#!/usr/bin/env python
#
# Copyright (c) 2006-2010 Benjamin Schweizer. All rights reserved.
#
# Abstract
# ~~~~~~~~
# This module provides methods for reading and writing LDIF files as
# specified in RFC 2849.
#
# Features
# ~~~~~~~~
# - read/write ldif files
# - folding/multi-line support
# - base64-latin1 support
#
# Authors
# ~~~~~~~
# Benjamin Schweizer, http://benjamin-schweizer.de/contact
#
# Changes
# ~~~~~~~
# 2010-03-19, benjamin: fixed multiline base64 output
# 2006-10-18, benjamin: fixed comments handling and empty lines
# 2006-04-03, benjamin: made module encapsulation
# 2006-03-03, benjamin: initial release
#
# Todo
# ~~~~
# - support for base64 chars
# - input validation
# - check zero length attributes
# - check multi line line length
# - testing
#

import base64

def parser(filehandle, encoding='utf-8'):
	"""a parser for ldif files implemented as generator"""
	record = {}
	key = False
	value = False
	is_base64 = False

	lineno = 0
	line = filehandle.readline()
	while line:
		line = line[:-1]
		lineno = lineno + 1

		if line == '':
			# end of record
			# append last entry
			if key:
				if is_base64:
					value = base64.decodestring(value).decode(encoding)
				if key in record.keys():
					record[key].append(value)
				else:
					record[key] = [value]
				key = False
				value = False
				is_base64 = False

			# yield if we have data
			if record:
				yield record
				record = {}
		elif line[0] == '#':
			# pass comments
			pass
		elif line[0] == ' ':
			# multi line
			value = value + line[1:]
		else:
			# append last entry
			if key:
				if is_base64:
					value = base64.decodestring(value).decode(encoding)
				if key in record.keys():
					record[key].append(value)
				else:
					record[key] = [value]
				key = False
				value = False
				is_base64 = False

			# process new pair
			if line.count(':: '):
				# base64
				key = line.split(':: ')[0]
				value = ':: '.join(line.split(':: ')[1:])
				is_base64 = True
			else:
				# ascii7
				key = line.split(': ')[0]
				value = ': '.join(line.split(': ')[1:])
				is_base64 = False

		line = filehandle.readline()
	# append last entry
	if key:
		if is_base64:
			value = base64.decodestring(value).decode(encoding)
		if key in record.keys():
			record[key].append(value)
		else:
			record[key] = [value]
		key = False
		value = False
		is_base64 = False
	# yield
	yield record
	yield False

class writer:
	"""a writer for ldif files"""
	def __init__(self, filehandle, encoding='utf-8'):
		self.filehandle = filehandle
		self.encoding = encoding

		self.filehandle.write("#\n# Generated with tinyparsers.ldif\n#\n\n")

	def __del__(self):
		self.filehandle.write("\n# eof.\n")

	def put(self, record):
		def flush(key, value):
			must_encode = False
			for c in value:
				if not ord(c) in range(32, 128):
					must_encode = True
			if must_encode:
				#str = "%s:: %s" % (key, base64.encodestring(value.encode(self.encoding))[:-1])
                str = ''.join(base64.encodestring(value.encode(self.encoding)).split('\n'))
                str = "%s:: %s" % (key, str)
			else:
				str = "%s: %s" % (key, value.encode('ascii'))

			self.filehandle.write(str[:78] + "\n")
			str = str[78:]
			while str:
				self.filehandle.write(' ' + str[:76] + "\n")
				str = str[76:]

		order = record.keys()
		order.remove('dn')
		order = ['dn'] + order
		if 'changetype' in order:
			order.remove('changetype')
			order = ['changetype'] + order

		for key in order:
			for element in record[key]:
				# supress empty lines
				if element:
					flush(key, element)
		self.filehandle.write("\n")

if __name__ == '__main__':
	import sys
	if not len(sys.argv) == 3:
		print """Copyright (c) 2006-2010 Benjamin Schweizer. All rights reserved.

usage:

	%s INPUT.ldif OUTPUT.ldif
""" % sys.argv[0]
		sys.exit(1)

	fh = open(sys.argv[1], 'r')
	fh2 = open(sys.argv[2], 'wb')
	ldifparser = parser(fh, encoding="latin1")
	ldifwriter = writer(fh2, encoding="utf-8")
	record = ldifparser.next()
	while record:
		print '.',
		ldifwriter.put(record)
		record = ldifparser.next()
	ldifparser = False
	fh.close()
	ldifwriter = False
	fh2.close()

# eof.