xenu_nntp/lib/nntp/tiny/mbox.py

104 lines
2.6 KiB
Python
Raw Normal View History

2024-11-08 15:47:21 -05:00
from nntp.tiny.message import Message
2024-11-06 23:34:51 -05:00
class MBoxReaderError(Exception):
pass
class MBoxReaderBuffer():
def __init__(self):
self.lines = [None, None, None, None]
self.index = None
def add(self, line: str):
if self.lines[0] is None:
self.lines[0] = line
self.index = 0
elif self.lines[1] is None:
self.lines[1] = line
self.index = 1
elif self.lines[2] is None:
self.lines[2] = line
self.index = 2
elif self.lines[3] is None:
self.lines[3] = line
self.index = 3
else:
self.lines[0] = self.lines[1]
self.lines[1] = self.lines[2]
self.lines[2] = self.lines[3]
self.lines[3] = line
self.index = 3
if self.index is None:
self.index = 0
elif self.index < 3:
self.index += 1
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):
2024-11-08 20:31:15 -05:00
return Message.RE_HEADER.match(self.lines[line]) is not None
2024-11-06 23:34:51 -05:00
def is_start(self):
if self.lines[0] is None or self.lines[1] is None:
return
if self.is_from_line(0) and self.is_header_line(1):
pass
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 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()
if line is None or line == '':
2024-11-06 23:34:51 -05:00
ret = self.message
self.message = None
return ret
self.line += 1
self.buf.add(line)
if self.buf.is_start():
if self.message is None:
2024-11-08 15:47:21 -05:00
self.message = Message()
2024-11-06 23:34:51 -05:00
else:
ret = self.message
2024-11-08 15:47:21 -05:00
self.message = Message()
2024-11-10 02:19:08 -05:00
self.message.read_line(line)
2024-11-06 23:34:51 -05:00
return ret
if self.message:
2024-11-10 02:19:08 -05:00
self.message.read_line(line)
2024-11-06 23:34:51 -05:00
def messages(self):
while True:
message = self.get_message()
if message is None:
break
yield message