fix: improve regex patterns for M3U parsing and attribute handling (#1353)

Author: Guovin

utils/constants.py

@@ -71,11 +71,11 @@
 m3u_pattern = re.compile(r"^#EXTINF:-1[\s+,，](?P<attributes>[^,，]+)[，,](?P<name>.*?)\n(?P<value>.+)$")
 
 multiline_m3u_pattern = re.compile(
-    r"^#EXTINF:-1(?:[\s+,，]*(?P<attributes>[^,，\r\n]+))?[,，](?P<name>.*?)[\r\n]+"
-    r"(?P<options>(?:#EXTVLCOPT:.*[\r\n]+)*)(?P<value>.*?)(?=\r?\n#EXTINF:-1|\Z)",
+    r"^#EXTINF:-1(?:[\s+,，]*(?P<attributes>(?:[^,，\r\n\"]+|\"[^\"\r\n]*\")*))?[,，](?P<name>.*?)[\r\n]+"
+    r"(?P<options>(?:(?:[ \t]*\r?\n)+|#EXTVLCOPT:[^\r\n]*(?:\r?\n|$))*)(?P<value>.*?)(?=\r?\n(?:[ \t]*\r?\n)*#EXTINF:-1|\Z)",
     re.MULTILINE | re.DOTALL)
 
-key_value_pattern = re.compile(r'(?P<key>\w+)=(?P<value>\S+)')
+key_value_pattern = re.compile(r'(?P<key>[\w-]+)=(?P<value>"[^"]*"|\'[^\']*\'|\S+)')
 
 sub_pattern = re.compile(
     r"-|_|\((.*?)\)|（(.*?)）|\[(.*?)]|「(.*?)」| |｜|频道|普清|标清|高清|HD|hd|超清|超高|超高清|4K|4k|中央|央视|电视台|台|电信|联通|移动")

utils/tools.py

@@ -653,8 +653,8 @@ def get_name_value(content, pattern, open_headers=False, check_value=True):
         if not name or (check_value and not value):
             continue
         data = {"name": name, "value": value}
-        attributes = {**get_headers_key_value(group_dict.get("attributes", "")),
-                      **get_headers_key_value(group_dict.get("options", ""))}
+        attributes = {**get_headers_key_value(group_dict.get("options", "")),
+                      **get_headers_key_value(group_dict.get("attributes", ""))}
         headers = {
             "User-Agent": attributes.get("useragent", ""),
             "Referer": attributes.get("referer", ""),