danschultzer · danschultzer · Apr 22, 2018 · Apr 16, 2018 · Apr 22, 2018
diff --git a/lib/premailex.ex b/lib/premailex.ex
@@ -3,6 +3,8 @@ defmodule Premailex do
   Documentation for Premailex.
   """
 
+  alias Premailex.HTMLParser
+
   @doc """
   Adds inline styles to an HTML string
 
@@ -29,7 +31,7 @@ defmodule Premailex do
   @spec to_text(String.t()) :: String.t()
   def to_text(html) do
     html
-    |> Floki.find("body")
+    |> HTMLParser.all("body")
     |> Premailex.HTMLToPlainText.process()
   end
 end
diff --git a/lib/premailex/html_inline_styles.ex b/lib/premailex/html_inline_styles.ex
@@ -39,7 +39,7 @@ defmodule Premailex.HTMLInlineStyles do
 
   defp add_rule_set_to_html(%{selector: selector, rules: rules, specificity: specificity}, html) do
     html
-    |> Floki.find(selector)
+    |> HTMLParser.all(selector)
     |> Enum.reduce(html, &update_style_for_html(&2, &1, rules, specificity))
   end
 
@@ -73,7 +73,7 @@ defmodule Premailex.HTMLInlineStyles do
 
   defp normalize_style(html) do
     html
-    |> Floki.find("[style]")
+    |> HTMLParser.all("[style]")
     |> Enum.reduce(html, &merge_styles(&2, &1))
   end
 

diff --git a/lib/premailex/html_parser.ex b/lib/premailex/html_parser.ex
@@ -4,6 +4,7 @@ defmodule Premailex.HTMLParser do
   """
 
   @default_parser Premailex.HTMLParser.Floki
+  @type html_tree :: tuple | list
 
   @doc """
   Parses a HTML string into an HTML tree.
@@ -13,7 +14,7 @@ defmodule Premailex.HTMLParser do
       iex> Premailex.HTMLParser.parse("<html><head></head><body><h1>Title</h1></body></html>")
       {"html", [], [{"head", [], []}, {"body", [], [{"h1", [], ["Title"]}]}]}
   """
-  @spec parse(String.t()) :: tuple
+  @spec parse(String.t()) :: html_tree
   def parse(html) do
     apply(parser(), :parse, [html])
   end
@@ -26,7 +27,7 @@ defmodule Premailex.HTMLParser do
       iex> Premailex.HTMLParser.all({"html", [], [{"head", [], []}, {"body", [], [{"h1", [], ["Title"]}]}]}, "h1")
       [{"h1", [], ["Title"]}]
   """
-  @spec all(tuple, String.t()) :: [tuple]
+  @spec all(html_tree, String.t()) :: [html_tree]
   def all(tree, selector) do
     apply(parser(), :all, [tree, selector])
   end
@@ -39,11 +40,24 @@ defmodule Premailex.HTMLParser do
       iex> Premailex.HTMLParser.to_string({"html", [], [{"head", [], []}, {"body", [], [{"h1", [], ["Title"]}]}]})
       "<html><head></head><body><h1>Title</h1></body></html>"
   """
-  @spec to_string(tuple) :: String.t()
+  @spec to_string(html_tree) :: String.t()
   def to_string(tree) do
     apply(parser(), :to_string, [tree])
   end
 
+  @doc """
+  Extracts text elements from the HTML tree.
+
+  ## Examples
+
+      iex> Premailex.HTMLParser.text({"html", [], [{"head", [], []}, {"body", [], [{"h1", [], ["Title"]}]}]})
+      "Title"
+  """
+  @spec text(html_tree) :: String.t()
+  def text(tree) do
+    apply(parser(), :text, [tree])
+  end
+
   defp parser() do
     Application.get_env(:premailex, :html_parser, @default_parser)
   end

diff --git a/lib/premailex/html_parser/floki.ex b/lib/premailex/html_parser/floki.ex
@@ -2,22 +2,28 @@ defmodule Premailex.HTMLParser.Floki do
   @moduledoc """
   API connection with Floki
   """
+  alias Premailex.HTMLParser
 
   @doc false
-  @spec parse(String.t()) :: tuple
+  @spec parse(String.t()) :: HTMLParser.html_tree()
   def parse(html) do
     Floki.parse(html)
   end
 
   @doc false
-  @spec all(tuple, String.t()) :: [tuple]
+  @spec all(HTMLParser.html_tree(), String.t()) :: [HTMLParser.html_tree()]
   def all(tree, selector) do
     Floki.find(tree, selector)
   end
 
   @doc false
-  @spec to_string(tuple) :: String.t()
+  @spec to_string(HTMLParser.html_tree()) :: String.t()
   def to_string(tree) do
     Floki.raw_html(tree)
   end
+
+  @spec text(HTMLParser.html_tree()) :: String.t()
+  def text(tree) do
+    Floki.text(tree)
+  end
 end
diff --git a/lib/premailex/html_parser/meeseeks.ex b/lib/premailex/html_parser/meeseeks.ex
@@ -3,10 +3,13 @@ defmodule Premailex.HTMLParser.Meeseeks do
   API connection with Meeseeks
   """
 
+  require Logger
   import Meeseeks.CSS
+  alias Premailex.HTMLParser
+  alias Meeseeks.{Selector.CSS.Parser.ParseError}
 
   @doc false
-  @spec parse(String.t()) :: tuple
+  @spec parse(String.t()) :: HTMLParser.html_tree()
   def parse(html) do
     html
     |> Meeseeks.parse()
@@ -15,21 +18,52 @@ defmodule Premailex.HTMLParser.Meeseeks do
       [html] -> html
       html -> html
     end
+    |> sanitize()
   end
 
   @doc false
-  @spec all(tuple, String.t()) :: [tuple]
+  @spec all(HTMLParser.html_tree(), String.t()) :: [HTMLParser.html_tree()]
   def all(tree, selector) do
-    tree
-    |> Meeseeks.all(css("#{selector}"))
-    |> Enum.map(&Meeseeks.tree/1)
+    try do
+      tree
+      |> Meeseeks.all(css("#{selector}"))
+      |> Enum.map(&Meeseeks.tree/1)
+    rescue
+      e in ParseError ->
+        Logger.warn("Meeseeks CSS ParseError: " <> e.message)
+        []
+    end
   end
 
   @doc false
-  @spec to_string(tuple) :: String.t()
+  @spec to_string(HTMLParser.html_tree()) :: String.t()
   def to_string(tree) do
     tree
     |> Meeseeks.parse()
     |> Meeseeks.html()
   end
+
+  @doc false
+  @spec text(HTMLParser.html_tree()) :: String.t()
+  def text(text) when is_binary(text), do: text
+  def text(list) when is_list(list), do: Enum.map_join(list, "", &text/1)
+  def text({_element, _attrs, children}), do: text(children)
+
+  defp sanitize(list) when is_list(list) do
+    list
+    |> Enum.map(&sanitize/1)
+    |> Enum.reject(&is_empty?/1)
+  end
+
+  defp sanitize({elem, attr, children}) do
+    {elem, attr, sanitize(children)}
+  end
+
+  defp sanitize(any), do: any
+
+  defp is_empty?(text) when is_binary(text) do
+    String.trim(text) == ""
+  end
+
+  defp is_empty?(_any), do: false
 end
diff --git a/lib/premailex/html_to_plain_text.ex b/lib/premailex/html_to_plain_text.ex
@@ -2,7 +2,7 @@ defmodule Premailex.HTMLToPlainText do
   @moduledoc """
   Module that converts HTML emails to plain text.
   """
-  alias Premailex.Util
+  alias Premailex.{HTMLParser, Util}
 
   @doc """
   Processes HTML string into a plain text string.
@@ -14,7 +14,11 @@ defmodule Premailex.HTMLToPlainText do
 
   """
   @spec process(String.t() | Util.html_tree()) :: String.t()
-  def process(html) when is_binary(html), do: html |> Floki.parse() |> process()
+  def process(html) when is_binary(html) do
+    html
+    |> HTMLParser.parse()
+    |> process()
+  end
 
   def process(html) do
     html
@@ -26,7 +30,7 @@ defmodule Premailex.HTMLToPlainText do
     |> unordered_lists()
     |> ordered_lists()
     |> tables()
-    |> Floki.text()
+    |> HTMLParser.text()
     |> wordwrap()
     |> clear_linebreaks()
     |> String.trim()
@@ -46,30 +50,30 @@ defmodule Premailex.HTMLToPlainText do
   defp headings(html), do: Util.traverse(html, Enum.map(1..6, &"h#{&1}"), &heading(&1))
 
   defp heading({type, _, content}) do
-    text = content |> Floki.text()
+    text = HTMLParser.text(content)
 
     length =
       text
       |> String.split("\n")
       |> Enum.map(&String.length(&1))
       |> Enum.max()
 
-    "\n\n#{heading(type, text, length)}\n\n"
+    "\n\n" <> heading(type, text, length) <> "\n\n"
   end
 
   defp heading("h1", text, length) do
     heading_line = String.duplicate("*", length)
-    "#{heading_line}\n#{text}\n#{heading_line}"
+    heading_line <> "\n" <> text <> "\n" <> heading_line
   end
 
   defp heading("h2", text, length) do
     heading_line = String.duplicate("-", length)
-    "#{heading_line}\n#{text}\n#{heading_line}"
+    heading_line <> "\n" <> text <> "\n" <> heading_line
   end
 
   defp heading(_, text, length) do
     heading_line = String.duplicate("-", length)
-    "#{text}\n#{heading_line}"
+    text <> "\n" <> heading_line
   end
 
   defp links(html), do: Util.traverse(html, "a", &link(&1))
@@ -81,7 +85,7 @@ defmodule Premailex.HTMLToPlainText do
       |> elem(1)
       |> String.replace("mailto:", "")
 
-    text = Floki.text(content)
+    text = HTMLParser.text(content)
 
     link(String.trim(url), String.trim(text))
   end
@@ -92,7 +96,7 @@ defmodule Premailex.HTMLToPlainText do
   defp link(url, text, false), do: "#{text} (#{url})"
 
   defp paragraphs(html), do: Util.traverse(html, "p", &paragraph(&1))
-  defp paragraph({_, _, content}), do: "#{Floki.text(content)}\n\n"
+  defp paragraph({_, _, content}), do: HTMLParser.text(content) <> "\n\n"
 
   defp unordered_lists(html), do: Util.traverse(html, "ul", &unordered_list_items(&1))
 
@@ -102,7 +106,9 @@ defmodule Premailex.HTMLToPlainText do
     |> Enum.join("")
   end
 
-  defp unordered_list_item({_, _, content}), do: "* #{Floki.text(content)}\n"
+  defp unordered_list_item({_, _, content}) do
+    "* " <> HTMLParser.text(content) <> "\n"
+  end
 
   defp ordered_lists(html), do: Util.traverse(html, "ol", &ordered_list_items(&1))
 
@@ -113,24 +119,32 @@ defmodule Premailex.HTMLToPlainText do
     |> Enum.join("")
   end
 
-  defp ordered_list_item({_, _, content}, acc), do: "#{acc + 1}. #{Floki.text(content)}\n"
+  defp ordered_list_item({_, _, content}, acc) do
+    "#{acc + 1}. " <> HTMLParser.text(content) <> "\n"
+  end
 
   defp tables(html), do: Util.traverse(html, "table", &table(&1))
 
   defp table({_, _, table_rows}) do
-    # Callings tables/1 to make sure all nested tables have been processed
+    # Calling tables/1 to make sure all nested tables have been processed
     table_rows
     |> tables()
+    |> flatten_table_body()
     |> Util.traverse("tr", &table_rows(&1))
     |> Enum.join("\n")
   end
 
   defp table_rows({_, _, table_cells}) do
     table_cells
-    |> Util.traverse("td", &Floki.text(&1))
+    |> Util.traverse("td", &HTMLParser.text(&1))
     |> Enum.join(" ")
   end
 
+  defp flatten_table_body([tree]), do: flatten_table_body(tree)
+  defp flatten_table_body(list) when is_list(list), do: Enum.map(list, &flatten_table_body/1)
+  defp flatten_table_body({"tbody", [], table_cells}), do: table_cells
+  defp flatten_table_body(elem), do: elem
+
   defp wordwrap(text) do
     text
     |> String.split("\n")

diff --git a/lib/premailex/util.ex b/lib/premailex/util.ex
@@ -3,8 +3,8 @@ defmodule Premailex.Util do
   Module that contains utility functions.
   """
 
-  @type html_tree :: tuple | list
-  @type needle :: binary | tuple | list
+  @type html_tree :: Premailex.HTMLParser.html_tree()
+  @type needle :: binary | html_tree
 
   @doc """
   Traverses tree searching for needle, and will call provided function on