From b3b6916f59c29dddb49367dec64ec175430f8efa Mon Sep 17 00:00:00 2001
From: Benjamin Skov Kaas-Hansen <epiben@hey.com>
Date: Tue, 23 Jan 2024 15:56:54 +0100
Subject: [PATCH] Gracefully handle tables with only empty rows (#360)

---
 NEWS.md                        |  2 ++
 R/table.R                      |  1 +
 tests/testthat/_snaps/table.md |  8 ++++++++
 tests/testthat/test-table.R    | 12 ++++++++++++
 4 files changed, 23 insertions(+)

diff --git a/NEWS.md b/NEWS.md
index 5a5468ba..3f668a76 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,5 +1,7 @@
 # rvest (development version)
 
+* `html_table()` discards rows without cells (@epiben, #360).
+
 # rvest 1.0.3
 
 * Re-document to fix HTML issues in `.Rd`.
diff --git a/R/table.R b/R/table.R
index eb2f72e0..bee96c2c 100644
--- a/R/table.R
+++ b/R/table.R
@@ -127,6 +127,7 @@ html_table.xml_node <- function(x,
   ns <- xml2::xml_ns(x)
   rows <- xml2::xml_find_all(x, ".//tr", ns = ns)
   cells <- lapply(rows, xml2::xml_find_all, ".//td|.//th", ns = ns)
+  cells <- compact(cells)
 
   if (length(cells) == 0) {
     return(tibble::tibble())
diff --git a/tests/testthat/_snaps/table.md b/tests/testthat/_snaps/table.md
index 607dcdae..5d7cef58 100644
--- a/tests/testthat/_snaps/table.md
+++ b/tests/testthat/_snaps/table.md
@@ -90,3 +90,11 @@
 
     # A tibble: 0 x 0
 
+# can handle tables consisting of a single empty row
+
+    # A tibble: 0 x 0
+
+# can handle tables consisting of only empty rows
+
+    # A tibble: 0 x 0
+
diff --git a/tests/testthat/test-table.R b/tests/testthat/test-table.R
index 079e8a04..fc3f35fd 100644
--- a/tests/testthat/test-table.R
+++ b/tests/testthat/test-table.R
@@ -195,3 +195,15 @@ test_that("can handle empty tables", {
   table <- html_table(html)[[1]]
   expect_snapshot_output(table)
 })
+
+test_that("can handle tables consisting of a single empty row", {
+  html <- minimal_html('<table><tr></tr></table>')
+  table <- html_table(html)[[1]]
+  expect_snapshot_output(table)
+})
+
+test_that("can handle tables consisting of only empty rows", {
+  html <- minimal_html('<table><tr></tr><tr></tr></table>')
+  table <- html_table(html)[[1]]
+  expect_snapshot_output(table)
+})