Merge branch 'auto_linker' into 'develop'

AutoLinker

Closes #609

See merge request pleroma/pleroma!839

config/config.exs

@@ -344,6 +344,16 @@ config :pleroma, Pleroma.Jobs,
   federator_outgoing: [max_jobs: 50],
   mailer: [max_jobs: 10]
 
+config :auto_linker,
+  opts: [
+    scheme: true,
+    extra: true,
+    class: false,
+    strip_prefix: false,
+    new_window: false,
+    rel: false
+  ]
+
 # Import environment specific config. This must remain at the bottom
 # of this file so it overrides the configuration defined above.
 import_config "#{Mix.env()}.exs"

docs/config.md

@@ -107,7 +107,7 @@ config :pleroma, Pleroma.Mailer,
 
 An example to enable ONLY ExSyslogger (f/ex in ``prod.secret.exs``) with info and debug suppressed:
 ```
-config :logger, 
+config :logger,
   backends: [{ExSyslogger, :ex_syslogger}]
 
 config :logger, :ex_syslogger,
@@ -301,3 +301,28 @@ For each pool, the options are:
 * `max_connections` - how much connections a pool can hold
 * `timeout` - retention duration for connections
 
+## :auto_linker
+
+Configuration for the `auto_linker` library:
+
+* `class: "auto-linker"` - specify the class to be added to the generated link. false to clear
+* `rel: "noopener noreferrer"` - override the rel attribute. false to clear
+* `new_window: true` - set to false to remove `target='_blank'` attribute
+* `scheme: false` - Set to true to link urls with schema `http://google.com`
+* `truncate: false` - Set to a number to truncate urls longer then the number. Truncated urls will end in `..`
+* `strip_prefix: true` - Strip the scheme prefix
+* `extra: false` - link urls with rarely used schemes (magnet, ipfs, irc, etc.)
+
+Example:
+
+```exs
+config :auto_linker,
+  opts: [
+    scheme: true,
+    extra: true,
+    class: false,
+    strip_prefix: false,
+    new_window: false,
+    rel: false
+  ]
+```

lib/pleroma/formatter.ex

@@ -8,33 +8,51 @@ defmodule Pleroma.Formatter do
   alias Pleroma.User
   alias Pleroma.Web.MediaProxy
 
-  @tag_regex ~r/((?<=[^&])|\A)(\#)(\w+)/u
   @markdown_characters_regex ~r/(`|\*|_|{|}|[|]|\(|\)|#|\+|-|\.|!)/
+  @link_regex ~r{((?:http(s)?:\/\/)?[\w.-]+(?:\.[\w\.-]+)+[\w\-\._~%:/?#[\]@!\$&'\(\)\*\+,;=.]+)|[0-9a-z+\-\.]+:[0-9a-z$-_.+!*'(),]+}ui
 
-  # Modified from https://www.w3.org/TR/html5/forms.html#valid-e-mail-address
+  @auto_linker_config hashtag: true,
-  @mentions_regex ~r/@[a-zA-Z0-9.!#$%&'*+\/=?^_`{|}~-]*@?[a-zA-Z0-9_-](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*/u
+                      hashtag_handler: &Pleroma.Formatter.hashtag_handler/4,
+                      mention: true,
+                      mention_handler: &Pleroma.Formatter.mention_handler/4
 
-  def parse_tags(text, data \\ %{}) do
+  def mention_handler("@" <> nickname, buffer, opts, acc) do
-    Regex.scan(@tag_regex, text)
+    case User.get_cached_by_nickname(nickname) do
-    |> Enum.map(fn ["#" <> tag = full_tag | _] -> {full_tag, String.downcase(tag)} end)
+      %User{id: id} = user ->
-    |> (fn map ->
+        ap_id = get_ap_id(user)
-          if data["sensitive"] in [true, "True", "true", "1"],
+        nickname_text = get_nickname_text(nickname, opts) |> maybe_escape(opts)
-            do: [{"#nsfw", "nsfw"}] ++ map,
+
-            else: map
+        link =
-        end).()
+          "<span class='h-card'><a data-user='#{id}' class='u-url mention' href='#{ap_id}'>@<span>#{
+            nickname_text
+          }</span></a></span>"
+
+        {link, %{acc | mentions: MapSet.put(acc.mentions, {"@" <> nickname, user})}}
+
+      _ ->
+        {buffer, acc}
+    end
   end
 
-  @doc "Parses mentions text and returns list {nickname, user}."
+  def hashtag_handler("#" <> tag = tag_text, _buffer, _opts, acc) do
-  @spec parse_mentions(binary()) :: list({binary(), User.t()})
+    tag = String.downcase(tag)
-  def parse_mentions(text) do
+    url = "#{Pleroma.Web.base_url()}/tag/#{tag}"
-    Regex.scan(@mentions_regex, text)
+    link = "<a class='hashtag' data-tag='#{tag}' href='#{url}' rel='tag'>#{tag_text}</a>"
-    |> List.flatten()
+
-    |> Enum.uniq()
+    {link, %{acc | tags: MapSet.put(acc.tags, {tag_text, tag})}}
-    |> Enum.map(fn nickname ->
+  end
-      with nickname <- String.trim_leading(nickname, "@"),
+
-           do: {"@" <> nickname, User.get_cached_by_nickname(nickname)}
+  @doc """
-    end)
+  Parses a text and replace plain text links with HTML. Returns a tuple with a result text, mentions, and hashtags.
-    |> Enum.filter(fn {_match, user} -> user end)
+  """
+  @spec linkify(String.t(), keyword()) ::
+          {String.t(), [{String.t(), User.t()}], [{String.t(), String.t()}]}
+  def linkify(text, options \\ []) do
+    options = options ++ @auto_linker_config
+    acc = %{mentions: MapSet.new(), tags: MapSet.new()}
+    {text, %{mentions: mentions, tags: tags}} = AutoLinker.link_map(text, acc, options)
+
+    {text, MapSet.to_list(mentions), MapSet.to_list(tags)}
   end
 
   def emojify(text) do
@@ -48,9 +66,7 @@ defmodule Pleroma.Formatter do
       emoji = HTML.strip_tags(emoji)
       file = HTML.strip_tags(file)
 
-      String.replace(
+      html =
-        text,
-        ":#{emoji}:",
         if not strip do
           "<img height='32px' width='32px' alt='#{emoji}' title='#{emoji}' src='#{
             MediaProxy.url(file)
@@ -58,8 +74,8 @@ defmodule Pleroma.Formatter do
         else
           ""
         end
-      )
+
-      |> HTML.filter_tags()
+      String.replace(text, ":#{emoji}:", html) |> HTML.filter_tags()
     end)
   end
 
@@ -75,12 +91,6 @@ defmodule Pleroma.Formatter do
 
   def get_emoji(_), do: []
 
-  @link_regex ~r/[0-9a-z+\-\.]+:[0-9a-z$-_.+!*'(),]+/ui
-
-  @uri_schemes Application.get_env(:pleroma, :uri_schemes, [])
-  @valid_schemes Keyword.get(@uri_schemes, :valid_schemes, [])
-
-  # TODO: make it use something other than @link_regex
   def html_escape(text, "text/html") do
     HTML.filter_tags(text)
   end
@@ -94,112 +104,6 @@ defmodule Pleroma.Formatter do
     |> Enum.join("")
   end
 
-  @doc """
-  Escapes a special characters in mention names.
-  """
-  @spec mentions_escape(String.t(), list({String.t(), any()})) :: String.t()
-  def mentions_escape(text, mentions) do
-    mentions
-    |> Enum.reduce(text, fn {name, _}, acc ->
-      escape_name = String.replace(name, @markdown_characters_regex, "\\\\\\1")
-      String.replace(acc, name, escape_name)
-    end)
-  end
-
-  @doc "changes scheme:... urls to html links"
-  def add_links({subs, text}) do
-    links =
-      text
-      |> String.split([" ", "\t", "<br>"])
-      |> Enum.filter(fn word -> String.starts_with?(word, @valid_schemes) end)
-      |> Enum.filter(fn word -> Regex.match?(@link_regex, word) end)
-      |> Enum.map(fn url -> {Ecto.UUID.generate(), url} end)
-      |> Enum.sort_by(fn {_, url} -> -String.length(url) end)
-
-    uuid_text =
-      links
-      |> Enum.reduce(text, fn {uuid, url}, acc -> String.replace(acc, url, uuid) end)
-
-    subs =
-      subs ++
-        Enum.map(links, fn {uuid, url} ->
-          {uuid, "<a href=\"#{url}\">#{url}</a>"}
-        end)
-
-    {subs, uuid_text}
-  end
-
-  @doc "Adds the links to mentioned users"
-  def add_user_links({subs, text}, mentions, options \\ []) do
-    mentions =
-      mentions
-      |> Enum.sort_by(fn {name, _} -> -String.length(name) end)
-      |> Enum.map(fn {name, user} -> {name, user, Ecto.UUID.generate()} end)
-
-    uuid_text =
-      mentions
-      |> Enum.reduce(text, fn {match, _user, uuid}, text ->
-        String.replace(text, match, uuid)
-      end)
-
-    subs =
-      subs ++
-        Enum.map(mentions, fn {match, %User{id: id, ap_id: ap_id, info: info}, uuid} ->
-          ap_id =
-            if is_binary(info.source_data["url"]) do
-              info.source_data["url"]
-            else
-              ap_id
-            end
-
-          nickname =
-            if options[:format] == :full do
-              User.full_nickname(match)
-            else
-              User.local_nickname(match)
-            end
-
-          {uuid,
-           "<span class='h-card'><a data-user='#{id}' class='u-url mention' href='#{ap_id}'>" <>
-             "@<span>#{nickname}</span></a></span>"}
-        end)
-
-    {subs, uuid_text}
-  end
-
-  @doc "Adds the hashtag links"
-  def add_hashtag_links({subs, text}, tags) do
-    tags =
-      tags
-      |> Enum.sort_by(fn {name, _} -> -String.length(name) end)
-      |> Enum.map(fn {name, short} -> {name, short, Ecto.UUID.generate()} end)
-
-    uuid_text =
-      tags
-      |> Enum.reduce(text, fn {match, _short, uuid}, text ->
-        String.replace(text, ~r/((?<=[^&])|(\A))#{match}/, uuid)
-      end)
-
-    subs =
-      subs ++
-        Enum.map(tags, fn {tag_text, tag, uuid} ->
-          url =
-            "<a class='hashtag' data-tag='#{tag}' href='#{Pleroma.Web.base_url()}/tag/#{tag}' rel='tag'>#{
-              tag_text
-            }</a>"
-
-          {uuid, url}
-        end)
-
-    {subs, uuid_text}
-  end
-
-  def finalize({subs, text}) do
-    Enum.reduce(subs, text, fn {uuid, replacement}, result_text ->
-      String.replace(result_text, uuid, replacement)
-    end)
-  end
-
   def truncate(text, max_length \\ 200, omission \\ "...") do
     # Remove trailing whitespace
     text = Regex.replace(~r/([^ \t\r\n])([ \t]+$)/u, text, "\\g{1}")
@@ -211,4 +115,16 @@ defmodule Pleroma.Formatter do
       String.slice(text, 0, length_with_omission) <> omission
     end
   end
+
+  defp get_ap_id(%User{info: %{source_data: %{"url" => url}}}) when is_binary(url), do: url
+  defp get_ap_id(%User{ap_id: ap_id}), do: ap_id
+
+  defp get_nickname_text(nickname, %{mentions_format: :full}), do: User.full_nickname(nickname)
+  defp get_nickname_text(nickname, _), do: User.local_nickname(nickname)
+
+  defp maybe_escape(str, %{mentions_escape: true}) do
+    String.replace(str, @markdown_characters_regex, "\\\\\\1")
+  end
+
+  defp maybe_escape(str, _), do: str
 end

lib/pleroma/user.ex

@@ -1193,9 +1193,6 @@ defmodule Pleroma.User do
   def parse_bio(bio, _user) when bio == "", do: bio
 
   def parse_bio(bio, user) do
-    mentions = Formatter.parse_mentions(bio)
-    tags = Formatter.parse_tags(bio)
-
     emoji =
       (user.info.source_data["tag"] || [])
       |> Enum.filter(fn %{"type" => t} -> t == "Emoji" end)
@@ -1204,7 +1201,8 @@ defmodule Pleroma.User do
       end)
 
     bio
-    |> CommonUtils.format_input(mentions, tags, "text/plain", user_links: [format: :full])
+    |> CommonUtils.format_input("text/plain", mentions_format: :full)
+    |> elem(0)
     |> Formatter.emojify(emoji)
   end
 

lib/pleroma/web/common_api/common_api.ex

@@ -82,40 +82,20 @@ defmodule Pleroma.Web.CommonAPI do
 
   def get_visibility(_), do: "public"
 
-  defp get_content_type(content_type) do
-    if Enum.member?(Pleroma.Config.get([:instance, :allowed_post_formats]), content_type) do
-      content_type
-    else
-      "text/plain"
-    end
-  end
-
   def post(user, %{"status" => status} = data) do
     visibility = get_visibility(data)
     limit = Pleroma.Config.get([:instance, :limit])
 
     with status <- String.trim(status),
          attachments <- attachments_from_ids(data),
-         mentions <- Formatter.parse_mentions(status),
          inReplyTo <- get_replied_to_activity(data["in_reply_to_status_id"]),
-         {to, cc} <- to_for_user_and_mentions(user, mentions, inReplyTo, visibility),
+         {content_html, mentions, tags} <-
-         tags <- Formatter.parse_tags(status, data),
-         content_html <-
            make_content_html(
              status,
-             mentions,
              attachments,
-             tags,
+             data
-             get_content_type(data["content_type"]),
-             Enum.member?(
-               [true, "true"],
-               Map.get(
-                 data,
-                 "no_attachment_links",
-                 Pleroma.Config.get([:instance, :no_attachment_links], false)
-               )
-             )
            ),
+         {to, cc} <- to_for_user_and_mentions(user, mentions, inReplyTo, visibility),
          context <- make_context(inReplyTo),
          cw <- data["spoiler_text"],
          full_payload <- String.trim(status <> (data["spoiler_text"] || "")),
@@ -247,7 +227,7 @@ defmodule Pleroma.Web.CommonAPI do
   def report(user, data) do
     with {:account_id, %{"account_id" => account_id}} <- {:account_id, data},
          {:account, %User{} = account} <- {:account, User.get_by_id(account_id)},
-         {:ok, content_html} <- make_report_content_html(data["comment"]),
+         {:ok, {content_html, _, _}} <- make_report_content_html(data["comment"]),
          {:ok, statuses} <- get_report_statuses(account, data),
          {:ok, activity} <-
            ActivityPub.flag(%{

lib/pleroma/web/common_api/utils.ex

@@ -10,7 +10,7 @@ defmodule Pleroma.Web.CommonAPI.Utils do
   alias Pleroma.Object
   alias Pleroma.Repo
   alias Pleroma.User
-  alias Pleroma.Web
+  alias Pleroma.Config
   alias Pleroma.Web.Endpoint
   alias Pleroma.Web.MediaProxy
   alias Pleroma.Web.ActivityPub.Utils
@@ -100,24 +100,45 @@ defmodule Pleroma.Web.CommonAPI.Utils do
 
   def make_content_html(
         status,
-        mentions,
         attachments,
-        tags,
+        data
-        content_type,
-        no_attachment_links \\ false
       ) do
+    no_attachment_links =
+      data
+      |> Map.get("no_attachment_links", Config.get([:instance, :no_attachment_links]))
+      |> Kernel.in([true, "true"])
+
+    content_type = get_content_type(data["content_type"])
+
     status
-    |> format_input(mentions, tags, content_type)
+    |> format_input(content_type)
     |> maybe_add_attachments(attachments, no_attachment_links)
+    |> maybe_add_nsfw_tag(data)
   end
 
+  defp get_content_type(content_type) do
+    if Enum.member?(Config.get([:instance, :allowed_post_formats]), content_type) do
+      content_type
+    else
+      "text/plain"
+    end
+  end
+
+  defp maybe_add_nsfw_tag({text, mentions, tags}, %{"sensitive" => sensitive})
+       when sensitive in [true, "True", "true", "1"] do
+    {text, mentions, [{"#nsfw", "nsfw"} | tags]}
+  end
+
+  defp maybe_add_nsfw_tag(data, _), do: data
+
   def make_context(%Activity{data: %{"context" => context}}), do: context
   def make_context(_), do: Utils.generate_context_id()
 
-  def maybe_add_attachments(text, _attachments, true = _no_links), do: text
+  def maybe_add_attachments(parsed, _attachments, true = _no_links), do: parsed
 
-  def maybe_add_attachments(text, attachments, _no_links) do
+  def maybe_add_attachments({text, mentions, tags}, attachments, _no_links) do
-    add_attachments(text, attachments)
+    text = add_attachments(text, attachments)
+    {text, mentions, tags}
   end
 
   def add_attachments(text, attachments) do
@@ -135,56 +156,39 @@ defmodule Pleroma.Web.CommonAPI.Utils do
     Enum.join([text | attachment_text], "<br>")
   end
 
-  def format_input(text, mentions, tags, format, options \\ [])
+  def format_input(text, format, options \\ [])
 
   @doc """
   Formatting text to plain text.
   """
-  def format_input(text, mentions, tags, "text/plain", options) do
+  def format_input(text, "text/plain", options) do
     text
     |> Formatter.html_escape("text/plain")
-    |> String.replace(~r/\r?\n/, "<br>")
+    |> Formatter.linkify(options)
-    |> (&{[], &1}).()
+    |> (fn {text, mentions, tags} ->
-    |> Formatter.add_links()
+          {String.replace(text, ~r/\r?\n/, "<br>"), mentions, tags}
-    |> Formatter.add_user_links(mentions, options[:user_links] || [])
+        end).()
-    |> Formatter.add_hashtag_links(tags)
-    |> Formatter.finalize()
   end
 
   @doc """
   Formatting text to html.
   """
-  def format_input(text, mentions, _tags, "text/html", options) do
+  def format_input(text, "text/html", options) do
     text
     |> Formatter.html_escape("text/html")
-    |> (&{[], &1}).()
+    |> Formatter.linkify(options)
-    |> Formatter.add_user_links(mentions, options[:user_links] || [])
-    |> Formatter.finalize()
   end
 
   @doc """
   Formatting text to markdown.
   """
-  def format_input(text, mentions, tags, "text/markdown", options) do
+  def format_input(text, "text/markdown", options) do
+    options = Keyword.put(options, :mentions_escape, true)
+
     text
-    |> Formatter.mentions_escape(mentions)
-    |> Earmark.as_html!()
     |> Formatter.html_escape("text/html")
-    |> (&{[], &1}).()
+    |> Formatter.linkify(options)
-    |> Formatter.add_user_links(mentions, options[:user_links] || [])
+    |> (fn {text, mentions, tags} -> {Earmark.as_html!(text), mentions, tags} end).()
-    |> Formatter.add_hashtag_links(tags)
-    |> Formatter.finalize()
-  end
-
-  def add_tag_links(text, tags) do
-    tags =
-      tags
-      |> Enum.sort_by(fn {tag, _} -> -String.length(tag) end)
-
-    Enum.reduce(tags, text, fn {full, tag}, text ->
-      url = "<a href='#{Web.base_url()}/tag/#{tag}' rel='tag'>##{tag}</a>"
-      String.replace(text, full, url)
-    end)
   end
 
   def make_note_data(
@@ -323,13 +327,13 @@ defmodule Pleroma.Web.CommonAPI.Utils do
 
   def maybe_extract_mentions(_), do: []
 
-  def make_report_content_html(nil), do: {:ok, nil}
+  def make_report_content_html(nil), do: {:ok, {nil, [], []}}
 
   def make_report_content_html(comment) do
     max_size = Pleroma.Config.get([:instance, :max_report_comment_size], 1000)
 
     if String.length(comment) <= max_size do
-      {:ok, format_input(comment, [], [], "text/plain")}
+      {:ok, format_input(comment, "text/plain")}
     else
       {:error, "Comment must be up to #{max_size} characters"}
     end

mix.exs

@@ -90,7 +90,10 @@ defmodule Pleroma.Mixfile do
       {:websocket_client, git: "https://github.com/jeremyong/websocket_client.git", only: :test},
       {:floki, "~> 0.20.0"},
       {:ex_syslogger, github: "slashmili/ex_syslogger", tag: "1.4.0"},
-      {:timex, "~> 3.5"}
+      {:timex, "~> 3.5"},
+      {:auto_linker,
+       git: "https://git.pleroma.social/pleroma/auto_linker.git",
+       ref: "94193ca5f97c1f9fdf3d1469653e2d46fac34bcd"}
     ]
   end
 

mix.lock

@@ -1,4 +1,5 @@
 %{
+  "auto_linker": {:git, "https://git.pleroma.social/pleroma/auto_linker.git", "94193ca5f97c1f9fdf3d1469653e2d46fac34bcd", [ref: "94193ca5f97c1f9fdf3d1469653e2d46fac34bcd"]},
   "base64url": {:hex, :base64url, "0.0.1", "36a90125f5948e3afd7be97662a1504b934dd5dac78451ca6e9abf85a10286be", [:rebar], [], "hexpm"},
   "bunt": {:hex, :bunt, "0.2.0", "951c6e801e8b1d2cbe58ebbd3e616a869061ddadcc4863d0a2182541acae9a38", [:mix], [], "hexpm"},
   "cachex": {:hex, :cachex, "3.0.2", "1351caa4e26e29f7d7ec1d29b53d6013f0447630bbf382b4fb5d5bad0209f203", [:mix], [{:eternal, "~> 1.2", [hex: :eternal, repo: "hexpm", optional: false]}, {:unsafe, "~> 1.0", [hex: :unsafe, repo: "hexpm", optional: false]}], "hexpm"},

test/formatter_test.exs

@@ -21,22 +21,16 @@ defmodule Pleroma.FormatterTest do
       expected_text =
         "I love <a class='hashtag' data-tag='cofe' href='http://localhost:4001/tag/cofe' rel='tag'>#cofe</a> and <a class='hashtag' data-tag='2hu' href='http://localhost:4001/tag/2hu' rel='tag'>#2hu</a>"
 
-      tags = Formatter.parse_tags(text)
+      assert {^expected_text, [], _tags} = Formatter.linkify(text)
-
-      assert expected_text ==
-               Formatter.add_hashtag_links({[], text}, tags) |> Formatter.finalize()
     end
 
     test "does not turn html characters to tags" do
-      text = "Fact #3: pleroma does what mastodon't"
+      text = "#fact_3: pleroma does what mastodon't"
 
       expected_text =
-        "Fact <a class='hashtag' data-tag='3' href='http://localhost:4001/tag/3' rel='tag'>#3</a>: pleroma does what mastodon't"
+        "<a class='hashtag' data-tag='fact_3' href='http://localhost:4001/tag/fact_3' rel='tag'>#fact_3</a>: pleroma does what mastodon't"
 
-      tags = Formatter.parse_tags(text)
+      assert {^expected_text, [], _tags} = Formatter.linkify(text)
-
-      assert expected_text ==
-               Formatter.add_hashtag_links({[], text}, tags) |> Formatter.finalize()
     end
   end
 
@@ -47,79 +41,79 @@ defmodule Pleroma.FormatterTest do
       expected =
         "Hey, check out <a href=\"https://www.youtube.com/watch?v=8Zg1-TufF%20zY?x=1&y=2#blabla\">https://www.youtube.com/watch?v=8Zg1-TufF%20zY?x=1&y=2#blabla</a> ."
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://mastodon.social/@lambadalambda"
 
       expected =
         "<a href=\"https://mastodon.social/@lambadalambda\">https://mastodon.social/@lambadalambda</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://mastodon.social:4000/@lambadalambda"
 
       expected =
         "<a href=\"https://mastodon.social:4000/@lambadalambda\">https://mastodon.social:4000/@lambadalambda</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "@lambadalambda"
       expected = "@lambadalambda"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "http://www.cs.vu.nl/~ast/intel/"
       expected = "<a href=\"http://www.cs.vu.nl/~ast/intel/\">http://www.cs.vu.nl/~ast/intel/</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://forum.zdoom.org/viewtopic.php?f=44&t=57087"
 
       expected =
         "<a href=\"https://forum.zdoom.org/viewtopic.php?f=44&t=57087\">https://forum.zdoom.org/viewtopic.php?f=44&t=57087</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul"
 
       expected =
         "<a href=\"https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul\">https://en.wikipedia.org/wiki/Sophia_(Gnosticism)#Mythos_of_the_soul</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://www.google.co.jp/search?q=Nasim+Aghdam"
 
       expected =
         "<a href=\"https://www.google.co.jp/search?q=Nasim+Aghdam\">https://www.google.co.jp/search?q=Nasim+Aghdam</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://en.wikipedia.org/wiki/Duff's_device"
 
       expected =
         "<a href=\"https://en.wikipedia.org/wiki/Duff's_device\">https://en.wikipedia.org/wiki/Duff's_device</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "https://pleroma.com https://pleroma.com/sucks"
 
       expected =
         "<a href=\"https://pleroma.com\">https://pleroma.com</a> <a href=\"https://pleroma.com/sucks\">https://pleroma.com/sucks</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text = "xmpp:contact@hacktivis.me"
 
       expected = "<a href=\"xmpp:contact@hacktivis.me\">xmpp:contact@hacktivis.me</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
 
       text =
         "magnet:?xt=urn:btih:7ec9d298e91d6e4394d1379caf073c77ff3e3136&tr=udp%3A%2F%2Fopentor.org%3A2710&tr=udp%3A%2F%2Ftracker.blackunicorn.xyz%3A6969&tr=udp%3A%2F%2Ftracker.ccc.de%3A80&tr=udp%3A%2F%2Ftracker.coppersurfer.tk%3A6969&tr=udp%3A%2F%2Ftracker.leechers-paradise.org%3A6969&tr=udp%3A%2F%2Ftracker.openbittorrent.com%3A80&tr=wss%3A%2F%2Ftracker.btorrent.xyz&tr=wss%3A%2F%2Ftracker.fastcast.nz&tr=wss%3A%2F%2Ftracker.openwebtorrent.com"
 
       expected = "<a href=\"#{text}\">#{text}</a>"
 
-      assert Formatter.add_links({[], text}) |> Formatter.finalize() == expected
+      assert {^expected, [], []} = Formatter.linkify(text)
     end
   end
 
@@ -136,12 +130,9 @@ defmodule Pleroma.FormatterTest do
 
       archaeme_remote = insert(:user, %{nickname: "archaeme@archae.me"})
 
-      mentions = Pleroma.Formatter.parse_mentions(text)
+      {text, mentions, []} = Formatter.linkify(text)
 
-      {subs, text} = Formatter.add_user_links({[], text}, mentions)
+      assert length(mentions) == 3
-
-      assert length(subs) == 3
-      Enum.each(subs, fn {uuid, _} -> assert String.contains?(text, uuid) end)
 
       expected_text =
         "<span class='h-card'><a data-user='#{gsimg.id}' class='u-url mention' href='#{
@@ -152,7 +143,7 @@ defmodule Pleroma.FormatterTest do
           archaeme_remote.id
         }' class='u-url mention' href='#{archaeme_remote.ap_id}'>@<span>archaeme</span></a></span>"
 
-      assert expected_text == Formatter.finalize({subs, text})
+      assert expected_text == text
     end
 
     test "gives a replacement for user links when the user is using Osada" do
@@ -160,48 +151,35 @@ defmodule Pleroma.FormatterTest do
 
       text = "@mike@osada.macgirvin.com test"
 
-      mentions = Formatter.parse_mentions(text)
+      {text, mentions, []} = Formatter.linkify(text)
 
-      {subs, text} = Formatter.add_user_links({[], text}, mentions)
+      assert length(mentions) == 1
-
-      assert length(subs) == 1
-      Enum.each(subs, fn {uuid, _} -> assert String.contains?(text, uuid) end)
 
       expected_text =
         "<span class='h-card'><a data-user='#{mike.id}' class='u-url mention' href='#{mike.ap_id}'>@<span>mike</span></a></span> test"
 
-      assert expected_text == Formatter.finalize({subs, text})
+      assert expected_text == text
     end
 
     test "gives a replacement for single-character local nicknames" do
       text = "@o hi"
       o = insert(:user, %{nickname: "o"})
 
-      mentions = Formatter.parse_mentions(text)
+      {text, mentions, []} = Formatter.linkify(text)
 
-      {subs, text} = Formatter.add_user_links({[], text}, mentions)
+      assert length(mentions) == 1
-
-      assert length(subs) == 1
-      Enum.each(subs, fn {uuid, _} -> assert String.contains?(text, uuid) end)
 
       expected_text =
         "<span class='h-card'><a data-user='#{o.id}' class='u-url mention' href='#{o.ap_id}'>@<span>o</span></a></span> hi"
 
-      assert expected_text == Formatter.finalize({subs, text})
+      assert expected_text == text
     end
 
     test "does not give a replacement for single-character local nicknames who don't exist" do
       text = "@a hi"
 
-      mentions = Formatter.parse_mentions(text)
-
-      {subs, text} = Formatter.add_user_links({[], text}, mentions)
-
-      assert Enum.empty?(subs)
-      Enum.each(subs, fn {uuid, _} -> assert String.contains?(text, uuid) end)
-
       expected_text = "@a hi"
-      assert expected_text == Formatter.finalize({subs, text})
+      assert {^expected_text, [] = _mentions, [] = _tags} = Formatter.linkify(text)
     end
   end
 
@@ -209,14 +187,14 @@ defmodule Pleroma.FormatterTest do
     test "parses tags in the text" do
       text = "Here's a #Test. Maybe these are #working or not. What about #漢字? And #は｡"
 
-      expected = [
+      expected_tags = [
         {"#Test", "test"},
         {"#working", "working"},
-        {"#漢字", "漢字"},
+        {"#は", "は"},
-        {"#は", "は"}
+        {"#漢字", "漢字"}
       ]
 
-      assert Formatter.parse_tags(text) == expected
+      assert {_text, [], ^expected_tags} = Formatter.linkify(text)
     end
   end
 
@@ -230,15 +208,15 @@ defmodule Pleroma.FormatterTest do
     archaeme = insert(:user, %{nickname: "archaeme"})
     archaeme_remote = insert(:user, %{nickname: "archaeme@archae.me"})
 
-    expected_result = [
+    expected_mentions = [
-      {"@gsimg", gsimg},
       {"@archaeme", archaeme},
       {"@archaeme@archae.me", archaeme_remote},
-      {"@o", o},
+      {"@gsimg", gsimg},
-      {"@jimm", jimm}
+      {"@jimm", jimm},
+      {"@o", o}
     ]
 
-    assert Formatter.parse_mentions(text) == expected_result
+    assert {_text, ^expected_mentions, []} = Formatter.linkify(text)
   end
 
   test "it adds cool emoji" do
@@ -281,22 +259,10 @@ defmodule Pleroma.FormatterTest do
     assert Formatter.get_emoji(text) == []
   end
 
-  describe "/mentions_escape" do
+  test "it escapes HTML in plain text" do
-    test "it returns text with escaped mention names" do
+    text = "hello & world google.com/?a=b&c=d \n http://test.com/?a=b&c=d 1"
-      text = """
+    expected = "hello &amp; world google.com/?a=b&c=d \n http://test.com/?a=b&c=d 1"
-      @a_breakin_glass@cybre.space
-      (also, little voice inside my head thinking "maybe this will encourage people
-      pronouncing it properly instead of saying _raKEWdo_ ")
-      """
 
-      escape_text = """
+    assert Formatter.html_escape(text, "text/plain") == expected
-      @a\\_breakin\\_glass@cybre\\.space
-      (also, little voice inside my head thinking \"maybe this will encourage people
-      pronouncing it properly instead of saying _raKEWdo_ \")
-      """
-
-      mentions = [{"@a_breakin_glass@cybre.space", %{}}]
-      assert Formatter.mentions_escape(text, mentions) == escape_text
-    end
   end
 end

test/web/common_api/common_api_utils_test.exs

@@ -57,19 +57,19 @@ defmodule Pleroma.Web.CommonAPI.UtilsTest do
     assert expected == Utils.emoji_from_profile(user)
   end
 
-  describe "format_input/4" do
+  describe "format_input/3" do
     test "works for bare text/plain" do
       text = "hello world!"
       expected = "hello world!"
 
-      output = Utils.format_input(text, [], [], "text/plain")
+      {output, [], []} = Utils.format_input(text, "text/plain")
 
       assert output == expected
 
       text = "hello world!\n\nsecond paragraph!"
       expected = "hello world!<br><br>second paragraph!"
 
-      output = Utils.format_input(text, [], [], "text/plain")
+      {output, [], []} = Utils.format_input(text, "text/plain")
 
       assert output == expected
     end
@@ -78,14 +78,14 @@ defmodule Pleroma.Web.CommonAPI.UtilsTest do
       text = "<p>hello world!</p>"
       expected = "<p>hello world!</p>"
 
-      output = Utils.format_input(text, [], [], "text/html")
+      {output, [], []} = Utils.format_input(text, "text/html")
 
       assert output == expected
 
       text = "<p>hello world!</p>\n\n<p>second paragraph</p>"
       expected = "<p>hello world!</p>\n\n<p>second paragraph</p>"
 
-      output = Utils.format_input(text, [], [], "text/html")
+      {output, [], []} = Utils.format_input(text, "text/html")
 
       assert output == expected
     end
@@ -94,14 +94,32 @@ defmodule Pleroma.Web.CommonAPI.UtilsTest do
       text = "**hello world**"
       expected = "<p><strong>hello world</strong></p>\n"
 
-      output = Utils.format_input(text, [], [], "text/markdown")
+      {output, [], []} = Utils.format_input(text, "text/markdown")
 
       assert output == expected
 
       text = "**hello world**\n\n*another paragraph*"
       expected = "<p><strong>hello world</strong></p>\n<p><em>another paragraph</em></p>\n"
 
-      output = Utils.format_input(text, [], [], "text/markdown")
+      {output, [], []} = Utils.format_input(text, "text/markdown")
+
+      assert output == expected
+    end
+
+    test "works for text/markdown with mentions" do
+      {:ok, user} =
+        UserBuilder.insert(%{nickname: "user__test", ap_id: "http://foo.com/user__test"})
+
+      text = "**hello world**\n\n*another @user__test and @user__test google.com paragraph*"
+
+      expected =
+        "<p><strong>hello world</strong></p>\n<p><em>another <span class='h-card'><a data-user='#{
+          user.id
+        }' class='u-url mention' href='http://foo.com/user__test'>@<span>user__test</span></a></span> and <span class='h-card'><a data-user='#{
+          user.id
+        }' class='u-url mention' href='http://foo.com/user__test'>@<span>user__test</span></a></span> <a href=\"http://google.com\">google.com</a> paragraph</em></p>\n"
+
+      {output, _, _} = Utils.format_input(text, "text/markdown")
 
       assert output == expected
     end