2019-07-10 05:13:23 +00:00
|
|
|
# Pleroma: A lightweight social networking server
|
|
|
|
# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
|
|
|
|
# SPDX-License-Identifier: AGPL-3.0-only
|
|
|
|
|
2019-01-10 18:09:56 +00:00
|
|
|
defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
|
|
|
|
def parse(html, data, prefix, error_message, key_name, value_name \\ "content") do
|
2019-06-22 13:12:57 +00:00
|
|
|
meta_data =
|
|
|
|
html
|
|
|
|
|> get_elements(key_name, prefix)
|
|
|
|
|> Enum.reduce(data, fn el, acc ->
|
|
|
|
attributes = normalize_attributes(el, prefix, key_name, value_name)
|
2019-06-13 08:02:46 +00:00
|
|
|
|
2019-06-22 13:12:57 +00:00
|
|
|
Map.merge(acc, attributes)
|
|
|
|
end)
|
|
|
|
|> maybe_put_title(html)
|
|
|
|
|
|
|
|
if Enum.empty?(meta_data) do
|
|
|
|
{:error, error_message}
|
2019-06-21 11:36:32 +00:00
|
|
|
else
|
2019-06-22 13:12:57 +00:00
|
|
|
{:ok, meta_data}
|
2019-01-10 18:09:56 +00:00
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
defp get_elements(html, key_name, prefix) do
|
|
|
|
html |> Floki.find("meta[#{key_name}^='#{prefix}:']")
|
|
|
|
end
|
|
|
|
|
|
|
|
defp normalize_attributes(html_node, prefix, key_name, value_name) do
|
|
|
|
{_tag, attributes, _children} = html_node
|
|
|
|
|
|
|
|
data =
|
|
|
|
Enum.into(attributes, %{}, fn {name, value} ->
|
|
|
|
{name, String.trim_leading(value, "#{prefix}:")}
|
|
|
|
end)
|
|
|
|
|
|
|
|
%{String.to_atom(data[key_name]) => data[value_name]}
|
|
|
|
end
|
2019-06-22 13:12:57 +00:00
|
|
|
|
|
|
|
defp maybe_put_title(%{title: _} = meta, _), do: meta
|
|
|
|
|
2019-06-22 14:22:59 +00:00
|
|
|
defp maybe_put_title(meta, html) when meta != %{} do
|
2019-06-22 13:12:57 +00:00
|
|
|
case get_page_title(html) do
|
|
|
|
"" -> meta
|
|
|
|
title -> Map.put_new(meta, :title, title)
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
2019-06-22 14:22:59 +00:00
|
|
|
defp maybe_put_title(meta, _), do: meta
|
|
|
|
|
2019-06-22 13:12:57 +00:00
|
|
|
defp get_page_title(html) do
|
|
|
|
Floki.find(html, "title") |> Floki.text()
|
|
|
|
end
|
2019-01-10 18:09:56 +00:00
|
|
|
end
|