fix: handle nested nsvod player payload
This commit is contained in:
@@ -228,6 +228,19 @@ class TestNSVodSpider(unittest.TestCase):
|
||||
self.assertEqual(result["url"], "https://cdn.example/direct.m3u8")
|
||||
self.assertEqual(result["header"]["Referer"], "https://nsvod.me/index.php/vod/play/id/70/sid/1/nid/1.html")
|
||||
|
||||
@patch.object(Spider, "_request_html")
|
||||
def test_player_content_supports_nested_player_aaaa_object(self, mock_request_html):
|
||||
mock_request_html.return_value = """
|
||||
<script type="text/javascript">
|
||||
var player_aaaa={"flag":"play","encrypt":0,"vod_data":{"vod_name":"阿爸请你嘛嘛吼","vod_actor":"甲,乙","vod_class":"剧情"},"url":"https:\\/\\/vip.dytt-kan.com\\/20260423\\/13858_751d8d46\\/index.m3u8","from":"dyttm3u8","id":"22663","sid":1,"nid":1}
|
||||
</script>
|
||||
"""
|
||||
result = self.spider.playerContent("DY线路", "/index.php/vod/play/id/22663/sid/1/nid/1.html", {})
|
||||
self.assertEqual(result["parse"], 0)
|
||||
self.assertEqual(result["jx"], 0)
|
||||
self.assertEqual(result["url"], "https://vip.dytt-kan.com/20260423/13858_751d8d46/index.m3u8")
|
||||
self.assertEqual(result["header"]["Referer"], "https://nsvod.me/index.php/vod/play/id/22663/sid/1/nid/1.html")
|
||||
|
||||
@patch.object(Spider, "_request_html")
|
||||
def test_player_content_falls_back_to_inline_m3u8(self, mock_request_html):
|
||||
mock_request_html.return_value = '<script>var now="https://cdn.example/fallback.m3u8?token=1"</script>'
|
||||
|
||||
+46
-2
@@ -261,12 +261,56 @@ class Spider(BaseSpider):
|
||||
return {"list": []}
|
||||
return {"list": [vod]}
|
||||
|
||||
def _extract_balanced_object(self, text, start_index):
|
||||
raw = self._stringify(text)
|
||||
if start_index < 0 or start_index >= len(raw) or raw[start_index] != "{":
|
||||
return ""
|
||||
|
||||
depth = 0
|
||||
in_string = False
|
||||
quote_char = ""
|
||||
escaped = False
|
||||
|
||||
for index in range(start_index, len(raw)):
|
||||
char = raw[index]
|
||||
if in_string:
|
||||
if escaped:
|
||||
escaped = False
|
||||
elif char == "\\":
|
||||
escaped = True
|
||||
elif char == quote_char:
|
||||
in_string = False
|
||||
continue
|
||||
|
||||
if char in ('"', "'"):
|
||||
in_string = True
|
||||
quote_char = char
|
||||
continue
|
||||
|
||||
if char == "{":
|
||||
depth += 1
|
||||
continue
|
||||
|
||||
if char == "}":
|
||||
depth -= 1
|
||||
if depth == 0:
|
||||
return raw[start_index : index + 1]
|
||||
|
||||
return ""
|
||||
|
||||
def _extract_player_data(self, html):
|
||||
matched = re.search(r"var\s+player_aaaa\s*=\s*(\{[\s\S]*?\})", self._stringify(html), re.I)
|
||||
text = self._stringify(html)
|
||||
matched = re.search(r"player_aaaa\s*=", text, re.I)
|
||||
if not matched:
|
||||
return {}
|
||||
object_start = text.find("{", matched.end())
|
||||
if object_start < 0:
|
||||
return {}
|
||||
raw_object = self._extract_balanced_object(text, object_start)
|
||||
if not raw_object:
|
||||
return {}
|
||||
try:
|
||||
return json.loads(matched.group(1))
|
||||
return json.loads(raw_object)
|
||||
except Exception:
|
||||
return {}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user