htmlparser unclosed script tag causes data loss

[waylan]

BPO	41989
Nosy	@terryjreedy, @ezio-melotti, @waylan
PRs	gh-86155: Fix htmlparser "unclosed script tag causes data loss" #22658
Files	test_html.py: A simple test

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/ezio-melotti'
closed_at = None
created_at = <Date 2020-10-10.01:08:29.514>
labels = ['type-bug', 'library', '3.10']
title = 'htmlparser unclosed script tag causes data loss'
updated_at = <Date 2020-10-16.20:42:44.634>
user = 'https://github.com/waylan'

bugs.python.org fields:

activity = <Date 2020-10-16.20:42:44.634>
actor = 'terry.reedy'
assignee = 'ezio.melotti'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)']
creation = <Date 2020-10-10.01:08:29.514>
creator = 'waylan'
dependencies = []
files = ['49505']
hgrepos = []
issue_num = 41989
keywords = ['patch']
message_count = 2.0
messages = ['378359', '378748']
nosy_count = 3.0
nosy_names = ['terry.reedy', 'ezio.melotti', 'waylan']
pr_nums = ['22658']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue41989'
versions = ['Python 3.10']

Linked PRs

• [3.14] gh-86155: Fix data loss after unclosed script or style tag in HTMLParser (GH-22658) #133844
• [3.13] gh-86155: Fix data loss after unclosed script or style tag in HTMLParser (GH-22658) #133845

[waylan]

When the close method of the HtmlParser is called, any cached text data is generally flushed and passed to a data event; except when in data_mode. Specifically, if an unclosed script or style tag has been encountered, a call to close does not flush the data.

A simple test which demonstrates the issue is attached.

I see that in Lib/html/parser.py#L244-L249 there are two nested if statements which both check for not self.cdata_elem. Obviously, if we got past the first one, that situation will never exist for the nested one. Somehow this block of code needs a branch for when self.cdata_elem is True.

I should note that the input is invalid HTML. However, the existing behavior results in data loss. Within any other unclosed tag (other than script or style) any data is still flushed and passed to a data event. I would expect the same behavior here. Although, the data escaping behavior should perhaps be applied as it is with data within properly closed tags.

[terryjreedy]

Waylan, 3.7 and before only get security fixes.

To me, this might be considered an enhancement rather than bug fix, but I will leave that to Ezio.

[serhiy-storchaka]

Thank you for your contribution, @waylan. Sorry it took so long to review your PR.