root/fedd/federation/authorizer.py

#/usr/local/bin/python

from tempfile import mkstemp
from subprocess import call
from threading import Lock

from string import join, hexdigits

from deter import fedid
from remote_service import service_caller
from service_error import service_error
from util import abac_pem_type, abac_split_cert
from proof import proof


import ABAC
import Creddy
import pickle

import sys
import os, os.path
import re

class authorizer_base:
    """
    Classes based on this one keep track of authorization attributes for the
    various modules running.  This base class holds some utility functions that
    they all potentially use.
    """

    # general error exception for badly formed names.  
    class bad_name(RuntimeError): pass
    # difficulty creating an attribute
    class attribute_error(RuntimeError): pass

    @staticmethod
    def auth_name(name):
        """
        Helper to convert a non-unicode local name to a unicode string.  Mixed
        representations can needlessly confuse the authorizer.
        """
        if isinstance(name, basestring):
            if not isinstance(name, unicode): return unicode(name)
            else: return name
        else: return name

    @staticmethod
    def valid_name(name):
        """
        Ensure that the given name is valid.  A valid name can either be a
        triple of strings and fedids representing one of our generalized Emulab
        names or a single fedid.  Compound names can include wildcards (None)
        and must anchor to a fedid at their highest level (unless they're all
        None)

        This either returns True or throws an exception.  More an assertion
        than a function.
        """
        if isinstance(name, tuple) and len(name) == 3:
            for n in name:
                if n: 
                    if not (isinstance(n, basestring) or isinstance(n, fedid)):
                        raise authorizer_base.bad_name(
                                "names must be either a triple or a fedid")
            for n in name: 
                if n:
                    if isinstance(n, fedid):
                        return True
                    else:
                        raise authorizer_base.bad_name(
                                "Compound names must be " + \
                                "rooted in fedids: %s" % str(name))

            return True
        elif isinstance(name, fedid):
            return True
        else:
            raise authorizer_base.bad_name(
                    "Names must be a triple or a fedid (%s)" % name)


class authorizer(authorizer_base):
    """
    This class keeps track of authorization attributes for the various modules
    running.  When it gets smarter it will be the basis for a real
    attribute-based authentication system.
    """
    def __init__(self, def_attr="testbed"):
        self.attrs = { }
        self.globals=set()

    def set_attribute(self, name, attr):
        """
        Attach attr to name.  Multiple attrs can be attached.
        """
        self.valid_name(name)
        if isinstance(name, tuple):
            aname = tuple([ self.auth_name(n) for n in name])
        else:
            aname = self.auth_name(name)

        if not self.attrs.has_key(aname):
            self.attrs[aname] = set()
        self.attrs[aname].add(attr)

    def unset_attribute(self, name, attr):
        """
        Remove an attribute from name
        """
        self.valid_name(name)
        if isinstance(name, tuple):
            aname = tuple([ self.auth_name(n) for n in name])
        else:
            aname = self.auth_name(name)

        attrs = self.attrs.get(aname, None)
        if attrs: attrs.discard(attr)

    def check_attribute(self, name, attr, with_proof=False):
        """
        Return True if name has attr (or if attr is global).  Tuple names match
        any tuple name that matches all names present and has None entries in
        other fileds.  For tuple names True implies that there is a matching
        tuple name with the attribute.
        """
        def tup(tup, i, p):
            mask = 1 << i
            if p & mask : return authorizer.auth_name(tup[i])
            else: return None

        self.valid_name(name)
        if attr in self.globals:
            if with_proof: return True, proof("me", name, attr)
            else: return True

        if isinstance(name, tuple):
            for p in range(0,8):
                lookup = ( tup(name, 0, p), tup(name,1, p), tup(name,2,p))
                if self.attrs.has_key(lookup):
                    if attr in self.attrs[lookup]:
                        if with_proof: return True, proof("me", name, attr)
                        else: return True
                # Drop through
                if with_proof: return False, proof("me", name, attr)
                else: return False
        else:
            if with_proof:
                return attr in self.attrs.get(self.auth_name(name), set()), \
                        proof("me", name, attr)
            else:
                return attr in self.attrs.get(self.auth_name(name), set())

    def set_global_attribute(self, attr):
        """
        Set a global attribute.  All names, even those otherwise unknown to the
        authorizer have this attribute.
        """
        self.globals.add(attr)

    def unset_global_attribute(self, attr):
        """
        Remove a global attribute
        """

        self.globals.discard(attr)

    def import_credentials(self, file_list=None, data_list=None):
        return False

    def __str__(self):
        rv = ""
        rv += "attrs %s\n" % self.attrs
        rv += "globals %s" % self.globals
        return rv

    def clone(self):
        rv = authorizer()
        rv.attrs = self.attrs.copy()
        rv.globals = self.globals.copy()
        return rv

    def save(self, fn=None):
        if fn:
            f = open(fn, "w")
            pickle.dump(self, f)
            f.close()

    def load(self, fn=None):
        if fn:
            f = open(fn, "r")
            a = pickle.load(f)
            f.close()
            self.attrs = a.attrs
            self.globals = a.globals
        

class abac_authorizer(authorizer_base):
    """
    Use the ABAC authorization system to make attribute decisions.
    """

    clean_attr_re = re.compile('[^A-Za-z0-9_]+')
    clean_prefix_attr_re = re.compile('^_+')
    cred_file_re = re.compile('.*\.der$')
    bad_name = authorizer_base.bad_name
    attribute_error = authorizer_base.attribute_error
    class no_file_error(RuntimeError): pass
    class bad_cert_error(RuntimeError): pass

    def __init__(self, certs=None, me=None, key=None, load=None, save=None):
        self.creddy_id = None
        self.globals = set()
        self.lock = Lock()
        self.me = me
        self.save_dir = load or save
        self.local_files = False
        if self.save_dir:
            self.save_dir = os.path.abspath(self.save_dir)
        # If the me parameter is a combination certificate, split it into the
        # abac_authorizer save directory (if any) for use with creddy.
        if self.me is not None and abac_pem_type(self.me) == 'both':
            if self.save_dir:
                keyfile="%s/key.pem" % self.save_dir 
                certfile = "%s/cert.pem" % self.save_dir

                # Clear a spot for the new key and cert files.
                for fn in (keyfile, certfile):
                    if os.access(fn, os.F_OK):
                        os.unlink(fn)

                self.key, self.me = abac_split_cert(self.me, keyfile, certfile)
                self.local_files = True
                self.init_libcreddy_id()
            else:
                raise abac_authorizer.bad_cert_error("Combination " + \
                        "certificate and nowhere to split it");
        else:
            self.key = key
        self.context = ABAC.Context()
        if me:
            self.fedid = fedid(file=self.me)
            rv = self.context.load_id_file(self.me)
            if rv != 0:
                raise abac_authorizer.bad_name(
                        'Cannot load identity from %s' % me)
        else:
            self.fedid = None

        if isinstance(certs, basestring):
            certs = [ certs ] 

        for dir in certs or []:
            self.context.load_directory(dir)

        if load:
            self.load(load)

    def init_libcreddy_id(self):
        self.creddy_id = None
        try:
            self.creddy_id = Creddy.ID(self.me)
        except:
            raise abac_authorizer.bad_cert_error('Cannot load cert %s' \
                    % self.me)

        try:
            self.creddy_id.load_privkey(self.key)
        except:
            self.creddy_id = None
            raise abac_authorized_bad_cert_error('Cannot load key %s' \
                    % self.key)



    # Modify the pickling operations so that the context and lock are not
    # pickled

    def __getstate__(self):
        d = self.__dict__.copy()
        del d['lock']
        del d['context']
        del d['creddy_id']
        return d

    def __setstate__(self, d):
        # Import everything from the pickle dict (except what we excluded in
        # __getstate__)
        self.__dict__.update(d)
        # Initialize the unpicklables
        self.context = ABAC.Context()
        self.lock = Lock()

    @staticmethod
    def clean_attr(attr):
        '''
        Convert anything not in A-Z_ into an _ and make sure that the string
        begins with a non-_.  Convert to s string to keep swig happy.
        '''
        a = abac_authorizer.clean_attr_re.sub('_', attr)
        return str(abac_authorizer.clean_prefix_attr_re.sub('', a))


    def import_credentials(self, file_list=None, data_list=None):
        '''
        Import a list of credentials.  Be careful about dependencies and only
        use one logic loop.  if files are given, append them to the list of
        chunks and then try to read each chunk as an ID and then the failed
        ones as attributes.
        '''
        if file_list:
            if data_list is None: data_list = []
            for fn in file_list:
                # Try to parse file as a Creddy ID, so we can import PEM files
                try:
                    cid = Creddy.ID(fn)
                    data_list.append(cid.cert_chunk())
                    continue
                except:
                    # Not an ID
                    pass
                try:
                    f = open(fn, 'r')
                    data_list.append(f.read())
                    f.close()
                except EnvironmentError, e:
                    # Ignore unreadable files
                    pass
        # Now one way or another data_list is ready
        if data_list:
            rv = []
            attr_list = [ ]
            for d in data_list:
                if self.context.load_id_chunk(d) != ABAC.ABAC_CERT_SUCCESS:
                    attr_list.append(d)
                else:
                    rv.append(True)

            for d in attr_list:
                v = self.context.load_attribute_chunk(d) 
                rv.append(v == ABAC.ABAC_CERT_SUCCESS)
            return all(rv)
        else:
            return False

    def import_credential(self, file=None, data=None):
        if data:
            if self.context.load_id_chunk(data) != ABAC.ABAC_CERT_SUCCESS:
                return self.context.load_attribute_chunk(data) == \
                        ABAC.ABAC_CERT_SUCCESS
            else:
                return True
        elif file:
            if self.context.load_id_file(file) != ABAC.ABAC_CERT_SUCCESS:
                return self.context.load_attribute_file(file) == \
                        ABAC.ABAC_CERT_SUCCESS
            else:
                return True
        else:
            return False

    def set_attribute(self, name=None, attr=None, cert=None):
        if name and attr:
            if isinstance(name, tuple):
                raise abac_authorizer.bad_name(
                        "ABAC doesn't understand three-names")
            # Convert non-string attributes to strings
            if not isinstance(attr, basestring):
                attr = "%s" % attr

            if self.creddy_id:
                # Create a credential and insert it into context
                # This will simplify when we have libcreddy
                try:
                    attrcert = Creddy.Attribute(self.creddy_id, 
                            self.clean_attr(attr), 3600 * 24 * 365 * 10)
                    attrcert.principal("%s" % name)
                    attrcert.bake()
                except:
                    raise abac_authorizer.bad_cert_error(
                            "Cannot create attribute cert")
                self.lock.acquire()
                # load it to context and remove the file
                rv = self.context.load_attribute_chunk(attrcert.cert_chunk())
                self.lock.release()
                return rv
            else:
                raise abac_authorizer.attribute_error(
                        "Identity and key not specified on creation")
        elif cert:
            # Insert this credential into the context
            self.lock.acquire()
            rv = self.context.load_attribute_chunk(cert)
            self.lock.release()
            return rv
        else:
            raise abac_authorizer.attribute_error(
                    "Neither name/attr nor cert is set")

    def unset_attribute(self, name, attr):
        if isinstance(name, tuple):
            raise abac_authorizer.bad_name(
                    "ABAC doesn't understand three-names")
        # Convert non-string attributes to strings
        if not isinstance(attr, basestring):
            attr = "%s" % attr
        cattr = self.clean_attr(attr)
        self.lock.acquire()
        ctxt = ABAC.Context()
        ids = set()
        for c in self.context.credentials():
            h = c.head()
            t = c.tail()
            if h.is_role() and t.is_principal():
                if t.principal() == '%s' % name and \
                        h.principal() == '%s' % self.fedid and \
                        h.role_name() == cattr:
                    continue

            id = c.issuer_cert()
            if id not in ids:
                ctxt.load_id_chunk(id)
                ids.add(id)
            ctxt.load_attribute_chunk(c.attribute_cert())
        self.context = ctxt
        self.lock.release()

    @staticmethod
    def starts_with_fedid(attr):
        """
        Return true if the first 40 characters of the string are hex digits
        followed by a dot.  False otherwise.  Used in check_attribute.
        """
        if attr.find('.') == 40:
            return all([ x in hexdigits for x in attr[0:40]])
        else:
            return False


    def check_attribute(self, name, attr, with_proof=False):
        if isinstance(name, tuple):
            raise abac_authorizer.bad_name(
                    "ABAC doesn't understand three-names")
        else:
            # Convert non-string attributes to strings
            if not isinstance(attr, basestring):
                attr = "%s" % attr
            # Attributes that start with a fedid only have the part of the
            # attribute after the dot cleaned.  Others are completely cleaned
            # and have the owner fedid attached.
            if self.starts_with_fedid(attr):
                r, a = attr.split('.',1)
                a = "%s.%s" % ( r, self.clean_attr(a))
            else: 
                a = "%s.%s" % (self.fedid, self.clean_attr(attr))

            a = str(a)
            n = str("%s" % name)

            self.lock.acquire()
            # Sigh. Unicode vs swig and swig seems to lose.  Make sure
            # everything we pass into ABAC is a str not a unicode.
            rv, p = self.context.query(a, n)
            # XXX delete soon
            if not rv and attr in self.globals: 
                rv = True
                p = None
            self.lock.release()
            if with_proof: return rv, proof(self.fedid, name, a, p)
            else: return rv

    def set_global_attribute(self, attr):
        """
        Set a global attribute.  All names, even those otherwise unknown to the
        authorizer have this attribute.
        """
        self.lock.acquire()
        self.globals.add(self.clean_attr(attr))
        self.lock.release()

    def unset_global_attribute(self, attr):
        """
        Remove a global attribute
        """

        self.lock.acquire()
        self.globals.discard(self.clean_attr(attr))
        self.lock.release()

    def clone(self):
        self.lock.acquire()
        rv = abac_authorizer(me=self.me, key=self.key)
        rv.globals = self.globals.copy()
        rv.context = ABAC.Context(self.context)
        self.lock.release()
        return rv

    def copy_file(self, src, dest, mode=0600):
        '''
        Copy src to dest with file mode mode.  May raise exceptions on file ops
        '''
        same_file = False
        try:
            ss = os.stat(src)
            ds = os.stat(dest)

            if  ds.st_ino == ss.st_ino: same_file = True
        except EnvironmentError:
            pass

        if not same_file:
            d = open(dest, 'w')
            s = open(src, 'r')
            d.write(s.read())
            s.close()
            d.close()
        os.chmod(dest, mode)



    def save(self, dir=None):
        self.lock.acquire()
        if dir:
            self.save_dir = os.path.abspath(dir)
        else:
            dir = self.save_dir
        if dir is None:
            self.lock.release()
            raise abac_authorizer.no_file_error("No load directory specified")
        try:
            if not os.access(dir, os.F_OK):
                os.mkdir(dir)

            # if self.key and self.me were split, copy the split files into the
            # new directory.
            if self.local_files:
                self.copy_file(self.key, '%s/key.pem' % dir)
                self.copy_file(self.me, '%s/cert.pem' % dir)
                # Point the key and me members to the new locations for
                # pickling
                hold_key = self.key
                hold_me = self.me
                self.key = '%s/key.pem' % dir
                self.me = '%s/cert.pem' % dir

            f = open("%s/state" % dir, "w")
            pickle.dump(self, f)
            f.close()

            if self.local_files:
                self.key = hold_key
                self.me = hold_me

            if not os.access("%s/certs" %dir, os.F_OK):
                os.mkdir("%s/certs" % dir)

            # Clear the certs subdir
            for fn in [ f for f in os.listdir("%s/certs" % dir) \
                    if abac_authorizer.cred_file_re.match(f)]:
                os.unlink('%s/certs/%s' % (dir, fn))

            # Save the context
            ii = 0
            ai = 0
            seenid = set()
            seenattr = set()
            for c in self.context.credentials():
                id = c.issuer_cert()
                attr = c.attribute_cert()
                # NB: file naming conventions matter here.  The trailing_ID and
                # _attr are required by ABAC.COntext.load_directory()
                if id and id not in seenid:
                    f = open("%s/certs/ID_%05d_ID.der" % (dir, ii), "w")
                    f.write(id)
                    f.close()
                    ii += 1
                    seenid.add(id)
                if attr and attr not in seenattr:
                    f = open("%s/certs/attr_%05d_attr.der" % (dir, ai), "w")
                    f.write(attr)
                    f.close()
                    ai += 1
                    seenattr.add(attr)
        except EnvironmentError, e:
            self.lock.release()
            raise e
        except pickle.PickleError, e:
            self.lock.release()
            raise e
        self.lock.release()

    def load(self, dir=None):
        self.lock.acquire()
        if dir:
            self.save_dir = dir
        else:
            dir = self.save_dir
        if dir is None:
            self.lock.release()
            raise abac_authorizer.no_file_error("No load directory specified")
        try:
            if os.access("%s/state" % dir, os.R_OK):
                f = open("%s/state" % dir, "r")
                st = pickle.load(f)
                f.close()
                # Copy the useful attributes from the pickled state
                for a in ('globals', 'key', 'me', 'cert', 'fedid'):
                    setattr(self, a, getattr(st, a, None))

            # Initialize the new context with the new identity
            self.context = ABAC.Context()
            if self.me:
                self.context.load_id_file(self.me)
                if self.key:
                    self.init_libcreddy_id()
            self.context.load_directory("%s/certs" % dir)
            self.save_dir = dir
        except EnvironmentError, e:
            self.lock.release()
            raise e
        except pickle.PickleError, e:
            self.lock.release()
            raise e
        self.lock.release()

    @staticmethod
    def encode_credential(c):
        return '%s <- %s' % (c.head().string(), c.tail().string())

    def get_creds_for_principal(self, fid):
        look_for = set(["%s" % fid])
        found_attrs = set()
        next_look = set()
        found = set([])

        self.lock.acquire()
        while look_for:
            for c in self.context.credentials():
                tail = c.tail()
                # XXX: This needs to be more aggressive for linked stuff
                if tail.string() in look_for and c not in found:
                    found.add(c)
                    next_look.add(c.head().string())

            look_for = next_look
            next_look = set()
        self.lock.release()
        
        return found

    def __str__(self):

        self.lock.acquire()
        rv = "%s" % self.fedid
        add = join([abac_authorizer.encode_credential(c)
            for c in self.context.credentials()], '\n');
        if add: rv += "\n%s" % add
        self.lock.release()
        return rv