gistfile1.py

import re
import string

# find the first quote in a string
quotere = re.compile(
    r"""(?P<quote>"[^\"\\]*(?:\\"|[^"])*") # Quote, possibly containing encoded
                                   # quotation mark
        \s*(?P<rest>.*)$           """,
    re.VERBOSE)

def debug(s):
    print s

def imapsplit(imapstring):
    """Takes a string from an IMAP conversation and returns a list containing
    its components"""

    if not isinstance(imapstring, basestring):
        debug("imapsplit() got a non-string input; working around.")
        # Sometimes, imaplib will throw us a tuple if the input
        # contains a literal.  See Python bug
        # #619732 at https://sourceforge.net/tracker/index.php?func=detail&aid=619732&group_id=5470&atid=105470
        # One example is:
        # result[0] = '() "\\\\" Admin'
        # result[1] = ('() "\\\\" {19}', 'Folder\\2')
        #
        # This function will effectively get result[0] or result[1], so
        # if we get the result[1] version, we need to parse apart the tuple
        # and figure out what to do with it.  Each even-numbered
        # part of it should end with the {} number, and each odd-numbered
        # part should be directly a part of the result.  We'll
        # artificially quote it to help out.
        retval = []
        for i in range(len(imapstring)):
            if i % 2:                   # Odd: quote then append.
                arg = imapstring[i]
                # Quote code lifted from imaplib
                arg = arg.replace('\\', '\\\\')
                arg = arg.replace('"', '\\"')
                arg = '"%s"' % arg
                debug("imapsplit() non-string [%d]: Appending %s" %\
                      (i, arg))
                retval.append(arg)
            else:
                # Even -- we have a string that ends with a literal
                # size specifier.  We need to strip off that, then run
                # what remains through the regular imapsplit parser.
                # Recursion to the rescue.
                arg = imapstring[i]
                arg = re.sub('\{\d+\}$', '', arg)
                debug("imapsplit() non-string [%d]: Feeding %s to recursion" %\
                      (i, arg))
                retval.extend(imapsplit(arg))
        debug("imapsplit() non-string: returning %s" % str(retval))
        return retval

    workstr = imapstring.strip()
    retval = []
    while len(workstr):
        # handle parenthized fragments (...()...)
        if workstr[0] == '(':
            rparenc = 1 # count of right parenthesis to match
            rpareni = 1 # position to examine
            while rparenc: # Find the end of the group.
                if workstr[rpareni] == ')':  # end of a group
                    rparenc -= 1
                elif workstr[rpareni] == '(':  # start of a group
                    rparenc += 1
                rpareni += 1  # Move to next character.
            parenlist = workstr[0:rpareni]
            workstr = workstr[rpareni:].lstrip()
            retval.append(parenlist)
        elif workstr[0] == '"':
            # quoted fragments '"...\"..."'
            m = quotere.match(workstr)
            retval.append(m.group('quote'))
            workstr = m.group('rest')
        else:
            splits = string.split(workstr, maxsplit = 1)
            splitslen = len(splits)
            # The unquoted word is splits[0]; the remainder is splits[1]
            if splitslen == 2:
                # There's an unquoted word, and more string follows.
                retval.append(splits[0])
                workstr = splits[1]    # split will have already lstripped it
                continue
            elif splitslen == 1:
                # We got a last unquoted word, but nothing else
                retval.append(splits[0])
                # Nothing remains.  workstr would be ''
                break
            elif splitslen == 0:
                # There was not even an unquoted word.
                break
    return retval

# In:  X-GM-THRID 1329675145120615090 X-GM-MSGID 1329675145120615090 X-GM-LABELS ("Smiley Label :)") UID 236563 FLAGS (\\Seen)
# Out: ['X-GM-THRID', '1329675145120615090', 'X-GM-MSGID', '1329675145120615090', 'X-GM-LABELS', '("Smiley Label :)")', 'UID', '236563', 'FLAGS', '(\\Seen)']
# In:  (\\HasNoChildren) "." "INBOX.Sent"
# Out: ['(\\HasNoChildren)', '"."', '"INBOX.Sent"']