From: Oleg Broytman <phd@phdru.name>
Date: Mon, 20 Nov 2023 17:49:22 +0000 (+0300)
Subject: Fix(Py3): `html.parser` cannot parse bytes
X-Git-Tag: 5.0.0~5
X-Git-Url: https://git.phdru.name/?p=bookmarks_db.git;a=commitdiff_plain;h=0478809ea1e8ae324e324e99cbb2e25bd0bf31b3

Fix(Py3): `html.parser` cannot parse bytes

Decode to unicode from a known encoding.
---

diff --git a/parse_html/bkmk_ph_htmlparser.py b/parse_html/bkmk_ph_htmlparser.py
index c0f89b4..d11a2ff 100644
--- a/parse_html/bkmk_ph_htmlparser.py
+++ b/parse_html/bkmk_ph_htmlparser.py
@@ -91,6 +91,11 @@ class HTMLParser(_HTMLParser):
 def parse_html(html_text, charset=None, log=None):
     if not html_text:
         return None
+    if charset is None and isinstance(html_text, bytes):
+        return None  # html.parser cannot parse bytes
+    if charset and isinstance(html_text, bytes):
+        html_text = html_text.decode(charset)
+
     parser = HTMLParser(charset)
 
     try: