From 181f6dcdbcd0e25d0d5bf059744f765192a77f7b Mon Sep 17 00:00:00 2001
From: yozachar <38415384+yozachar@users.noreply.github.com>
Date: Thu, 18 Apr 2024 05:30:40 +0530
Subject: [PATCH] fix: rfc cases in the `domain` validator

---
 src/validators/domain.py | 14 +++++++++-----
 tests/test_domain.py     |  4 ++++
 2 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/src/validators/domain.py b/src/validators/domain.py
index 0e8c0c48..43ed981f 100644
--- a/src/validators/domain.py
+++ b/src/validators/domain.py
@@ -59,17 +59,21 @@ def domain(
         return False
 
     try:
-        return not re.search(r"\s", value) and re.match(
+
+        service_record = r"_" if rfc_2782 else ""
+        trailing_dot = r"\.?$" if rfc_1034 else r"$"
+
+        return not re.search(r"\s|__+", value) and re.match(
             # First character of the domain
-            rf"^(?:[a-z0-9{r'_?'if rfc_2782 else ''}]"
+            rf"^(?:[a-z0-9{service_record}]"
             # Sub-domain
-            + rf"(?:[a-z0-9-{r'_?'if rfc_2782 else ''}]{{0,61}}"
+            + rf"(?:[a-z0-9-{service_record}]{{0,61}}"
             # Hostname
-            + rf"[a-z0-9{r'_?'if rfc_2782 else ''}])?\.)"
+            + rf"[a-z0-9{service_record}])?\.)"
             # First 61 characters of the gTLD
             + r"+[a-z0-9][a-z0-9-_]{0,61}"
             # Last character of the gTLD
-            + rf"[a-z]{r'.?$' if rfc_1034 else r'$'}",
+            + rf"[a-z]{trailing_dot}",
             value.encode("idna").decode("utf-8"),
             re.IGNORECASE,
         )
diff --git a/tests/test_domain.py b/tests/test_domain.py
index 21aadf90..6d8e8675 100644
--- a/tests/test_domain.py
+++ b/tests/test_domain.py
@@ -20,6 +20,7 @@
         ("3.cn.", True, False),
         ("_example.com", False, True),
         ("example_.com", False, True),
+        ("_exa_mple_.com", False, True),
         ("a.cn", False, False),
         ("sub1.sub2.sample.co.uk", False, False),
         ("somerandomexample.xn--fiqs8s", False, False),
@@ -67,6 +68,9 @@ def test_returns_true_on_valid_top_level_domain(
         ("_example._com", False, False),
         ("example_.com", False, False),
         ("example", False, False),
+        ("example.com!", True, False),
+        ("example?.com", True, False),
+        ("__exa__mple__.com", False, True),
         ("a......b.com", False, False),
         ("a.123", False, False),
         ("123.123", False, False),