <- jaccard_right_join(dataset_1, dataset_2, threshold = .6, n_bands = 300)
   )
 
-
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2))
   expect_identical(sort(test$id_2), c(1, 2, 3))
 })
 
+test_that("jaccard_anti_join works on tiny dataset", {
+  capture_messages(
+    test <- jaccard_anti_join(dataset_1, dataset_2, threshold = .6, n_bands = 300)
+  )
+
+  expect_true(all(test$id_1 == test$id_2, na.rm = T))
+  expect_identical(sort(test$id_1, na.last = TRUE), c(3, NA))
+  expect_identical(sort(test$id_2, na.last = TRUE), c(3, NA))
+})
+
 test_that("jaccard_inner_join gives same results as stringdist_inner_join", {
   for (i in 1:20) {
     capture_messages({
@@ -165,7 +173,6 @@ test_that("hamming_inner_join works on tiny dataset", {
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2))
   expect_identical(sort(test$id_2), c(1, 2))
 })
@@ -176,7 +183,6 @@ test_that("hamming_full_join works on tiny dataset", {
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2, 3))
   expect_identical(sort(test$id_2), c(1, 2, 3))
 })
@@ -187,7 +193,6 @@ test_that("hamming_left_join works on tiny dataset", {
   )
 
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2, 3))
   expect_identical(sort(test$id_2), c(1, 2))
 })
@@ -197,13 +202,21 @@ test_that("hamming_right_join works on tiny dataset", {
     test <- hamming_right_join(dataset_1, dataset_2, threshold = 3, band_width = 1, n_bands = 300)
   )
 
-
   expect_true(all(test$id_1 == test$id_2, na.rm = T))
-
   expect_identical(sort(test$id_1), c(1, 2))
   expect_identical(sort(test$id_2), c(1, 2, 3))
 })
 
+test_that("hamming_anti_join works on tiny dataset", {
+  capture_messages(
+    test <- hamming_anti_join(dataset_1, dataset_2, threshold = 3, band_width = 1, n_bands = 300)
+  )
+
+  expect_true(all(test$id_1 == test$id_2, na.rm = T))
+  expect_identical(sort(test$id_1, na.last = TRUE), c(3, NA))
+  expect_identical(sort(test$id_2, na.last = TRUE), c(3, NA))
+})
+
 test_that("jaccard_inner_join gives same results as stringdist_inner_join", {
   for (i in 1:20) {
     capture_messages({
@@ -218,6 +231,17 @@ test_that("jaccard_inner_join gives same results as stringdist_inner_join", {
   }
 })
 
+test_that("hamming: arg 'similarity_column' works", {
+  capture_messages(
+    test <- hamming_inner_join(
+      dataset_1, dataset_2,
+      threshold = 3, band_width = 1, n_bands = 300,
+      similarity_column = "sim"
+    )
+  )
+  expect_equal(test$sim, hamming_distance(test$string.x, test$string.y))
+})
+
 
 test_that("seed works for hamming joins", {
   for (i in 1:15) {