Annotation of pkgsrc/lang/python27/patches/patch-Lib_urlparse.py, revision 1.2.2.1

1.2.2.1 ! bsiegert    1: $NetBSD$
1.1       gutterid    2: 
                      3: Fix CVE-2021-23336: Add `separator` argument to parse_qs; warn with default
                      4: Via Fedora:
                      5: https://src.fedoraproject.org/rpms/python2.7/blob/rawhide/f/00359-CVE-2021-23336.patch
                      6: 
1.2.2.1 ! bsiegert    7: Fix CVE-2022-0391: urlparse does not sanitize URLs containing ASCII newline and tabs
        !             8: Via Fedora:
        !             9: https://src.fedoraproject.org/rpms/python2.7/raw/40dd05e5d77dbfa81777c9f84b704bc2239bf710/f/00377-CVE-2022-0391.patch
        !            10: 
1.1       gutterid   11: --- Lib/urlparse.py.orig       2020-04-19 21:13:39.000000000 +0000
                     12: +++ Lib/urlparse.py
                     13: @@ -29,6 +29,7 @@ test_urlparse.py provides a good indicat
                     14:  """
                     15:  
                     16:  import re
                     17: +import os
                     18:  
                     19:  __all__ = ["urlparse", "urlunparse", "urljoin", "urldefrag",
                     20:             "urlsplit", "urlunsplit", "parse_qs", "parse_qsl"]
1.2.2.1 ! bsiegert   21: @@ -62,6 +63,9 @@ scheme_chars = ('abcdefghijklmnopqrstuvw
        !            22:                  '0123456789'
        !            23:                  '+-.')
        !            24:  
        !            25: +# Unsafe bytes to be removed per WHATWG spec
        !            26: +_UNSAFE_URL_BYTES_TO_REMOVE = ['\t', '\r', '\n']
        !            27: +
        !            28:  MAX_CACHE_SIZE = 20
        !            29:  _parse_cache = {}
        !            30:  
        !            31: @@ -184,12 +188,19 @@ def _checknetloc(netloc):
        !            32:                               "under NFKC normalization"
        !            33:                               % netloc)
        !            34:  
        !            35: +def _remove_unsafe_bytes_from_url(url):
        !            36: +    for b in _UNSAFE_URL_BYTES_TO_REMOVE:
        !            37: +        url = url.replace(b, "")
        !            38: +    return url
        !            39: +
        !            40:  def urlsplit(url, scheme='', allow_fragments=True):
        !            41:      """Parse a URL into 5 components:
        !            42:      <scheme>://<netloc>/<path>?<query>#<fragment>
        !            43:      Return a 5-tuple: (scheme, netloc, path, query, fragment).
        !            44:      Note that we don't break the components up in smaller bits
        !            45:      (e.g. netloc is a single string) and we don't expand % escapes."""
        !            46: +    url = _remove_unsafe_bytes_from_url(url)
        !            47: +    scheme = _remove_unsafe_bytes_from_url(scheme)
        !            48:      allow_fragments = bool(allow_fragments)
        !            49:      key = url, scheme, allow_fragments, type(url), type(scheme)
        !            50:      cached = _parse_cache.get(key, None)
        !            51: @@ -382,7 +393,8 @@ def unquote(s):
1.1       gutterid   52:              append(item)
                     53:      return ''.join(res)
                     54:  
                     55: -def parse_qs(qs, keep_blank_values=0, strict_parsing=0, max_num_fields=None):
                     56: +def parse_qs(qs, keep_blank_values=0, strict_parsing=0, max_num_fields=None,
                     57: +             separator=None):
                     58:      """Parse a query given as a string argument.
                     59:  
                     60:          Arguments:
1.2.2.1 ! bsiegert   61: @@ -405,14 +417,23 @@ def parse_qs(qs, keep_blank_values=0, st
1.1       gutterid   62:      """
                     63:      dict = {}
                     64:      for name, value in parse_qsl(qs, keep_blank_values, strict_parsing,
                     65: -                                 max_num_fields):
                     66: +                                 max_num_fields, separator):
                     67:          if name in dict:
                     68:              dict[name].append(value)
                     69:          else:
                     70:              dict[name] = [value]
                     71:      return dict
                     72:  
                     73: -def parse_qsl(qs, keep_blank_values=0, strict_parsing=0, max_num_fields=None):
                     74: +class _QueryStringSeparatorWarning(RuntimeWarning):
                     75: +    """Warning for using default `separator` in parse_qs or parse_qsl"""
                     76: +
                     77: +# The default "separator" for parse_qsl can be specified in a config file.
                     78: +# It's cached after first read.
1.2       gutterid   79: +_QS_SEPARATOR_CONFIG_FILENAME = '@PKG_SYSCONFDIR@/python/urllib.cfg'
1.1       gutterid   80: +_default_qs_separator = None
                     81: +
                     82: +def parse_qsl(qs, keep_blank_values=0, strict_parsing=0, max_num_fields=None,
                     83: +              separator=None):
                     84:      """Parse a query given as a string argument.
                     85:  
                     86:      Arguments:
1.2.2.1 ! bsiegert   87: @@ -434,15 +455,72 @@ def parse_qsl(qs, keep_blank_values=0, s
1.1       gutterid   88:  
                     89:      Returns a list, as G-d intended.
                     90:      """
                     91: +
                     92: +    if (not separator or (not isinstance(separator, (str, bytes)))) and separator is not None:
                     93: +        raise ValueError("Separator must be of type string or bytes.")
                     94: +
                     95: +    # Used when both "&" and ";" act as separators. (Need a non-string value.)
                     96: +    _legacy = object()
                     97: +
                     98: +    if separator is None:
                     99: +        global _default_qs_separator
                    100: +        separator = _default_qs_separator
                    101: +        envvar_name = 'PYTHON_URLLIB_QS_SEPARATOR'
                    102: +        if separator is None:
                    103: +            # Set default separator from environment variable
                    104: +            separator = os.environ.get(envvar_name)
                    105: +            config_source = 'environment variable'
                    106: +        if separator is None:
                    107: +            # Set default separator from the configuration file
                    108: +            try:
                    109: +                file = open(_QS_SEPARATOR_CONFIG_FILENAME)
                    110: +            except EnvironmentError:
                    111: +                pass
                    112: +            else:
                    113: +                with file:
                    114: +                    import ConfigParser
                    115: +                    config = ConfigParser.ConfigParser()
                    116: +                    config.readfp(file)
                    117: +                    separator = config.get('parse_qs', envvar_name)
                    118: +                    _default_qs_separator = separator
                    119: +                config_source = _QS_SEPARATOR_CONFIG_FILENAME
                    120: +        if separator is None:
                    121: +            # The default is '&', but warn if not specified explicitly
                    122: +            if ';' in qs:
                    123: +                from warnings import warn
                    124: +                warn("The default separator of urlparse.parse_qsl and "
                    125: +                    + "parse_qs was changed to '&' to avoid a web cache "
                    126: +                    + "poisoning issue (CVE-2021-23336). "
                    127: +                    + "By default, semicolons no longer act as query field "
                    128: +                    + "separators. "
                    129: +                    + "See https://access.redhat.com/articles/5860431 for "
                    130: +                    + "more details.",
                    131: +                    _QueryStringSeparatorWarning, stacklevel=2)
                    132: +            separator = '&'
                    133: +        elif separator == 'legacy':
                    134: +            separator = _legacy
                    135: +        elif len(separator) != 1:
                    136: +            raise ValueError(
                    137: +                '{} (from {}) must contain '.format(envvar_name, config_source)
                    138: +                + '1 character, or "legacy". See '
                    139: +                + 'https://access.redhat.com/articles/5860431 for more details.'
                    140: +            )
                    141: +
                    142:      # If max_num_fields is defined then check that the number of fields
                    143:      # is less than max_num_fields. This prevents a memory exhaustion DOS
                    144:      # attack via post bodies with many fields.
                    145:      if max_num_fields is not None:
                    146: -        num_fields = 1 + qs.count('&') + qs.count(';')
                    147: +        if separator is _legacy:
                    148: +            num_fields = 1 + qs.count('&') + qs.count(';')
                    149: +        else:
                    150: +            num_fields = 1 + qs.count(separator)
                    151:          if max_num_fields < num_fields:
                    152:              raise ValueError('Max number of fields exceeded')
                    153:  
                    154: -    pairs = [s2 for s1 in qs.split('&') for s2 in s1.split(';')]
                    155: +    if separator is _legacy:
                    156: +        pairs = [s2 for s1 in qs.split('&') for s2 in s1.split(';')]
                    157: +    else:
                    158: +        pairs = [s1 for s1 in qs.split(separator)]
                    159:      r = []
                    160:      for name_value in pairs:
                    161:          if not name_value and not strict_parsing:

CVSweb <webmaster@jp.NetBSD.org>