#!/usr/bin/python
# BeOS to mbox converter
#
# Written by: Colin Stewart <colin at owlfish.com> Copyright 2003
#
# WARNING: This software might eat your email - use after taking backups
# of the relevant files!

import sys, os, os.path, re

# Get a list of all files from the given directory
# Read in each file
# Retrieve the X-From_: line for use as the 'From ' seperator
# In a loop check for a match of the regex '^>*From '.  If present then escape.

# If your mail is already mostly escaped (mine was) then leave as is, otherwise
# use the first regex and comment out the second one.
#ESCAPE_REGEX=re.compile ('^>*From ')
ESCAPE_REGEX=re.compile ('^From ')

XFROM_LINE_REGEX=re.compile ('^X-From_: ')
DATE_LINE_REGEX=re.compile ('^Date: ')
RECIEVED_DATE_REGEX=re.compile ('Received: [^;]*; ')

class BeMailToMbox:
	def __init__ (self, sourceDir, destFile):
		self.sourceDir = sourceDir
		self.destFile = destFile
		
		if (not os.path.isdir (sourceDir)):
			raise Exception ("Provided source directory is not a directory!")
		
	def convert (self):
		print "Starting conversion process."
		candidateFiles = os.listdir (self.sourceDir)
		self.outputFile = open (self.destFile, 'w')
		count = 0
		for candidate in candidateFiles:
			beosMailFile = os.path.join (self.sourceDir, candidate)
			if (candidate == "Mail-It.cache"):
				print "Skipping cache file."
			elif (not os.path.isfile (beosMailFile)):
				print "WARNING: %s is not a file - skipping." % beosMailFile
			else:
				self.process (beosMailFile)
				count += 1
		self.outputFile.close()
		print "Finished - wrote %s messages to mbox" % str (count)
	
	def process (self, mailFilename):
		mailFile = open (mailFilename, 'r')
		firstLine = 1
		mailLines = []
		fromLine = None
		for line in mailFile.readlines():
			if (firstLine):
				match = XFROM_LINE_REGEX.match (line)
				if (match is not None):
					fromLine = line[match.end():]
				firstLine = 0
			elif (fromLine is None):
				match = DATE_LINE_REGEX.match (line)
				if (match is not None):
					fromLine = 'BeosMailToMbox@localhost %s' % line[match.end():]
				else:
					match = RECIEVED_DATE_REGEX.match (line)
					if (match is not None):
						fromLine = 'BeosMailToMbox@localhost %s' % line[match.end():]
			if (ESCAPE_REGEX.match (line)):
				print "Escaping line: %s" % line
				mailLines.append ('>%s' % line)
			else:
				mailLines.append (line)
		# Finished this mail
		if (fromLine is None):
			raise Exception ("No from line found in email %s" % mailFilename)
		self.outputFile.write ('From %s' % fromLine)
		self.outputFile.write (''.join (mailLines))
		mailFile.close()
		
# Usage: BeosMailToMbox.py dir mboxfile
if (len (sys.argv) != 3):
	print "Usage: BeosMailToMbox.py dir mboxfile"
	sys.exit (0)

try:
	conv = BeMailToMbox (sys.argv[1], sys.argv[2])
	conv.convert()
except Exception, e:
	print "Error during conversion: %s" % str (e)