131 lines
3.4 KiB
Python
131 lines
3.4 KiB
Python
from xenu_nntp.message import Message
|
|
|
|
class MBoxReaderError(Exception):
|
|
pass
|
|
|
|
class MBoxReaderBuffer():
|
|
def __init__(self):
|
|
self.lines = [None, None, None, None]
|
|
|
|
def add(self, line: str):
|
|
if self.lines[0] is None:
|
|
self.lines[0] = line
|
|
elif self.lines[1] is None:
|
|
self.lines[1] = line
|
|
elif self.lines[2] is None:
|
|
self.lines[2] = line
|
|
elif self.lines[3] is None:
|
|
self.lines[3] = line
|
|
else:
|
|
self.lines[0] = self.lines[1]
|
|
self.lines[1] = self.lines[2]
|
|
self.lines[2] = self.lines[3]
|
|
self.lines[3] = line
|
|
|
|
def is_empty_line(self, line):
|
|
return self.lines[line] == '\n'
|
|
|
|
def is_from_line(self, line):
|
|
return self.lines[line][0:5] == 'From '
|
|
|
|
def is_header_line(self, line):
|
|
return Message.RE_HEADER.match(self.lines[line]) is not None
|
|
|
|
def is_start(self):
|
|
if self.lines[0] is None or self.lines[1] is None:
|
|
return False
|
|
|
|
if self.is_from_line(0) \
|
|
and self.is_header_line(1) \
|
|
and self.lines[2] is None \
|
|
and self.lines[3] is None:
|
|
return True
|
|
elif self.is_empty_line(0) \
|
|
and self.is_empty_line(1) \
|
|
and self.is_from_line(2) \
|
|
and self.is_header_line(3):
|
|
return True
|
|
else:
|
|
return False
|
|
|
|
def is_end(self):
|
|
return self.lines[0] is not None \
|
|
and self.is_empty_line(1) \
|
|
and self.is_empty_line(2) \
|
|
and self.is_from_line(3)
|
|
|
|
class MBoxReader():
|
|
__slots__ = 'path', 'fh', 'line', 'buf', 'message',
|
|
|
|
def __init__(self, path: str):
|
|
self.path = path
|
|
self.fh = open(path, 'r', newline='')
|
|
self.line = 0
|
|
self.buf = MBoxReaderBuffer()
|
|
self.message = None
|
|
|
|
def get_message(self):
|
|
while True:
|
|
line = self.fh.readline()
|
|
|
|
#
|
|
# We have reached the final message in the stream.
|
|
#
|
|
if line is None or line == '':
|
|
if self.message:
|
|
ret = self.message
|
|
ret.finish()
|
|
|
|
self.message = None
|
|
|
|
return ret
|
|
|
|
return None
|
|
|
|
line = line.replace('\x00', '')
|
|
|
|
self.line += 1
|
|
|
|
#
|
|
# Load the line in the circular mbox state buffer.
|
|
#
|
|
self.buf.add(line)
|
|
|
|
ret = None
|
|
|
|
#
|
|
# If we are at the start of a new message...
|
|
#
|
|
if self.buf.is_start():
|
|
if self.message is not None:
|
|
#
|
|
# If we already have a message, return it.
|
|
#
|
|
ret = self.message
|
|
|
|
#
|
|
# ...Prepare a new message object.
|
|
#
|
|
self.message = Message()
|
|
elif not self.buf.is_end():
|
|
if self.message is not None:
|
|
#
|
|
# Buffer the line into the current message.
|
|
#
|
|
self.message.readline(line)
|
|
|
|
#
|
|
# Return the message if we have one.
|
|
#
|
|
if ret:
|
|
ret.finish()
|
|
return ret
|
|
|
|
def messages(self):
|
|
while True:
|
|
message = self.get_message()
|
|
|
|
if message is None:
|
|
break
|
|
|
|
yield message
|