| 1 | n/a | import binascii |
|---|
| 2 | n/a | import email.charset |
|---|
| 3 | n/a | import email.message |
|---|
| 4 | n/a | import email.errors |
|---|
| 5 | n/a | from email import quoprimime |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | class ContentManager: |
|---|
| 8 | n/a | |
|---|
| 9 | n/a | def __init__(self): |
|---|
| 10 | n/a | self.get_handlers = {} |
|---|
| 11 | n/a | self.set_handlers = {} |
|---|
| 12 | n/a | |
|---|
| 13 | n/a | def add_get_handler(self, key, handler): |
|---|
| 14 | n/a | self.get_handlers[key] = handler |
|---|
| 15 | n/a | |
|---|
| 16 | n/a | def get_content(self, msg, *args, **kw): |
|---|
| 17 | n/a | content_type = msg.get_content_type() |
|---|
| 18 | n/a | if content_type in self.get_handlers: |
|---|
| 19 | n/a | return self.get_handlers[content_type](msg, *args, **kw) |
|---|
| 20 | n/a | maintype = msg.get_content_maintype() |
|---|
| 21 | n/a | if maintype in self.get_handlers: |
|---|
| 22 | n/a | return self.get_handlers[maintype](msg, *args, **kw) |
|---|
| 23 | n/a | if '' in self.get_handlers: |
|---|
| 24 | n/a | return self.get_handlers[''](msg, *args, **kw) |
|---|
| 25 | n/a | raise KeyError(content_type) |
|---|
| 26 | n/a | |
|---|
| 27 | n/a | def add_set_handler(self, typekey, handler): |
|---|
| 28 | n/a | self.set_handlers[typekey] = handler |
|---|
| 29 | n/a | |
|---|
| 30 | n/a | def set_content(self, msg, obj, *args, **kw): |
|---|
| 31 | n/a | if msg.get_content_maintype() == 'multipart': |
|---|
| 32 | n/a | # XXX: is this error a good idea or not? We can remove it later, |
|---|
| 33 | n/a | # but we can't add it later, so do it for now. |
|---|
| 34 | n/a | raise TypeError("set_content not valid on multipart") |
|---|
| 35 | n/a | handler = self._find_set_handler(msg, obj) |
|---|
| 36 | n/a | msg.clear_content() |
|---|
| 37 | n/a | handler(msg, obj, *args, **kw) |
|---|
| 38 | n/a | |
|---|
| 39 | n/a | def _find_set_handler(self, msg, obj): |
|---|
| 40 | n/a | full_path_for_error = None |
|---|
| 41 | n/a | for typ in type(obj).__mro__: |
|---|
| 42 | n/a | if typ in self.set_handlers: |
|---|
| 43 | n/a | return self.set_handlers[typ] |
|---|
| 44 | n/a | qname = typ.__qualname__ |
|---|
| 45 | n/a | modname = getattr(typ, '__module__', '') |
|---|
| 46 | n/a | full_path = '.'.join((modname, qname)) if modname else qname |
|---|
| 47 | n/a | if full_path_for_error is None: |
|---|
| 48 | n/a | full_path_for_error = full_path |
|---|
| 49 | n/a | if full_path in self.set_handlers: |
|---|
| 50 | n/a | return self.set_handlers[full_path] |
|---|
| 51 | n/a | if qname in self.set_handlers: |
|---|
| 52 | n/a | return self.set_handlers[qname] |
|---|
| 53 | n/a | name = typ.__name__ |
|---|
| 54 | n/a | if name in self.set_handlers: |
|---|
| 55 | n/a | return self.set_handlers[name] |
|---|
| 56 | n/a | if None in self.set_handlers: |
|---|
| 57 | n/a | return self.set_handlers[None] |
|---|
| 58 | n/a | raise KeyError(full_path_for_error) |
|---|
| 59 | n/a | |
|---|
| 60 | n/a | |
|---|
| 61 | n/a | raw_data_manager = ContentManager() |
|---|
| 62 | n/a | |
|---|
| 63 | n/a | |
|---|
| 64 | n/a | def get_text_content(msg, errors='replace'): |
|---|
| 65 | n/a | content = msg.get_payload(decode=True) |
|---|
| 66 | n/a | charset = msg.get_param('charset', 'ASCII') |
|---|
| 67 | n/a | return content.decode(charset, errors=errors) |
|---|
| 68 | n/a | raw_data_manager.add_get_handler('text', get_text_content) |
|---|
| 69 | n/a | |
|---|
| 70 | n/a | |
|---|
| 71 | n/a | def get_non_text_content(msg): |
|---|
| 72 | n/a | return msg.get_payload(decode=True) |
|---|
| 73 | n/a | for maintype in 'audio image video application'.split(): |
|---|
| 74 | n/a | raw_data_manager.add_get_handler(maintype, get_non_text_content) |
|---|
| 75 | n/a | |
|---|
| 76 | n/a | |
|---|
| 77 | n/a | def get_message_content(msg): |
|---|
| 78 | n/a | return msg.get_payload(0) |
|---|
| 79 | n/a | for subtype in 'rfc822 external-body'.split(): |
|---|
| 80 | n/a | raw_data_manager.add_get_handler('message/'+subtype, get_message_content) |
|---|
| 81 | n/a | |
|---|
| 82 | n/a | |
|---|
| 83 | n/a | def get_and_fixup_unknown_message_content(msg): |
|---|
| 84 | n/a | # If we don't understand a message subtype, we are supposed to treat it as |
|---|
| 85 | n/a | # if it were application/octet-stream, per |
|---|
| 86 | n/a | # tools.ietf.org/html/rfc2046#section-5.2.4. Feedparser doesn't do that, |
|---|
| 87 | n/a | # so do our best to fix things up. Note that it is *not* appropriate to |
|---|
| 88 | n/a | # model message/partial content as Message objects, so they are handled |
|---|
| 89 | n/a | # here as well. (How to reassemble them is out of scope for this comment :) |
|---|
| 90 | n/a | return bytes(msg.get_payload(0)) |
|---|
| 91 | n/a | raw_data_manager.add_get_handler('message', |
|---|
| 92 | n/a | get_and_fixup_unknown_message_content) |
|---|
| 93 | n/a | |
|---|
| 94 | n/a | |
|---|
| 95 | n/a | def _prepare_set(msg, maintype, subtype, headers): |
|---|
| 96 | n/a | msg['Content-Type'] = '/'.join((maintype, subtype)) |
|---|
| 97 | n/a | if headers: |
|---|
| 98 | n/a | if not hasattr(headers[0], 'name'): |
|---|
| 99 | n/a | mp = msg.policy |
|---|
| 100 | n/a | headers = [mp.header_factory(*mp.header_source_parse([header])) |
|---|
| 101 | n/a | for header in headers] |
|---|
| 102 | n/a | try: |
|---|
| 103 | n/a | for header in headers: |
|---|
| 104 | n/a | if header.defects: |
|---|
| 105 | n/a | raise header.defects[0] |
|---|
| 106 | n/a | msg[header.name] = header |
|---|
| 107 | n/a | except email.errors.HeaderDefect as exc: |
|---|
| 108 | n/a | raise ValueError("Invalid header: {}".format( |
|---|
| 109 | n/a | header.fold(policy=msg.policy))) from exc |
|---|
| 110 | n/a | |
|---|
| 111 | n/a | |
|---|
| 112 | n/a | def _finalize_set(msg, disposition, filename, cid, params): |
|---|
| 113 | n/a | if disposition is None and filename is not None: |
|---|
| 114 | n/a | disposition = 'attachment' |
|---|
| 115 | n/a | if disposition is not None: |
|---|
| 116 | n/a | msg['Content-Disposition'] = disposition |
|---|
| 117 | n/a | if filename is not None: |
|---|
| 118 | n/a | msg.set_param('filename', |
|---|
| 119 | n/a | filename, |
|---|
| 120 | n/a | header='Content-Disposition', |
|---|
| 121 | n/a | replace=True) |
|---|
| 122 | n/a | if cid is not None: |
|---|
| 123 | n/a | msg['Content-ID'] = cid |
|---|
| 124 | n/a | if params is not None: |
|---|
| 125 | n/a | for key, value in params.items(): |
|---|
| 126 | n/a | msg.set_param(key, value) |
|---|
| 127 | n/a | |
|---|
| 128 | n/a | |
|---|
| 129 | n/a | # XXX: This is a cleaned-up version of base64mime.body_encode (including a bug |
|---|
| 130 | n/a | # fix in the calculation of unencoded_bytes_per_line). It would be nice to |
|---|
| 131 | n/a | # drop both this and quoprimime.body_encode in favor of enhanced binascii |
|---|
| 132 | n/a | # routines that accepted a max_line_length parameter. |
|---|
| 133 | n/a | def _encode_base64(data, max_line_length): |
|---|
| 134 | n/a | encoded_lines = [] |
|---|
| 135 | n/a | unencoded_bytes_per_line = max_line_length // 4 * 3 |
|---|
| 136 | n/a | for i in range(0, len(data), unencoded_bytes_per_line): |
|---|
| 137 | n/a | thisline = data[i:i+unencoded_bytes_per_line] |
|---|
| 138 | n/a | encoded_lines.append(binascii.b2a_base64(thisline).decode('ascii')) |
|---|
| 139 | n/a | return ''.join(encoded_lines) |
|---|
| 140 | n/a | |
|---|
| 141 | n/a | |
|---|
| 142 | n/a | def _encode_text(string, charset, cte, policy): |
|---|
| 143 | n/a | lines = string.encode(charset).splitlines() |
|---|
| 144 | n/a | linesep = policy.linesep.encode('ascii') |
|---|
| 145 | n/a | def embedded_body(lines): return linesep.join(lines) + linesep |
|---|
| 146 | n/a | def normal_body(lines): return b'\n'.join(lines) + b'\n' |
|---|
| 147 | n/a | if cte==None: |
|---|
| 148 | n/a | # Use heuristics to decide on the "best" encoding. |
|---|
| 149 | n/a | try: |
|---|
| 150 | n/a | return '7bit', normal_body(lines).decode('ascii') |
|---|
| 151 | n/a | except UnicodeDecodeError: |
|---|
| 152 | n/a | pass |
|---|
| 153 | n/a | if (policy.cte_type == '8bit' and |
|---|
| 154 | n/a | max(len(x) for x in lines) <= policy.max_line_length): |
|---|
| 155 | n/a | return '8bit', normal_body(lines).decode('ascii', 'surrogateescape') |
|---|
| 156 | n/a | sniff = embedded_body(lines[:10]) |
|---|
| 157 | n/a | sniff_qp = quoprimime.body_encode(sniff.decode('latin-1'), |
|---|
| 158 | n/a | policy.max_line_length) |
|---|
| 159 | n/a | sniff_base64 = binascii.b2a_base64(sniff) |
|---|
| 160 | n/a | # This is a little unfair to qp; it includes lineseps, base64 doesn't. |
|---|
| 161 | n/a | if len(sniff_qp) > len(sniff_base64): |
|---|
| 162 | n/a | cte = 'base64' |
|---|
| 163 | n/a | else: |
|---|
| 164 | n/a | cte = 'quoted-printable' |
|---|
| 165 | n/a | if len(lines) <= 10: |
|---|
| 166 | n/a | return cte, sniff_qp |
|---|
| 167 | n/a | if cte == '7bit': |
|---|
| 168 | n/a | data = normal_body(lines).decode('ascii') |
|---|
| 169 | n/a | elif cte == '8bit': |
|---|
| 170 | n/a | data = normal_body(lines).decode('ascii', 'surrogateescape') |
|---|
| 171 | n/a | elif cte == 'quoted-printable': |
|---|
| 172 | n/a | data = quoprimime.body_encode(normal_body(lines).decode('latin-1'), |
|---|
| 173 | n/a | policy.max_line_length) |
|---|
| 174 | n/a | elif cte == 'base64': |
|---|
| 175 | n/a | data = _encode_base64(embedded_body(lines), policy.max_line_length) |
|---|
| 176 | n/a | else: |
|---|
| 177 | n/a | raise ValueError("Unknown content transfer encoding {}".format(cte)) |
|---|
| 178 | n/a | return cte, data |
|---|
| 179 | n/a | |
|---|
| 180 | n/a | |
|---|
| 181 | n/a | def set_text_content(msg, string, subtype="plain", charset='utf-8', cte=None, |
|---|
| 182 | n/a | disposition=None, filename=None, cid=None, |
|---|
| 183 | n/a | params=None, headers=None): |
|---|
| 184 | n/a | _prepare_set(msg, 'text', subtype, headers) |
|---|
| 185 | n/a | cte, payload = _encode_text(string, charset, cte, msg.policy) |
|---|
| 186 | n/a | msg.set_payload(payload) |
|---|
| 187 | n/a | msg.set_param('charset', |
|---|
| 188 | n/a | email.charset.ALIASES.get(charset, charset), |
|---|
| 189 | n/a | replace=True) |
|---|
| 190 | n/a | msg['Content-Transfer-Encoding'] = cte |
|---|
| 191 | n/a | _finalize_set(msg, disposition, filename, cid, params) |
|---|
| 192 | n/a | raw_data_manager.add_set_handler(str, set_text_content) |
|---|
| 193 | n/a | |
|---|
| 194 | n/a | |
|---|
| 195 | n/a | def set_message_content(msg, message, subtype="rfc822", cte=None, |
|---|
| 196 | n/a | disposition=None, filename=None, cid=None, |
|---|
| 197 | n/a | params=None, headers=None): |
|---|
| 198 | n/a | if subtype == 'partial': |
|---|
| 199 | n/a | raise ValueError("message/partial is not supported for Message objects") |
|---|
| 200 | n/a | if subtype == 'rfc822': |
|---|
| 201 | n/a | if cte not in (None, '7bit', '8bit', 'binary'): |
|---|
| 202 | n/a | # http://tools.ietf.org/html/rfc2046#section-5.2.1 mandate. |
|---|
| 203 | n/a | raise ValueError( |
|---|
| 204 | n/a | "message/rfc822 parts do not support cte={}".format(cte)) |
|---|
| 205 | n/a | # 8bit will get coerced on serialization if policy.cte_type='7bit'. We |
|---|
| 206 | n/a | # may end up claiming 8bit when it isn't needed, but the only negative |
|---|
| 207 | n/a | # result of that should be a gateway that needs to coerce to 7bit |
|---|
| 208 | n/a | # having to look through the whole embedded message to discover whether |
|---|
| 209 | n/a | # or not it actually has to do anything. |
|---|
| 210 | n/a | cte = '8bit' if cte is None else cte |
|---|
| 211 | n/a | elif subtype == 'external-body': |
|---|
| 212 | n/a | if cte not in (None, '7bit'): |
|---|
| 213 | n/a | # http://tools.ietf.org/html/rfc2046#section-5.2.3 mandate. |
|---|
| 214 | n/a | raise ValueError( |
|---|
| 215 | n/a | "message/external-body parts do not support cte={}".format(cte)) |
|---|
| 216 | n/a | cte = '7bit' |
|---|
| 217 | n/a | elif cte is None: |
|---|
| 218 | n/a | # http://tools.ietf.org/html/rfc2046#section-5.2.4 says all future |
|---|
| 219 | n/a | # subtypes should be restricted to 7bit, so assume that. |
|---|
| 220 | n/a | cte = '7bit' |
|---|
| 221 | n/a | _prepare_set(msg, 'message', subtype, headers) |
|---|
| 222 | n/a | msg.set_payload([message]) |
|---|
| 223 | n/a | msg['Content-Transfer-Encoding'] = cte |
|---|
| 224 | n/a | _finalize_set(msg, disposition, filename, cid, params) |
|---|
| 225 | n/a | raw_data_manager.add_set_handler(email.message.Message, set_message_content) |
|---|
| 226 | n/a | |
|---|
| 227 | n/a | |
|---|
| 228 | n/a | def set_bytes_content(msg, data, maintype, subtype, cte='base64', |
|---|
| 229 | n/a | disposition=None, filename=None, cid=None, |
|---|
| 230 | n/a | params=None, headers=None): |
|---|
| 231 | n/a | _prepare_set(msg, maintype, subtype, headers) |
|---|
| 232 | n/a | if cte == 'base64': |
|---|
| 233 | n/a | data = _encode_base64(data, max_line_length=msg.policy.max_line_length) |
|---|
| 234 | n/a | elif cte == 'quoted-printable': |
|---|
| 235 | n/a | # XXX: quoprimime.body_encode won't encode newline characters in data, |
|---|
| 236 | n/a | # so we can't use it. This means max_line_length is ignored. Another |
|---|
| 237 | n/a | # bug to fix later. (Note: encoders.quopri is broken on line ends.) |
|---|
| 238 | n/a | data = binascii.b2a_qp(data, istext=False, header=False, quotetabs=True) |
|---|
| 239 | n/a | data = data.decode('ascii') |
|---|
| 240 | n/a | elif cte == '7bit': |
|---|
| 241 | n/a | # Make sure it really is only ASCII. The early warning here seems |
|---|
| 242 | n/a | # worth the overhead...if you care write your own content manager :). |
|---|
| 243 | n/a | data.encode('ascii') |
|---|
| 244 | n/a | elif cte in ('8bit', 'binary'): |
|---|
| 245 | n/a | data = data.decode('ascii', 'surrogateescape') |
|---|
| 246 | n/a | msg.set_payload(data) |
|---|
| 247 | n/a | msg['Content-Transfer-Encoding'] = cte |
|---|
| 248 | n/a | _finalize_set(msg, disposition, filename, cid, params) |
|---|
| 249 | n/a | for typ in (bytes, bytearray, memoryview): |
|---|
| 250 | n/a | raw_data_manager.add_set_handler(typ, set_bytes_content) |
|---|
