Merge pull request #331 from pragdave/i228-block-level-html

Fixes: #228

lib/earmark/line_scanner.ex

@@ -155,9 +155,9 @@ defmodule Earmark.LineScanner do
         [_, tag] = match
         %Line.HtmlOpenTag{tag: tag, content: line, indent: 0}
 
-      match = !recursive && Regex.run(~r/\A<\/([-\w]+?)>/, line) ->
-        [_, tag] = match
-        %Line.HtmlCloseTag{tag: tag, indent: 0}
+      match = !recursive && Regex.run(~r/\A(\s{0,3})<\/([-\w]+?)>/, line) ->
+        [_, leading_spaces, tag] = match
+        %Line.HtmlCloseTag{tag: tag, indent: String.length(leading_spaces)}
 
       match = Regex.run(@id_re, line) ->
         [_, leading, id, url | title] = match

lib/earmark/parser.ex

@@ -220,13 +220,12 @@ defmodule Earmark.Parser do
   # HTML block #
   ##############
   defp _parse([ opener = %Line.HtmlOpenTag{tag: tag, lnb: lnb} | rest], result, options, recursive) do
-    {html_lines, rest, unclosed} = html_match_to_closing(opener, rest)
+    {html_lines, rest1, unclosed} = _html_match_to_closing(opener, rest) 
     options1 = add_messages(options,
                             unclosed
                             |> Enum.map(fn %{lnb: lnb1, tag: tag} -> {:warning, lnb1, "Failed to find closing <#{tag}>"} end))
-
-    html = (for line <- Enum.reverse(html_lines), do: line.line)
-    _parse(rest, [ %Block.Html{tag: tag, html: html, lnb: lnb} | result ], options1, recursive)
+    html = Enum.reverse(html_lines)
+    _parse(rest1, [ %Block.Html{tag: tag, html: html, lnb: lnb} | result ], options1, recursive)
   end
 
   ####################
@@ -351,7 +350,7 @@ defmodule Earmark.Parser do
   end
 
   defp _consolidate_para( [line | rest] = lines, result, pending ) do
-    case inline_or_text?( line, pending ) do
+    case _inline_or_text?( line, pending ) do
       %{pending: still_pending, continue: true} -> _consolidate_para( rest, [line | result], still_pending )
       _                                         -> {result, lines, @not_pending}
     end
@@ -492,46 +491,46 @@ defmodule Earmark.Parser do
   # Consume HTML, taking care of nesting. Assumes one tag per line. #
   ###################################################################
 
-  defp html_match_to_closing(opener, rest), do: find_closing_tags([opener], rest, [opener])
+  defp _html_match_to_closing(opener, rest), do: _find_closing_tags([opener], rest, [String.trim_leading(opener.line)])
 
+  defp _find_closing_tags(needed, input, html_lines)
   # No more open tags, happy case
-  defp find_closing_tags([], rest, html_lines), do: {html_lines, rest, []}
-
+  defp _find_closing_tags([], rest, html_lines), do: {html_lines, rest, []}
   # run out of input, unhappy case
-  defp find_closing_tags(needed, [], html_lines), do: {html_lines, [], needed}
-
+  defp _find_closing_tags(needed, [], html_lines), do: {html_lines, [], needed}
   # still more lines, still needed closing
-  defp find_closing_tags(needed = [needed_hd|needed_tl], [rest_hd|rest_tl], html_lines) do
+  defp _find_closing_tags(needed = [needed_hd|needed_tl], [rest_hd|rest_tl], html_lines) do
     cond do
-      closes_tag?(rest_hd, needed_hd) -> find_closing_tags(needed_tl, rest_tl, [rest_hd|html_lines])
-      opens_tag?(rest_hd)             -> find_closing_tags([rest_hd|needed], rest_tl, [rest_hd|html_lines])
-      true                            -> find_closing_tags(needed, rest_tl, [rest_hd|html_lines])
+      _closes_tag?(rest_hd, needed_hd) -> _find_closing_tags(needed_tl, rest_tl, [String.trim_leading(rest_hd.line)|html_lines])
+      _opens_tag?(rest_hd)             -> _find_closing_tags([rest_hd|needed], rest_tl, [String.trim_leading(rest_hd.line)|html_lines])
+      true                             -> _find_closing_tags(needed, rest_tl, [rest_hd.line|html_lines])
     end
   end
 
   ###########
   # Helpers #
   ###########
 
-  defp closes_tag?(%Line.HtmlCloseTag{tag: ctag}, %Line.HtmlOpenTag{tag: otag}), do: ctag == otag
-  defp closes_tag?(_, _), do: false
+  defp _closes_tag?(%Line.HtmlCloseTag{tag: ctag}, %Line.HtmlOpenTag{tag: otag}) do 
+    ctag == otag
+  end
+  defp _closes_tag?(_, _), do: false
 
-  defp opens_tag?(%Line.HtmlOpenTag{}), do: true
-  defp opens_tag?(_), do: false
+  defp _opens_tag?(%Line.HtmlOpenTag{}), do: true
+  defp _opens_tag?(_), do: false
 
 
-  # (_,{'nil' | binary(),number()}) -> #{}jj
-  defp inline_or_text?(line, pending)
-  defp inline_or_text?(line = %Line.Text{}, @not_pending) do
+  defp _inline_or_text?(line, pending)
+  defp _inline_or_text?(line = %Line.Text{}, @not_pending) do
     pending = opens_inline_code(line)
     %{pending: pending, continue: true}
   end
-  defp inline_or_text?(line = %Line.TableLine{}, @not_pending) do
+  defp _inline_or_text?(line = %Line.TableLine{}, @not_pending) do
     pending = opens_inline_code(line)
     %{pending: pending, continue: true}
   end
-  defp inline_or_text?( _line, @not_pending), do: %{pending: @not_pending, continue: false}
-  defp inline_or_text?( line, pending ) do
+  defp _inline_or_text?( _line, @not_pending), do: %{pending: @not_pending, continue: false}
+  defp _inline_or_text?( line, pending ) do
     pending = still_inline_code(line, pending)
     %{pending: pending, continue: true}
   end

test/acceptance/ast/html_test.exs → test/acceptance/ast/html/block_test.exs

@@ -1,4 +1,4 @@
-defmodule Acceptance.Ast.HtmlBlocksTest do
+defmodule Acceptance.Ast.Html.BlockTest do
   use ExUnit.Case, async: true
   import Support.Helpers, only: [as_ast: 1, parse_html: 1]
   import Support.AstHelpers, only: [p: 1, void_tag: 1]
@@ -246,20 +246,7 @@ defmodule Acceptance.Ast.HtmlBlocksTest do
       assert as_ast(markdown) == {:ok, ast, messages}
     end
   end
-
-  describe "arbitrary tags" do
-    # Needs a fix with issue [#326](https://github.com/pragdave/earmark/issues/326)
-    @tag :wip
-    test "mixture of tags (was regtest #103)" do 
-      markdown = "<x>a\n<y></y>\n<y>\n<z>\n</z>\n<z>\n</x>"
-      html     = "<x>a\n<y></y>\n<y>\n<z>\n</z>\n<z>\n</x>"
-      ast      = parse_html(html)
-      messages = Enum.zip(1..3, ~w[x y z])
-                 |> Enum.map(fn {lnb, tag} -> {:warning, lnb, "Failed to find closing <#{tag}>"} end)
-
-      assert as_ast(markdown) == {:error, [ast], messages}
-    end
-  end
 end
 
 # SPDX-License-Identifier: Apache-2.0
+

test/acceptance/ast/html/permissive_test.exs

@@ -0,0 +1,40 @@
+defmodule Acceptance.Ast.Html.PermissiveTest do
+  use ExUnit.Case, async: true
+  import Support.Helpers, only: [as_ast: 1]
+
+  @verbatim %{meta: %{verbatim: true}}
+
+  describe "some nesting" do
+    test "simple case" do
+      markdown = "<div>\ncontent\n </div>"
+      ast      = [{"div", [], ["content"], @verbatim}]
+      messages = []
+
+      assert as_ast(markdown) == {:ok, ast, messages}
+    end
+
+    test "more nesting" do
+      markdown = "<div>\n<section>\n<span>content</span>\n</section>\n</div>"
+      ast      = [{"div", [],
+          ["<section>", "<span>content</span>", "</section>"], @verbatim}]
+      messages = []
+
+      assert as_ast(markdown) == {:ok, ast, messages}
+    end
+  end
+
+  describe "arbitrary tags" do
+    # Needs a fix with issue [#326](https://github.com/pragdave/earmark/issues/326)
+    test "mixture of tags (was regtest #103)" do 
+      markdown = "<x>a\n<y></y>\n<y>\n<z>\n</z>\n<z>\n</x>"
+      ast      = [{"x", '', ["a", "<y></y>", "<y>", "<z>", "</z>", "<z>"], @verbatim}]
+      messages = Enum.zip([1, 3, 6], ~w[x y z])
+                 |> Enum.map(fn {lnb, tag} -> {:warning, lnb, "Failed to find closing <#{tag}>"} end)
+
+      assert as_ast(markdown) == {:error, ast, messages}
+    end
+  end
+end
+
+# SPDX-License-Identifier: Apache-2.0
+

test/functional/scanner/line_type_test.exs

@@ -85,6 +85,7 @@ defmodule Functional.Scanner.LineTypeTest do
     { "<pre>",             %Line.HtmlOpenTag{tag: "pre", content: "<pre>"} },
     { "<pre class='123'>", %Line.HtmlOpenTag{tag: "pre", content: "<pre class='123'>"} },
     { "</pre>",            %Line.HtmlCloseTag{tag: "pre"} },
+    { "   </pre>",            %Line.HtmlCloseTag{indent: 3, tag: "pre"} },
 
     { "<pre>a</pre>",      %Line.HtmlOneLine{tag: "pre", content: "<pre>a</pre>"} },
 

test/performance/longer_input_test.exs

@@ -4,7 +4,7 @@ defmodule Performance.LongerInputTest do
   describe "some test data" do
     test "medium" do
       ast = convert_file("medium.md", :ast, 100)
-      assert Enum.count(ast) == 8300
+      assert Enum.count(ast) == 8500
     end
 
     test "show data" do