# MH interface -- purely object-oriented (well, almost)
#
# Executive summary:
#
# import mhlib
#
# mh = mhlib.MH()         # use default mailbox directory and profile
# mh = mhlib.MH(mailbox)  # override mailbox location (default from profile)
# mh = mhlib.MH(mailbox, profile) # override mailbox and profile
#
# mh.error(format, ...)   # print error message -- can be overridden
# s = mh.getprofile(key)  # profile entry (None if not set)
# path = mh.getpath()     # mailbox pathname
# name = mh.getcontext()  # name of current folder
# mh.setcontext(name)     # set name of current folder
#
# list = mh.listfolders() # names of top-level folders
# list = mh.listallfolders() # names of all folders, including subfolders
# list = mh.listsubfolders(name) # direct subfolders of given folder
# list = mh.listallsubfolders(name) # all subfolders of given folder
#
# mh.makefolder(name)     # create new folder
# mh.deletefolder(name)   # delete folder -- must have no subfolders
#
# f = mh.openfolder(name) # new open folder object
#
# f.error(format, ...)    # same as mh.error(format, ...)
# path = f.getfullname()  # folder's full pathname
# path = f.getsequencesfilename() # full pathname of folder's sequences file
# path = f.getmessagefilename(n)  # full pathname of message n in folder
#
# list = f.listmessages() # list of messages in folder (as numbers)
# n = f.getcurrent()      # get current message
# f.setcurrent(n)         # set current message
# list = f.parsesequence(seq)     # parse msgs syntax into list of messages
# n = f.getlast()         # get last message (0 if no messagse)
# f.setlast(n)            # set last message (internal use only)
#
# dict = f.getsequences() # dictionary of sequences in folder {name: list}
# f.putsequences(dict)    # write sequences back to folder
#
# f.removemessages(list)  # remove messages in list from folder
# f.refilemessages(list, tofolder) # move messages in list to other folder
# f.movemessage(n, tofolder, ton)  # move one message to a given destination
# f.copymessage(n, tofolder, ton)  # copy one message to a given destination
#
# m = f.openmessage(n)    # new open message object (costs a file descriptor)
# m is a derived class of mimetools.Message(rfc822.Message), with:
# s = m.getheadertext()   # text of message's headers
# s = m.getheadertext(pred) # text of message's headers, filtered by pred
# s = m.getbodytext()     # text of message's body, decoded
# s = m.getbodytext(0)    # text of message's body, not decoded
#
# XXX To do, functionality:
# - annotate messages
# - create, send messages
#
# XXX To do, organization:
# - move IntSet to separate file
# - move most Message functionality to module mimetools


# Customizable defaults

MH_PROFILE = '~/.mh_profile'
PATH = '~/Mail'
MH_SEQUENCES = '.mh_sequences'
FOLDER_PROTECT = 0700


# Imported modules

import os
import sys
from stat import ST_NLINK
import regex
import string
import mimetools
import multifile
import shutil


# Exported constants

Error = 'mhlib.Error'


# Class representing a particular collection of folders.
# Optional constructor arguments are the pathname for the directory
# containing the collection, and the MH profile to use.
# If either is omitted or empty a default is used; the default
# directory is taken from the MH profile if it is specified there.

class MH:

	# Constructor
	def __init__(self, path = None, profile = None):
		if not profile: profile = MH_PROFILE
		self.profile = os.path.expanduser(profile)
		if not path: path = self.getprofile('Path')
		if not path: path = PATH
		if not os.path.isabs(path) and path[0] != '~':
			path = os.path.join('~', path)
		path = os.path.expanduser(path)
		if not os.path.isdir(path): raise Error, 'MH() path not found'
		self.path = path

	# String representation
	def __repr__(self):
		return 'MH(%s, %s)' % (`self.path`, `self.profile`)

	# Routine to print an error.  May be overridden by a derived class
	def error(self, msg, *args):
		sys.stderr.write('MH error: %s\n' % (msg % args))

	# Return a profile entry, None if not found
	def getprofile(self, key):
		return pickline(self.profile, key)

	# Return the path (the name of the collection's directory)
	def getpath(self):
		return self.path

	# Return the name of the current folder
	def getcontext(self):
		context = pickline(os.path.join(self.getpath(), 'context'),
			  'Current-Folder')
		if not context: context = 'inbox'
		return context

	# Set the name of the current folder
	def setcontext(self, context):
		fn = os.path.join(self.getpath(), 'context')
		f = open(fn, "w")
		f.write("Current-Folder: %s\n" % context)
		f.close()

	# Return the names of the top-level folders
	def listfolders(self):
		folders = []
		path = self.getpath()
		for name in os.listdir(path):
			if name in (os.curdir, os.pardir): continue
			fullname = os.path.join(path, name)
			if os.path.isdir(fullname):
				folders.append(name)
		folders.sort()
		return folders

	# Return the names of the subfolders in a given folder
	# (prefixed with the given folder name)
	def listsubfolders(self, name):
		fullname = os.path.join(self.path, name)
		# Get the link count so we can avoid listing folders
		# that have no subfolders.
		st = os.stat(fullname)
		nlinks = st[ST_NLINK]
		if nlinks <= 2:
			return []
		subfolders = []
		subnames = os.listdir(fullname)
		for subname in subnames:
			if subname in (os.curdir, os.pardir): continue
			fullsubname = os.path.join(fullname, subname)
			if os.path.isdir(fullsubname):
				name_subname = os.path.join(name, subname)
				subfolders.append(name_subname)
				# Stop looking for subfolders when
				# we've seen them all
				nlinks = nlinks - 1
				if nlinks <= 2:
					break
		subfolders.sort()
		return subfolders

	# Return the names of all folders, including subfolders, recursively
	def listallfolders(self):
		return self.listallsubfolders('')

	# Return the names of subfolders in a given folder, recursively
	def listallsubfolders(self, name):
		fullname = os.path.join(self.path, name)
		# Get the link count so we can avoid listing folders
		# that have no subfolders.
		st = os.stat(fullname)
		nlinks = st[ST_NLINK]
		if nlinks <= 2:
			return []
		subfolders = []
		subnames = os.listdir(fullname)
		for subname in subnames:
			if subname in (os.curdir, os.pardir): continue
			if subname[0] == ',' or isnumeric(subname): continue
			fullsubname = os.path.join(fullname, subname)
			if os.path.isdir(fullsubname):
				name_subname = os.path.join(name, subname)
				subfolders.append(name_subname)
				if not os.path.islink(fullsubname):
					subsubfolders = self.listallsubfolders(
						  name_subname)
					subfolders = subfolders + subsubfolders
				# Stop looking for subfolders when
				# we've seen them all
				nlinks = nlinks - 1
				if nlinks <= 2:
					break
		subfolders.sort()
		return subfolders

	# Return a new Folder object for the named folder
	def openfolder(self, name):
		return Folder(self, name)

	# Create a new folder.  This raises os.error if the folder
	# cannot be created
	def makefolder(self, name):
		protect = pickline(self.profile, 'Folder-Protect')
		if protect and isnumeric(protect):
			mode = string.atoi(protect, 8)
		else:
			mode = FOLDER_PROTECT
		os.mkdir(os.path.join(self.getpath(), name), mode)

	# Delete a folder.  This removes files in the folder but not
	# subdirectories.  If deleting the folder itself fails it
	# raises os.error
	def deletefolder(self, name):
		fullname = os.path.join(self.getpath(), name)
		for subname in os.listdir(fullname):
			if subname in (os.curdir, os.pardir): continue
			fullsubname = os.path.join(fullname, subname)
			try:
				os.unlink(fullsubname)
			except os.error:
				self.error('%s not deleted, continuing...' %
					  fullsubname)
		os.rmdir(fullname)


# Class representing a particular folder

numericprog = regex.compile('[1-9][0-9]*')
def isnumeric(str):
	return numericprog.match(str) == len(str)

class Folder:

	# Constructor
	def __init__(self, mh, name):
		self.mh = mh
		self.name = name
		if not os.path.isdir(self.getfullname()):
			raise Error, 'no folder %s' % name

	# String representation
	def __repr__(self):
		return 'Folder(%s, %s)' % (`self.mh`, `self.name`)

	# Error message handler
	def error(self, *args):
		apply(self.mh.error, args)

	# Return the full pathname of the folder
	def getfullname(self):
		return os.path.join(self.mh.path, self.name)

	# Return the full pathname of the folder's sequences file
	def getsequencesfilename(self):
		return os.path.join(self.getfullname(), MH_SEQUENCES)

	# Return the full pathname of a message in the folder
	def getmessagefilename(self, n):
		return os.path.join(self.getfullname(), str(n))

	# Return list of direct subfolders
	def listsubfolders(self):
		return self.mh.listsubfolders(self.name)

	# Return list of all subfolders
	def listallsubfolders(self):
		return self.mh.listallsubfolders(self.name)

	# Return the list of messages currently present in the folder.
	# As a side effect, set self.last to the last message (or 0)
	def listmessages(self):
		messages = []
		for name in os.listdir(self.getfullname()):
			if name[0] != "," and \
			   numericprog.match(name) == len(name):
				messages.append(string.atoi(name))
		messages.sort()
		if messages:
			self.last = max(messages)
		else:
			self.last = 0
		return messages

	# Return the set of sequences for the folder
	def getsequences(self):
		sequences = {}
		fullname = self.getsequencesfilename()
		try:
			f = open(fullname, 'r')
		except IOError:
			return sequences
		while 1:
			line = f.readline()
			if not line: break
			fields = string.splitfields(line, ':')
			if len(fields) <> 2:
				self.error('bad sequence in %s: %s' %
					  (fullname, string.strip(line)))
			key = string.strip(fields[0])
			value = IntSet(string.strip(fields[1]), ' ').tolist()
			sequences[key] = value
		return sequences

	# Write the set of sequences back to the folder
	def putsequences(self, sequences):
		fullname = self.getsequencesfilename()
		f = None
		for key in sequences.keys():
			s = IntSet('', ' ')
			s.fromlist(sequences[key])
			if not f: f = open(fullname, 'w')
			f.write('%s: %s\n' % (key, s.tostring()))
		if not f:
			try:
				os.unlink(fullname)
			except os.error:
				pass
		else:
			f.close()

	# Return the current message.  Raise KeyError when there is none
	def getcurrent(self):
		return min(self.getsequences()['cur'])

	# Set the current message
	def setcurrent(self, n):
		updateline(self.getsequencesfilename(), 'cur', str(n), 0)

	# Parse an MH sequence specification into a message list.
	def parsesequence(self, seq):
	    if seq == "all":
		return self.listmessages()
	    try:
		n = string.atoi(seq, 10)
	    except string.atoi_error:
		n = 0
	    if n > 0:
		return [n]
	    if regex.match("^last:", seq) >= 0:
		try:
		    n = string.atoi(seq[5:])
		except string.atoi_error:
		    n = 0
		if n > 0:
		    return self.listmessages()[-n:]
	    if regex.match("^first:", seq) >= 0:
		try:
		    n = string.atoi(seq[6:])
		except string.atoi_error:
		    n = 0
		if n > 0:
		    return self.listmessages()[:n]
	    dict = self.getsequences()
	    if dict.has_key(seq):
		return dict[seq]
	    context = self.mh.getcontext()
	    # Complex syntax -- use pick
	    pipe = os.popen("pick +%s %s 2>/dev/null" % (self.name, seq))
	    data = pipe.read()
	    sts = pipe.close()
	    self.mh.setcontext(context)
	    if sts:
		    raise Error, "unparseable sequence %s" % `seq`
	    items = string.split(data)
	    return map(string.atoi, items)

	# Open a message -- returns a Message object
	def openmessage(self, n):
		return Message(self, n)

	# Remove one or more messages -- may raise os.error
	def removemessages(self, list):
		errors = []
		deleted = []
		for n in list:
			path = self.getmessagefilename(n)
			commapath = self.getmessagefilename(',' + str(n))
			try:
				os.unlink(commapath)
			except os.error:
				pass
			try:
				os.rename(path, commapath)
			except os.error, msg:
				errors.append(msg)
			else:
				deleted.append(n)
		if deleted:
			self.removefromallsequences(deleted)
		if errors:
			if len(errors) == 1:
				raise os.error, errors[0]
			else:
				raise os.error, ('multiple errors:', errors)

	# Refile one or more messages -- may raise os.error.
	# 'tofolder' is an open folder object
	def refilemessages(self, list, tofolder):
		errors = []
		refiled = []
		for n in list:
			ton = tofolder.getlast() + 1
			path = self.getmessagefilename(n)
			topath = tofolder.getmessagefilename(ton)
			try:
				os.rename(path, topath)
			except os.error:
				# Try copying
				try:
					shutil.copy2(path, topath)
					os.unlink(path)
				except (IOError, os.error), msg:
					errors.append(msg)
					try:
						os.unlink(topath)
					except os.error:
						pass
					continue
			tofolder.setlast(ton)
			refiled.append(n)
		if refiled:
			self.removefromallsequences(refiled)
		if errors:
			if len(errors) == 1:
				raise os.error, errors[0]
			else:
				raise os.error, ('multiple errors:', errors)

	# Move one message over a specific destination message,
	# which may or may not already exist.
	def movemessage(self, n, tofolder, ton):
		path = self.getmessagefilename(n)
		# Open it to check that it exists
		f = open(path)
		f.close()
		del f
		topath = tofolder.getmessagefilename(ton)
		backuptopath = tofolder.getmessagefilename(',%d' % ton)
		try:
			os.rename(topath, backuptopath)
		except os.error:
			pass
		try:
			os.rename(path, topath)
		except os.error:
			# Try copying
			ok = 0
			try:
				tofolder.setlast(None)
				shutil.copy2(path, topath)
				ok = 1
			finally:
				if not ok:
					try:
						os.unlink(topath)
					except os.error:
						pass
			os.unlink(path)
		self.removefromallsequences([n])

	# Copy one message over a specific destination message,
	# which may or may not already exist.
	def copymessage(self, n, tofolder, ton):
		path = self.getmessagefilename(n)
		# Open it to check that it exists
		f = open(path)
		f.close()
		del f
		topath = tofolder.getmessagefilename(ton)
		backuptopath = tofolder.getmessagefilename(',%d' % ton)
		try:
			os.rename(topath, backuptopath)
		except os.error:
			pass
		ok = 0
		try:
			tofolder.setlast(None)
			shutil.copy2(path, topath)
			ok = 1
		finally:
			if not ok:
				try:
					os.unlink(topath)
				except os.error:
					pass

	# Remove one or more messages from all sequeuces (including last)
	def removefromallsequences(self, list):
		if hasattr(self, 'last') and self.last in list:
			del self.last
		sequences = self.getsequences()
		changed = 0
		for name, seq in sequences.items():
			for n in list:
				if n in seq:
					seq.remove(n)
					changed = 1
					if not seq:
						del sequences[name]
		if changed:
			self.putsequences(sequences)

	# Return the last message number
	def getlast(self):
		if not hasattr(self, 'last'):
			messages = self.listmessages()
		return self.last

	# Set the last message number
	def setlast(self, last):
		if last is None:
			if hasattr(self, 'last'):
				del self.last
		else:
			self.last = last

class Message(mimetools.Message):

	# Constructor
	def __init__(self, f, n, fp = None):
		self.folder = f
		self.number = n
		if not fp:
			path = f.getmessagefilename(n)
			fp = open(path, 'r')
		mimetools.Message.__init__(self, fp)

	# String representation
	def __repr__(self):
		return 'Message(%s, %s)' % (repr(self.folder), self.number)

	# Return the message's header text as a string.  If an
	# argument is specified, it is used as a filter predicate to
	# decide which headers to return (its argument is the header
	# name converted to lower case).
	def getheadertext(self, pred = None):
		if not pred:
			return string.joinfields(self.headers, '')
		headers = []
		hit = 0
		for line in self.headers:
			if line[0] not in string.whitespace:
				i = string.find(line, ':')
				if i > 0:
					hit = pred(string.lower(line[:i]))
			if hit: headers.append(line)
		return string.joinfields(headers, '')

	# Return the message's body text as string.  This undoes a
	# Content-Transfer-Encoding, but does not interpret other MIME
	# features (e.g. multipart messages).  To suppress to
	# decoding, pass a 0 as argument
	def getbodytext(self, decode = 1):
		self.fp.seek(self.startofbody)
		encoding = self.getencoding()
		if not decode or encoding in ('7bit', '8bit', 'binary'):
			return self.fp.read()
		from StringIO import StringIO
		output = StringIO()
		mimetools.decode(self.fp, output, encoding)
		return output.getvalue()

	# Only for multipart messages: return the message's body as a
	# list of SubMessage objects.  Each submessage object behaves
	# (almost) as a Message object.
	def getbodyparts(self):
		if self.getmaintype() != 'multipart':
			raise Error, \
			      'Content-Type is not multipart/*'
		bdry = self.getparam('boundary')
		if not bdry:
			raise Error, 'multipart/* without boundary param'
		self.fp.seek(self.startofbody)
		mf = multifile.MultiFile(self.fp)
		mf.push(bdry)
		parts = []
		while mf.next():
			n = str(self.number) + '.' + `1 + len(parts)`
			part = SubMessage(self.folder, n, mf)
			parts.append(part)
		mf.pop()
		return parts

	# Return body, either a string or a list of messages
	def getbody(self):
		if self.getmaintype() == 'multipart':
			return self.getbodyparts()
		else:
			return self.getbodytext()


class SubMessage(Message):

	# Constructor
	def __init__(self, f, n, fp):
		Message.__init__(self, f, n, fp)
		if self.getmaintype() == 'multipart':
			self.body = Message.getbodyparts(self)
		else:
			self.body = Message.getbodytext(self)
			# XXX If this is big, should remember file pointers

	# String representation
	def __repr__(self):
		f, n, fp = self.folder, self.number, self.fp
		return 'SubMessage(%s, %s, %s)' % (f, n, fp)

	def getbodytext(self):
		if type(self.body) == type(''):
			return self.body

	def getbodyparts(self):
		if type(self.body) == type([]):
			return self.body

	def getbody(self):
		return self.body


# Class implementing sets of integers.
#
# This is an efficient representation for sets consisting of several
# continuous ranges, e.g. 1-100,200-400,402-1000 is represented
# internally as a list of three pairs: [(1,100), (200,400),
# (402,1000)].  The internal representation is always kept normalized.
#
# The constructor has up to three arguments:
# - the string used to initialize the set (default ''),
# - the separator between ranges (default ',')
# - the separator between begin and end of a range (default '-')
# The separators may be regular expressions and should be different.
#
# The tostring() function yields a string that can be passed to another
# IntSet constructor; __repr__() is a valid IntSet constructor itself.
#
# XXX The default begin/end separator means that negative numbers are
#     not supported very well.
#
# XXX There are currently no operations to remove set elements.

class IntSet:

	def __init__(self, data = None, sep = ',', rng = '-'):
		self.pairs = []
		self.sep = sep
		self.rng = rng
		if data: self.fromstring(data)

	def reset(self):
		self.pairs = []

	def __cmp__(self, other):
		return cmp(self.pairs, other.pairs)

	def __hash__(self):
		return hash(self.pairs)

	def __repr__(self):
		return 'IntSet(%s, %s, %s)' % (`self.tostring()`,
			  `self.sep`, `self.rng`)

	def normalize(self):
		self.pairs.sort()
		i = 1
		while i < len(self.pairs):
			alo, ahi = self.pairs[i-1]
			blo, bhi = self.pairs[i]
			if ahi >= blo-1:
				self.pairs[i-1:i+1] = [
				      (alo, max(ahi, bhi))]
			else:
				i = i+1

	def tostring(self):
		s = ''
		for lo, hi in self.pairs:
			if lo == hi: t = `lo`
			else: t = `lo` + self.rng + `hi`
			if s: s = s + (self.sep + t)
			else: s = t
		return s

	def tolist(self):
		l = []
		for lo, hi in self.pairs:
			m = range(lo, hi+1)
			l = l + m
		return l

	def fromlist(self, list):
		for i in list:
			self.append(i)

	def clone(self):
		new = IntSet()
		new.pairs = self.pairs[:]
		return new

	def min(self):
		return self.pairs[0][0]

	def max(self):
		return self.pairs[-1][-1]

	def contains(self, x):
		for lo, hi in self.pairs:
			if lo <= x <= hi: return 1
		return 0

	def append(self, x):
		for i in range(len(self.pairs)):
			lo, hi = self.pairs[i]
			if x < lo: # Need to insert before
				if x+1 == lo:
					self.pairs[i] = (x, hi)
				else:
					self.pairs.insert(i, (x, x))
				if i > 0 and x-1 == self.pairs[i-1][1]:
					# Merge with previous
					self.pairs[i-1:i+1] = [
					        (self.pairs[i-1][0],
					         self.pairs[i][1])
					      ]
				return
			if x <= hi: # Already in set
				return
		i = len(self.pairs) - 1
		if i >= 0:
			lo, hi = self.pairs[i]
			if x-1 == hi:
				self.pairs[i] = lo, x
				return
		self.pairs.append((x, x))

	def addpair(self, xlo, xhi):
		if xlo > xhi: return
		self.pairs.append((xlo, xhi))
		self.normalize()

	def fromstring(self, data):
		import string, regsub
		new = []
		for part in regsub.split(data, self.sep):
			list = []
			for subp in regsub.split(part, self.rng):
				s = string.strip(subp)
				list.append(string.atoi(s))
			if len(list) == 1:
				new.append((list[0], list[0]))
			elif len(list) == 2 and list[0] <= list[1]:
				new.append((list[0], list[1]))
			else:
				raise ValueError, 'bad data passed to IntSet'
		self.pairs = self.pairs + new
		self.normalize()


# Subroutines to read/write entries in .mh_profile and .mh_sequences

def pickline(file, key, casefold = 1):
	try:
		f = open(file, 'r')
	except IOError:
		return None
	pat = key + ':'
	if casefold:
		prog = regex.compile(pat, regex.casefold)
	else:
		prog = regex.compile(pat)
	while 1:
		line = f.readline()
		if not line: break
		if prog.match(line) >= 0:
			text = line[len(key)+1:]
			while 1:
				line = f.readline()
				if not line or \
				   line[0] not in string.whitespace:
					break
				text = text + line
			return string.strip(text)
	return None

def updateline(file, key, value, casefold = 1):
	try:
		f = open(file, 'r')
		lines = f.readlines()
		f.close()
	except IOError:
		lines = []
	pat = key + ':\(.*\)\n'
	if casefold:
		prog = regex.compile(pat, regex.casefold)
	else:
		prog = regex.compile(pat)
	if value is None:
		newline = None
	else:
		newline = '%s: %s\n' % (key, value)
	for i in range(len(lines)):
		line = lines[i]
		if prog.match(line) == len(line):
			if newline is None:
				del lines[i]
			else:
				lines[i] = newline
			break
	else:
		if newline is not None:
			lines.append(newline)
	tempfile = file + "~"
	f = open(tempfile, 'w')
	for line in lines:
		f.write(line)
	f.close()
	os.rename(tempfile, file)


# Test program

def test():
	global mh, f
	os.system('rm -rf $HOME/Mail/@test')
	mh = MH()
	def do(s): print s; print eval(s)
	do('mh.listfolders()')
	do('mh.listallfolders()')
	testfolders = ['@test', '@test/test1', '@test/test2',
		       '@test/test1/test11', '@test/test1/test12',
		       '@test/test1/test11/test111']
	for t in testfolders: do('mh.makefolder(%s)' % `t`)
	do('mh.listsubfolders(\'@test\')')
	do('mh.listallsubfolders(\'@test\')')
	f = mh.openfolder('@test')
	do('f.listsubfolders()')
	do('f.listallsubfolders()')
	do('f.getsequences()')
	seqs = f.getsequences()
	seqs['foo'] = IntSet('1-10 12-20', ' ').tolist()
	print seqs
	f.putsequences(seqs)
	do('f.getsequences()')
	testfolders.reverse()
	for t in testfolders: do('mh.deletefolder(%s)' % `t`)
	do('mh.getcontext()')
	context = mh.getcontext()
	f = mh.openfolder(context)
	do('f.listmessages()')
	do('f.getcurrent()')


if __name__ == '__main__':
	test()