| 1 | """unicode email support""" |
|---|
| 2 | |
|---|
| 3 | import email |
|---|
| 4 | from email.Utils import parseaddr, parsedate |
|---|
| 5 | from email.Header import decode_header |
|---|
| 6 | |
|---|
| 7 | try: |
|---|
| 8 | from mx.DateTime import DateTime |
|---|
| 9 | except ImportError: |
|---|
| 10 | def DateTime(*args): return None |
|---|
| 11 | |
|---|
| 12 | def decode_QP(string): |
|---|
| 13 | parts = [] |
|---|
| 14 | for decoded, charset in decode_header(string): |
|---|
| 15 | if charset is None: |
|---|
| 16 | charset = 'iso-8859-15' |
|---|
| 17 | parts.append(unicode(decoded, charset, 'replace')) |
|---|
| 18 | |
|---|
| 19 | return u' '.join(parts) |
|---|
| 20 | |
|---|
| 21 | def message_from_file(fd): |
|---|
| 22 | try: |
|---|
| 23 | return UMessage(email.message_from_file(fd)) |
|---|
| 24 | except email.Errors.MessageParseError: |
|---|
| 25 | return '' |
|---|
| 26 | |
|---|
| 27 | def message_from_string(string): |
|---|
| 28 | try: |
|---|
| 29 | return UMessage(email.message_from_string(string)) |
|---|
| 30 | except email.Errors.MessageParseError: |
|---|
| 31 | return '' |
|---|
| 32 | |
|---|
| 33 | class UMessage: |
|---|
| 34 | """Encapsulates an email.Message instance and returns only unicode objects""" |
|---|
| 35 | |
|---|
| 36 | def __init__(self, message): |
|---|
| 37 | self.message = message |
|---|
| 38 | |
|---|
| 39 | # email.Message interface ################################################# |
|---|
| 40 | |
|---|
| 41 | def get(self, header, default=None): |
|---|
| 42 | value = self.message.get(header, default) |
|---|
| 43 | if value: |
|---|
| 44 | return decode_QP(value) |
|---|
| 45 | return value |
|---|
| 46 | |
|---|
| 47 | def get_all(self, header, default=()): |
|---|
| 48 | return [decode_QP(val) for val in self.message.get_all(header, default) |
|---|
| 49 | if val is not None] |
|---|
| 50 | |
|---|
| 51 | def get_payload(self, index=None, decode=False): |
|---|
| 52 | message = self.message |
|---|
| 53 | if index is None: |
|---|
| 54 | payload = message.get_payload(index, decode) |
|---|
| 55 | if isinstance(payload, list): |
|---|
| 56 | return [UMessage(msg) for msg in payload] |
|---|
| 57 | if message.get_content_maintype() != 'text': |
|---|
| 58 | return payload |
|---|
| 59 | |
|---|
| 60 | charset = message.get_content_charset() or 'iso-8859-1' |
|---|
| 61 | if charset == 'unknown-8bit': |
|---|
| 62 | charset = 'iso-8859-1' |
|---|
| 63 | return unicode(payload or '', charset) |
|---|
| 64 | else: |
|---|
| 65 | payload = UMessage(message.get_payload(index, decode)) |
|---|
| 66 | return payload |
|---|
| 67 | |
|---|
| 68 | def is_multipart(self): |
|---|
| 69 | return self.message.is_multipart() |
|---|
| 70 | |
|---|
| 71 | def get_boundary(self): |
|---|
| 72 | return self.message.get_boundary() |
|---|
| 73 | |
|---|
| 74 | def walk(self): |
|---|
| 75 | for part in self.message.walk(): |
|---|
| 76 | yield UMessage(part) |
|---|
| 77 | |
|---|
| 78 | def get_content_maintype(self): |
|---|
| 79 | return unicode(self.message.get_content_maintype()) |
|---|
| 80 | |
|---|
| 81 | def get_content_type(self): |
|---|
| 82 | return unicode(self.message.get_content_type()) |
|---|
| 83 | |
|---|
| 84 | def get_filename(self, failobj=None): |
|---|
| 85 | value = self.message.get_filename(failobj) |
|---|
| 86 | if value is failobj: |
|---|
| 87 | return value |
|---|
| 88 | try: |
|---|
| 89 | return unicode(value) |
|---|
| 90 | except UnicodeDecodeError: |
|---|
| 91 | return u'error decoding filename' |
|---|
| 92 | |
|---|
| 93 | # other convenience methods ############################################### |
|---|
| 94 | |
|---|
| 95 | def headers(self): |
|---|
| 96 | """return an unicode string containing all the message's headers""" |
|---|
| 97 | values = [] |
|---|
| 98 | for header in self.message.keys(): |
|---|
| 99 | values.append(u'%s: %s' % (header, self.get(header))) |
|---|
| 100 | return '\n'.join(values) |
|---|
| 101 | |
|---|
| 102 | def multi_addrs(self, header): |
|---|
| 103 | """return a list of 2-uple (name, address) for the given address (which |
|---|
| 104 | is exepected to be an header containing address such as from, to, cc...) |
|---|
| 105 | """ |
|---|
| 106 | persons = [] |
|---|
| 107 | for person in self.get_all(header, ()): |
|---|
| 108 | name, mail = parseaddr(person) |
|---|
| 109 | persons.append((name, mail)) |
|---|
| 110 | return persons |
|---|
| 111 | |
|---|
| 112 | def date(self): |
|---|
| 113 | """return a mx.DateTime object for the email's date or None if no date is |
|---|
| 114 | set or if it can't be parsed |
|---|
| 115 | """ |
|---|
| 116 | value = self.get('date') |
|---|
| 117 | if value: |
|---|
| 118 | datetuple = parsedate(value) |
|---|
| 119 | if datetuple: |
|---|
| 120 | return DateTime(*datetuple[:6]) |
|---|
| 121 | return None |
|---|
| 122 | |
|---|
| 123 | |
|---|
| 124 | |
|---|
| 125 | |
|---|