Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.11/site-packages/werkzeug/sansio/utils.py: 47%

1from __future__ import annotations 

2 

3import re 

4import typing as t 

5from urllib.parse import quote 

6 

7from .._internal import _plain_int 

8from ..exceptions import SecurityError 

9from ..http import parse_set_header 

10from ..urls import uri_to_iri 

11 

12_host_re = re.compile( 

13 r""" 

14 ( 

15 [a-z0-9.-]+ # domain or ipv4 

16 | 

17 \[[a-f0-9]*:[a-f0-9.:]+] # ipv6 

18 ) 

19 (?::[0-9]+)? # optional port 

20 """, 

21 flags=re.ASCII | re.IGNORECASE | re.VERBOSE, 

22) 

23 

24 

25def host_is_trusted( 

26 hostname: str | None, trusted_list: t.Collection[str] | None = None 

27) -> bool: 

28 """Perform some checks on a ``Host`` header ``host:port``. The host must be 

29 made up of valid characters, but this does not check validity beyond that. 

30 If a list of trusted domains is given, the domain must match one. 

31 

32 :param hostname: The ``Host`` header ``host:port`` to check. 

33 :param trusted_list: A list of trusted domains to match. These should 

34 already be IDNA encoded, but will be encoded if needed. The port is 

35 ignored for this check. If a name starts with a dot it will match as a 

36 suffix, accepting all subdomains. If empty or ``None``, all domains are 

37 allowed. 

38 

39 .. versionchanged:: 3.2 

40 The value's characters are validated. 

41 

42 .. versionchanged:: 3.2 

43 ``trusted_list`` defaults to ``None``. 

44 

45 .. versionadded:: 0.9 

46 """ 

47 if not hostname: 

48 return False 

49 

50 if _host_re.fullmatch(hostname) is None: 

51 return False 

52 

53 hostname = hostname.partition(":")[0] 

54 

55 if not trusted_list: 

56 return True 

57 

58 if isinstance(trusted_list, str): 

59 trusted_list = [trusted_list] 

60 

61 for ref in trusted_list: 

62 if ref.startswith("."): 

63 ref = ref[1:] 

64 suffix_match = True 

65 else: 

66 suffix_match = False 

67 

68 try: 

69 ref = ref.partition(":")[0].encode("idna").decode("ascii") 

70 except UnicodeEncodeError: 

71 return False 

72 

73 if ref == hostname or (suffix_match and hostname.endswith(f".{ref}")): 

74 return True 

75 

76 return False 

77 

78 

79def get_host( 

80 scheme: str, 

81 host_header: str | None, 

82 server: tuple[str, int | None] | None = None, 

83 trusted_hosts: t.Collection[str] | None = None, 

84) -> str: 

85 """Get and validate a request's ``host:port`` based on the given values. 

86 

87 The ``Host`` header sent by the client is preferred. Otherwise, the server's 

88 configured address is used. The port is omitted if it matches the standard 

89 HTTP or HTTPS ports. 

90 

91 The value is passed through :func:`host_is_trusted`. The host must be made 

92 up of valid characters, but this does not check validity beyond that. If a 

93 list of trusted domains is given, the domain must match one. 

94 

95 :param scheme: The protocol of the request. Used to omit the standard ports 

96 80 and 443. 

97 :param host_header: The ``Host`` header value. 

98 :param server: The server's configured address ``(host, port)``. The server 

99 may be using a Unix socket and give ``(path, None)``; this is ignored as 

100 it would not produce a useful host value. 

101 :param trusted_hosts: A list of trusted domains to match. These should 

102 already be IDNA encoded, but will be encoded if needed. The port is 

103 ignored for this check. If a name starts with a dot it will match as a 

104 suffix, accepting all subdomains. If empty or ``None``, all domains are 

105 allowed. 

106 

107 :return: Host, with port if necessary. 

108 :raise .SecurityError: If the host is not trusted. 

109 

110 .. versionchanged:: 3.2 

111 The characters of the host value are validated. The empty string is no 

112 longer allowed if no header value is available. 

113 

114 .. versionchanged:: 3.2 

115 When using the server address, Unix sockets are ignored. 

116 

117 .. versionchanged:: 3.1.3 

118 If ``SERVER_NAME`` is IPv6, it is wrapped in ``[]``. 

119 """ 

120 if host_header is not None: 

121 host = host_header 

122 # The port server[1] will be None for a Unix socket. Ignore in that case. 

123 elif server is not None and server[1] is not None: 

124 host = server[0] 

125 

126 # If SERVER_NAME is IPv6, wrap it in [] to match Host header. 

127 # Check for : because domain or IPv4 can't have that. 

128 if ":" in host and host[0] != "[": 

129 host = f"[{host}]" 

130 

131 host = f"{host}:{server[1]}" 

132 else: 

133 host = "" 

134 

135 if scheme in {"http", "ws"}: 

136 host = host.removesuffix(":80") 

137 elif scheme in {"https", "wss"}: 

138 host = host.removesuffix(":443") 

139 

140 if not host_is_trusted(host, trusted_hosts): 

141 raise SecurityError(f"Host {host!r} is not trusted.") 

142 

143 return host 

144 

145 

146def get_current_url( 

147 scheme: str, 

148 host: str, 

149 root_path: str | None = None, 

150 path: str | None = None, 

151 query_string: bytes | None = None, 

152) -> str: 

153 """Recreate the URL for a request. If an optional part isn't 

154 provided, it and subsequent parts are not included in the URL. 

155 

156 The URL is an IRI, not a URI, so it may contain Unicode characters. 

157 Use :func:`~werkzeug.urls.iri_to_uri` to convert it to ASCII. 

158 

159 :param scheme: The protocol the request used, like ``"https"``. 

160 :param host: The host the request was made to. See :func:`get_host`. 

161 :param root_path: Prefix that the application is mounted under. This 

162 is prepended to ``path``. 

163 :param path: The path part of the URL after ``root_path``. 

164 :param query_string: The portion of the URL after the "?". 

165 """ 

166 url = [scheme, "://", host] 

167 

168 if root_path is None: 

169 url.append("/") 

170 return uri_to_iri("".join(url)) 

171 

172 # safe = https://url.spec.whatwg.org/#url-path-segment-string 

173 # as well as percent for things that are already quoted 

174 url.append(quote(root_path.rstrip("/"), safe="!$&'()*+,/:;=@%")) 

175 url.append("/") 

176 

177 if path is None: 

178 return uri_to_iri("".join(url)) 

179 

180 url.append(quote(path.lstrip("/"), safe="!$&'()*+,/:;=@%")) 

181 

182 if query_string: 

183 url.append("?") 

184 url.append(quote(query_string, safe="!$&'()*+,/:;=?@%")) 

185 

186 return uri_to_iri("".join(url)) 

187 

188 

189def get_content_length( 

190 http_content_length: str | None = None, 

191 http_transfer_encoding: str | None = None, 

192) -> int | None: 

193 """Return the ``Content-Length`` header value as an int. If the header is not given 

194 or the ``Transfer-Encoding`` header is ``chunked``, ``None`` is returned to indicate 

195 a streaming request. If the value is not an integer, or negative, 0 is returned. 

196 

197 :param http_content_length: The Content-Length HTTP header. 

198 :param http_transfer_encoding: The Transfer-Encoding HTTP header. 

199 

200 .. versionadded:: 2.2 

201 """ 

202 if ( 

203 http_transfer_encoding is not None 

204 and "chunked" in parse_set_header(http_transfer_encoding) 

205 ) or http_content_length is None: 

206 return None 

207 

208 try: 

209 return max(0, _plain_int(http_content_length)) 

210 except ValueError: 

211 return 0