routerbox/utils/he_ipv6/config.py

import collections
import copy
import os
import re
##
import requests
import requests.auth
from lxml import etree
##
from . import tunnel


def create_default_cfg():
    # Create a stripped sample config.
    ws_re = re.compile(r'^\s*$')
    cur_dir = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
    xtbxml = os.path.join(cur_dir, 'example.tunnelbroker.xml')
    with open(xtbxml, 'rb') as fh:
        xml = etree.fromstring(fh.read())
    # Create a stripped sample config.
    # First we strip comments (and fix the ensuing whitespace).
    # etree has a .canonicalize(), but it chokes on a default namespace.
    # https://bugs.launchpad.net/lxml/+bug/1869455
    # So everything we do is kind of a hack.
    # for c in xml.xpath("//comment()"):
    #    parent = c.getparent()
    #    parent.remove(c)
    xmlstr = etree.tostring(xml, with_comments = False, method = 'c14n', pretty_print = True).decode('utf-8')
    newstr = []
    for line in xmlstr.splitlines():
        r = ws_re.search(line)
        if not r:
            newstr.append(line.strip())
    xml = etree.fromstring(''.join(newstr).encode('utf-8'))
    # Remove text and attr text.
    xpathq = "descendant-or-self::*[namespace-uri()!='']"
    for e in xml.xpath(xpathq):
        if e.tag == '{{{0}}}heIPv6'.format(xml.nsmap[None]):
            continue
        if e.text is not None and e.text.strip() != '':
            e.text = ''
        for k, v in e.attrib.items():
            if v is not None:
                e.attrib[k] = ''
    # Remove multiple children of same type to simplify.
    for e in xml.xpath(xpathq):
        if e.tag == '{{{0}}}heIPv6'.format(xml.nsmap[None]):
            continue
        parent = e.getparent()
        try:
            for idx, child in enumerate(parent.findall(e.tag)):
                if idx == 0:
                    continue
                parent.remove(child)
        except AttributeError:
            pass
    # And add a comment pointing them to the fully commented config.
    xml.insert(0, etree.Comment(('\n  Please reference the fully commented example.tunnelbroker.xml found either '
                                 'at:\n  '
                                 '  * {0}\n    * https://git.square-r00t.net/RouterBox/tree/utils/he_ipv6/'
                                 'example.tunnelbroker.xml\n  and then configure this according to those '
                                 'instructions.\n  ').format(xtbxml)))
    return(etree.tostring(xml,
                          pretty_print = True,
                          with_comments = True,
                          with_tail = True,
                          encoding = 'UTF-8',
                          xml_declaration = True))


class Credential(object):
    def __init__(self, cred_xml):
        self.xml = cred_xml
        self.id = None
        self.user = None
        self.password = None
        self.parse()

    def _id(self):
        _id = self.xml.attrib.get('id')
        if not _id:
            raise ValueError('Missing required id attribute')
        self.id = _id.strip()
        return(None)

    def _password(self):
        _key_xml = self.xml.find('password')
        if _key_xml is None:
            raise ValueError('Missing required password element')
        _key_txt = _key_xml.text
        if not _key_txt:
            raise ValueError('password element is empty')
        self.password = _key_txt.strip()
        return(None)

    def _user(self):
        _user_xml = self.xml.find('user')
        if _user_xml is None:
            raise ValueError('Missing required user element')
        _user_txt = _user_xml.text
        if not _user_txt:
            raise ValueError('user element is empty')
        self.user = _user_txt.strip()
        return(None)

    def parse(self):
        self._id()
        self._user()
        self._password()
        return(None)


class BaseConfig(object):
    default_xsd = None

    def __init__(self, xml_raw, *args, **kwargs):
        self.raw = xml_raw
        self.tree = None
        self.ns_tree = None
        self.xml = None
        self.ns_xml = None
        self.xsd = None
        self.defaults_parser = None
        self.obj = None
        self.parse_xml()

    def get_xsd(self):
        raw_xsd = None
        base_url = None
        xsi = self.xml.nsmap.get('xsi', 'http://www.w3.org/2001/XMLSchema-instance')
        schemaLocation = '{{{0}}}schemaLocation'.format(xsi)
        schemaURL = self.xml.attrib.get(schemaLocation, self.default_xsd)
        split_url = schemaURL.split()
        if len(split_url) == 2:  # a properly defined schemaLocation
            schemaURL = split_url[1]
        else:
            schemaURL = split_url[0]  # a LAZY schemaLocation
        if schemaURL.startswith('file://'):
            schemaURL = re.sub(r'^file://', r'', schemaURL)
            with open(schemaURL, 'rb') as fh:
                raw_xsd = fh.read()
            base_url = os.path.dirname(schemaURL) + '/'
        else:
            req = requests.get(schemaURL)
            if not req.ok:
                raise RuntimeError('Could not download XSD')
            raw_xsd = req.content
            base_url = os.path.split(req.url)[0] + '/'  # This makes me feel dirty.
        self.xsd = etree.XMLSchema(etree.XML(raw_xsd, base_url = base_url))
        return(None)

    def parse_xml(self):
        self.parse_raw()
        self.get_xsd()
        self.populate_defaults()
        self.validate()
        return(None)

    def parse_raw(self, parser = None):
        self.xml = etree.fromstring(self.raw, parser = parser)
        self.ns_xml = etree.fromstring(self.raw, parser = parser)
        self.tree = self.xml.getroottree()
        self.ns_tree = self.ns_xml.getroottree()
        self.tree.xinclude()
        self.ns_tree.xinclude()
        self.strip_ns()
        return(None)

    def populate_defaults(self):
        if not self.xsd:
            self.get_xsd()
        if not self.defaults_parser:
            self.defaults_parser = etree.XMLParser(schema = self.xsd, attribute_defaults = True)
        self.parse_raw(parser = self.defaults_parser)
        return(None)

    def remove_defaults(self):
        self.parse_raw()
        return(None)

    def strip_ns(self, obj = None):
        # https://stackoverflow.com/questions/30232031/how-can-i-strip-namespaces-out-of-an-lxml-tree/30233635#30233635
        xpathq = "descendant-or-self::*[namespace-uri()!='']"
        if not obj:
            for x in (self.tree, self.xml):
                for e in x.xpath(xpathq):
                    e.tag = etree.QName(e).localname
        elif isinstance(obj, (etree._Element, etree._ElementTree)):
            obj = copy.deepcopy(obj)
            for e in obj.xpath(xpathq):
                e.tag = etree.QName(e).localname
            return(obj)
        else:
            raise ValueError('Did not know how to parse obj parameter')
        return(None)

    def validate(self):
        if not self.xsd:
            self.get_xsd()
        self.xsd.assertValid(self.ns_tree)
        return(None)


class Config(BaseConfig):
    default_xsd = 'http://schema.xml.r00t2.io/projects/router/he_ipv6.xsd'

    def __init__(self, xml_path, *args, **kwargs):
        self.xml_path = os.path.abspath(os.path.expanduser(xml_path))
        if not os.path.isfile(self.xml_path):
            with open(self.xml_path, 'wb') as fh:
                fh.write(create_default_cfg())
            raise ValueError('xml_path does not exist; '
                             'a sample configuration has been generated (be sure to configure it)')
        else:
            with open(xml_path, 'rb') as fh:
                raw_xml = fh.read()
        super().__init__(raw_xml, *args, **kwargs)
        self.creds = {}
        self.tunnels = collections.OrderedDict()
        self.subparse()

    def _creds(self):
        creds_xml = self.xml.find('creds')
        for cred_xml in creds_xml.findall('cred'):
            cred = Credential(cred_xml)
            self.creds[cred.id] = cred
        return(None)

    def _tunnels(self):
        tunnels_xml = self.xml.find('tunnels')
        for tun_xml in tunnels_xml.findall('tunnel'):
            tun_id = int(tun_xml.attrib['id'].strip())
            tun_creds_id = tun_xml.attrib['creds']
            creds = self.creds[tun_creds_id]
            update_key = tun_xml.find('updateKey').text.strip()
            # TODO: do I instead want to use HEConfig() and fetch the single unified config?
            # Pros:
            #   * I wouldn't completely die on a misconfigured tunnel in the user config.
            # Cons:
            #   * We'd have to skip missing tunnels (bad auth at HE, etc.)
            #   * We would use more memory and take more time during init.
            he_conf = HETunnelConfig(tun_id, creds, update_key)
            tun = tunnel.Tunnel(tun_xml, he_conf, self.creds[tun_creds_id])
            self.tunnels[tun_id] = tun
        return(None)

    def subparse(self):
        self._creds()
        self._tunnels()
        return(None)


class HEBaseConfig(BaseConfig):
    default_xsd = 'http://schema.xml.r00t2.io/projects/tunnelbroker.xsd'
    nsmap = {None: 'https://tunnelbroker.net/tunnelInfo.php',
             'xsi': 'http://www.w3.org/2001/XMLSchema-instance'}
    attr_qname = etree.QName('http://www.w3.org/2001/XMLSchema-instance', 'schemaLocation')
    schema_loc = 'https://tunnelbroker.net/tunnelInfo.php {0}'.format(default_xsd)
    url = ''

    def __init__(self, creds, *args, **kwargs):
        self.creds = creds
        super().__init__(self._fetch(), *args, **kwargs)

    def _add_ns(self, raw_xml):
        # https://mailman-mail5.webfaction.com/pipermail/lxml/20100323/013260.html
        _xml = etree.fromstring(raw_xml)
        _nsmap = copy.deepcopy(_xml.nsmap)
        _nsmap.update(self.nsmap)
        mod_xml = etree.Element(_xml.tag,
                                {self.attr_qname: self.schema_loc},
                                nsmap = _nsmap)
        mod_xml[:] = _xml[:]
        return(etree.tostring(mod_xml,
                              encoding = 'UTF-8',
                              xml_declaration = True,
                              pretty_print = True,
                              with_tail = True,
                              with_comments = True))

    def _fetch(self):
        req = requests.get(self.url,
                           auth = requests.auth.HTTPBasicAuth(self.creds.user,
                                                              self.creds.password))
        if not req.ok:
            raise RuntimeError('Could not fetch remote tunnel information')
        raw_xml = self._add_ns(req.content)
        return(raw_xml)


# This isn't really used.
class HEConfig(HEBaseConfig):
    default_xsd = 'http://schema.xml.r00t2.io/projects/tunnelbroker.xsd'
    nsmap = {None: 'https://tunnelbroker.net/tunnelInfo.php',
             'xsi': 'http://www.w3.org/2001/XMLSchema-instance'}
    attr_qname = etree.QName('http://www.w3.org/2001/XMLSchema-instance', 'schemaLocation')
    schema_loc = 'https://tunnelbroker.net/tunnelInfo.php {0}'.format(default_xsd)
    url = 'https://tunnelbroker.net/tunnelInfo.php'

    def __init__(self, creds, *args, **kwargs):
        super().__init__(creds, *args, **kwargs)
        self.tunnels = {}

    def add_tunnel(self, tun_id, update_key):
        self.tunnels[tun_id] = HETunnelConfig(tun_id, self.creds, update_key)
        return(None)


class HETunnelConfig(HEBaseConfig):
    # default_xsd = 'http://schema.xml.r00t2.io/projects/tunnelbroker.tun.xsd'
    # nsmap = {None: 'https://tunnelbroker.net/tunnelInfo.php?tid',
    #          'xsi': 'http://www.w3.org/2001/XMLSchema-instance'}
    # attr_qname = etree.QName('http://www.w3.org/2001/XMLSchema-instance', 'schemaLocation')
    # schema_loc = 'https://tunnelbroker.net/tunnelInfo.php?tid {0}'.format(default_xsd)
    url = 'https://tunnelbroker.net/tunnelInfo.php?tid={0}'

    def __init__(self, tun_id, creds, update_key, *args, **kwargs):
        self.tun_id = int(tun_id)
        self.url = self.url.format(self.tun_id)
        self.creds = copy.deepcopy(creds)
        self.creds.password = update_key
        super().__init__(self.creds, *args, **kwargs)
        self.id = None
        self.description = None
        self.client = None  # Client IPv6
        self.server = None  # Server IPv6
        self.endpoint = None  # Server IPv4
        self.my_ip = None  # Client IPv4 (not necessary; we locally cache Tunnel.my_ip)
        self.allocations = {}  # keys are 64 and 48
        self.rdns = []  # Also not necessary, but it's in the XML so why not.
        # Will only return a single <tunnel> for this URL.
        # TODO: I can probably consolidate all this into HECond instead?
        self.tun_xml = self.xml.find('tunnel')
        self.parse()

    def _alloc(self):
        for a in ('64', '48'):
            _alloc = self.tun_xml.find('routed{0}'.format(a))
            if _alloc is not None and _alloc.text.strip() != '':
                self.allocations[int(a)] = tunnel.Allocation(_alloc.text.strip())
        return(None)

    def _client(self):
        _client = self.tun_xml.find('clientv6').text
        if _client is not None and _client.strip() != '':
            self.client = tunnel.IP6(_client.strip(), 64)
        return(None)

    def _desc(self):
        _desc = self.tun_xml.find('description').text
        if _desc is not None and _desc.strip() != '':
            self.description = _desc.strip()
        return(None)

    def _endpoint(self):
        self.endpoint = tunnel.IP4(self.tun_xml.find('serverv4').text.strip(), 32)
        return(None)

    def _id(self):
        self.id = int(self.tun_xml.attrib['id'])
        return(None)

    def _my_ip(self):
        _ip = self.tun_xml.find('clientv4').text
        if _ip is not None and _ip.strip() != '':
            self.my_ip = tunnel.IP4(_ip.strip(), 32)
        return(None)

    def _rdns(self):
        self.rdns = []
        for r in range(1, 6):
            _rdns = self.tun_xml.find('rdns{0}'.format(r))
            if _rdns is not None and _rdns.text.strip() != '':
                self.rdns.append(_rdns.text.strip())
        self.rdns = tuple(self.rdns)
        return(None)

    def _server(self):
        self.server = tunnel.IP6(self.tun_xml.find('serverv6').text.strip(), 128)
        return(None)

    def parse(self):
        self._id()
        self._client()
        self._desc()
        self._endpoint()
        self._server()
        self._my_ip()
        self._alloc()
        self._rdns()
        return(None)