From 0d3c08d5ec2d49cca3dc6604bbd08ac22319be49 Mon Sep 17 00:00:00 2001
From: Alexander Rossell Hayes <alexander@rossellhayes.com>
Date: Mon, 1 Apr 2024 22:48:45 -0700
Subject: [PATCH] test(census_geo_api): add test that age/sex subsets sum to
 race totals

---
 tests/testthat/test-census_geo_api.R | 49 +++++++++++++++++++++++++++-
 1 file changed, 48 insertions(+), 1 deletion(-)

diff --git a/tests/testthat/test-census_geo_api.R b/tests/testthat/test-census_geo_api.R
index d4d3534..8588bf6 100644
--- a/tests/testthat/test-census_geo_api.R
+++ b/tests/testthat/test-census_geo_api.R
@@ -2,7 +2,6 @@ skip_if_not(nzchar(Sys.getenv("CENSUS_API_KEY")))
 
 test_that("snapshot", {
   # TODO: Test that sub-geographies sum to match pooled geographies (e.g. blocks sum to block groups, sum to tracts, sum to counties)
-  # TODO: Test that age/sex subsets sum to race totals
   
   # These snapshots were generated using the calculations in v2.0.0
   # and verified that the calculations resulted in the same numbers for PR #120.
@@ -23,3 +22,51 @@ test_that("snapshot", {
     style = "deparse"
   )
 })
+
+expect_subset_sums_equal_overall_total <- function(data) {
+  `%>%` <- dplyr::`%>%`
+  
+  sums <- data %>%
+    dplyr::select(-dplyr::starts_with("r_")) %>%
+    tidyr::pivot_longer(dplyr::starts_with("P")) %>%
+    dplyr::mutate(
+      subset = dplyr::case_when(
+        grepl("001", name) ~ "overall",
+        .default = "subset"
+      ),
+      name = sub("_.+", "", name)
+    ) %>%
+    dplyr::summarise(
+      value = sum(value),
+      .by = -value
+    ) %>%
+    dplyr::summarize(
+      are_equal = length(unique(value)) <= 1,
+      .by = c(-subset, -value)
+    )
+  
+  expect_true(all(sums$are_equal))
+}
+
+test_that("sums", {
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2020", sex = TRUE)
+  )
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2020", age = TRUE)
+  )
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2020", age = TRUE, sex = TRUE)
+  )
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2010", sex = TRUE)
+  )
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2010", age = TRUE)
+  )
+  expect_subset_sums_equal_overall_total(
+    census_geo_api(state = "DE", geo = "county", year = "2010", age = TRUE, sex = TRUE)
+  )
+})
+
+# TODO: Test that all variables sum to total population of geography