source: fedd/fedd_proj.py @ bb3769a

#!/usr/local/bin/python

import os,sys

from BaseHTTPServer import BaseHTTPRequestHandler
from ZSI import *
from M2Crypto import SSL
from M2Crypto.m2xmlrpclib import SSL_Transport
from M2Crypto.SSL.SSLServer import SSLServer
import M2Crypto.httpslib
import xmlrpclib

import re
import random
import string
import subprocess
import tempfile
import copy

from fedd_services import *
from fedd_util import *

if False:
    _RequestAccessFault_typecode = Struct(pname=("http://www.isi.edu/faber/fedd.wsdl", 'RequestAccessFault'), ofwhat=[ns0.faultType_Def( pname=("http://www.isi.edu/faber/fedd.wsdl", "RequestAccessFaultBody"), aname="_RequestAccessFaultBody", minoccurs=0, maxoccurs=1, nillable=True, encoded=None, typed=True)], pyclass=None)
    class RequestAccessFault:
        typecode = _RequestAccessFault_typecode
        __metaclass__ = pyclass_type
        def __init__(self, code=0, str="str"):
            self._RequestAccessFaultBody = None
    RequestAccessFault.typecode.pyclass=RequestAccessFault

class fedd_proj:
    """
    The implementation of access control based on mapping users to projects.

    Users can be mapped to existing projects or have projects created
    dynamically.  This implements both direct requests and proxies.
    """
    # Attributes that can be parsed from the configuration file
    bool_attrs = ("dynamic_projects", "project_priority")
    emulab_attrs = ("boss", "ops", "domain", "fileserver", "eventserver")
    id_attrs = ("testbed", "cert_file", "trusted_certs", "proxy",
            "proxy_trusted_certs", "cert_pwd")

    # Used by the SOAP caller
    soap_namespace = 'http://www.isi.edu/faber/fedd.wsdl'
    soap_methods = { 'RequestAccess': 'soap_RequestAccess' }
    xmlrpc_methods = { 'RequestAccess': 'xmlrpc_RequestAccess' }

    class access_project:
        """
        A project description used to grant access to this testbed.

        The description includes a name and a list of node types to which the
        project will be granted access.
        """
        def __init__(self, name, nt):
            self.name = name
            self.node_types = list(nt)

        def __repr__(self):
            if len(self.node_types) > 0:
                return "access_proj('%s', ['%s'])" % \
                        (self.name, str("','").join(self.node_types))
            else:
                return "access_proj('%s', [])" % self.name

    # These are used to make the error reporting independent of the transport.
    # The XMLRPC and SOAP dispatchers will convert them to transport-specific
    # errors
    class server_error(RuntimeError): pass
    class client_error(RuntimeError): pass
    # Error parsing config
    class parse_error(RuntimeError): pass


    def __init__(self, config=None):
        """
        Initializer.  Parses a configuration if one is given.
        """

        # Create instance attributes from the static lists
        for a in fedd_proj.bool_attrs:
            setattr(self, a, False)

        for a in fedd_proj.emulab_attrs + fedd_proj.id_attrs:
            setattr(self, a, None)

        # Other attributes
        self.attrs = {}
        self.access = {}
        self.fedid_category = {}
        self.fedid_default = "user"
        self.restricted = []
        self.wap = '/usr/testbed/sbin/wap'
        self.newproj = '/usr/testbed/sbin/newproj'
        self.mkproj = '/usr/testbed/sbin/mkproj'
        self.grantnodetype = '/usr/testbed/sbin/grantnodetype'

        # Read the configuration
        if config != None: 
            self.read_config(config)

    def dump_state(self):
        """
        Dump the state read from a configuration file.  Mostly for debugging.
        """
        for a in fedd_proj.bool_attrs:
            print "%s: %s" % (a, getattr(self, a ))
        for a in fedd_proj.emulab_attrs + fedd_proj.id_attrs:
            print "%s: %s" % (a, getattr(self, a))
        for k, v in self.attrs.iteritems():
            print "%s %s" % (k, v)
        print "Access DB:"
        for k, v in self.access.iteritems():
            print "%s %s" % (k, v)
        print "Trust DB:"
        for k, v in self.fedid_category.iteritems():
            print "%s %s" % (k, v)
        print "Restricted: %s" % str(',').join(sorted(self.restricted))

    def get_id(self, id):
        """
        Utility to get an object from the polymorphic IDType.
        
        Only fedids and usernames are currently understood.  If neither is
        present None is returned.  If both are present (which is a bug) the
        fedid is returned.
        """
        if id == None: 
            return None
        elif getattr(id, "get_element_fedid", None) != None:
            return fedid(id.get_element_fedid())
        elif getattr(id, "get_element_username", None) != None:
            return id.get_element_username()
        else:
            return None

    def get_users(self, obj):
        """
        Return a list of the IDs of the users in dict
        """
        if obj.has_key('user'):
            return [ unpack_id(u['userID']) \
                    for u in obj['user'] if u.has_key('userID') ]
        else:
            return None

    def random_string(self, s, n=3):
        """Append n random ASCII characters to s and return the string"""
        rv = s
        for i in range(0,n):
            rv += random.choice(string.ascii_letters)
        return rv

    def write_attr_xml(self, file, root, lines):
        """
        Write an emulab config file for a dynamic project.

        Format is <root><attribute name=lines[0]>lines[1]</attribute></root>
        """
        # Convert a pair to an attribute line
        out_attr = lambda a,v : \
                '<attribute name="%s"><value>%s</value></attribute>' % (a, v)

        f = os.fdopen(file, "w")
        f.write("<%s>\n" % root)
        f.write("\n".join([out_attr(*l) for l in lines]))
        f.write("</%s>\n" % root)
        f.close()


    def dynamic_project(self, found, ssh):
        """Create a dynamic project with ssh access"""
        user_fields = [
                ("name", "Federation User %s" % found[1]),
                ("email", "%s-fed@isi.deterlab.net" % found[1]),
                ("password", self.random_string("", 8)),
                ("login", found[1]),
                ("address", "4676 Admiralty"),
                ("city", "Marina del Rey"),
                ("state", "CA"),
                ("zip", "90292"),
                ("country", "USA"),
                ("phone", "310-448-9190"),
                ("title", "None"),
                ("affiliation", "USC/ISI")
        ]

        proj_fields = [
                ("name", found[0].name),
                ("short description", "dynamic federated project"),
                ("URL", "http://www.isi.edu/~faber"),
                ("funders", "USC/USU"),
                ("long description", "Federation access control"),
                ("public", "1"),
                ("num_pcs", "100"),
                ("linkedtous", "1")
        ]

        # tempfiles for the parameter files
        uf, userfile = tempfile.mkstemp(prefix="usr", suffix=".xml",
                dir="/tmp")
        pf, projfile = tempfile.mkstemp(prefix="proj", suffix=".xml",
                dir="/tmp")

        # A few more dynamic fields
        for s in ssh:
            user_fields.append(("pubkey", s))
        proj_fields.append(("newuser_xml", userfile))

        # Write out the files
        self.write_attr_xml(uf, "user", user_fields)
        self.write_attr_xml(pf, "project", proj_fields)

        # Generate the commands (only grantnodetype's are dynamic)
        cmds = [
                (self.wap, self.newproj, projfile),
                (self.wap, self.mkproj, found[0].name)
                ]
        for nt in found[0].node_types:
            cmds.append((self.wap, self.grantnodetype, '-p', found[0].name, nt))

        # Create the projects
        rc = 0
        for cmd in cmds:
            if self.dynamic_projects:
                try:
                    rc = subprocess.call(cmd)
                except OSerror, e:
                    raise fedd_proj.server_error(
                            "Dynamic project subprocess creation error "+ \
                                    "[%s] (%s)" %  (cmd[1], e.strerror))
            else:
                print >>sys.stdout, str(" ").join(cmd)

            if rc != 0: 
                raise fedd_proj.server_error(
                        "Dynamic project subprocess error " +\
                                "[%s] (%d)" % (cmd[1], rc))
        # Clean up tempfiles
        os.unlink(userfile)
        os.unlink(projfile)

    
    def strip_unicode(self, obj):
        """Loosly de-unicode an object"""
        if isinstance(obj, dict):
            for k in obj.keys():
                obj[k] = self.strip_unicode(obj[k])
            return obj
        elif isinstance(obj, basestring):
            return str(obj)
        elif getattr(obj, "__iter__", None):
            return [ self.strip_unicode(x) for x in obj]
        else:
            return obj

    def proxy_xmlrpc_request(self, dt, req):
        """Send an XMLRPC proxy request.  Called if the SOAP RPC fails"""
        tc = self.proxy_trusted_certs or self.trusted_certs

        # No retry loop here.  Proxy servers must correctly authenticate
        # themselves without help
        try:
            ctx = fedd_ssl_context(self.cert_file, tc, password=self.cert_pwd)
        except SSL.SSLError:
            raise fedd_proj.server_error("Server certificates misconfigured")

        # Of all the dumbass things.  The XMLRPC library in use here won't
        # properly encode unicode strings, so we make a copy of req with the
        # unicode objects converted.  We also convert the destination testbed
        # to a basic string if it isn't one already.
        if isinstance(dt, str): url = dt
        else: url = str(dt)

        r = copy.deepcopy(req)
        self.strip_unicode(r)
        
        transport = SSL_Transport(ctx)
        port = xmlrpclib.ServerProxy(url, transport=transport)

        # Reconstruct the full request message
        try:
            resp = port.RequestAccess(
                    { "RequestAccessRequestBody": r})
            resp, method = xmlrpclib.loads(resp)
        except xmlrpclib.Error, e:
            raise fedd_proj.server_error("Remote XMLRPC Fault: %s" % e)
        
        if resp[0].has_key('RequestAccessResponseBody'):
            return resp[0]['RequestAccessResponseBody']
        else:
            raise fedd_proj.server_error("Bad proxy response")

    def proxy_request(self, dt, req):
        """
        Send req on to the real destination in dt and return the response

        Req is just the requestType object.  This function re-wraps it.  It
        also rethrows any faults.
        """
        tc = self.proxy_trusted_certs or self.trusted_certs

        # No retry loop here.  Proxy servers must correctly authenticate
        # themselves without help
        try:
            ctx = fedd_ssl_context(self.cert_file, tc, password=self.cert_pwd)
        except SSL.SSLError:
            raise fedd_proj.server_error("Server certificates misconfigured")

        loc = feddServiceLocator();
        port = loc.getfeddPortType(dt,
                transport=M2Crypto.httpslib.HTTPSConnection, 
                transdict={ 'ssl_context' : ctx })

        # Reconstruct the full request message
        msg = RequestAccessRequestMessage()
        msg.set_element_RequestAccessRequestBody(
                pack_soap(msg, "RequestAccessRequestBody", req))
        try:
            resp = port.RequestAccess(msg)
        except ZSI.ParseException, e:
            raise fedd_proj.server_error("Bad format message (XMLRPC??): %s" %
                    str(e))
        except ZSI.FaultException, e:
            raise e.fault

        r = unpack_soap(resp)

        if r.has_key('RequestAccessResponseBody'):
            return r['RequestAccessResponseBody']
        else:
            raise fedd_proj.server_error("Bad proxy response")

    def permute_wildcards(self, a, p):
        """Return a copy of a with various fields wildcarded.

        The bits of p control the wildcards.  A set bit is a wildcard
        replacement with the lowest bit being user then project then testbed.
        """
        if p & 1: user = ["<any>"]
        else: user = a[2]
        if p & 2: proj = "<any>"
        else: proj = a[1]
        if p & 4: tb = "<any>"
        else: tb = a[0]

        return (tb, proj, user)

    def find_access(self, search):
        """
        Search the access DB for a match on this tuple.  Return the matching
        access tuple and the user that matched.
        
        NB, if the initial tuple fails to match we start inserting wildcards in
        an order determined by self.project_priority.  Try the list of users in
        order (when wildcarded, there's only one user in the list).
        """
        if self.project_priority: perm = (0, 1, 2, 3, 4, 5, 6, 7)
        else: perm = (0, 2, 1, 3, 4, 6, 5, 7)

        for p in perm: 
            s = self.permute_wildcards(search, p)
            # s[2] is None on an anonymous, unwildcarded request
            if s[2] != None:
                for u in s[2]:
                    if self.access.has_key((s[0], s[1], u)):
                        return (self.access[(s[0], s[1], u)], u)
            else:
                if self.access.has_key(s):
                    return (self.access[s], None)
        return None

    def lookup_access(self, req, fid):
        """
        Determine the allowed access for this request.  Return the access and
        which fields are dynamic.

        The fedid is needed to construct the request
        """
        tb = None
        project = None
        user = None

        principal_type = self.fedid_category.get(fid, self.fedid_default)

        if principal_type == "testbed": tb = fid

        if req.has_key('project'):
            p = req['project']
            if p.has_key('name'):
                project = unpack_id(p['name'])
            user = self.get_users(p)
        else:
            user = self.get_users(req)

        # Now filter by prinicpal type
        if principal_type == "user":
            if user != None:
                fedids = [ u for u in user if isinstance(u, type(fid))]
                if len(fedids) > 1:
                    raise fedd_proj.client_error(
                            "User asserting multiple fedids")
                elif len(fedids) == 1 and fedids[0] != fid:
                    raise fedd_proj.client_error(
                            "User asserting different fedid")
            project = None
            tb = None
        elif principal_type == "project":
            if isinstance(project, type(fid)) and fid != project:
                raise fedd_proj.client_error(
                        "Project asserting different fedid")
            tb = None

        # Ready to look up access
        print "Lookup %s %s %s: " % (tb, project, user)
        found, user_match = self.find_access((tb, project, user))
        print "Found: ", found
        
        if found == None: raise fedd_proj.server_error("Access denied")

        # resolve <dynamic> and <same> in found
        dyn_proj = False
        dyn_user = False

        if found[0].name == "<same>":
            if project != None:
                found[0].name = project
            else : 
                raise fedd_proj.server_error(
                        "Project matched <same> when no project given")
        elif found[0].name == "<dynamic>":
            found[0].name = self.random_string("project", 3)
            dyn_proj = True

        if found[1] == "<same>":
            if user_match == "<any>":
                if user != None: found = (found[0], user[0])
                else: raise fedd_proj.server_error(
                        "Matched <same> on anonymous request")
            else:
                found = (found[0], user_match)
        elif found[1] == "<dynamic>":
            found = (found[0], self.random_string("user", 4))
            dyn_user = True
        
        return found, (dyn_user, dyn_proj)

    def build_response(self, alloc_id, ap, ssh):
        """
        Create the SOAP response.

        Build the dictionary description of the response and use
        fedd_utils.pack_soap to create the soap message.  NB that alloc_id is
        a fedd_services_types.IDType_Holder pulled from the incoming message
        """
        # Because alloc_id is already a fedd_services_types.IDType_Holder,
        # there's no need to repack it
        msg = { 
                'allocID': alloc_id,
                'emulab': { 
                    'domain': self.domain,
                    'boss': self.boss,
                    'ops': self.ops,
                    'fileServer': self.fileserver,
                    'eventServer': self.eventserver,
                    'project': {
                        'name': pack_id(ap[0].name),
                        'user': [ {
                            'userID': pack_id(ap[1]),
                            'access' : [ { 'sshPubkey': x } for x in ssh ],
                            }
                        ]
                    }
                },
            }
        if len(self.attrs) > 0:
            msg['emulab']['fedAttr'] = \
                [ { 'attribute': x, 'value' : y } \
                        for x,y in self.attrs.iteritems()]
        return msg

    def RequestAccess(self, req, fid):

        if req.has_key('RequestAccessRequestBody'):
            req = req['RequestAccessRequestBody']
        else:
            raise fedd_proj.client_error("No request!?")

        if req.has_key('destinationTestbed'):
            dt = unpack_id(req['destinationTestbed'])
        
        if dt == None or dt == self.testbed:
            # Request for this fedd
            found, dyn = self.lookup_access(req, fid)

            # Check for access to restricted nodes
            if req.has_key('resources') and req['resources'].has_key('node'):
                resources = req['resources']
                inaccessible = [ t for n in resources['node'] \
                                if n.has_key('hardware') != None \
                                    for t in n['hardware'] \
                                        if t in self.restricted and \
                                                t not in found[0].node_types]
                if len(inaccessible) > 0:
                    raise fedd_proj.server_error(
                            "Access denied (nodetypes %s)" % \
                            str(', ').join(inaccessible))

            ssh = [ x['sshPubkey'] \
                    for x in req['access'] if x.has_key('sshPubkey')]

            if len(ssh) > 0: 
                if dyn[1]: self.dynamic_project(found, ssh)
                else: pass    # SSH key additions
            else:
                raise fedd_proj.client_error("SSH access parameters required")

            resp = self.build_response(req['allocID'], found, ssh)
            return resp
        else:
            p_fault = None      # Any SOAP failure (sent unless XMLRPC works)
            try:
                # Proxy the request using SOAP
                return self.proxy_request(dt, req)
            except RuntimeError, f:
                p_fault = f
          
           
            # This runs if SOAP failed
            try:
                return self.proxy_xmlrpc_request(dt, req)
                print "here"
            except RuntimeError:
                # Both failed, return the SOAP error.  If somehow that
                # exception is gone, return the XMLRPC one
                if p_fault != None: raise p_fault
                else: raise

    def soap_RequestAccess(self, ps, fid):
        req = ps.Parse(RequestAccessRequestMessage.typecode)

        msg = self.RequestAccess(unpack_soap(req), fedid)
        
        resp = RequestAccessResponseMessage()
        resp.set_element_RequestAccessResponseBody(
                pack_soap(resp, "RequestAccessResponseBody", msg))

        return resp

    def xmlrpc_RequestAccess(self, params, fid):
        msg = self.RequestAccess(params[0], fedid)

        if msg != None:
            return xmlrpclib.dumps(({ "RequestAccessResponseBody": msg },))
        else:
            raise server_error("No response generated?!");

    def read_trust(self, trust):
        """
        Read a trust file that splits fedids into testbeds, users or projects

        Format is:

        [type]
        fedid
        fedid
        default: type
        """
        lineno = 0;
        cat = None
        cat_re = re.compile("\[(user|testbed|project)\]$", re.IGNORECASE)
        fedid_re = re.compile("[" + string.hexdigits + "]+$")
        default_re = re.compile("default:\s*(user|testbed|project)$", 
                re.IGNORECASE)

        f = open(trust, "r")
        for line in f:
            lineno += 1
            line = line.strip()
            if len(line) == 0 or line.startswith("#"):
                continue
            # Category line
            m = cat_re.match(line)
            if m != None:
                cat = m.group(1).lower()
                continue
            # Fedid line
            m = fedid_re.match(line)
            if m != None:
                if cat != None:
                    self.fedid_category[fedid(hexstr=m.string)] = cat
                else:
                    raise fedd_proj.parse_error(\
                            "Bad fedid in trust file (%s) line: %d" % \
                            (trust, lineno))
                continue
            # default line
            m = default_re.match(line)
            if m != None:
                self.fedid_default = m.group(1).lower()
                continue
            # Nothing matched - bad line, raise exception
            f.close()
            raise fedd_proj.parse_error(\
                    "Unparsable line in trustfile %s line %d" % (trust, lineno))
        f.close()

    def read_config(self, config):
        """
        Read a configuration file and set internal parameters.

        The format is more complex than one might hope.  The basic format is
        attribute value pairs separated by colons(:) on a signle line.  The
        attributes in bool_attrs, emulab_attrs and id_attrs can all be set
        directly using the name: value syntax.  E.g. 
        boss: hostname
        sets self.boss to hostname.  In addition, there are access lines of the
        form (tb, proj, user) -> (aproj, auser) that map the first tuple of
        names to the second for access purposes.  Names in the key (left side)
        can include "<NONE> or <ANY>" to act as wildcards or to require the
        fields to be empty.  Similarly aproj or auser can be <SAME> or
        <DYNAMIC> indicating that either the matching key is to be used or a
        dynamic user or project will be created.  These names can also be
        federated IDs (fedid's) if prefixed with fedid:.  Finally, the aproj
        can be followed with a colon-separated list of node types to which that
        project has access (or will have access if dynamic).
        Testbed attributes outside the forms above can be given using the
        format attribute: name value: value.  The name is a single word and the
        value continues to the end of the line.  Empty lines and lines startin
        with a # are ignored.

        Parsing errors result in a parse_error exception being raised.
        """
        lineno=0
        name_expr = "["+string.ascii_letters + string.digits + "\.\-_]+"
        fedid_expr = "fedid:[" + string.hexdigits + "]+"
        key_name = "(<ANY>|<NONE>|"+fedid_expr + "|"+ name_expr + ")"
        access_proj = "(<DYNAMIC>(?::" + name_expr +")*|"+ \
                "<SAME>" + "(?::" + name_expr + ")*|" + \
                fedid_expr + "(?::" + name_expr + ")*|" + \
                name_expr + "(?::" + name_expr + ")*)"
        access_name = "(<DYNAMIC>|<SAME>|" + fedid_expr + "|"+ name_expr + ")"

        bool_re = re.compile('(' + '|'.join(fedd_proj.bool_attrs) + 
                '):\s+(true|false)', re.IGNORECASE)
        string_re = re.compile( "(" + \
                '|'.join(fedd_proj.emulab_attrs + fedd_proj.id_attrs) + \
                '):\s*(.*)', re.IGNORECASE)
        attr_re = re.compile('attribute:\s*([\._\-a-z0-9]+)\s+value:\s*(.*)',
                re.IGNORECASE)
        access_re = re.compile('\('+key_name+'\s*,\s*'+key_name+'\s*,\s*'+
                key_name+'\s*\)\s*->\s*\('+access_proj + '\s*,\s*' + 
                access_name + '\s*\)', re.IGNORECASE)
        trustfile_re = re.compile("trustfile:\s*(.*)", re.IGNORECASE)
        restricted_re = re.compile("restricted:\s*(.*)", re.IGNORECASE)

        def parse_name(n):
            if n.startswith('fedid:'): return fedid(n[len('fedid:'):])
            else: return n

        f = open(config, "r");
        for line in f:
            lineno += 1
            line = line.strip();
            if len(line) == 0 or line.startswith('#'):
                continue

            # Boolean attribute line
            m = bool_re.match(line);
            if m != None:
                attr, val = m.group(1,2)
                setattr(self, attr.lower(), bool(val.lower() == "true"))
                continue

            # String attribute line
            m = string_re.match(line)
            if m != None:
                attr, val = m.group(1,2)
                setattr(self, attr.lower(), val)
                continue

            # Extended (attribute: x value: y) attribute line
            m = attr_re.match(line)
            if m != None:
                attr, val = m.group(1,2)
                self.attrs[attr] = val
                continue

            # Access line (t, p, u) -> (ap, au) line
            m = access_re.match(line)
            if m != None:
                access_key = tuple([ parse_name(x) for x in m.group(1,2,3)])
                aps = m.group(4).split(":");
                if aps[0] == 'fedid:':
                    del aps[0]
                    aps[0] = fedid(hexstr=aps[0])

                au = m.group(5)
                if au.startswith("fedid:"):
                    au = fedid(hexstr=aus[len("fedid:"):])

                access_val = (fedd_proj.access_project(aps[0], aps[1:]), au)

                self.access[access_key] = access_val
                continue

            # Trustfile inclusion
            m = trustfile_re.match(line)
            if m != None:
                self.read_trust(m.group(1))
                continue
            # Restricted node types

            m = restricted_re.match(line)
            if m != None:
                self.restricted.append(m.group(1))
                continue

            # Nothing matched to here: unknown line - raise exception
            f.close()
            raise fedd_proj.parse_error("Unknown statement at line %d of %s" % \
                    (lineno, config))
        f.close()

    def soap_dispatch(self, method, req, fid):
        if fedd_proj.soap_methods.has_key(method):
            try:
                return getattr(self, fedd_proj.soap_methods[method])(req, fid)
            except fedd_proj.server_error, e:
                de = ns0.faultType_Def((ns0.faultType_Def.schema,"RequestAccessFaultBody")).pyclass()
                de._code=100
                de._desc="temp"
                f = Fault(Fault.Server, "Server Test", detail=de)
                raise f
            except fedd_proj.client_error, e:
                raise Fault(Fault.Client, str(e))
        else:
            raise Fault(Fault.Client, "Unknown method: %s" % method)

    def xmlrpc_dispatch(self, method, req, fid):
        if fedd_proj.xmlrpc_methods.has_key(method):
            try:
                return getattr(self, fedd_proj.xmlrpc_methods[method])(req, fid)
            except fedd_proj.server_error, e:
                raise xmlrpclib.Fault("server", str(e))
            except fedd_proj.client_error, e:
                raise xmlrpclib.Fault("client", str(e))
        else:
            raise xmlrpclib.Fault(100, "Unknown method: %s" % method)

def new_feddservice(configfile):
    return fedd_proj(configfile)