diff options
| author | NavaL <ayoyo@thoughtworks.com> | 2016-10-21 21:11:22 +0200 |
|---|---|---|
| committer | NavaL <ayoyo@thoughtworks.com> | 2016-10-21 21:11:22 +0200 |
| commit | aaed875ecce770bde03a2066ade40944cd1ab31d (patch) | |
| tree | 8b1808dd803a1b45e1d3be394b97ac9010d9be96 /service/pixelated/adapter | |
| parent | 2dcdbc47e6232309329b4ec53f5daaab201d3d5b (diff) | |
[#797] reinstating use of the provided email body charstet encoding
Diffstat (limited to 'service/pixelated/adapter')
| -rw-r--r-- | service/pixelated/adapter/mailstore/body_parser.py | 5 | ||||
| -rw-r--r-- | service/pixelated/adapter/mailstore/leap_mailstore.py | 3 |
2 files changed, 5 insertions, 3 deletions
diff --git a/service/pixelated/adapter/mailstore/body_parser.py b/service/pixelated/adapter/mailstore/body_parser.py index 5408bd9d..8cac75cf 100644 --- a/service/pixelated/adapter/mailstore/body_parser.py +++ b/service/pixelated/adapter/mailstore/body_parser.py @@ -31,13 +31,14 @@ def _parse_charset_header(content_type_and_charset_header, default_charset='us-a class BodyParser(object): - def __init__(self, content, content_type='text/plain; charset="us-ascii"', content_transfer_encoding=None): + def __init__(self, content, content_type='text/plain; charset="us-ascii"', content_transfer_encoding=None, charset=None): self._content = content self._content_type = content_type self._content_transfer_encoding = content_transfer_encoding + self._charset = charset def parsed_content(self): - charset = _parse_charset_header(self._content_type) + charset = self._charset or _parse_charset_header(self._content_type) text = self._serialize_for_parser(charset) decoded_body = self._parse_and_decode(text) diff --git a/service/pixelated/adapter/mailstore/leap_mailstore.py b/service/pixelated/adapter/mailstore/leap_mailstore.py index 900a7cbb..288223dd 100644 --- a/service/pixelated/adapter/mailstore/leap_mailstore.py +++ b/service/pixelated/adapter/mailstore/leap_mailstore.py @@ -330,7 +330,8 @@ class LeapMailStore(MailStore): parser = BodyParser('', content_type='text/plain', content_transfer_encoding='UTF-8') # It fix the problem when leap doesn'r found body_phash and returns empty string if not isinstance(content_doc, str): - parser = BodyParser(content_doc.raw, content_type=content_doc.content_type, content_transfer_encoding=content_doc.content_transfer_encoding) + parser = BodyParser(content_doc.raw, content_type=content_doc.content_type, + content_transfer_encoding=content_doc.content_transfer_encoding, charset=content_doc.charset) defer.returnValue(parser.parsed_content()) |