From c8026cd16c37079e3b2468f06f37f7dcd6aa3f95 Mon Sep 17 00:00:00 2001
From: rinpatch <rinpatch@sdf.org>
Date: Mon, 17 Jun 2019 23:06:31 +0300
Subject: [PATCH] Add an option to disable TLD validation

Added `validate_tld` option, which can be set to false to disable tld
validation and to :no_scheme to disable tld validation when a scheme is
present

Closes #8
---
 lib/auto_linker.ex        |  1 +
 lib/auto_linker/parser.ex | 16 +++++++----
 test/parser_test.exs      | 60 ++++++++++++++++++++++++++++++++++++++-
 3 files changed, 70 insertions(+), 7 deletions(-)

diff --git a/lib/auto_linker.ex b/lib/auto_linker.ex
index 222cd79..d07e6df 100644
--- a/lib/auto_linker.ex
+++ b/lib/auto_linker.ex
@@ -48,6 +48,7 @@ defmodule AutoLinker do
   * `hashtag_prefix: nil` - a prefix to build a link for a hashtag (example: `https://example.com/tag/`)
   * `hashtag_handler: nil` - a custom handler to validate and formart a hashtag
   * `extra: false` - link urls with rarely used schemes (magnet, ipfs, irc, etc.)
+  * `validate_tld: true` - Set to false to disable TLD validation for urls, also can be set to :no_scheme to validate TLDs only for urls without a scheme (e.g `example.com` will be validated, but `http://example.loki` won't)
 
   Each of the above options can be specified when calling `link(text, opts)`
   or can be set in the `:auto_linker`'s configuration. For example:
diff --git a/lib/auto_linker/parser.ex b/lib/auto_linker/parser.ex
index 739173f..0ca906d 100644
--- a/lib/auto_linker/parser.ex
+++ b/lib/auto_linker/parser.ex
@@ -42,7 +42,7 @@ defmodule AutoLinker.Parser do
 
   @tlds "./priv/tlds.txt" |> File.read!() |> String.split("\n", trim: true) |> MapSet.new()
 
-  @default_opts ~w(url)a
+  @default_opts ~w(url validate_tld)a
 
   @doc """
   Parse the given string, identifying items to link.
@@ -262,7 +262,7 @@ defmodule AutoLinker.Parser do
   def check_and_link(buffer, opts, _user_acc) do
     str = strip_parens(buffer)
 
-    if url?(str, opts[:scheme]) do
+    if url?(str, opts[:scheme], opts[:validate_tld]) do
       case parse_link(str, opts) do
         ^buffer -> link_url(buffer, opts)
         url -> String.replace(buffer, url, link_url(url, opts))
@@ -315,12 +315,16 @@ defmodule AutoLinker.Parser do
   end
 
   # @doc false
-  def url?(buffer, true) do
-    valid_url?(buffer) && Regex.match?(@match_scheme, buffer) && valid_tld?(buffer)
+  def url?(buffer, scheme, validate_tld \\ true)
+
+  def url?(buffer, true, validate_tld) do
+    valid_url?(buffer) && Regex.match?(@match_scheme, buffer) &&
+      (!validate_tld or validate_tld == :no_scheme || valid_tld?(buffer))
   end
 
-  def url?(buffer, _) do
-    valid_url?(buffer) && Regex.match?(@match_url, buffer) && valid_tld?(buffer)
+  def url?(buffer, _, validate_tld) do
+    valid_url?(buffer) && Regex.match?(@match_url, buffer) &&
+      (validate_tld == false || valid_tld?(buffer))
   end
 
   def email?(buffer) do
diff --git a/test/parser_test.exs b/test/parser_test.exs
index 851fe3e..2b49a7c 100644
--- a/test/parser_test.exs
+++ b/test/parser_test.exs
@@ -4,7 +4,7 @@ defmodule AutoLinker.ParserTest do
 
   import AutoLinker.Parser
 
-  describe "url?/2" do
+  describe "url?/3" do
     test "valid scheme true" do
       valid_scheme_urls()
       |> Enum.each(fn url ->
@@ -32,6 +32,48 @@ defmodule AutoLinker.ParserTest do
         refute url?(url, false)
       end)
     end
+
+    test "checks the tld for url with a scheme when validate_tld: true" do
+      custom_tld_scheme_urls()
+      |> Enum.each(fn url ->
+        refute url?(url, true, true)
+      end)
+    end
+
+    test "does not check the tld for url with a scheme when validate_tld: false" do
+      custom_tld_scheme_urls()
+      |> Enum.each(fn url ->
+        assert url?(url, true, false)
+      end)
+    end
+
+    test "does not check the tld for url with a scheme when validate_tld: :no_scheme" do
+      custom_tld_scheme_urls()
+      |> Enum.each(fn url ->
+        assert url?(url, true, :no_scheme)
+      end)
+    end
+
+    test "checks the tld for url without a scheme when validate_tld: true" do
+      custom_tld_non_scheme_urls()
+      |> Enum.each(fn url ->
+        refute url?(url, false, true)
+      end)
+    end
+
+    test "checks the tld for url without a scheme when validate_tld: :no_scheme" do
+      custom_tld_non_scheme_urls()
+      |> Enum.each(fn url ->
+        refute url?(url, false, :no_scheme)
+      end)
+    end
+
+    test "does not check the tld for url without a scheme when validate_tld: false" do
+      custom_tld_non_scheme_urls()
+      |> Enum.each(fn url ->
+        assert url?(url, false, false)
+      end)
+    end
   end
 
   describe "match_phone" do
@@ -216,4 +258,20 @@ defmodule AutoLinker.ParserTest do
       "x5",
       "(555) 555-55"
     ]
+
+  def custom_tld_scheme_urls,
+    do: [
+      "http://whatever.null/",
+      "https://example.o/index.html",
+      "http://pleroma.i2p/test",
+      "http://misskey.loki"
+    ]
+
+  def custom_tld_non_scheme_urls,
+    do: [
+      "whatever.null/",
+      "example.o/index.html",
+      "pleroma.i2p/test",
+      "misskey.loki"
+    ]
 end