Page MenuHomePhorge
Files F85712930

No OneTemporary
Actions

Size
10 KB
Referenced Files
None
Subscribers
None

View Options

diff --git a/lib/pleroma/web/rich_media/parser.ex b/lib/pleroma/web/rich_media/parser.ex
index 17a6ad617..837182d8a 100644
--- a/lib/pleroma/web/rich_media/parser.ex
+++ b/lib/pleroma/web/rich_media/parser.ex
@@ -1,169 +1,169 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser do
require Logger
@cachex Pleroma.Config.get([:cachex, :provider], Cachex)
defp parsers do
Pleroma.Config.get([:rich_media, :parsers])
end
def parse(nil), do: {:error, "No URL provided"}
if Pleroma.Config.get(:env) == :test do
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url), do: parse_url(url)
else
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url) do
with {:ok, data} <- get_cached_or_parse(url),
{:ok, _} <- set_ttl_based_on_image(data, url) do
{:ok, data}
end
end
defp get_cached_or_parse(url) do
case @cachex.fetch(:rich_media_cache, url, fn ->
case parse_url(url) do
{:ok, _} = res ->
{:commit, res}
{:error, reason} = e ->
# Unfortunately we have to log errors here, instead of doing that
# along with ttl setting at the bottom. Otherwise we can get log spam
# if more than one process was waiting for the rich media card
# while it was generated. Ideally we would set ttl here as well,
# so we don't override it number_of_waiters_on_generation
# times, but one, obviously, can't set ttl for not-yet-created entry
# and Cachex doesn't support returning ttl from the fetch callback.
log_error(url, reason)
{:commit, e}
end
end) do
{action, res} when action in [:commit, :ok] ->
case res do
{:ok, _data} = res ->
res
{:error, reason} = e ->
if action == :commit, do: set_error_ttl(url, reason)
e
end
{:error, e} ->
{:error, {:cachex_error, e}}
end
end
defp set_error_ttl(_url, :body_too_large), do: :ok
defp set_error_ttl(_url, {:content_type, _}), do: :ok
# The TTL is not set for the errors above, since they are unlikely to change
# with time
defp set_error_ttl(url, _reason) do
ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
@cachex.expire(:rich_media_cache, url, ttl)
:ok
end
defp log_error(url, {:invalid_metadata, data}) do
Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
end
defp log_error(url, reason) do
Logger.warning(fn -> "Rich media error for #{url}: #{inspect(reason)}" end)
end
end
@doc """
Set the rich media cache based on the expiration time of image.
Adopt behaviour `Pleroma.Web.RichMedia.Parser.TTL`
## Example
defmodule MyModule do
@behaviour Pleroma.Web.RichMedia.Parser.TTL
def ttl(data, url) do
image_url = Map.get(data, :image)
# do some parsing in the url and get the ttl of the image
# and return ttl is unix time
parse_ttl_from_url(image_url)
end
end
Define the module in the config
config :pleroma, :rich_media,
ttl_setters: [MyModule]
"""
@spec set_ttl_based_on_image(map(), String.t()) ::
{:ok, integer() | :noop} | {:error, :no_key}
def set_ttl_based_on_image(data, url) do
case get_ttl_from_image(data, url) do
- {:ok, ttl} when is_number(ttl) ->
+ ttl when is_number(ttl) ->
ttl = ttl * 1000
case @cachex.expire_at(:rich_media_cache, url, ttl) do
{:ok, true} -> {:ok, ttl}
{:ok, false} -> {:error, :no_key}
end
_ ->
{:ok, :noop}
end
end
defp get_ttl_from_image(data, url) do
[:rich_media, :ttl_setters]
|> Pleroma.Config.get()
|> Enum.reduce({:ok, nil}, fn
module, {:ok, _ttl} ->
module.ttl(data, url)
_, error ->
error
end)
end
def parse_url(url) do
with {:ok, %Tesla.Env{body: html}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url),
{:ok, html} <- Floki.parse_document(html) do
html
|> maybe_parse()
|> Map.put("url", url)
|> clean_parsed_data()
|> check_parsed_data()
end
end
defp maybe_parse(html) do
Enum.reduce_while(parsers(), %{}, fn parser, acc ->
case parser.parse(html, acc) do
data when data != %{} -> {:halt, data}
_ -> {:cont, acc}
end
end)
end
defp check_parsed_data(%{"title" => title} = data)
when is_binary(title) and title != "" do
{:ok, data}
end
defp check_parsed_data(data) do
{:error, {:invalid_metadata, data}}
end
defp clean_parsed_data(data) do
data
|> Enum.reject(fn {key, val} ->
not match?({:ok, _}, Jason.encode(%{key => val}))
end)
|> Map.new()
end
end
diff --git a/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
index fa41c160d..c8c869792 100644
--- a/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
+++ b/lib/pleroma/web/rich_media/parser/ttl/aws_signed_url.ex
@@ -1,50 +1,50 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl do
@behaviour Pleroma.Web.RichMedia.Parser.TTL
@impl true
def ttl(data, _url) do
image = Map.get(data, :image)
if is_aws_signed_url(image) do
image
|> parse_query_params()
|> format_query_params()
|> get_expiration_timestamp()
else
{:error, "Not aws signed url #{inspect(image)}"}
end
end
defp is_aws_signed_url(image) when is_binary(image) and image != "" do
%URI{host: host, query: query} = URI.parse(image)
String.contains?(host, "amazonaws.com") and String.contains?(query, "X-Amz-Expires")
end
defp is_aws_signed_url(_), do: nil
defp parse_query_params(image) do
%URI{query: query} = URI.parse(image)
query
end
defp format_query_params(query) do
query
|> String.split(~r/&|=/)
|> Enum.chunk_every(2)
|> Map.new(fn [k, v] -> {k, v} end)
end
defp get_expiration_timestamp(params) when is_map(params) do
{:ok, date} =
params
|> Map.get("X-Amz-Date")
|> Timex.parse("{ISO:Basic:Z}")
- {:ok, Timex.to_unix(date) + String.to_integer(Map.get(params, "X-Amz-Expires"))}
+ Timex.to_unix(date) + String.to_integer(Map.get(params, "X-Amz-Expires"))
end
end
diff --git a/test/pleroma/web/rich_media/parser/ttl/aws_signed_url_test.exs b/test/pleroma/web/rich_media/parser/ttl/aws_signed_url_test.exs
index 59b3330ba..b90f7d9e2 100644
--- a/test/pleroma/web/rich_media/parser/ttl/aws_signed_url_test.exs
+++ b/test/pleroma/web/rich_media/parser/ttl/aws_signed_url_test.exs
@@ -1,81 +1,81 @@
# Pleroma: A lightweight social networking server
# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrlTest do
# Relies on Cachex, needs to be synchronous
use Pleroma.DataCase
test "s3 signed url is parsed correct for expiration time" do
url = "https://pleroma.social/amz"
{:ok, timestamp} =
Timex.now()
|> DateTime.truncate(:second)
|> Timex.format("{ISO:Basic:Z}")
# in seconds
valid_till = 30
metadata = construct_metadata(timestamp, valid_till, url)
expire_time =
Timex.parse!(timestamp, "{ISO:Basic:Z}") |> Timex.to_unix() |> Kernel.+(valid_till)
- assert {:ok, expire_time} == Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl.ttl(metadata, url)
+ assert expire_time == Pleroma.Web.RichMedia.Parser.TTL.AwsSignedUrl.ttl(metadata, url)
end
test "s3 signed url is parsed and correct ttl is set for rich media" do
url = "https://pleroma.social/amz"
{:ok, timestamp} =
Timex.now()
|> DateTime.truncate(:second)
|> Timex.format("{ISO:Basic:Z}")
# in seconds
valid_till = 30
metadata = construct_metadata(timestamp, valid_till, url)
body = """
<meta name="twitter:card" content="Pleroma" />
<meta name="twitter:site" content="Pleroma" />
<meta name="twitter:title" content="Pleroma" />
<meta name="twitter:description" content="Pleroma" />
<meta name="twitter:image" content="#{Map.get(metadata, :image)}" />
"""
Tesla.Mock.mock(fn
%{
method: :get,
url: "https://pleroma.social/amz"
} ->
%Tesla.Env{status: 200, body: body}
end)
Cachex.put(:rich_media_cache, url, metadata)
Pleroma.Web.RichMedia.Parser.set_ttl_based_on_image(metadata, url)
{:ok, cache_ttl} = Cachex.ttl(:rich_media_cache, url)
# as there is delay in setting and pulling the data from cache we ignore 1 second
# make it 2 seconds for flakyness
assert_in_delta(valid_till * 1000, cache_ttl, 2000)
end
defp construct_s3_url(timestamp, valid_till) do
"https://pleroma.s3.ap-southeast-1.amazonaws.com/sachin%20%281%29%20_a%20-%25%2Aasdasd%20BNN%20bnnn%20.png?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIBLWWK6RGDQXDLJQ%2F20190716%2Fap-southeast-1%2Fs3%2Faws4_request&X-Amz-Date=#{timestamp}&X-Amz-Expires=#{valid_till}&X-Amz-Signature=04ffd6b98634f4b1bbabc62e0fac4879093cd54a6eed24fe8eb38e8369526bbf&X-Amz-SignedHeaders=host"
end
defp construct_metadata(timestamp, valid_till, url) do
%{
image: construct_s3_url(timestamp, valid_till),
site: "Pleroma",
title: "Pleroma",
description: "Pleroma",
url: url
}
end
end

File Metadata

Mime Type
text/x-diff
Expires
Sat, Sep 19, 1:42 PM (19 h, 24 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
1769533
Default Alt Text
(10 KB)

Event Timeline