Gracefully handle tables with only empty rows (#360)

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`.

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())

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
+

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)
+})