Fix markdown table parser consuming lines without pipes as table rows (#1626)

The TableRow rule in the PEG grammar allowed rows with zero pipe
characters (`TableItem2*`). This caused lines like `<br>` immediately
following a table to be parsed as single-cell table rows, producing
spurious `<td><br></td>` in the rendered HTML.

Change `TableItem2*` to `TableItem2+` so that rows not starting with `|`
must contain at least one `|` to be recognized as table rows.

Author: st0012

lib/rdoc/markdown.kpeg

@@ -1270,13 +1270,13 @@ Table = &{ github? }
 TableHead = TableItem2+:items "|"? @Newline
             { items }
 
-TableRow = ( ( TableItem:item1 TableItem2*:items { [item1, *items] } ):row | TableItem2+:row ) "|"? @Newline
+TableRow = ( ( TableItem:item1 TableItem2+:items { [item1, *items] } ):row | TableItem2+:row ) "|"? @Newline
 	   { row }
 TableItem2 = "|" TableItem
 TableItem = < /(?:\\.|[^|\n])+/ >
 	    { text.strip.gsub(/\\([|])/, '\1')  }
 
-TableLine = ( ( TableAlign:align1 TableAlign2*:aligns {[align1, *aligns] } ):line | TableAlign2+:line ) "|"? @Newline
+TableLine = ( ( TableAlign:align1 TableAlign2+:aligns {[align1, *aligns] } ):line | TableAlign2+:line ) "|"? @Newline
 	    { line }
 TableAlign2 = "|" @Sp TableAlign
 TableAlign = < /:?-+:?/ > @Sp

lib/rdoc/markdown.rb

@@ -15937,7 +15937,7 @@ def _TableHead
     return _tmp
   end
 
-  # TableRow = ((TableItem:item1 TableItem2*:items { [item1, *items] }):row | TableItem2+:row) "|"? @Newline { row }
+  # TableRow = ((TableItem:item1 TableItem2+:items { [item1, *items] }):row | TableItem2+:row) "|"? @Newline { row }
   def _TableRow
 
     _save = self.pos
@@ -15954,14 +15954,21 @@ def _TableRow
             self.pos = _save2
             break
           end
+          _save3 = self.pos
           _ary = []
-          while true
-            _tmp = apply(:_TableItem2)
-            _ary << @result if _tmp
-            break unless _tmp
+          _tmp = apply(:_TableItem2)
+          if _tmp
+            _ary << @result
+            while true
+              _tmp = apply(:_TableItem2)
+              _ary << @result if _tmp
+              break unless _tmp
+            end
+            _tmp = true
+            @result = _ary
+          else
+            self.pos = _save3
           end
-          _tmp = true
-          @result = _ary
           items = @result
           unless _tmp
             self.pos = _save2
@@ -16077,7 +16084,7 @@ def _TableItem
     return _tmp
   end
 
-  # TableLine = ((TableAlign:align1 TableAlign2*:aligns {[align1, *aligns] }):line | TableAlign2+:line) "|"? @Newline { line }
+  # TableLine = ((TableAlign:align1 TableAlign2+:aligns {[align1, *aligns] }):line | TableAlign2+:line) "|"? @Newline { line }
   def _TableLine
 
     _save = self.pos
@@ -16094,14 +16101,21 @@ def _TableLine
             self.pos = _save2
             break
           end
+          _save3 = self.pos
           _ary = []
-          while true
-            _tmp = apply(:_TableAlign2)
-            _ary << @result if _tmp
-            break unless _tmp
+          _tmp = apply(:_TableAlign2)
+          if _tmp
+            _ary << @result
+            while true
+              _tmp = apply(:_TableAlign2)
+              _ary << @result if _tmp
+              break unless _tmp
+            end
+            _tmp = true
+            @result = _ary
+          else
+            self.pos = _save3
           end
-          _tmp = true
-          @result = _ary
           aligns = @result
           unless _tmp
             self.pos = _save2
@@ -16666,10 +16680,10 @@ def _DefinitionListDefinition
   Rules[:_CodeFence] = rule_info("CodeFence", "&{ github? } Ticks3 (@Sp StrChunk:format)? Spnl < ((!\"`\" Nonspacechar)+ | !Ticks3 /`+/ | Spacechar | @Newline)+ > Ticks3 @Sp @Newline* { verbatim = RDoc::Markup::Verbatim.new text               verbatim.format = format.intern if format.instance_of?(String)               verbatim             }")
   Rules[:_Table] = rule_info("Table", "&{ github? } TableHead:header TableLine:line TableRow+:body {           table = RDoc::Markup::Table.new(header, line, body)           parse_table_cells(table)         }")
   Rules[:_TableHead] = rule_info("TableHead", "TableItem2+:items \"|\"? @Newline { items }")
-  Rules[:_TableRow] = rule_info("TableRow", "((TableItem:item1 TableItem2*:items { [item1, *items] }):row | TableItem2+:row) \"|\"? @Newline { row }")
+  Rules[:_TableRow] = rule_info("TableRow", "((TableItem:item1 TableItem2+:items { [item1, *items] }):row | TableItem2+:row) \"|\"? @Newline { row }")
   Rules[:_TableItem2] = rule_info("TableItem2", "\"|\" TableItem")
   Rules[:_TableItem] = rule_info("TableItem", "< /(?:\\\\.|[^|\\n])+/ > { text.strip.gsub(/\\\\([|])/, '\\1')  }")
-  Rules[:_TableLine] = rule_info("TableLine", "((TableAlign:align1 TableAlign2*:aligns {[align1, *aligns] }):line | TableAlign2+:line) \"|\"? @Newline { line }")
+  Rules[:_TableLine] = rule_info("TableLine", "((TableAlign:align1 TableAlign2+:aligns {[align1, *aligns] }):line | TableAlign2+:line) \"|\"? @Newline { line }")
   Rules[:_TableAlign2] = rule_info("TableAlign2", "\"|\" @Sp TableAlign")
   Rules[:_TableAlign] = rule_info("TableAlign", "< /:?-+:?/ > @Sp {                 text.start_with?(\":\") ?                 (text.end_with?(\":\") ? :center : :left) :                 (text.end_with?(\":\") ? :right : nil)               }")
   Rules[:_DefinitionList] = rule_info("DefinitionList", "&{ definition_lists? } DefinitionListItem+:list { RDoc::Markup::List.new :NOTE, *list.flatten }")

test/rdoc/rdoc_markdown_test.rb

@@ -1310,6 +1310,43 @@ def test_gfm_table_with_links_and_code
     assert_equal expected, doc
   end
 
+  def test_gfm_table_does_not_consume_following_line_without_pipe
+    doc = parse <<~MD
+    | Command | Shows |
+    |---------|-------|
+    | ri File | Document for Ruby class File. |
+    <br>
+
+    Following paragraph.
+    MD
+
+    head = %w[Command Shows]
+    align = [nil, nil]
+    body = [
+      ['ri File', 'Document for Ruby class File.'],
+    ]
+    expected = doc(
+      @RM::Table.new(head, align, body),
+      para('<br>'),
+      para('Following paragraph.')
+    )
+
+    assert_equal expected, doc
+  end
+
+  def test_gfm_table_separator_without_pipe_does_not_form_table
+    doc = parse <<~MD
+    | Command | Shows |
+    ---------
+    | ri File | Document for Ruby class File. |
+    MD
+
+    # The separator line has no pipe, so this should NOT parse as a table.
+    # "| Command | Shows |" becomes a paragraph, "---" becomes a paragraph,
+    # and "| ri File | ..." becomes a paragraph.
+    assert(doc.parts.none? { |part| part.is_a?(@RM::Table) })
+  end
+
   def test_gfm_table_with_backslashes_in_code_spans
     doc = parse <<~MD
     Plain text: `$\\\\` and `$\\\\ ` should work.