feat: use python html insteadof beautifulsoup

blinkpy/api.py

@@ -2,16 +2,15 @@
 
 import logging
 import string
-import json
 from json import dumps
 from asyncio import sleep
 from urllib.parse import urlencode, urlparse, parse_qs
-from bs4 import BeautifulSoup
 from blinkpy.helpers.util import (
     get_time,
     Throttle,
     local_storage_clip_url_template,
 )
+from blinkpy.helpers.oauth_parser import OAuthArgsParser
 from blinkpy.helpers.constants import (
     TIMEOUT,
     DEFAULT_USER_AGENT,
@@ -812,18 +811,10 @@ async def oauth_get_signin_page(auth):
 
     # Extract CSRF token from oauth-args script tag
     try:
-        # Parse HTML
-        soup = BeautifulSoup(html, "html.parser")
-
-        # Find oauth-args script
-        oauth_script = soup.find(
-            "script", {"id": "oauth-args", "type": "application/json"}
-        )
-        if oauth_script and oauth_script.string:
-            oauth_data = json.loads(oauth_script.string)
-            csrf_token = oauth_data.get("csrf-token")
-            if csrf_token:
-                return csrf_token
+        parser = OAuthArgsParser()
+        parser.feed(html)
+        if parser.csrf_token:
+            return parser.csrf_token
     except Exception as error:
         _LOGGER.error("Failed to extract CSRF token: %s", error)
 

blinkpy/helpers/oauth_parser.py

@@ -0,0 +1,39 @@
+"""HTML parser for OAuth CSRF token extraction."""
+
+import json
+from html.parser import HTMLParser
+
+
+class OAuthArgsParser(HTMLParser):
+    """HTML parser to extract CSRF token from oauth-args script tag."""
+
+    def __init__(self):
+        """Initialize parser."""
+        super().__init__()
+        self.csrf_token = None
+        self._in_oauth_script = False
+
+    def handle_starttag(self, tag, attrs):
+        """Handle opening tags."""
+        if tag == "script":
+            attrs_dict = dict(attrs)
+            if (
+                attrs_dict.get("id") == "oauth-args"
+                and attrs_dict.get("type") == "application/json"
+            ):
+                self._in_oauth_script = True
+
+    def handle_data(self, data):
+        """Handle tag content."""
+        if self._in_oauth_script:
+            try:
+                oauth_data = json.loads(data)
+                self.csrf_token = oauth_data.get("csrf-token")
+            except (json.JSONDecodeError, AttributeError):
+                pass
+            self._in_oauth_script = False
+
+    def handle_endtag(self, tag):
+        """Handle closing tags."""
+        if tag == "script":
+            self._in_oauth_script = False

requirements.txt

@@ -3,5 +3,4 @@ requests>=2.24.0
 python-slugify>=4.0.1
 sortedcontainers~=2.4.0
 aiohttp>=3.8.4
-aiofiles>=23.1.0
-beautifulsoup4>=4.12.0
+aiofiles>=23.1.0