small fix

Author: marcburri

R/revisions.R

@@ -2347,7 +2347,7 @@ get_nth_release <- function(df, n = 0, diagonal = FALSE) {
             dplyr::summarise(max_time = max(.data$time), .groups = "drop"),
           by = c("id", "min_pub_date" = "pub_date")
         ) %>%
-        dplyr::select(.data$id, .data$max_time)
+        dplyr::select("id", "max_time")
 
       nth_release <- nth_release %>%
         dplyr::left_join(diagonal_thresholds, by = "id") %>%
@@ -2456,10 +2456,28 @@ get_first_release <- function(df, diagonal = FALSE) {
   }
 
   if (diagonal) {
-    df <- df %>%
-      dplyr::filter(
-        !dplyr::lead(.data$pub_date) == (.data$pub_date)
-      )
+    if ("id" %in% colnames(df)) {
+      diagonal_thresholds <- df %>%
+        dplyr::group_by(.data$id) %>%
+        dplyr::summarise(min_pub_date = min(.data$pub_date), .groups = "drop") %>%
+        dplyr::left_join(
+          df %>%
+            dplyr::group_by(.data$id, .data$pub_date) %>%
+            dplyr::summarise(max_time = max(.data$time), .groups = "drop"),
+          by = c("id", "min_pub_date" = "pub_date")
+        ) %>%
+        dplyr::select("id", "max_time")
+
+      df <- df %>%
+        dplyr::left_join(diagonal_thresholds, by = "id") %>%
+        dplyr::filter(.data$time >= .data$max_time) %>%
+        dplyr::select(-"max_time")
+    } else {
+      min_pub_date <- min(df$pub_date)
+      max_time <- max(df$time[df$pub_date == min_pub_date])
+      df <- df %>%
+        dplyr::filter(.data$time >= max_time)
+    }
   }
 
   # Add the class only if it is not already present

R/utils.R

@@ -71,15 +71,17 @@ vintages_long <- function(df, names_to = "pub_date", keep_na = FALSE) {
       check <- vintages_check(df[[id]])
       if (check == "long") {
         rlang::warn("The input data is already in long format.")
-        return(df[[id]])
+        long_df_tmp <- df[[id]] %>%
+          dplyr::mutate(id = .env$id)
+      } else {
+        long_df_tmp <- df[[id]] %>%
+          tidyr::pivot_longer(
+            cols = -"time",
+            names_to = names_to,
+            values_to = "value"
+          ) %>%
+          dplyr::mutate(id = .env$id)
       }
-      long_df_tmp <- df[[id]] %>%
-        tidyr::pivot_longer(
-          cols = -"time",
-          names_to = names_to,
-          values_to = "value"
-        ) %>%
-        dplyr::mutate(id = id)
 
       if (keep_na) {
         return(long_df_tmp)

tests/testthat/test-revisions.R

@@ -379,6 +379,21 @@ test_that("get_first_release returns first release", {
   expect_true(all(result$release == "release_0"))
 })
 
+test_that("get_first_release diagonal keeps the real first-release diagonal", {
+  df_single_wide <- data.frame(
+    time = as.Date(c("2020-01-01", "2020-02-01", "2020-03-01")),
+    `2020-03-01` = c(1, 2, 3),
+    check.names = FALSE
+  )
+
+  result <- get_first_release(vintages_long(df_single_wide), diagonal = TRUE)
+
+  expect_equal(nrow(result), 1)
+  expect_equal(result$time, as.Date("2020-03-01"))
+  expect_equal(result$pub_date, as.Date("2020-03-01"))
+  expect_true(all(result$release == "release_0"))
+})
+
 # ===== Tests for get_latest_release =====
 
 test_that("get_latest_release returns latest release", {

tests/testthat/test-utils.R

@@ -96,6 +96,19 @@ test_that("vintages_long handles list input", {
   expect_equal(nrow(result), 48)  # 12 * 2 vintages * 2 ids
 })
 
+test_that("vintages_long keeps ids for long-format list input", {
+  long_list <- list(US = df_long, EA = df_long)
+
+  expect_warning(
+    result <- vintages_long(long_list, names_to = "pub_date"),
+    "already in long format"
+  )
+
+  expect_true("id" %in% colnames(result))
+  expect_equal(sort(unique(result$id)), c("EA", "US"))
+  expect_equal(unname(as.integer(table(result$id))), c(nrow(df_long), nrow(df_long)))
+})
+
 test_that("vintages_long validates names_to parameter", {
   expect_error(
     vintages_long(df_wide, names_to = "invalid"),
@@ -591,4 +604,4 @@ test_that("conversion handles tibbles", {
 
   long_again <- vintages_long(wide, names_to = "pub_date")
   expect_s3_class(long_again, "tbl_df")
-})
+})