2019-01-01 20:26:40 +00:00
|
|
|
defmodule Pleroma.Web.RichMedia.Parsers.OGP do
|
|
|
|
def parse(html, data) do
|
|
|
|
with elements = [_ | _] <- get_elements(html),
|
|
|
|
ogp_data =
|
|
|
|
Enum.reduce(elements, data, fn el, acc ->
|
|
|
|
attributes = normalize_attributes(el)
|
|
|
|
|
|
|
|
Map.merge(acc, attributes)
|
|
|
|
end) do
|
|
|
|
{:ok, ogp_data}
|
|
|
|
else
|
|
|
|
_e -> {:error, "No OGP metadata found"}
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
defp get_elements(html) do
|
|
|
|
html |> Floki.find("meta[property^='og:']")
|
|
|
|
end
|
|
|
|
|
2019-01-01 20:29:47 +00:00
|
|
|
defp normalize_attributes(html_node) do
|
|
|
|
{_tag, attributes, _children} = html_node
|
2019-01-01 20:26:40 +00:00
|
|
|
|
|
|
|
data =
|
|
|
|
Enum.into(attributes, %{}, fn {name, value} ->
|
|
|
|
{name, String.trim_leading(value, "og:")}
|
|
|
|
end)
|
|
|
|
|
|
|
|
%{String.to_atom(data["property"]) => data["content"]}
|
|
|
|
end
|
|
|
|
end
|