html.parse() truncates attribute value

[Bob Kline]

If you run the following code with the attached repro.html in the current working directory, you will see from the output that the title attribute of the last img element on the page is truncated.

#!/usr/bin/env python

from re import search
from lxml import html

page = open("repro.html", encoding="utf-8").read()
img = search('<img id="_426"[^>]+.', page).group(0)
print("SERIALIZED IMG ELEMENT:", img)
title = search('title="([^"]+)"', img).group(1)
print("SERIALIZED TITLE ATTRIBUTE:", title)

for node in html.parse("repro.html").iter("img"):
    if node.get("id") == "_426":
        print("PARSED (AND TRUNCATED) TITLE ATTRIBUTE:", node.get("title"))

Requested environment information:

Python : sys.version_info(major=3, minor=7, micro=6, releaselevel='final', serial=0)
lxml.etree : (4, 4, 2, 0)
libxml used : (2, 9, 10)
libxml compiled : (2, 9, 10)
libxslt used : (1, 1, 34)
libxslt compiled : (1, 1, 34)

That's on MacOS. Also reproducible on Linux and Windows, Python 3.8.0, lxml 4.5.0.