New regex to exclude leading dashes per #62

lightswitch05 · Jan 21, 2019 · e9bb24f · e9bb24f
1 parent a01a183
commit e9bb24f
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 6 deletions.
diff --git a/HostsTools/hosts_tools.py b/HostsTools/hosts_tools.py
@@ -10,7 +10,8 @@
 import time
 
 STRIP_COMMENTS_PATTERN = re.compile(r"^([^#]+)")
-ALLOWED_DOMAIN_PATTERN = re.compile("^[^\*\?\[\]{}\|\\\/&^%$#@!+=~`\s\.<>,\"']+$", re.IGNORECASE)
+EXCLUDE_DOMAIN_PATTERN = re.compile(r"^[-]", re.IGNORECASE)
+ALLOWED_DOMAIN_PATTERN = re.compile(r"^[^\*\?\[\]{}\|\\/&^%$#@!+=~`\s\.<>,\"']+$", re.IGNORECASE)
 FILE_HEADER = """
 # Collection of Analytics, Ads, and tracking hosts to block.
 #
@@ -129,7 +130,7 @@ def is_valid_domain(domain: str) -> bool:
         return False
     if domain[-1] == ".":
         domain = domain[:-1]  # strip exactly one dot from the right, if present
-    return all(ALLOWED_DOMAIN_PATTERN.match(x) for x in domain.split("."))
+    return all(ALLOWED_DOMAIN_PATTERN.match(x) and not EXCLUDE_DOMAIN_PATTERN.match(x) for x in domain.split("."))
 
 
 def filter_whitelist(domains: Set[str], whitelist: Set[Pattern] = {}):

diff --git a/ads-and-tracking-extended.txt b/ads-and-tracking-extended.txt
@@ -1,8 +1,8 @@
 # Collection of Analytics, Ads, and tracking hosts to block.
 #
 # Title: Lightswitch05's ads-and-tracking-extended.txt
-# Released: 2019-01-21T13:42:14+00:00
-# Count: 32130 domains
+# Released: 2019-01-21T22:42:38+00:00
+# Count: 32129 domains
 # Details: https://github.com/lightswitch05/hosts
 # Issues: https://github.com/lightswitch05/hosts/issues
 # Source: https://raw.githubusercontent.com/lightswitch05/hosts/master/ads-and-tracking-extended.txt
@@ -20509,7 +20509,6 @@
 0.0.0.0 www.vilynx.com
 0.0.0.0 www2.vilynx.com
 0.0.0.0 vindicosuite.com
-0.0.0.0 -x3.vindicosuite.com
 0.0.0.0 1.vindicosuite.com
 0.0.0.0 adcreative.vindicosuite.com
 0.0.0.0 addirector.vindicosuite.com

diff --git a/ads-and-tracking.txt b/ads-and-tracking.txt
@@ -1,7 +1,7 @@
 # Collection of Analytics, Ads, and tracking hosts to block.
 #
 # Title: Lightswitch05's ads-and-tracking.txt
-# Released: 2019-01-21T13:42:14+00:00
+# Released: 2019-01-21T22:42:37+00:00
 # Count: 2183 domains
 # Details: https://github.com/lightswitch05/hosts
 # Issues: https://github.com/lightswitch05/hosts/issues

diff --git a/tests/unit/test_hosts_tools.py b/tests/unit/test_hosts_tools.py
@@ -64,6 +64,14 @@ def test_trailing_dot_is_a_valid_domain(self):
         is_valid = hosts_tools.is_valid_domain("www.example.com.")
         assert is_valid
 
+    def test_leading_dash_is_not_a_valid_domain(self):
+        is_valid = hosts_tools.is_valid_domain("-example.com")
+        assert not is_valid
+
+    def test_middle_dash_is_a_valid_domain(self):
+        is_valid = hosts_tools.is_valid_domain("my-example.com")
+        assert is_valid
+
     def test_extract_basic(self):
         extracted = hosts_tools.extract_domain("0.0.0.0 example.com")
         assert extracted == "example.com"