Pass arguments to parser modules (#446)

* Add optional arguments to pass to the parser module when parsing documents and fragments

* Add tests for parse_document and parse_fragment with arguments

---------

Co-authored-by: Rick Littel <[email protected]>

lib/floki/html_parser.ex

@@ -21,15 +21,19 @@ defmodule Floki.HTMLParser do
 
  @default_parser Floki.HTMLParser.Mochiweb
 
- @callback parse_document(binary()) :: {:ok, Floki.html_tree()} | {:error, String.t()}
- @callback parse_fragment(binary()) :: {:ok, Floki.html_tree()} | {:error, String.t()}
+ @callback parse_document(binary(), list()) :: {:ok, Floki.html_tree()} | {:error, String.t()}
+ @callback parse_fragment(binary(), list()) :: {:ok, Floki.html_tree()} | {:error, String.t()}
 
  def parse_document(html, opts \\ []) do
- parser(opts).parse_document(html)
+ parser_args = opts[:parser_args] || []
+
+ parser(opts).parse_document(html, parser_args)
  end
 
  def parse_fragment(html, opts \\ []) do
- parser(opts).parse_fragment(html)
+ parser_args = opts[:parser_args] || []
+
+ parser(opts).parse_fragment(html, parser_args)
  end
 
  defp parser(opts) do

lib/floki/html_parser/fast_html.ex

@@ -3,13 +3,13 @@ defmodule Floki.HTMLParser.FastHtml do
  @moduledoc false
 
  @impl true
- def parse_document(html) do
- execute_with_module(fn module -> module.decode(html) end)
+ def parse_document(html, args) do
+ execute_with_module(fn module -> module.decode(html, args) end)
  end
 
  @impl true
- def parse_fragment(html) do
- execute_with_module(fn module -> module.decode_fragment(html) end)
+ def parse_fragment(html, args) do
+ execute_with_module(fn module -> module.decode_fragment(html, args) end)
  end
 
  defp execute_with_module(fun) do

lib/floki/html_parser/html5ever.ex

@@ -4,7 +4,7 @@ defmodule Floki.HTMLParser.Html5ever do
  @moduledoc false
 
  @impl true
- def parse_document(html) do
+ def parse_document(html, _args) do
  case Code.ensure_loaded(Html5ever) do
  {:module, module} ->
  case module.parse(html) do
@@ -22,5 +22,5 @@ defmodule Floki.HTMLParser.Html5ever do
 
  # NOTE: html5ever does not implement parse_fragment yet.
  @impl true
- def parse_fragment(html), do: parse_document(html)
+ def parse_fragment(html, args), do: parse_document(html, args)
 end

lib/floki/html_parser/mochiweb.ex

@@ -5,7 +5,7 @@ defmodule Floki.HTMLParser.Mochiweb do
  @root_node "floki"
 
  @impl true
- def parse_document(html) do
+ def parse_document(html, _args) do
  html = "<#{@root_node}>#{html}</#{@root_node}>"
  {@root_node, [], parsed} = :floki_mochi_html.parse(html)
 
@@ -14,5 +14,5 @@ defmodule Floki.HTMLParser.Mochiweb do
 
  # NOTE: mochi_html cannot make a distinction of a fragment and document.
  @impl true
- def parse_fragment(html), do: parse_document(html)
+ def parse_fragment(html, args), do: parse_document(html, args)
 end

test/floki_test.exs

@@ -175,6 +175,80 @@ defmodule FlokiTest do
 
  Enum.each(@plain_text_tags, validate_html)
  end
+
+ @tag only_parser: FastHtml
+ test "parses all elements as strings by default" do
+ html = html_body(~s(<div><p>Content</p><custom>Custom</custom></div>))
+
+ {:ok, parsed} = Floki.parse_document(html)
+
+ assert [
+ {
+ "html",
+ [],
+ [
+ {"head", [], []},
+ {
+ "body",
+ [],
+ [
+ {"div", [], [{"p", [], ["Content"]}, {"custom", [], ["Custom"]}]}
+ ]
+ }
+ ]
+ }
+ ] = parsed
+ end
+
+ @tag only_parser: FastHtml
+ test "parses known elements as atoms when :html_atoms format argument is given" do
+ html = html_body(~s(<div><p>Content</p><custom>Custom</custom></div>))
+
+ {:ok, parsed} = Floki.parse_document(html, parser_args: [format: [:html_atoms]])
+
+ assert [
+ {
+ :html,
+ [],
+ [
+ {:head, [], []},
+ {
+ :body,
+ [],
+ [
+ {:div, [], [{:p, [], ["Content"]}, {"custom", [], ["Custom"]}]}
+ ]
+ }
+ ]
+ }
+ ] == parsed
+ end
+ end
+
+ describe "parse_fragment/2" do
+ @tag only_parser: FastHtml
+ test "does not parse a table row with missing parent table tag by default" do
+ html = "<tr><td>Column 1</td><td>Column 2</td></tr>"
+
+ {:ok, parsed} = Floki.parse_fragment(html)
+
+ assert ["Column 1Column 2"] == parsed
+ end
+
+ @tag only_parser: FastHtml
+ test "parses a table row with missing parent table tag when table context is given" do
+ html = "<tr><td>1</td><td>2</td></tr>"
+
+ {:ok, parsed} = Floki.parse_fragment(html, parser_args: [context: "table"])
+
+ assert [
+ {
+ "tbody",
+ [],
+ [{"tr", [], [{"td", [], ["1"]}, {"td", [], ["2"]}]}]
+ }
+ ] == parsed
+ end
  end
 
  # Floki.raw_html/2