fix: bugs with braces and semicolons in quoted strings

Author: peakwinter

nginx.py

@@ -444,43 +444,44 @@ def loads(data, conf=True):
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*location\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        n = re.compile(r'(?!\B"[^"]*);(?![^"]*"\B)')
+        m = re.compile(r'^\s*location\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             l = Location(m.group(1))
             lopen.insert(0, l)
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*if\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        m = re.compile(r'^\s*if\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             ifs = If(m.group(1))
             lopen.insert(0, ifs)
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*upstream\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        m = re.compile(r'^\s*upstream\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             u = Upstream(m.group(1))
             lopen.insert(0, u)
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*geo\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        m = re.compile(r'^\s*geo\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             g = Geo(m.group(1))
             lopen.insert(0, g)
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*map\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        m = re.compile(r'^\s*map\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             g = Map(m.group(1))
             lopen.insert(0, g)
             index += m.end()
             continue
 
-        m = re.compile(r'^\s*limit_except\s*([^;]*?)\s*{', re.S).search(data[index:])
-        if m:
+        m = re.compile(r'^\s*limit_except\s+(.*?)\s*(?!\B"[^"]*){(?![^"]*"\B)', re.S).search(data[index:])
+        if m and not n.search(m.group()):
             l = LimitExcept(m.group(1))
             lopen.insert(0, l)
             index += m.end()

tests.py

@@ -167,6 +167,10 @@
 location test9 {
     add_header X-XSS-Protection "1;mode-block";
 }
+
+if ( $http_user_agent = "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)" ) {
+    return 403;
+}
 """
 
 TESTBLOCK_CASE_10 = """
@@ -194,6 +198,21 @@
 }
 """
 
+TESTBLOCK_CASE_12 = """
+server {
+    listen 80;
+    server_name test.example.com;
+
+    location ~ "^/(test|[0-9a-zA-Z]{6})$" {
+        if ($query_string ~ pid=(111)) {
+            return 403;
+        }
+
+        proxy_pass http://127.0.0.1:81;
+    }
+}
+"""
+
 
 class TestPythonNginx(unittest.TestCase):
     def test_basic_load(self):
@@ -298,6 +317,8 @@ def test_semicolon_in_second_key_value(self):
         self.assertEqual(len(location_children), 1)
         self.assertEqual(location_children[0].name, "add_header")
         self.assertEqual(location_children[0].value, 'X-XSS-Protection "1;mode-block"')
+        self.assertEqual(len(inp_data.filter("If")), 1)
+        self.assertEqual(inp_data.filter("If")[0].value, "( $http_user_agent = \"Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)\" )")
 
     def test_types_block(self):
         inp_data = nginx.loads(TESTBLOCK_CASE_10)
@@ -311,6 +332,11 @@ def test_missing_semi_colon(self):
         with pytest.raises(nginx.ParseError) as e:
             nginx.loads(TESTBLOCK_CASE_11)
         self.assertEqual(str(e.value), "Config syntax, missing ';' at index: 189")
+    
+    def test_brace_inside_block_param(self):
+        inp_data = nginx.loads(TESTBLOCK_CASE_12)
+        self.assertEqual(len(inp_data.server.filter("Location")), 1)
+        self.assertEqual(inp_data.server.filter("Location")[0].value, "~ \"^/(test|[0-9a-zA-Z]{6})$\"")
 
 
 if __name__ == '__main__':