path: root/src/bin/stats/stats_httpd.py.in

#!@PYTHON@

# Copyright (C) 2011-2012  Internet Systems Consortium.
#
# Permission to use, copy, modify, and distribute this software for any
# purpose with or without fee is hereby granted, provided that the above
# copyright notice and this permission notice appear in all copies.
#
# THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SYSTEMS CONSORTIUM
# DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL
# INTERNET SYSTEMS CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT,
# INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING
# FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT,
# NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION
# WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

"""
A standalone HTTP server for HTTP/XML interface of statistics in BIND 10

"""
import sys; sys.path.append ('@@PYTHONPATH@@')
import os
import time
import errno
import select
from optparse import OptionParser, OptionValueError
import http.server
import socket
import string
import xml.etree.ElementTree
import urllib.parse
import re

import isc.cc
import isc.config
import isc.util.process
import isc.util.traceback_handler
from isc.util.address_formatter import AddressFormatter

import isc.log
from isc.log_messages.stats_httpd_messages import *

isc.log.init("b10-stats-httpd", buffer=True)
logger = isc.log.Logger("stats-httpd")

# Some constants for debug levels.
DBG_STATSHTTPD_INIT = logger.DBGLVL_START_SHUT
DBG_STATSHTTPD_MESSAGING = logger.DBGLVL_COMMAND

# If B10_FROM_SOURCE is set in the environment, we use data files
# from a directory relative to that, otherwise we use the ones
# installed on the system
if "B10_FROM_SOURCE" in os.environ:
    BASE_LOCATION = os.environ["B10_FROM_SOURCE"] + os.sep + \
        "src" + os.sep + "bin" + os.sep + "stats"
else:
    PREFIX = "@prefix@"
    DATAROOTDIR = "@datarootdir@"
    BASE_LOCATION = "@datadir@" + os.sep + "@PACKAGE@"
    BASE_LOCATION = BASE_LOCATION.replace("${datarootdir}", DATAROOTDIR).replace("${prefix}", PREFIX)
SPECFILE_LOCATION = BASE_LOCATION + os.sep + "stats-httpd.spec"
XML_TEMPLATE_LOCATION = BASE_LOCATION + os.sep + "stats-httpd-xml.tpl"
XSD_TEMPLATE_LOCATION = BASE_LOCATION + os.sep + "stats-httpd-xsd.tpl"
XSL_TEMPLATE_LOCATION = BASE_LOCATION + os.sep + "stats-httpd-xsl.tpl"

# These variables are paths part of URL.
# eg. "http://${address}" + XXX_URL_PATH
XML_URL_PATH = '/bind10/statistics/xml'
XSD_URL_PATH = '/bind10/statistics.xsd'
XSL_URL_PATH = '/bind10/statistics.xsl'
# TODO: This should be considered later.
XSD_NAMESPACE = 'http://bind10.isc.org/bind10'
XMLNS_XSI = "http://www.w3.org/2001/XMLSchema-instance"

# constant parameter of XML
XML_ROOT_ELEMENT = 'bind10:statistics'
XML_ROOT_ATTRIB = { 'xsi:schemaLocation' : '%s %s' % (XSD_NAMESPACE, XSD_URL_PATH),
                    'xmlns:bind10' : XSD_NAMESPACE,
                    'xmlns:xsi' : XMLNS_XSI }

# Assign this process name
isc.util.process.rename()

def item_name_list(element, identifier):
    """Return a list of strings. The strings are string expression of
    the first argument element which is dict type. The second argument
    identifier is a string for specifying the strings which are
    returned from this method as a list. For example, if we specify as

      item_name_list({'a': {'aa': [0, 1]}, 'b': [0, 1]}, 'a/aa'),

    then it returns

      ['a/aa', 'a/aa[0]', 'a/aa[1]'].

    If an empty string is specified in the second argument, all
    possible strings are returned as a list.  In that example if we
    specify an empty string in the second argument, then it returns

      ['a', 'a/aa', 'a/aa[0]', 'a/aa[1]', 'b', 'b[0]', 'b[1]'].

    The key name of element which is in the first argument is sorted.
    Even if we specify as

      item_name_list({'xx': 0, 'a': 1, 'x': 2}, ''),

    then it returns

      ['a', 'x', 'xx'].

    This method internally invokes isc.cc.data.find(). The arguments
    of this method are passed to isc.cc.data.find(). So an exception
    DataNotFoundError or DataTypeError might be raised via
    isc.cc.data.find() depending on the arguments. See details of
    isc.cc.data.find() for more information about exceptions"""
    elem = isc.cc.data.find(element, identifier)
    ret = []
    ident = identifier
    if ident:
        ret.append(ident)
    if type(elem) is dict:
        if ident:
            ident = ident + '/'
        for key in sorted(elem.keys()):
            idstr = '%s%s' % (ident, key)
            ret += item_name_list(element, idstr)
    elif type(elem) is list:
        for i in range(0, len(elem)):
            idstr = '%s[%d]' % (ident, i)
            ret += item_name_list(element, idstr)
    return ret

class HttpHandler(http.server.BaseHTTPRequestHandler):
    """HTTP handler class for HttpServer class. The class inherits the super
    class http.server.BaseHTTPRequestHandler. It implements do_GET()
    and do_HEAD() and overrides log_message()"""
    def do_GET(self):
        body = self.send_head()
        if body is not None:
            self.wfile.write(body.encode())

    def do_HEAD(self):
        self.send_head()

    def send_head(self):
        try:
            req_path = self.path
            req_path = urllib.parse.urlsplit(req_path).path
            req_path = urllib.parse.unquote(req_path)
            body = None
            # In case that the requested path (req_path),
            # e.g. /bind10/statistics/Auth/, is started with
            # XML_URL_PATH + '/'
            if req_path.find('%s/' % XML_URL_PATH) == 0:
                # remove './' from the path if there is
                req_path = os.path.normpath(req_path)
                # get the strings tailing after XML_URL_PATH
                req_path = req_path.lstrip('%s/' % XML_URL_PATH)
                # remove empty dir names from the path if there are
                path_dirs = req_path.split('/')
                path_dirs = [ d for d in filter(None, path_dirs) ]
                req_path = '/'.join(path_dirs)
                # pass the complete requested path,
                # e.g. Auth/queries.upd, to xml_handler()
                body = self.server.xml_handler(req_path)
            # In case that the requested path (req_path) is exactly
            # matched with XSD_URL_PATH
            elif req_path == XSD_URL_PATH:
                body = self.server.xsd_handler()
            # In case that the requested path (req_path) is exactly
            # matched with XSL_URL_PATH
            elif req_path == XSL_URL_PATH:
                body = self.server.xsl_handler()
            else:
                if 'Host' in self.headers.keys() and \
                        ( req_path == '/' or req_path == XML_URL_PATH ):
                    # redirect to XML URL only when requested with '/' or XML_URL_PATH
                    toloc = "http://%s%s/" % (self.headers.get('Host'), XML_URL_PATH)
                    self.send_response(302)
                    self.send_header("Location", toloc)
                    self.end_headers()
                    return None
                else:
                    # Couldn't find HOST
                    self.send_error(404)
                    return None
        except StatsHttpdDataError as err:
            # Couldn't find neither specified module name nor
            # specified item name
            self.send_error(404)
            logger.error(STATSHTTPD_SERVER_DATAERROR, err)
            return None
        except Exception as err:
            self.send_error(500)
            logger.error(STATSHTTPD_SERVER_ERROR, err)
            return None
        else:
            self.send_response(200)
            self.send_header("Content-type", "text/xml")
            self.send_header("Content-Length", len(body))
            self.send_header("Last-Modified", self.date_time_string(time.time()))
            self.end_headers()
            return body

    def log_message(self, format, *args):
        """overrides the parent method log_message()
        to use the bind10 logging framework.
        """
        logger.debug(DBG_STATSHTTPD_MESSAGING, STATSHTTPD_HTTPLOG,
                     self.address_string(),
                     format%args)

class HttpServerError(Exception):
    """Exception class for HttpServer class. It is intended to be
    passed from the HttpServer object to the StatsHttpd object."""
    pass

class HttpServer(http.server.HTTPServer):
    """HTTP Server class. The class inherits the super
    http.server.HTTPServer. Some parameters are specified as
    arguments, which are xml_handler, xsd_handler, xsl_handler, and
    log_writer. These all are parameters which the StatsHttpd object
    has. The handler parameters are references of functions which
    return body of each document. The last parameter log_writer is
    reference of writer function to just write to
    sys.stderr.write. They are intended to be referred by HttpHandler
    object."""
    def __init__(self, server_address, handler,
                 xml_handler, xsd_handler, xsl_handler, log_writer):
        self.server_address = server_address
        self.xml_handler = xml_handler
        self.xsd_handler = xsd_handler
        self.xsl_handler = xsl_handler
        self.log_writer = log_writer
        http.server.HTTPServer.__init__(self, server_address, handler)

class StatsHttpdError(Exception):
    """Exception class for StatsHttpd class. It is intended to be
    thrown from the the StatsHttpd object to the HttpHandler object or
    main routine."""
    pass

class StatsHttpdDataError(Exception):
    """Exception class for StatsHttpd class. The reason seems to be
    due to the data. It is intended to be thrown from the the
    StatsHttpd object to the HttpHandler object or main routine."""
    pass

class StatsHttpd:
    """The main class of HTTP server of HTTP/XML interface for
    statistics module. It handles HTTP requests, and command channel
    and config channel CC session. It uses select.select function
    while waiting for clients requests."""
    def __init__(self):
        self.running = False
        self.poll_intval = 0.5
        self.write_log = sys.stderr.write
        self.mccs = None
        self.httpd = []
        self.open_mccs()
        self.config = {}
        self.load_config()
        self.http_addrs = []
        self.mccs.start()
        try:
            self.open_httpd()
        except HttpServerError:
            # if some exception, e.g. address in use, is raised, then it closes mccs and httpd
            self.close_mccs()
            raise

    def open_mccs(self):
        """Opens a ModuleCCSession object"""
        # create ModuleCCSession
        logger.debug(DBG_STATSHTTPD_INIT, STATSHTTPD_STARTING_CC_SESSION)
        self.mccs = isc.config.ModuleCCSession(
            SPECFILE_LOCATION, self.config_handler, self.command_handler)
        self.cc_session = self.mccs._session

    def close_mccs(self):
        """Closes a ModuleCCSession object"""
        if self.mccs is None:
            return
        self.mccs.send_stopping()

        logger.debug(DBG_STATSHTTPD_INIT, STATSHTTPD_CLOSING_CC_SESSION)
        self.mccs.close()
        self.mccs = None

    def load_config(self, new_config={}):
        """Loads configuration from spec file or new configuration
        from the config manager"""
        # load config
        if len(self.config) == 0:
            self.config = dict([
                (itm['item_name'], self.mccs.get_value(itm['item_name'])[0])
                for itm in self.mccs.get_module_spec().get_config_spec()
                ])
        self.config.update(new_config)
        # set addresses and ports for HTTP
        addrs = []
        if 'listen_on' in self.config:
            for cf in self.config['listen_on']:
                if 'address' in cf and 'port' in cf:
                    addrs.append((cf['address'], cf['port']))
        self.http_addrs = addrs

    def open_httpd(self):
        """Opens sockets for HTTP. Iterating each HTTP address to be
        configured in spec file"""
        for addr in self.http_addrs:
            self.httpd.append(self._open_httpd(addr))

    def _open_httpd(self, server_address):
        httpd = None
        try:
            # get address family for the server_address before
            # creating HttpServer object. If a specified address is
            # not numerical, gaierror may be thrown.
            address_family = socket.getaddrinfo(
                server_address[0], server_address[1], 0,
                socket.SOCK_STREAM, socket.IPPROTO_TCP, socket.AI_NUMERICHOST
                )[0][0]
            HttpServer.address_family = address_family
            httpd = HttpServer(
                server_address, HttpHandler,
                self.xml_handler, self.xsd_handler, self.xsl_handler,
                self.write_log)
            logger.info(STATSHTTPD_STARTED,
                        AddressFormatter(server_address, address_family))
            return httpd
        except (socket.gaierror, socket.error,
                OverflowError, TypeError) as err:
           if httpd:
                httpd.server_close()
           raise HttpServerError(
               "Invalid address %s, port %s: %s: %s" %
               (server_address[0], server_address[1],
                err.__class__.__name__, err))

    def close_httpd(self):
        """Closes sockets for HTTP"""
        while len(self.httpd)>0:
            ht = self.httpd.pop()
            logger.info(STATSHTTPD_CLOSING,
                        AddressFormatter(ht.server_address))
            ht.server_close()

    def start(self):
        """Starts StatsHttpd objects to run. Waiting for client
        requests by using select.select functions"""
        self.running = True
        while self.running:
            try:
                (rfd, wfd, xfd) = select.select(
                    self.get_sockets(), [], [], self.poll_intval)
            except select.error as err:
                # select.error exception is caught only in the case of
                # EINTR, or in other cases it is just thrown.
                if err.args[0] == errno.EINTR:
                    (rfd, wfd, xfd) = ([], [], [])
                else:
                    raise
            # FIXME: This module can handle only one request at a
            # time. If someone sends only part of the request, we block
            # waiting for it until we time out.
            # But it isn't so big issue for administration purposes.
            for fd in rfd + xfd:
                if fd == self.mccs.get_socket():
                    self.mccs.check_command(nonblock=False)
                    continue
                for ht in self.httpd:
                    if fd == ht.socket:
                        ht.handle_request()
                        break
        self.stop()

    def stop(self):
        """Stops the running StatsHttpd objects. Closes CC session and
        HTTP handling sockets"""
        logger.info(STATSHTTPD_SHUTDOWN)
        self.close_httpd()
        self.close_mccs()
        self.running = False

    def get_sockets(self):
        """Returns sockets to select.select"""
        sockets = []
        if self.mccs is not None:
            sockets.append(self.mccs.get_socket())
        if len(self.httpd) > 0:
            for ht in self.httpd:
                sockets.append(ht.socket)
        return sockets

    def config_handler(self, new_config):
        """Config handler for the ModuleCCSession object. It resets
        addresses and ports to listen HTTP requests on."""
        logger.debug(DBG_STATSHTTPD_MESSAGING, STATSHTTPD_HANDLE_CONFIG,
                   new_config)
        errors = []
        if not self.mccs.get_module_spec().\
                validate_config(False, new_config, errors):
                return isc.config.ccsession.create_answer(
                    1, ", ".join(errors))
        # backup old config
        old_config = self.config.copy()
        self.load_config(new_config)
        # If the http sockets aren't opened or
        # if new_config doesn't have 'listen_on', it returns
        if len(self.httpd) == 0 or 'listen_on' not in new_config:
            return isc.config.ccsession.create_answer(0)
        self.close_httpd()
        try:
            self.open_httpd()
        except HttpServerError as err:
            logger.error(STATSHTTPD_SERVER_INIT_ERROR, err)
            # restore old config
            self.load_config(old_config)
            self.open_httpd()
            return isc.config.ccsession.create_answer(1, str(err))
        else:
            return isc.config.ccsession.create_answer(0)

    def command_handler(self, command, args):
        """Command handler for the ModuleCCSesson object. It handles
        "status" and "shutdown" commands."""
        if command == "status":
            logger.debug(DBG_STATSHTTPD_MESSAGING,
                         STATSHTTPD_RECEIVED_STATUS_COMMAND)
            return isc.config.ccsession.create_answer(
                0, "Stats Httpd is up. (PID " + str(os.getpid()) + ")")
        elif command == "shutdown":
            logger.debug(DBG_STATSHTTPD_MESSAGING,
                         STATSHTTPD_RECEIVED_SHUTDOWN_COMMAND)
            self.running = False
            return isc.config.ccsession.create_answer(0)
        else:
            logger.debug(DBG_STATSHTTPD_MESSAGING,
                         STATSHTTPD_RECEIVED_UNKNOWN_COMMAND, command)
            return isc.config.ccsession.create_answer(
                1, "Unknown command: " + str(command))

    def get_stats_data(self, owner=None, name=None):
        """Requests statistics data to the Stats daemon and returns
        the data which obtains from it. The first argument is the
        module name which owns the statistics data, the second
        argument is one name of the statistics items which the the
        module owns. The second argument cannot be specified when the
        first argument is not specified. It returns the statistics
        data of the specified module or item. When the session timeout
        or the session error is occurred, it raises
        StatsHttpdError. When the stats daemon returns none-zero
        value, it raises StatsHttpdDataError."""
        param = {}
        if owner is None and name is None:
            param = None
        if owner is not None:
            param['owner'] = owner
        if name is not None:
            param['name'] = name
        try:
            return self.mccs.rpc_call('show', 'Stats', params=param)
        except (isc.cc.session.SessionTimeout,
                isc.cc.session.SessionError,
                isc.config.RPCRecipientMissing) as err:
            raise StatsHttpdError("%s: %s" %
                                  (err.__class__.__name__, err))
        except isc.config.RPCError as e:
            raise StatsHttpdDataError("Stats module: %s" % str(e))

    def get_stats_spec(self, owner=None, name=None):
        """Requests statistics data to the Stats daemon and returns
        the data which obtains from it. The first argument is the
        module name which owns the statistics data, the second
        argument is one name of the statistics items which the the
        module owns. The second argument cannot be specified when the
        first argument is not specified. It returns the statistics
        specification of the specified module or item. When the
        session timeout or the session error is occurred, it raises
        StatsHttpdError. When the stats daemon returns none-zero
        value, it raises StatsHttpdDataError."""
        param = {}
        if owner is None and name is None:
            param = None
        if owner is not None:
            param['owner'] = owner
        if name is not None:
            param['name'] = name
        try:
            return self.mccs.rpc_call('showschema', 'Stats', params=param)
        except isc.config.RPCError as e:
            raise StatsHttpdDataError("Stats module: %s" % str(e))
        except (isc.cc.session.SessionTimeout,
                isc.cc.session.SessionError) as err:
            raise StatsHttpdError("%s: %s" %
                                  (err.__class__.__name__, err))


    def xml_handler(self, path=''):
        """Requests the specified statistics data and specification by
        using the functions get_stats_data and get_stats_spec
        respectively and loads the XML template file and returns the
        string of the XML document.The argument is a path in the
        requested URI. For example, the path is assumed to be like
        ${module_name}/${top_level_item_name}/${second_level_item_name}/..."""

        dirs = [ d for d in path.split("/") if len(d) > 0 ]
        module_name = None
        item_name = None
        if len(dirs) > 0:
            module_name = dirs[0]
        if len(dirs) > 1:
            item_name = dirs[1]
            # removed an index string when list-type value is
            # requested. Because such a item name can be accept by the
            # stats module currently.
            item_name = re.sub('\[\d+\]$', '', item_name)
        stats_spec = self.get_stats_spec(module_name, item_name)
        stats_data = self.get_stats_data(module_name, item_name)
        path_list = []
        try:
            path_list = item_name_list(stats_data, path)
        except isc.cc.data.DataNotFoundError as err:
            raise StatsHttpdDataError(
                "%s: %s" % (err.__class__.__name__, err))
        item_list = []
        for path in path_list:
            dirs = path.split("/")
            if len(dirs) < 2: continue
            item = {}
            item['identifier'] = path
            value = isc.cc.data.find(stats_data, path)
            if type(value) is bool:
                value = str(value).lower()
            if type(value) is dict or type(value) is list:
                value = None
            if value is not None:
                item['value'] = str(value)
            owner = dirs[0]
            item['owner'] = owner
            item['uri'] = urllib.parse.quote('%s/%s' % (XML_URL_PATH, path))
            item_path = '/'.join(dirs[1:])
            spec = isc.config.find_spec_part(stats_spec[owner], item_path)
            for key in ['name', 'type', 'description', 'title', \
                            'optional', 'default', 'format']:
                value = spec.get('item_%s' % key)
                if type(value) is bool:
                    value = str(value).lower()
                if type(value) is dict or type(value) is list:
                    value = None
                if value is not None:
                    item[key] = str(value)
            item_list.append(item)
        xml_elem = xml.etree.ElementTree.Element(
            XML_ROOT_ELEMENT, attrib=XML_ROOT_ATTRIB)
        for item in item_list:
            item_elem = xml.etree.ElementTree.Element('item', attrib=item)
            xml_elem.append(item_elem)
        # The coding conversion is tricky. xml..tostring() of Python 3.2
        # returns bytes (not string) regardless of the coding, while
        # tostring() of Python 3.1 returns a string.  To support both
        # cases transparently, we first make sure tostring() returns
        # bytes by specifying utf-8 and then convert the result to a
        # plain string (code below assume it).
        # FIXME: Non-ASCII characters might be lost here. Consider how
        # the whole system should handle non-ASCII characters.
        xml_string = str(xml.etree.ElementTree.tostring(xml_elem, encoding='utf-8'),
                         encoding='us-ascii')
        self.xml_body = self.open_template(XML_TEMPLATE_LOCATION).substitute(
            xml_string=xml_string, xsl_url_path=XSL_URL_PATH)
        return self.xml_body

    def xsd_handler(self):
        """Loads the XSD template file, replaces the variable strings,
        and returns the string of the XSD document."""
        self.xsd_body = self.open_template(XSD_TEMPLATE_LOCATION).substitute(
            xsd_namespace=XSD_NAMESPACE)
        return self.xsd_body

    def xsl_handler(self):
        """Loads the XSL template file, replaces the variable strings,
        and returns the string of the XSL document."""
        self.xsl_body = self.open_template(XSL_TEMPLATE_LOCATION).substitute(
            xsd_namespace=XSD_NAMESPACE)
        return self.xsl_body

    def open_template(self, file_name):
        """It opens a template file, and it loads all lines to a
        string variable and returns string. Template object includes
        the variable. Limitation of a file size isn't needed there. XXXX"""
        lines = None
        try:
            with open(file_name, 'r') as f:
                lines = "".join(f.readlines())
        except IOError as err:
            raise StatsHttpdDataError(
                "%s: %s" % (err.__class__.__name__, err))
        return string.Template(lines)

def main():
    try:
        parser = OptionParser()
        parser.add_option(
            "-v", "--verbose", dest="verbose", action="store_true",
            help="enable maximum debug logging")
        (options, args) = parser.parse_args()
        if options.verbose:
            isc.log.init("b10-stats-httpd", "DEBUG", 99, buffer=True)
        stats_httpd = StatsHttpd()
        stats_httpd.start()
    except OptionValueError as ove:
        logger.fatal(STATSHTTPD_BAD_OPTION_VALUE, ove)
        sys.exit(1)
    except isc.cc.session.SessionError as se:
        logger.fatal(STATSHTTPD_CC_SESSION_ERROR, se)
        sys.exit(1)
    except HttpServerError as hse:
        logger.fatal(STATSHTTPD_START_SERVER_INIT_ERROR, hse)
        sys.exit(1)
    except KeyboardInterrupt as kie:
        logger.info(STATSHTTPD_STOPPED_BY_KEYBOARD)

    logger.info(STATSHTTPD_EXITING)

if __name__ == "__main__":
    isc.util.traceback_handler.traceback_handler(main)