+
+
+ very deep content
+
+
+
+ 
+
+ Hello
" + + assert Html5ever.parse(html) == + {:ok, + [ + {"html", [], + [ + {"head", [], []}, + {"body", [], [{"h1", [], ["Hello"]}, {:comment, " my comment "}]} + ]} + ]} + end + + test "does not parse with not valid UTF8 binary" do + invalid = + <<98, 29, 104, 122, 46, 145, 14, 37, 122, 155, 227, 121, 49, 120, 108, 209, 155, 113, 229, + 98, 90, 181, 146>> + + assert Html5ever.parse(invalid) == + {:error, "cannot transform bytes from binary to a valid UTF8 string"} end test "flat parse basic html" do @@ -38,6 +55,41 @@ defmodule Html5everTest do assert Html5ever.flat_parse(html) == ret end + test "does not flat parse with not valid UTF8 binary" do + invalid = + <<98, 29, 104, 122, 46, 145, 14, 37, 122, 155, 227, 121, 49, 120, 108, 209, 155, 113, 229, + 98, 90, 181, 146>> + + assert Html5ever.flat_parse(invalid) == + {:error, "cannot transform bytes from binary to a valid UTF8 string"} + end + + test "flat parse basic html with attributes as maps" do + # Duplicated attribute is removed. + html = "" + + ret = + {:ok, + %{ + nodes: %{ + 0 => %{id: 0, parent: nil, type: :document}, + 1 => %{children: [2, 3], id: 1, parent: 0, type: :element, attrs: %{}, name: "html"}, + 2 => %{children: [], id: 2, parent: 1, type: :element, attrs: %{}, name: "head"}, + 3 => %{ + children: [], + id: 3, + parent: 1, + type: :element, + attrs: %{"test" => "woo", "class" => "content"}, + name: "body" + } + }, + root: 0 + }} + + assert Html5ever.flat_parse_with_attributes_as_maps(html) == ret + end + test "parse example.com html" do html = read_html("example.html") assert {:ok, _} = Html5ever.parse(html) @@ -147,4 +199,155 @@ defmodule Html5everTest do ]} ]} = parsed end + + test "reasonably deep html with attributes as maps" do + html = """ + + + +
+
+
+
+
+
+
+ """
+
+ parsed = Html5ever.parse_with_attributes_as_maps(html)
+
+ assert {:ok,
+ [
+ {:doctype, "html", "", ""},
+ {"html", %{},
+ [
+ {"head", %{}, ["\n", " ", {"title", %{}, ["Test"]}, "\n", " "]},
+ "\n",
+ " ",
+ {"body", %{},
+ [
+ "\n",
+ " ",
+ {"div", %{"class" => "content"},
+ [
+ "\n",
+ " ",
+ {"span", %{},
+ [
+ "\n",
+ " ",
+ {"div", %{},
+ [
+ "\n",
+ " ",
+ {"span", %{},
+ [
+ "\n",
+ " ",
+ {"small", %{},
+ ["\n", " very deep content", "\n", " "]},
+ "\n",
+ " "
+ ]},
+ "\n",
+ " "
+ ]},
+ "\n",
+ " ",
+ {"img", %{"src" => "file.jpg"}, []},
+ "\n",
+ " "
+ ]},
+ "\n",
+ " "
+ ]},
+ "\n",
+ " ",
+ "\n",
+ "\n"
+ ]}
+ ]}
+ ]} = parsed
+ end
+
+ test "parse html with a template tag ignores template content" do
+ html = """
+
+
+
+
+ Document
+ +Flower
+
+
+
+
+ """
+
+ assert Html5ever.parse(html) ==
+ {:ok,
+ [
+ {:doctype, "html", "", ""},
+ {"html", [],
+ [
+ {"head", [], [{"title", [], ["With template"]}]},
+ "\n",
+ {"body", [],
+ ["\n", {"h1", [], ["Document"]}, "\n", {"template", [], []}, "\n", "\n", "\n"]}
+ ]}
+ ]}
+ end
+
+ test "parse html starting with a XML tag" do
+ html = """
+
+
+
+
+