Fixed aggregate functions, added tests

R/src_h2.r

@@ -18,16 +18,16 @@ NULL
 #' @export
 src_h2 <- function(x) UseMethod("src_h2")
 
-#' @rdname src_h2
 #' @export
+#' @rdname src_h2
 src_h2.character <- function(x, ...) {
   assert_that(requireNamespace("dbj.h2", quietly = TRUE))
   con <- DBI::dbConnect(dbj.h2::driver(), x, ...)
   src_h2(con, ...)
 }
 
-#' @rdname src_h2
 #' @export
+#' @rdname src_h2
 src_h2.H2Connection <- function(x, ...) {
   assert_that(requireNamespace("dbj.h2", quietly = TRUE))
   info <- DBI::dbGetInfo(x)
@@ -41,18 +41,23 @@ tbl.src_h2 <- function(src, from, ...) {
 }
 
 #' @export
+#' @rdname src_h2
 src_desc.src_h2 <- function(x) {
-  info <- x$info
-
-  sprintf("H2 %s [%s@%s]", x$database_product_version, x$user_name, x$url)
+  paste0(x$info$dbname, x$info$db.version, " [", x$info$url, "]")
 }
 
 #' @export
+#' @rdname src_h2
 src_translate_env.src_h2 <- function(x) {
   sql_variant(
     base_scalar,
     sql_translator(.parent = base_agg,
-      n = function(arg="*") { build_sql("COUNT(", ident(arg), ")") }
+      n = function(arg="*") { build_sql("COUNT(", ifelse(arg == "*", arg, ident(arg)), ")") },
+      n_distinct = function(x) {
+        build_sql("COUNT(DISTINCT ", ident(x), ")")
+      },
+      sd =  sql_prefix("stddev_samp"),
+      var = sql_prefix("var_samp")
     )
   )
 }

tests/testthat.R

@@ -1,3 +1,4 @@
 library(testthat)
+library(dplyr.src.h2)
 
 test_check("dplyr.src.h2")

tests/testthat/test-integration-data-nycflights13.r

@@ -1,12 +1,71 @@
 context("data-nycflights13")
 
+library(nycflights13)
+
 test_that("nycflights13_h2 has flights table", {
   # when
-  library(nycflights13)
   flights_db <- tbl(nycflights13_h2(), "flights")
 
   # then
-  expect_that(flights_db, is_a("tbl_h2"))
+  expect_is(flights_db, "tbl_h2")
   expect_equal(nrow(flights_db), 336776)
   expect_equal(ncol(flights_db), 16)
 })
+
+test_that("filter", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  filtered <- filter(flights_db, month == 1, day == 1)
+
+  expect_is(filtered, "tbl_h2")
+})
+
+test_that("arrange", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  arranged <- arrange(flights_db, year, month, day)
+
+  expect_is(arranged, "tbl_h2")
+})
+
+test_that("select", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  selected <- select(flights_db, year, month, day)
+
+  expect_is(selected, "tbl_h2")
+})
+
+test_that("distinct", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  distinct_tailnum <- distinct(select(flights_db, tailnum))
+
+  expect_is(distinct_tailnum, "tbl_h2")
+})
+
+test_that("group_by", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  grouped <- group_by(flights_db, tailnum)
+
+  expect_is(grouped, "tbl_h2")
+})
+
+test_that("summarize", {
+  flights_db <- tbl(nycflights13_h2(), "flights")
+
+  summarized <- summarise(flights_db,
+    count = n(),
+    distinct = n_distinct("flight"),
+    dist = mean(distance),
+    delay = mean(arr_delay),
+    min_delay = min(arr_delay),
+    max_delay = max(arr_delay),
+    sum_delay = sum(arr_delay),
+    sd_delay = sd(arr_delay)
+  )
+
+  expect_is(summarized, "tbl_h2")
+})
+