def __init__(self, path, prefix='', filename=None):
"""
:param path: path to the msg file in the system or is the raw msg file.
:param prefix: used for extracting embedded msg files
inside the main one. Do not set manually unless
you know what you are doing.
"""
logger.log(5, 'prefix: {}'.format(prefix))
self.__path = path
olefile.OleFileIO.__init__(self, path)
prefixl = []
tmp_condition = prefix != ''
if tmp_condition:
prefix = inputToString(prefix, 'utf-8')
try:
prefix = '/'.join(prefix)
except:
raise TypeError('Invalid prefix type: ' + str(type(prefix)) +
'\n(This was probably caused by you setting it manually).')
prefix = prefix.replace('\\', '/')
g = prefix.split('/')
if g[-1] == '':
g.pop()
prefixl = g
if prefix[-1] != '/':
prefix += '/'
self.__prefix = prefix
self.__prefixList = prefixl
if tmp_condition:
filename = self._getStringStream(prefixl[:-1] + ['__substg1.0_3001'], prefix=False)
if filename is not None:
self.filename = filename
else:
logger.log(5, ':param path: has __len__ attribute?: {}'.format(has_len(path)))
if has_len(path):
if len(path) < 1536:
self.filename = path
logger.log(5, ':param path: length is {}; Using :param path: as file path'.format(len(path)))
else:
logger.log(5, ':param path: length is {}; Using :param path: as raw msg stream'.format(len(path)))
self.filename = None
else:
self.filename = None
self.mainProperties
recipientDirs = []
for dir_ in self.listDir():
if dir_[len(self.__prefixList)].startswith('__recip') and\
dir_[len(self.__prefixList)] not in recipientDirs:
recipientDirs.append(dir_[len(self.__prefixList)])
self.recipients
self.attachments
self.date
def body(self):
"""
Returns the message body, if it exists.
"""
try:
return self._body
except AttributeError:
self._body = self._getStringStream('__substg1.0_1000')
if self._body:
self._body = inputToString(self._body, 'utf-8')
a = re.search('\n', self._body)
if a is not None:
if re.search('\r\n', self._body) is not None:
self.__crlf = '\r\n'
return self._body
def save(self,
contentId=False,
json=False,
useFileName=False,
raw=False,
customPath=None,
customFilename=None
): #, html = False, rtf = False, allowFallback = False):
# Check if the user has specified a custom filename
filename = None
if customFilename is not None and customFilename != '':
filename = customFilename
else:
# If not...
# Check if user wants to save the file under the Content-id
if contentId:
filename = self.cid
# If filename is None at this point, use long filename as first preference
if filename is None:
filename = self.longFilename
# Otherwise use the short filename
if filename is None:
filename = self.shortFilename
# Otherwise just make something up!
if filename is None:
filename = 'UnknownFilename ' + \
''.join(random.choice(string.ascii_uppercase + string.digits)
for _ in range(5)) + '.bin'
# Someone managed to have a null character here, so let's get rid of that
filename = prepareFilename(
inputToString(filename, self.msg.stringEncoding))
if customPath is not None and customPath != '':
if customPath[-1] != '/' or customPath[-1] != '\\':
customPath += '/'
filename = customPath + filename
if self.__type == "data":
with open(filename, 'wb') as f:
f.write(self.__data)
else:
self.saveEmbededMessage(
contentId, json, useFileName, raw, customPath,
customFilename) #, html, rtf, allowFallback)
return filename
def __init__(self,
path,
prefix='',
attachmentClass=Attachment,
filename=None,
overrideEncoding=None):
"""
:param path: path to the msg file in the system or is the raw msg file.
:param prefix: used for extracting embeded msg files
inside the main one. Do not set manually unless
you know what you are doing.
:param attachmentClass: optional, the class the MSGFile object
will use for attachments. You probably should
not change this value unless you know what you
are doing.
:param filename: optional, the filename to be used by default when saving.
:param overrideEncoding: optional, an encoding to use instead of the one
specified by the msg file. Do not report encoding errors caused by this.
"""
# WARNING DO NOT MANUALLY MODIFY PREFIX. Let the program set it.
self.__path = path
self.__attachmentClass = attachmentClass
if overrideEncoding is not None:
codecs.lookup(overrideEncoding)
logger.warning(
'You have chosen to override the string encoding. Do not report encoding erros caused by this.'
)
self.__stringEncoding = overrideEncoding
self.__overrideEncoding = overrideEncoding
try:
olefile.OleFileIO.__init__(self, path)
except IOError as e: # py2 and py3 compatible
logger.error(e)
if str(e) == 'not an OLE2 structured storage file':
raise InvalidFileFormat(e)
else:
raise
prefixl = []
tmp_condition = prefix != ''
if tmp_condition:
try:
prefix = inputToString(prefix, 'utf-8')
except:
try:
prefix = '/'.join(prefix)
except:
raise TypeError(
'Invalid prefix type: ' + str(type(prefix)) +
'\n(This was probably caused by you setting it manually).'
)
prefix = prefix.replace('\\', '/')
g = prefix.split('/')
if g[-1] == '':
g.pop()
prefixl = g
if prefix[-1] != '/':
prefix += '/'
self.__prefix = prefix
self.__prefixList = prefixl
if tmp_condition:
filename = self._getStringStream(prefixl[:-1] +
['__substg1.0_3001'],
prefix=False)
if filename is not None:
self.filename = filename
elif has_len(path):
if len(path) < 1536:
self.filename = path
else:
self.filename = None
else:
self.filename = None
def __init__(self,
path,
prefix='',
attachmentClass=Attachment,
filename=None):
"""
:param path: path to the msg file in the system or is the raw msg file.
:param prefix: used for extracting embeded msg files
inside the main one. Do not set manually unless
you know what you are doing.
:param attachmentClass: optional, the class the Message object
will use for attachments. You probably should
not change this value unless you know what you
are doing.
:param filename: optional, the filename to be used by default when saving.
"""
# WARNING DO NOT MANUALLY MODIFY PREFIX. Let the program set it.
self.__path = path
self.__attachmentClass = attachmentClass
try:
olefile.OleFileIO.__init__(self, path)
except IOError as e: # py2 and py3 compatible
logger.error(e)
if str(e) == 'not an OLE2 structured storage file':
raise InvalidFileFormat(e)
else:
raise
prefixl = []
tmp_condition = prefix != ''
if tmp_condition:
try:
prefix = inputToString(prefix, 'utf-8')
except:
try:
prefix = '/'.join(prefix)
except:
raise TypeError(
'Invalid prefix type: ' + str(type(prefix)) +
'\n(This was probably caused by you setting it manually).'
)
prefix = prefix.replace('\\', '/')
g = prefix.split("/")
if g[-1] == '':
g.pop()
prefixl = g
if prefix[-1] != '/':
prefix += '/'
self.__prefix = prefix
self.__prefixList = prefixl
if tmp_condition:
filename = self._getStringStream(prefixl[:-1] +
['__substg1.0_3001'],
prefix=False)
if filename is not None:
self.filename = filename
elif has_len(path):
if len(path) < 1536:
self.filename = path
else:
self.filename = None
else:
self.filename = None
# Initialize properties in the order that is least likely to cause bugs.
# TODO have each function check for initialization of needed data so these
# lines will be unnecessary.
self.mainProperties
self.header
self.recipients
self.attachments
self.to
self.cc
self.sender
self.date
self.__crlf = '\n' # This variable keeps track of what the new line character should be
self.body
def save(self,
toJson=False,
useFileName=False,
raw=False,
ContentId=False,
customPath=None,
customFilename=None,
html=False,
rtf=False):
"""
Saves the message body and attachments found in the message. Setting toJson
to true will output the message body as JSON-formatted text. The body and
attachments are stored in a folder. Setting useFileName to true will mean that
the filename is used as the name of the folder; otherwise, the message's date
and subject are used as the folder name.
Here is the absolute order of prioity for the name of the folder:
1. customFilename
2. self.filename if useFileName
3. {date} {subject}
"""
crlf = inputToBytes(self.__crlf, 'utf-8')
if customFilename != None and customFilename != '':
dirName = customFilename
else:
if useFileName:
# strip out the extension
if self.filename is not None:
dirName = self.filename.split('/').pop().split('.')[0]
else:
ValueError(
'Filename must be specified, or path must have been an actual path, to save using filename'
)
else:
# Create a directory based on the date and subject of the message
d = self.parsedDate
if d is not None:
dirName = '{0:02d}-{1:02d}-{2:02d}_{3:02d}{4:02d}'.format(
*d)
else:
dirName = 'UnknownDate'
if self.subject is None:
subject = '[No subject]'
else:
subject = ''.join(i for i in self.subject
if i not in r'\/:*?"<>|')
dirName = dirName + ' ' + subject
if customPath != None and customPath != '':
if customPath[-1] != '/' or customPath[-1] != '\\':
customPath += '/'
dirName = customPath + dirName
try:
os.makedirs(dirName)
except Exception:
newDirName = addNumToDir(dirName)
if newDirName is not None:
dirName = newDirName
else:
raise Exception(
"Failed to create directory '%s'. Does it already exist?" %
dirName)
oldDir = os.getcwdu()
try:
os.chdir(dirName)
attachmentNames = []
# Save the attachments
for attachment in self.attachments:
attachmentNames.append(
attachment.save(ContentId,
toJson,
useFileName,
raw,
html=html,
rtf=rtf))
# Save the message body
fext = 'json' if toJson else 'txt'
useHtml = False
useRtf = False
#if html:
# if self.htmlBody is not None:
# useHtml = True
# fext = 'html'
#elif rtf:
# if self.htmlBody is not None:
# useRtf = True
# fext = 'rtf'
with open('message.' + fext, 'wb') as f:
if toJson:
emailObj = {
'from': inputToString(self.sender, 'utf-8'),
'to': inputToString(self.to, 'utf-8'),
'cc': inputToString(self.cc, 'utf-8'),
'subject': inputToString(self.subject, 'utf-8'),
'date': inputToString(self.date, 'utf-8'),
'attachments': attachmentNames,
'body': decode_utf7(self.body)
}
f.write(inputToBytes(json.dumps(emailObj), 'utf-8'))
else:
if useHtml:
# Do stuff
pass
elif useRtf:
# Do stuff
pass
else:
f.write(b'From: ' +
inputToBytes(self.sender, 'utf-8') + crlf)
f.write(b'To: ' + inputToBytes(self.to, 'utf-8') +
crlf)
f.write(b'CC: ' + inputToBytes(self.cc, 'utf-8') +
crlf)
f.write(b'Subject: ' +
inputToBytes(self.subject, 'utf-8') + crlf)
f.write(b'Date: ' + inputToBytes(self.date, 'utf-8') +
crlf)
f.write(b'-----------------' + crlf + crlf)
f.write(inputToBytes(self.body, 'utf-8'))
except Exception as e:
self.saveRaw()
raise
finally:
# Return to previous directory
os.chdir(oldDir)
def save(self,
toJson=False,
useFileName=False,
raw=False,
ContentId=False,
customPath=None,
customFilename=None
): #, html = False, rtf = False, allowFallback = False):
"""
Saves the message body and attachments found in the message. The body and
attachments are stored in a folder. Setting useFileName to true will mean that
the filename is used as the name of the folder; otherwise, the message's date
and subject are used as the folder name.
Here is the absolute order of prioity for the name of the folder:
1. customFilename
2. self.filename if useFileName
3. {date} {subject}
"""
#There are several parameters used to determine how the message will be saved.
#By default, the message will be saved as plain text. Setting one of the
#following parameters to True will change that:
# * :param html: will try to output the message in HTML format.
# * :param json: will output the message in JSON format.
# * :param raw: will output the message in a raw format.
# * :param rtf: will output the message in RTF format.
#
#Usage of more than one formatting parameter will raise an exception.
#
#Using HTML or RTF will raise an exception if they could not be retrieved
#unless you have :param allowFallback: set to True. Fallback will go in this
#order, starting at the top most format that is set:
# * HTML
# * RTF
# * Plain text
#"""
count = 1 if toJson else 0
#count += 1 if html else 0
#count += 1 if rtf else 0
count += 1 if raw else 0
if count > 1:
raise IncompatibleOptionsError(
'Only one of the following options may be used at a time: toJson, raw, html, rtf'
)
crlf = inputToBytes(self.crlf, 'utf-8')
if customFilename != None and customFilename != '':
dirName = customFilename
else:
if useFileName:
# strip out the extension
if self.filename is not None:
dirName = self.filename.split('/').pop().split('.')[0]
else:
ValueError(
'Filename must be specified, or path must have been an actual path, to save using filename'
)
else:
# Create a directory based on the date and subject of the message
d = self.parsedDate
if d is not None:
dirName = '{0:02d}-{1:02d}-{2:02d}_{3:02d}{4:02d}'.format(
*d)
else:
dirName = 'UnknownDate'
if self.subject is None:
subject = '[No subject]'
else:
subject = prepareFilename(self.subject)
dirName = dirName + ' ' + subject
if customPath != None and customPath != '':
if customPath[-1] != '/' or customPath[-1] != '\\':
customPath += '/'
dirName = customPath + dirName
try:
os.makedirs(dirName)
except Exception:
newDirName = addNumToDir(dirName)
if newDirName is not None:
dirName = newDirName
else:
raise Exception(
"Failed to create directory '%s'. Does it already exist?" %
dirName)
oldDir = os.getcwdu()
try:
os.chdir(dirName)
attachmentNames = []
# Save the attachments
for attachment in self.attachments:
attachmentNames.append(
attachment.save(ContentId, toJson, useFileName, raw)
) #, html = html, rtf = rtf, allowFallback = allowFallback))
# Save the message body
fext = 'json' if toJson else 'txt'
useHtml = False
useRtf = False
#if html:
# if self.htmlBody is not None:
# useHtml = True
# fext = 'html'
#elif not allowFallback:
# raise DataNotFoundError('Could not find the htmlBody')
#if rtf or (html and not useHtml):
# if self.rtfBody is not None:
# useRtf = True
# fext = 'rtf'
#elif not allowFallback:
# raise DataNotFoundError('Could not find the rtfBody')
with open('message.' + fext, 'wb') as f:
if toJson:
emailObj = {
'from': inputToString(self.sender, 'utf-8'),
'to': inputToString(self.to, 'utf-8'),
'cc': inputToString(self.cc, 'utf-8'),
'subject': inputToString(self.subject, 'utf-8'),
'date': inputToString(self.date, 'utf-8'),
'attachments': attachmentNames,
'body': decode_utf7(self.body)
}
f.write(inputToBytes(json.dumps(emailObj), 'utf-8'))
else:
if useHtml:
# Do stuff
pass
elif useRtf:
# Do stuff
pass
else:
f.write(b'From: ' +
inputToBytes(self.sender, 'utf-8') + crlf)
f.write(b'To: ' + inputToBytes(self.to, 'utf-8') +
crlf)
f.write(b'CC: ' + inputToBytes(self.cc, 'utf-8') +
crlf)
f.write(b'Subject: ' +
inputToBytes(self.subject, 'utf-8') + crlf)
f.write(b'Date: ' + inputToBytes(self.date, 'utf-8') +
crlf)
f.write(b'-----------------' + crlf + crlf)
f.write(inputToBytes(self.body, 'utf-8'))
except Exception as e:
self.saveRaw()
raise
finally:
# Return to previous directory
os.chdir(oldDir)
# Return the instance so that functions can easily be chained.
return self
