From 9eb19fac04037443d8c4de79ccb89c2b34ab9304 Mon Sep 17 00:00:00 2001
From: pictuga <contact@pictuga.com>
Date: Mon, 3 Jan 2022 16:26:17 +0000
Subject: [PATCH] readabilite: use custom html parser within bs4's lxml parser

Solves the following obscure error:
ValueError: Invalid PI name 'b'xml''
---
 morss/readabilite.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/morss/readabilite.py b/morss/readabilite.py
index 608dae4..a7ac92f 100644
--- a/morss/readabilite.py
+++ b/morss/readabilite.py
@@ -17,14 +17,20 @@
 
 import re
 
+import bs4.builder._lxml
 import lxml.etree
 import lxml.html
 import lxml.html.soupparser
 
 
+class CustomTreeBuilder(bs4.builder._lxml.LXMLTreeBuilder):
+    def default_parser(self, encoding):
+        return lxml.html.HTMLParser(remove_comments=True, encoding=encoding)
+
+
 def parse(data, encoding=None):
     kwargs = {'from_encoding': encoding} if encoding else {}
-    return lxml.html.soupparser.fromstring(data, features='lxml', **kwargs)
+    return lxml.html.soupparser.fromstring(data, features='lxml', builder=CustomTreeBuilder, **kwargs)
 
 
 def count_words(string):