Packages

earmark

1.4.5
1.5.0-pre1 retired 1.5.0-pre retired 1.4.49 retired 1.4.48 retired 1.4.47 retired 1.4.46 retired 1.4.45 retired 1.4.44 retired 1.4.43 retired 1.4.42 retired 1.4.41 retired 1.4.40 retired 1.4.39 retired 1.4.38 retired 1.4.37 retired 1.4.36 retired 1.4.35 retired 1.4.34 retired 1.4.33 retired 1.4.32 retired 1.4.31 retired 1.4.30 retired 1.4.29 retired 1.4.28 retired 1.4.27 retired 1.4.26 retired 1.4.25 retired 1.4.24 retired 1.4.23 retired 1.4.22 retired 1.4.21 retired 1.4.20 retired 1.4.19 retired 1.4.18 retired 1.4.17 retired 1.4.16 retired 1.4.16-pre2 retired 1.4.16-pre1 retired 1.4.16-pre retired 1.4.15 retired 1.4.14 retired 1.4.13 retired 1.4.12 retired 1.4.10 retired 1.4.9 retired 1.4.8 retired 1.4.7 retired 1.4.6 retired 1.4.5 retired 1.4.4 retired 1.4.3 retired 1.4.2 retired 1.4.1 retired 1.4.0 retired 1.3.6 retired 1.3.5 retired 1.3.4 retired 1.3.3 retired 1.3.2 retired 1.3.1 retired 1.3.0 retired 1.2.6 retired 1.2.5 retired 1.2.4 retired 1.2.3 retired 1.2.2 retired 1.2.1 retired 1.2.0 retired 1.1.1 retired 1.1.0 retired 1.0.3 retired 1.0.2 retired 1.0.1 retired 1.0.0 retired 0.2.1 retired 0.2.0 retired 0.1.19 retired 0.1.18 retired 0.1.17 retired 0.1.16 retired 0.1.15 retired 0.1.14 retired 0.1.13 retired 0.1.12 retired 0.1.10 retired 0.1.9 retired 0.1.8 retired 0.1.7 retired 0.1.6 retired 0.1.5 retired 0.1.4 retired 0.1.3 retired 0.1.2 retired 0.1.1 retired 0.1.0 retired

Earmark is a pure-Elixir Markdown converter. It is intended to be used as a library (just call Earmark.as_html), but can also be used as a command-line tool (run mix escript.build first). Output generation is pluggable.

Retired package: Deprecated - Earmark is no longer maintained. Migrate to a replacement, for example MDEx (https://hex.pm/packages/mdex).
Security advisory: This version has known vulnerabilities. View advisories
HexDocs

Current section

Files

Jump to
Readme
95 Versions
0 Dependencies
114 Dependants
1 Advisory
Files
Activity
earmark lib earmark helpers html_parser.ex
Raw

lib/earmark/helpers/html_parser.ex 

defmodule Earmark.Helpers.HtmlParser do



  @moduledoc false



  import Earmark.Helpers.StringHelpers, only: [behead: 2]

  import Earmark.LineScanner, only: [void_tag?: 1]



  def parse_html(lines)

  def parse_html([tag_line|rest]) do

    case _parse_tag(tag_line) do

      nil                   -> [tag_line|rest]

      { :ok, tag, "" }      -> [_parse_rest(rest, tag, [])]

      { :ok, tag, suffix }  -> [_parse_rest(rest, tag, [suffix])]

      { :ext, tag, "" }     -> [_parse_rest(rest, tag, [])]

      { :ext, tag, suffix } -> [_parse_rest(rest, tag, []), [suffix]]

    end

  end



  # Parse One Tag

  # -------------



  @attribute ~r{\A([-\w]+)=(["'])(.*?)\2\s*}

  defp _parse_atts(string, tag, atts) do

    case Regex.run(@attribute, string) do

      [all, name, _delim, value] -> _parse_atts(behead(string, all), tag, [{name, value}|atts])

      _                          -> _parse_tag_tail(string, tag, atts)

    end

  end



  # Are leading and trailing "-"s ok?

  @tag_head ~r{\A\s*<([-\w]+)\s*}

  defp _parse_tag(string) do

    case Regex.run(@tag_head, string) do

      [all, tag] -> _parse_atts(behead(string, all), tag, [])

      _          -> nil

    end

  end



  @tag_tail ~r{\A.*?(/?)>\s*(.*)\z}

  defp _parse_tag_tail(string, tag, atts) do

    case Regex.run(@tag_tail, string)  do

      [_, closing, suffix]  ->

        _close_tag_tail(tag, atts, closing != "", suffix)

      # [_, _, ""]            -> {:ok, {tag, Enum.reverse(atts)} }

      # [_, "", suffix]       -> {:ok, {tag, Enum.reverse(atts)}, suffix }

      # [_, _closing, suffix] -> {:ext, {tag, Enum.reverse(atts)}, suffix }

      _                     -> nil

    end

  end



  defp _close_tag_tail(tag, atts, closing?, suffix) do

    if closing? || void_tag?(tag) do

      {:ext, {tag, Enum.reverse(atts)}, suffix }

    else

      {:ok, {tag, Enum.reverse(atts)}, suffix }

    end

  end



  # 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

  defp _parse_rest([], tag_tpl, lines) do

    tag_tpl |> Tuple.append(Enum.reverse(lines))

  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])) |> 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

    _parse_rest(rest, tag_tpl, [inner_line|lines])

  end



end