From c38437d5653286ee39cfbe3154d5e96ea6ce2e7e Mon Sep 17 00:00:00 2001
From: Jon Hadfield <jon@lessknown.co.uk>
Date: Sun, 21 Apr 2024 10:47:03 +0100
Subject: [PATCH] adds compatibility with hosts file containing non-ASCII
 characters.

---
 python_hosts/hosts.py     |  2 +-
 python_hosts/utils.py     |  2 +-
 test_files/hosts3         |  4 ++++
 tests/test_hosts.py       | 11 +++++++++++
 tests/test_hosts_entry.py | 11 +++++++++++
 5 files changed, 28 insertions(+), 2 deletions(-)
 create mode 100644 test_files/hosts3

diff --git a/python_hosts/hosts.py b/python_hosts/hosts.py
index bbfa507..e0a5bad 100755
--- a/python_hosts/hosts.py
+++ b/python_hosts/hosts.py
@@ -362,7 +362,7 @@ def import_file(self, import_file_path=None):
         invalid_count = 0
         if is_readable(import_file_path):
             import_entries = []
-            with open(import_file_path, 'r') as infile:
+            with open(import_file_path, 'r', encoding='utf-8-sig') as infile:
                 for line in infile:
                     stripped_entry = line.strip()
                     if (not stripped_entry) or (stripped_entry.startswith('#')):
diff --git a/python_hosts/utils.py b/python_hosts/utils.py
index 890c036..32c2506 100644
--- a/python_hosts/utils.py
+++ b/python_hosts/utils.py
@@ -43,7 +43,7 @@ def valid_hostnames(hostname_list):
     for entry in hostname_list:
         if len(entry) > 255:
             return False
-        allowed = re.compile(r'(?!-)[A-Z\d-]{1,63}(?<!-)$', re.IGNORECASE)
+        allowed = re.compile(r'(?!-)[A-Z\w-]{1,63}(?<!-)$', re.UNICODE | re.IGNORECASE)
         if not all(allowed.match(x) for x in entry.split(".")):
             return False
     return True
diff --git a/test_files/hosts3 b/test_files/hosts3
new file mode 100644
index 0000000..eb9b79e
--- /dev/null
+++ b/test_files/hosts3
@@ -0,0 +1,4 @@
+5.5.5.5 test.com 测试
+0.0.0.0 bob.com
+0.0.0.0 jane.com
+0.0.0.0 simon.com asif.com
diff --git a/tests/test_hosts.py b/tests/test_hosts.py
index 9778c13..a5c0723 100755
--- a/tests/test_hosts.py
+++ b/tests/test_hosts.py
@@ -561,6 +561,17 @@ def test_import_file_returns_duplicate_correctly(tmpdir):
     assert write_result.get('ipv4_entries_written') == 2
 
 
+def test_import_file_with_chinese_simplified(tmpdir):
+    """
+    Test that importing of a file with simplified Chinese characters succeeds
+    """
+    hosts_file = tmpdir.mkdir("etc").join("hosts")
+    hosts_file.write("6.6.6.6\texample.com\n")
+    hosts = Hosts(path=hosts_file.strpath)
+    hosts.import_file(import_file_path=os.path.join("test_files", "hosts3"))
+    assert hosts.exists(address='5.5.5.5', names=['test.com', '测试'])
+
+
 def test_addition_of_ipv6_entry_where_matching_name_exists_and_force_false(tmpdir):
     """
     Test no replacement of an ipv6 entry where the address is different
diff --git a/tests/test_hosts_entry.py b/tests/test_hosts_entry.py
index 27f3158..38bd91b 100644
--- a/tests/test_hosts_entry.py
+++ b/tests/test_hosts_entry.py
@@ -18,6 +18,17 @@ def test_create_ipv4_instance():
     assert hosts_entry.comment == 'this is a comment'
 
 
+def test_create_ipv4_instance_with_chinese_simplified():
+    """ add an ipv4 type entry """
+    hosts_entry = HostsEntry(entry_type='ipv4', address='1.2.3.4',
+                             names=['example.com', 'example', '例'],
+                             comment='这是一个评论')
+    assert hosts_entry.entry_type == 'ipv4'
+    assert hosts_entry.address == '1.2.3.4'
+    assert hosts_entry.names == ['example.com', 'example', '例']
+    assert hosts_entry.comment == '这是一个评论'
+
+
 def test_str_to_hostentry_ipv4():
     str_entry = HostsEntry.str_to_hostentry(
         '10.10.10.10 example.com example.org example # another comment')