Refs: #288; Adding metadata for parsed html to AST and adapting Trans…

…form WIP [amend-me]

  - Some tests with manually annotated ast transformations

lib/earmark/helpers/html_parser.ex

@@ -66,6 +66,7 @@ defmodule Earmark.Helpers.HtmlParser do
  # Iterate over lines inside a tag
  # -------------------------------
 
+ @verbatim %{meta: %{verbatim: true}}
  defp _parse_rest(rest, tag_tpl, lines)
  # Hu? That should never have happened but let us return some
  # sensible data, allowing rendering after the corruped block
@@ -74,9 +75,9 @@ defmodule Earmark.Helpers.HtmlParser do
  end
  defp _parse_rest([last_line], {tag, _}=tag_tpl, lines) do
  case Regex.run(~r{\A</#{tag}>\s*(.*)}, last_line) do
- nil -> tag_tpl |> Tuple.append(Enum.reverse([last_line|lines]))
- [_, ""] -> tag_tpl |> Tuple.append(Enum.reverse(lines))
- [_, suffix] -> [tag_tpl |> Tuple.append(Enum.reverse(lines)), suffix]
+ nil -> tag_tpl |> Tuple.append(Enum.reverse([last_line|lines])) |> Tuple.append(@verbatim)
+ [_, ""] -> tag_tpl |> Tuple.append(Enum.reverse(lines)) |> Tuple.append(@verbatim)
+ [_, suffix] -> [tag_tpl |> Tuple.append(Enum.reverse(lines)) |> Tuple.append(@verbatim), suffix]
  end
  end
  defp _parse_rest([inner_line|rest], tag_tpl, lines) do

lib/earmark/transform.ex

@@ -41,51 +41,69 @@ defmodule Earmark.Transform do
  _to_html(ast, options, Map.get(options, :initial_indent, 0)) |> IO.iodata_to_binary
  end
 
- defp _to_html(ast, options, level)
- defp _to_html(elements, options, level) when is_list(elements) do
+ defp _to_html(ast, options, level, verbatim \\ false)
+ defp _to_html(elements, options, level, verbatim) when is_list(elements) do
  elements
- |> Enum.map(&_to_html(&1, options, level))
+ |> Enum.map(&_to_html(&1, options, level, verbatim))
  end
- defp _to_html(element, options, level) when is_binary(element) do
+ defp _to_html(element, options, level, false) when is_binary(element) do
  escape(element, options, level)
  end
+ defp _to_html(element, options, level, true) when is_binary(element) do
+ [make_indent(options, level), element]
+ end
  # Void tags: `area`, `br`, `hr`, `img`, and `wbr` are rendered slightly differently
- defp _to_html({"area", _, _}=tag, options, level), do: void_tag(tag, options, level)
- defp _to_html({"br", _, _}=tag, options, level), do: void_tag(tag, options, level)
- defp _to_html({"hr", _, _}=tag, options, level), do: void_tag(tag, options, level)
- defp _to_html({"img", _, _}=tag, options, level), do: void_tag(tag, options, level)
- defp _to_html({"wbr", _, _}=tag, options, level), do: void_tag(tag, options, level)
- defp _to_html({:comment, _, children}, options, level) do
+ defp _to_html({"area", _, _}=tag, options, level, _verbatim), do: void_tag(tag, options, level)
+ defp _to_html({"br", _, _}=tag, options, level, _verbatim), do: void_tag(tag, options, level)
+ defp _to_html({"hr", _, _}=tag, options, level, _verbatim), do: void_tag(tag, options, level)
+ defp _to_html({"img", _, _}=tag, options, level, _verbatim), do: void_tag(tag, options, level)
+ defp _to_html({"wbr", _, _}=tag, options, level, _verbatim), do: void_tag(tag, options, level)
+ defp _to_html({:comment, _, children}, options, level, _verbatim) do
  indent = make_indent(options, level)
  [ indent,
  "<!--", Enum.intersperse(children, ["\n", indent, " "]), "-->"]
  end
- defp _to_html({tag, atts, []}, options, level) do
+ defp _to_html({tag, atts, []}, options, level, _verbatim) do
  [ make_indent(options, level),
  open_tag(tag, atts),
  "</",
  tag,
  ">\n" ]
  end
- defp _to_html({"code", atts, children}, options, level) do
+ defp _to_html({"code", atts, children}, options, level, _verbatim) do
  [ make_indent(options, 0),
  open_tag("code", atts),
  Enum.join(children, "\n")|>Earmark.Helpers.escape(),
  "</code>"]
  end
- defp _to_html({"pre", atts, children}, options, level) do
+ defp _to_html({"pre", atts, children}, options, level, _verbatim) do
  [ make_indent(options, level),
  open_tag("pre", atts),
  _to_html(children, options, level),
  "</pre>\n"]
  end
- defp _to_html({tag, atts, children}, options, level) do
+ defp _to_html({"pre", atts, children, meta}, options, level, _verbatim) do
+ verbatim = Map.get(meta, :meta, %{}) |> Map.get(:verbatim, false)
+ [ make_indent(options, level),
+ open_tag("pre", atts),
+ _to_html(children, options, level, verbatim),
+ "</pre>\n"]
+ end
+ defp _to_html({tag, atts, children}, options, level, _verbatim) do
  [ make_indent(options, level),
  open_tag(tag, atts),
  "\n",
  _to_html(children, options, level+1),
  close_tag(tag, options, level)]
  end
+ defp _to_html({tag, atts, children, meta}, options, level, _verbatim) do
+ verbatim = Map.get(meta, :meta, %{}) |> Map.get(:verbatim, false)
+ [ make_indent(options, level),
+ open_tag(tag, atts),
+ "\n",
+ _to_html(children, options, level+1, verbatim),
+ close_tag(tag, options, level)]
+ end
 
  defp close_tag(tag, options, level) do
  [make_indent(options, level), "</", tag, ">\n"]

test/acceptance/ast/html_test.exs

@@ -5,11 +5,13 @@ defmodule Acceptance.Ast.HtmlBlocksTest do
 
  @moduletag :ast
 
+ @verbatim %{meta: %{verbatim: true}}
+
  describe "HTML blocks" do
  test "tables are just tables again (or was that mountains?)" do
  markdown = "<table>\n <tr>\n <td>\n hi\n </td>\n </tr>\n</table>\n\nokay.\n"
  ast = [
- {"table", [], [" <tr>", " <td>", " hi", " </td>", " </tr>"]},
+ {"table", [], [" <tr>", " <td>", " hi", " </td>", " </tr>"], @verbatim},
  {"p", [], ["okay."]}]
  messages = []
 
@@ -18,23 +20,23 @@ defmodule Acceptance.Ast.HtmlBlocksTest do
 
  test "div (ine?)" do
  markdown = "<div>\n *hello*\n <foo><a>\n</div>\n"
- ast = [{"div", [], [" *hello*", " <foo><a>"]}]
+ ast = [{"div", [], [" *hello*", " <foo><a>"], @verbatim}]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
 
  test "we are leaving html alone" do
  markdown = "<div>\n*Emphasized* text.\n</div>"
- ast = [{"div", [], ["*Emphasized* text."]}]
+ ast = [{"div", [], ["*Emphasized* text."], @verbatim}]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
 
  test "even block elements" do
  markdown = "<div>\n```elixir\ndefmodule Mine do\n```\n</div>"
- ast = [{"div", [], ["```elixir", "defmodule Mine do", "```"]}]
+ ast = [{"div", [], ["```elixir", "defmodule Mine do", "```"], @verbatim}]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
@@ -190,57 +192,57 @@ defmodule Acceptance.Ast.HtmlBlocksTest do
  describe "multiple tags in closing line" do
  test "FTF" do
  markdown = "<div class=\"my-div\">\nline\n</div>"
- ast = [{"div", [{"class", "my-div"}], ["line"]}]
+ ast = [{"div", [{"class", "my-div"}], ["line"], @verbatim}]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
  test "this is not closing" do
  markdown = "<div>\nline\n</hello></div>"
- ast = [{"div", [], ["line", "</hello></div>"]}]
+ ast = [{"div", [], ["line", "</hello></div>"], @verbatim}]
  messages = [{:warning, 1, "Failed to find closing <div>"}]
 
  assert as_ast(markdown) == {:error, ast, messages}
  end
  test "therefore the div continues" do
  markdown = "<div>\nline\n</hello></div>\n</div>"
- ast = [{"div", [], ["line", "</hello></div>"]}]
+ ast = [{"div", [], ["line", "</hello></div>"], @verbatim}]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
  test "...nor is this" do
  markdown = "<div>\nline\n<hello></div>"
- ast = [{"div", [], ["line", "<hello></div>"]}]
+ ast = [{"div", [], ["line", "<hello></div>"], @verbatim}]
  messages = [{:warning, 1, "Failed to find closing <div>"},
  {:warning, 3, "Failed to find closing <hello>"}]
 
  assert as_ast(markdown) == {:error, ast, messages}
  end
  test "however, this closes and keeps the garbage" do
  markdown = "<div>\nline\n</div><hello>"
- ast = [{"div", [], ["line"]}, "<hello>"]
+ ast = [{"div", [], ["line"], @verbatim}, "<hello>"]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
  test "however, this closes and keeps **whatever** garbage" do
  markdown = "<div>\nline\n</div> `garbage`"
- ast = [{"div", [], ["line"]}, "`garbage`"]
+ ast = [{"div", [], ["line"], @verbatim}, "`garbage`"]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
  test "however, this closes and keeps not even warnings" do
  markdown = "<div>\nline\n</div> `garbage"
- ast = [{"div", [], ["line"]}, "`garbage"]
+ ast = [{"div", [], ["line"], @verbatim}, "`garbage"]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}
  end
  test "however, this closes and kept garbage is not even inline formatted" do
  markdown = "<div>\nline\n</div> _garbage_"
- ast = [{"div", [], ["line"]}, "_garbage_"]
+ ast = [{"div", [], ["line"], @verbatim}, "_garbage_"]
  messages = []
 
  assert as_ast(markdown) == {:ok, ast, messages}

test/acceptance/html1/html_test.exs

@@ -7,29 +7,25 @@ defmodule Acceptance.Html1.Html1BlocksTest do
  @moduletag :html1
 
  describe "HTML blocks" do
- @tag :wip
  test "tables are just tables again (or was that mountains?)" do
  markdown = "<table>\n <tr>\n <td>\n hi\n </td>\n </tr>\n</table>\n\nokay.\n"
- html = construct([
- {:table, nil, {:tr, nil, {:td, nil, " hi"}}},
- {:p, nil, "okay."} ])
+ html = "<table>\n <tr> <td> hi </td> </tr></table>\n<p>\n okay.\n</p>\n"
  messages = []
 
  assert to_html1(markdown) == {:ok, html, messages}
  end
 
  test "div (ine?)" do
  markdown = "<div>\n *hello*\n <foo><a>\n</div>\n"
- html = construct({:div, nil, [" *hello*", " &lt;foo&gt;&lt;a&gt;"]})
-
+ html = "<div>\n *hello* <foo><a></div>\n"
  messages = []
 
  assert to_html1(markdown) == {:ok, html, messages}
  end
 
  test "we are leaving html alone" do
  markdown = "<div>\n*Emphasized* text.\n</div>"
- html = construct({:div, nil, "*Emphasized* text."})
+ html ="<div>\n *Emphasized* text.</div>\n"
  messages = []
 
  assert to_html1(markdown) == {:ok, html, messages}
@@ -159,18 +155,6 @@ defmodule Acceptance.Html1.Html1BlocksTest do
  assert to_html1(markdown) == {:ok, html, messages}
  end
 
- @tag :wip
- test "block elements close para" do
- markdown = "alpha\n<div></div>beta"
- html = construct([
- {:p, nil, "alpha"},
- "<div></div>beta"
- ])
- messages = []
-
- assert to_html1(markdown) == {:ok, html, messages}
- end
-
  test "block elements close para, atts do not matter" do
  markdown = "alpha\n<div class=\"first\"></div>beta"
  html = "<p>alpha</p>\n<div class=\"first\"></div>beta"