From c8735e58378bef65f534c30253b3f5e689950933 Mon Sep 17 00:00:00 2001
From: Thomas Citharel <tcit@tcit.fr>
Date: Tue, 18 Jan 2022 12:46:33 +0100
Subject: [PATCH] Fix parsing links with hashtag characters

Closes #1008

Signed-off-by: Thomas Citharel <tcit@tcit.fr>
---
 lib/service/formatter/formatter.ex        | 43 ++---------------------
 test/service/formatter/formatter_test.exs | 14 ++++++++
 2 files changed, 17 insertions(+), 40 deletions(-)

diff --git a/lib/service/formatter/formatter.ex b/lib/service/formatter/formatter.ex
index 32f515ffa..bf1e623b9 100644
--- a/lib/service/formatter/formatter.ex
+++ b/lib/service/formatter/formatter.ex
@@ -94,8 +94,7 @@ defmodule Mobilizon.Service.Formatter do
     options = linkify_opts() ++ options
 
     acc = %{mentions: MapSet.new(), tags: MapSet.new()}
-    {text, %{mentions: mentions}} = Linkify.link_map(text, acc, options)
-    {text, tags} = extract_tags(text)
+    {text, %{mentions: mentions, tags: tags}} = Linkify.link_map(text, acc, options)
 
     {text, MapSet.to_list(mentions), MapSet.to_list(tags)}
   end
@@ -157,46 +156,10 @@ defmodule Mobilizon.Service.Formatter do
   defp linkify_opts do
     Mobilizon.Config.get(__MODULE__) ++
       [
-        hashtag: false,
+        hashtag: true,
+        hashtag_handler: &__MODULE__.hashtag_handler/4,
         mention: true,
         mention_handler: &__MODULE__.mention_handler/4
       ]
   end
-
-  @match_hashtag ~r/(?:^|[^\p{L}\p{M}\p{Nd}\)])(?<tag>\#[[:word:]_]*[[:alpha:]_·][[:word:]_·\p{M}]*)/u
-
-  @spec extract_tags(String.t()) :: {String.t(), MapSet.t()}
-  def extract_tags(text) do
-    matches =
-      @match_hashtag
-      |> Regex.scan(text, capture: [:tag])
-      |> Enum.map(&hd/1)
-      |> Enum.map(&{&1, tag_text_strip(&1)})
-      |> MapSet.new()
-
-    text =
-      @match_hashtag
-      |> Regex.replace(text, &generate_tag_link/2)
-      |> String.trim()
-
-    {text, matches}
-  end
-
-  @spec generate_tag_link(String.t(), String.t()) :: String.t()
-  defp generate_tag_link(_, tag_text) do
-    tag = tag_text_strip(tag_text)
-    url = "#{Endpoint.url()}/tag/#{tag}"
-
-    Tag.content_tag(:a, tag_text,
-      class: "hashtag",
-      "data-tag": tag,
-      href: url,
-      rel: "tag ugc"
-    )
-    |> Phoenix.HTML.safe_to_string()
-    |> (&" #{&1}").()
-  end
-
-  @spec tag_text_strip(String.t()) :: String.t()
-  defp tag_text_strip(tag), do: tag |> String.trim("#") |> String.downcase()
 end
diff --git a/test/service/formatter/formatter_test.exs b/test/service/formatter/formatter_test.exs
index 3a5451354..556ea8dfb 100644
--- a/test/service/formatter/formatter_test.exs
+++ b/test/service/formatter/formatter_test.exs
@@ -112,6 +112,20 @@ defmodule Mobilizon.Service.FormatterTest do
         "<a href=\"#{text}\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#{text}</a>"
 
       assert {^expected, [], []} = Formatter.linkify(text)
+
+      text = "https://example.org/#foobar"
+
+      expected =
+        "<a href=\"#{text}\" target=\"_blank\" rel=\"noopener noreferrer ugc\">#{text}</a>"
+
+      assert {^expected, [], []} = Formatter.linkify(text)
+
+      text = "<p>An article tagged with a #tag.</p>"
+
+      expected =
+        "<p>An article tagged with a <a class=\"hashtag\" data-tag=\"tag\" href=\"http://mobilizon.test/tag/tag\" rel=\"tag ugc\">#tag</a>.</p>"
+
+      assert {^expected, [], [{"#tag", "tag"}]} = Formatter.linkify(text)
     end
   end