UnboundLocalError: local variable 'match' referenced before assignment

[jvoisin]

I'm getting the following stacktrace when running the following python script on the following input on beautifulsoup4, version: 4.9.1

```
$ python3 bs4_repro.py crash-54df4824926ab452e78144c7bdeb6b35397323db78f2919a54be6a643c83e224
/home/jvoisin/dev/pythonfuzz/ven/lib/python3.8/site-packages/bs4/builder/_htmlparser.py:102: UserWarning: unknown status keyword 'c_bc_h' in marked section
  warnings.warn(msg)
Traceback (most recent call last):
  File "bs4_repro.py", line 14, in <module>
    main()
  File "bs4_repro.py", line 12, in main
    BeautifulSoup(buf, features=parsers[idx]).prettify()
  File "/home/jvoisin/dev/pythonfuzz/ven/lib/python3.8/site-packages/bs4/__init__.py", line 345, in __init__
    self._feed()
  File "/home/jvoisin/dev/pythonfuzz/ven/lib/python3.8/site-packages/bs4/__init__.py", line 431, in _feed
    self.builder.feed(self.markup)
  File "/home/jvoisin/dev/pythonfuzz/ven/lib/python3.8/site-packages/bs4/builder/_htmlparser.py", line 377, in feed
    parser.feed(markup)
  File "/usr/lib/python3.8/html/parser.py", line 111, in feed
    self.goahead(0)
  File "/usr/lib/python3.8/html/parser.py", line 179, in goahead
    k = self.parse_html_declaration(i)
  File "/usr/lib/python3.8/html/parser.py", line 264, in parse_html_declaration
    return self.parse_marked_section(i)
  File "/usr/lib/python3.8/_markupbase.py", line 160, in parse_marked_section
    if not match:
UnboundLocalError: local variable 'match' referenced before assignment
zsh: exit 1 python3 bs4_repro.py
$
```

Reproducing script:

```
from bs4 import BeautifulSoup
import sys

def main ():
    with open(sys.argv[1], 'rb') as f:
        buf = f.read()
        parsers = ['lxml-xml', 'html5lib', 'html.parser', 'lxml']
        try:
            idx = int(buf[0]) % len(parsers)
        except ValueError:
            return
        BeautifulSoup(buf, features=parsers[idx]).prettify()

main()

```

Input file (use `xxd -r` to transform the hexdump into a file):

```
00000000: 1a00 746d 6c2f 3c21 5b63 5f62 635f 683b ..tml/<![c_bc_h;
00000010: 3c75 6c3e 3c2f 6cc6 9e69 bb0c 0005 3e3c <ul></l..i....><
00000020: 212d 2d20 2d29 193b 0108 2d3e 3c2f 215b !-- -).;..-></![
00000030: 635f 623d 3b69 7469 3e3c 212d c_b=;iti><!-
```

It looks like the following section of /usr/lib/python3.8/_markupbase.py is the problem:

```
    # Override this to handle MS-word extension syntax <![if word]>content<![endif]>
    def parse_marked_section(self, i, report=1):
        rawdata= self.rawdata
        assert rawdata[i:i+3] == '<![', "unexpected call to parse_marked_section()"
        sectName, j = self._scan_name( i+3, i )
        if j < 0:
            return j
        if sectName in {"temp", "cdata", "ignore", "include", "rcdata"}:
            # look for standard ]]> ending
            match= _markedsectionclose.search(rawdata, i+3)
        elif sectName in {"if", "else", "endif"}:
            # look for MS Office ]> ending
            match= _msmarkedsectionclose.search(rawdata, i+3)
        else:
            self.error('unknown status keyword %r in marked section' % rawdata[i+3:j])
        if not match:
            return -1
        if report:
            j = match.start(0)
            self.unknown_decl(rawdata[i+3: j])
        return match.end(0)
```

The `self.error` function isn't apparently supposed to return, but only to abort.

[Leonard Richardson (leonardr)]

Thanks for taking the time to investigate this issue. As with bug #1883104, this is a problem with Python's html.parser library. It was filed against Python as https://bugs.python.org/issue34480. I've added a comment to the Python bug referring to this issue.