def connect(self):
     host, port, user, pswd, path = \
         pump.parse_spec(self.opts, self.spec,
                         int(getattr(self.opts, "port", 11211)))
     if self.opts.ssl:
         port = couchbaseConstants.SSL_PORT
     return MCSink.connect_mc(host, port, user, pswd)

    def total_msgs(opts, source_bucket, source_node, source_map):
        source_name = source_node.get("hostname", None)
        if not source_name:
            return 0, None

        spec = source_map["spec"]
        name = source_bucket["name"]
        vbuckets_num = len(source_bucket["vBucketServerMap"]["vBucketMap"])
        if not vbuckets_num:
            return 0, None

        vbucket_list = getattr(opts, "vbucket_list", None)

        stats_vals = {}
        host, port, user, pswd, _ = pump.parse_spec(opts, spec, 8091)
        for stats in ["curr_items", "vb_active_resident_items_ratio"]:
            path = "/pools/default/buckets/%s/stats/%s" % (name, stats)
            err, json, data = pump.rest_request_json(host, int(port), user, pswd, opts.ssl, path, reason="total_msgs")
            if err:
                return 0, None

            nodeStats = data.get("nodeStats", None)
            if not nodeStats:
                return 0, None
            vals = nodeStats.get(source_name, None)
            if not vals:
                return 0, None
            stats_vals[stats] = vals[-1]

        total_msgs = stats_vals["curr_items"]
        resident_ratio = stats_vals["vb_active_resident_items_ratio"]
        if 0 < resident_ratio < 100:
            # for DGM case, server will transfer both in-memory items and backfill all items on disk
            total_msgs += (resident_ratio / 100.0) * stats_vals["curr_items"]
        return 0, int(total_msgs)

 def check(opts, spec, source_map):
     host, port, user, pswd, path = \
         pump.parse_spec(opts, spec, int(getattr(opts, "port", 11211)))
     rv, conn = MCSink.connect_mc(host, port, user, pswd)
     if rv != 0:
         return rv, None
     conn.close()
     return 0, None

 def connect(self) -> Tuple[couchbaseConstants.PUMP_ERROR, Optional[cb_bin_client.MemcachedClient]]:
     host, port, user, pswd, path = \
         pump.parse_spec(self.opts, self.spec,
                         int(getattr(self.opts, "port", 11211)))
     if self.opts.ssl:
         port = couchbaseConstants.SSL_PORT
     return MCSink.connect_mc(host, port, user, pswd, self.sink_map["name"],
                              self.opts.ssl, collections=self.opts.collection)

 def check(opts, spec: str, source_map) -> Tuple[couchbaseConstants.PUMP_ERROR, Optional[Dict[str, Any]]]:
     host, port, user, pswd, path = pump.parse_spec(opts, spec, int(getattr(opts, "port", 11211)))
     if opts.ssl:
         ports = couchbaseConstants.SSL_PORT
     rv, conn = MCSink.connect_mc(host, port, user, pswd, None, opts.ssl, opts.no_ssl_verify, opts.cacert)
     if rv != 0:
         return rv, None
     conn.close()  # type: ignore
     return 0, None

    def provide_design(opts, source_spec: str, source_bucket: Dict[str, Any], source_map) -> \
            Tuple[couchbaseConstants.PUMP_ERROR, Optional[str]]:
        # Ephemeral buckets do not have design docs
        if source_bucket['bucketType'] == 'ephemeral':
            return 0, None
        spec_parts = source_map.get('spec_parts')
        if not spec_parts:
            return "error: no design spec_parts", None
        host, port, user, pswd, path = spec_parts

        source_nodes = pump.filter_bucket_nodes(source_bucket, spec_parts)
        if not source_nodes:
            source_nodes = source_bucket['nodes']
            if not source_nodes:
                return f'error: no design source node; spec_parts: {spec_parts}', None

        couch_api_base = source_nodes[0].get('couchApiBase')
        if not couch_api_base:
            return 0, None  # No couchApiBase; probably not 2.0.

        err, ddocs_json, ddocs = \
            pump.rest_request_json(host, int(port), user, pswd, opts.ssl,
                                   "/pools/default/buckets/%s/ddocs" %
                                   (source_bucket['name']),
                                   reason="provide_design", verify=opts.no_ssl_verify, cacert=opts.cacert)
        if err and "response: 404" in err: # A 404/not-found likely means 2.0-DP4.
            ddocs_json = None
            ddocs_url = f'{couch_api_base}/_all_docs'
            ddocs_qry = "?startkey=\"_design/\"&endkey=\"_design0\"&include_docs=true"

            host, port, user, pswd, path = \
                pump.parse_spec(opts, ddocs_url, 8092)
            # Not using user/pwd as 2.0-DP4 CAPI did not support auth.
            err, ddocs_json, ddocs = \
                pump.rest_request_json(host, int(port), None, None, opts.ssl,
                                       path + ddocs_qry,
                                       reason="provide_design-2.0DP4", verify=opts.no_ssl_verify, cacert=opts.cacert)
        if err is not None:
            return err, None

        if ddocs is None:
            return 0, None

        if not ddocs.get('rows', None):
            return 0, None
        else:
            return 0, json.dumps(ddocs.get('rows', []))

    def provide_design(opts, source_spec, source_bucket, source_map):
        spec_parts = source_map.get("spec_parts")
        if not spec_parts:
            return "error: no design spec_parts", None
        host, port, user, pswd, path = spec_parts

        source_nodes = pump.filter_bucket_nodes(source_bucket, spec_parts)
        if not source_nodes:
            source_nodes = source_bucket["nodes"]
            if not source_nodes:
                return ("error: no design source node; spec_parts: %s" % (spec_parts,), None)

        couch_api_base = source_nodes[0].get("couchApiBase")
        if not couch_api_base:
            return 0, None  # No couchApiBase; probably not 2.0.

        err, ddocs_json, ddocs = pump.rest_request_json(
            host,
            int(port),
            user,
            pswd,
            opts.ssl,
            "/pools/default/buckets/%s/ddocs" % (source_bucket["name"]),
            reason="provide_design",
        )
        if err and "response: 404" in err:  # A 404/not-found likely means 2.0-DP4.
            ddocs_json = None
            ddocs_url = couch_api_base + "/_all_docs"
            ddocs_qry = '?startkey="_design/"&endkey="_design0"&include_docs=true'

            host, port, user, pswd, path = pump.parse_spec(opts, ddocs_url, 8092)
            # Not using user/pwd as 2.0-DP4 CAPI did not support auth.
            err, ddocs_json, ddocs = pump.rest_request_json(
                host, int(port), None, None, opts.ssl, path + ddocs_qry, reason="provide_design-2.0DP4"
            )
        if err:
            return err, None

        if not ddocs.get("rows", None):
            return 0, None
        else:
            return 0, json.dumps(ddocs.get("rows", []))

    def total_msgs(opts, source_bucket: Dict[str, Any], source_node, source_map: Dict[str, Any]) ->\
            Tuple[couchbaseConstants.PUMP_ERROR, Optional[int]]:
        source_name = source_node.get("hostname", None)
        if not source_name:
            return 0, None

        spec = source_map['spec']
        name = source_bucket['name']
        vbuckets_num = len(source_bucket['vBucketServerMap']['vBucketMap'])
        if not vbuckets_num:
            return 0, None

        vbucket_list = getattr(opts, "vbucket_list", None)

        stats_vals = {}
        host, port, user, pswd, _ = pump.parse_spec(opts, spec, 8091)
        for stats in ["curr_items", "vb_active_resident_items_ratio"]:
            path = f'/pools/default/buckets/{name}/stats/{stats}'
            err, json, data = pump.rest_request_json(host, int(port),
                                                     user, pswd, opts.ssl, path,
                                                     reason="total_msgs", verify=opts.no_ssl_verify, cacert=opts.cacert)
            if err:
                return 0, None

            if data is not None:
                nodeStats = data.get("nodeStats", None)

            if not nodeStats:
                return 0, None
            vals = nodeStats.get(source_name, None)
            if not vals:
                return 0, None
            stats_vals[stats] = vals[-1]

        total_msgs = stats_vals["curr_items"]
        resident_ratio = stats_vals["vb_active_resident_items_ratio"]
        if 0 < resident_ratio < 100:
            # for DGM case, server will transfer both in-memory items and
            # backfill all items on disk
            total_msgs += (resident_ratio/100.0) * stats_vals["curr_items"]
        return 0, int(total_msgs)

    def provide_design(opts, source_spec, source_bucket, source_map):
        spec_parts = source_map.get('spec_parts')
        if not spec_parts:
            return "error: no design spec_parts", None
        host, port, user, pswd, path = spec_parts

        source_nodes = pump.filter_bucket_nodes(source_bucket, spec_parts)
        if not source_nodes:
            if spec_parts[0] not in ['localhost', '127.0.0.1']:
                return ("error: no design source node; spec_parts: %s" %
                        (spec_parts,), None)
            else:
                source_nodes = source_bucket['nodes']

        couch_api_base = source_nodes[0].get('couchApiBase')
        if not couch_api_base:
            return 0, None # No couchApiBase; probably not 2.0.

        err, ddocs_json, ddocs = \
            pump.rest_request_json(host, int(port), user, pswd,
                                   "/pools/default/buckets/%s/ddocs" %
                                   (source_bucket['name']),
                                   reason="provide_design")
        if err and "response: 404" in err: # A 404/not-found likely means 2.0-DP4.
            ddocs_json = None
            ddocs_url = couch_api_base + "/_all_docs"
            ddocs_qry = "?startkey=\"_design/\"&endkey=\"_design0\"&include_docs=true"

            host, port, user, pswd, path = \
                pump.parse_spec(opts, ddocs_url, 8092)
            # Not using user/pwd as 2.0-DP4 CAPI did not support auth.
            err, ddocs_json, ddocs = \
                pump.rest_request_json(host, int(port), None, None,
                                       path + ddocs_qry,
                                       reason="provide_design-2.0DP4")
        if err:
            return err, None

        return 0, json.dumps(ddocs.get('rows', []))

    def total_msgs(opts, source_bucket, source_node, source_map):
        source_name = source_node.get("hostname", None)
        if not source_name:
            return 0, None

        spec = source_map["spec"]
        name = source_bucket["name"]
        path = "/pools/default/buckets/%s/stats/curr_items" % (name)
        host, port, user, pswd, _ = pump.parse_spec(opts, spec, 8091)
        err, json, data = pump.rest_request_json(host, int(port), user, pswd, path, reason="total_msgs")
        if err:
            return 0, None

        nodeStats = data.get("nodeStats", None)
        if not nodeStats:
            return 0, None

        curr_items = nodeStats.get(source_name, None)
        if not curr_items:
            return 0, None

        return 0, curr_items[-1]

 def connect(self):
     host, port, user, pswd, path = \
         pump.parse_spec(self.opts, self.spec,
                         int(getattr(self.opts, "port", 11211)))
     return MCSink.connect_mc(host, port, user, pswd)

        if not source_design:
            return 0
        try:
            sd = json.loads(source_design)
            if not sd:
               return 0
            print json.dumps(sd, indent=2)
        except ValueError, e:
            return "error: could not parse source design; exception: %s" % (e)

        err, index_server = pump.filter_server(opts, sink_spec, 'index')
        if err or not index_server:
            logging.error("could not find index server")
            return 0

        spec_parts = pump.parse_spec(opts, sink_spec, 8091)
        if not spec_parts:
            return "error: design sink no spec_parts: " + sink_spec
        host, port, user, pswd, path = spec_parts
        host,port = pump.hostport(index_server)
        sink_bucket = sink_map['buckets'][0]
        url = "/restoreIndexMetadata?bucket=%s" % sink_bucket['name']
        #post_headers = {"Content-type": "application/x-www-form-urlencoded"}
        err, conn, response = \
            pump.rest_request(host, couchbaseConstants.INDEX_PORT, user, pswd, opts.ssl,
                              url, method='POST',
                              #body=urllib.urlencode(sd),
                              body=json.dumps(sd),
                              #headers=post_headers,
                              reason='restore index')
        print response

    def encode_tap_connect_opts(opts, backfill=False, vblist=None):
        header = 0
        val = []

        for op in sorted(opts.keys()):
            header |= op
            if op in memcacheConstants.TAP_FLAG_TYPES:
                val.append(struct.pack(memcacheConstants.TAP_FLAG_TYPES[op],
                                       opts[op]))
            elif backfill and op == memcacheConstants.TAP_FLAG_CHECKPOINT:
                if opts[op][2] >= 0:
                    val.append(struct.pack(">HHQ",
                                           opts[op][0], opts[op][1], opts[op][2]))
            elif vblist and op == memcacheConstants.TAP_FLAG_LIST_VBUCKETS:
                val.apend(struct.pack(">H", len(vblist))
                vblist = vblist[1:-1].split(",")
                for v in vblist:
                    val.apend(struct.pack(">H", int(v)))
            else:
                val.append(opts[op])

        return struct.pack(">I", header), ''.join(val)

    @staticmethod
    def total_msgs(opts, source_bucket, source_node, source_map):
        source_name = source_node.get("hostname", None)
        if not source_name:
            return 0, None

        spec = source_map['spec']
        name = source_bucket['name']
        path = "/pools/default/buckets/%s/stats/curr_items" % (name)
        host, port, user, pswd, _ = pump.parse_spec(opts, spec, 8091)
        err, json, data = pump.rest_request_json(host, int(port),
                                                 user, pswd, path,
                                                 reason="total_msgs")
        if err:
            return 0, None

        nodeStats = data.get("nodeStats", None)
        if not nodeStats:
            return 0, None

        curr_items = nodeStats.get(source_name, None)
        if not curr_items:
            return 0, None

        return 0, curr_items[-1]

class TapSink(pump_mc.CBSink):
    """Smart client sink using tap protocal to couchbase cluster."""
    def __init__(self, opts, spec, source_bucket, source_node,
                 source_map, sink_map, ctl, cur):
        super(TapSink, self).__init__(opts, spec, source_bucket, source_node,
                                     source_map, sink_map, ctl, cur)

        self.tap_name = "".join(random.sample(string.letters, 16))

    @staticmethod
    def check_base(opts, spec):
        #allow destination vbucket state to be anything

        op = getattr(opts, "destination_operation", None)
        if not op in [None, 'set', 'add', 'get']:
            return ("error: --destination-operation unsupported value: %s" +
                    "; use set, add, get") % (op)

        return pump.EndPoint.check_base(opts, spec)

    def find_conn(self, mconns, vbucket_id):

        rc, conn = super(TapSink, self).find_conn(mconns, vbucket_id)
        if rc != 0:
            return rc, None

        tap_opts = {memcacheConstants.TAP_FLAG_SUPPORT_ACK: ''}

        conn.tap_fix_flag_byteorder = version.split(".") >= ["2", "0", "0"]
        if self.tap_conn.tap_fix_flag_byteorder:
            tap_opts[memcacheConstants.TAP_FLAG_TAP_FIX_FLAG_BYTEORDER] = ''

        ext, val = TapSink.encode_tap_connect_opts(tap_opts)

        conn._sendCmd(memcacheConstants.CMD_TAP_CONNECT,
                     self.tap_name, val, 0, ext)
        return rv, conn

    def send_msgs(self, conn, msgs, operation, vbucket_id=None):
        rv = super(TapSink, self).sendMsg(conn, msgs, operation, vbucket_id)
        if rv != 0:
            return rv

        #send vbucket recovery commit msg
        host, port, user, pwd, path = \
            pump.parse_spec(self.opts, self.sink, 8091)

        params={"vbucket", vbucket_id}
        err, conn = \
            pump.rest_request(host, int(port), user, pwd,
                              '/pools/default/buckets/%s/commitVbucketRecovery' % self.sink_bucket,
#.........这里部分代码省略.........

    def consume_design(opts, sink_spec: str, sink_map, source_bucket,
                       source_map, source_design: Union[str, bytes]) -> couchbaseConstants.PUMP_ERROR:
        if not source_design:
            return 0
        try:
            sd = json.loads(source_design)
        except ValueError as e:
            return f'error: could not parse source design; exception: {e!s}'
        if not sd:
            return 0

        if (not sink_map['buckets'] or
            len(sink_map['buckets']) != 1 or
            not sink_map['buckets'][0] or
            not sink_map['buckets'][0]['name']):
            return "error: design sink incorrect sink_map bucket"
        spec_parts = pump.parse_spec(opts, sink_spec, 8091)
        if not spec_parts:
            return "error: design sink no spec_parts: " + sink_spec
        sink_bucket = sink_map['buckets'][0]
        sink_nodes = pump.filter_bucket_nodes(sink_bucket, spec_parts) or \
            sink_bucket['nodes']
        if not sink_nodes:
            return "error: design sink nodes missing"
        couch_api_base = sink_nodes[0].get('couchApiBase')
        if not couch_api_base:
            return f'error: cannot restore bucket design on a couchbase cluster that does not support couch API;' \
                f' the couchbase cluster may be an older, pre-2.0 version; please check your cluster URL: {sink_spec}'
        host, port, user, pswd, path = \
            pump.parse_spec(opts, couch_api_base, 8092)
        if user is None:
            user = spec_parts[2] # Default to the main REST user/pwsd.
            pswd = spec_parts[3]

        if opts.username_dest is not None and opts.password_dest is not None:
            user = opts.username_dest
            user = opts.password_dest
        if type(sd) is dict:

            id = sd.get('_id', None)
            if id:
                str_source = _to_string(source_design)
                err, conn, response = \
                    pump.rest_request(host, int(port), user, pswd, opts.ssl,
                                      f'{path}/{id}', method='PUT', body=str_source,
                                      reason="consume_design", verify=opts.no_ssl_verify, ca_cert=opts.cacert)
                if conn:
                    conn.close()
                if err:
                    return f'error: could not restore design doc id: {id}; response: {response}; err: {err}'
            else:
                stmts = sd.get('statements', [])
                hostname = f'http://{spec_parts[0]}:{spec_parts[1]!s}'
                cm = ClusterManager(hostname, user, pswd, opts.ssl)
                try:
                    for stmt in stmts:
                        result, errors = cm.n1ql_query(stmt['statement'], stmt.get('args', None))
                        if errors:
                            logging.error(f'N1QL query {stmt["statement"]} failed due to {errors}')

                        if result and 'errors' in result:
                            for error in result['errors']:
                                logging.error(f'N1QL query {stmt["statement"]} failed due to error `{error["msg"]}`')
                except ServiceNotAvailableException as e:
                    logging.error("Failed to restore indexes, cluster does not contain a query node")
        elif type(sd) is list:
            for row in sd:
                logging.debug(f'design_doc row: {row!s}')

                doc = row.get('doc', None)
                if not doc:
                    stmt = row.get('statement', None)
                    if not stmt:
                        return f'error: missing design doc or index statement in row: {row}'
                    else:
                        #publish index
                        return 0

                if 'json' in doc and 'meta' in doc:
                    js = doc['json']
                    id = doc['meta'].get('id', None)
                    if not id:
                        return f'error: missing id for design doc: {row}'
                else:
                    # Handle design-doc from 2.0DP4.
                    js = doc
                    if '_rev' in js:
                        del js['_rev']
                    id = row.get('id', None)
                    if not id:
                        return f'error: missing id for row: {row}'

                js_doc = json.dumps(js)
                if id.startswith(CBSink.DDOC_HEAD):
                    id = CBSink.DDOC_HEAD + urllib.parse.quote(id[len(CBSink.DDOC_HEAD):], '')
                else:
                    id = urllib.parse.quote(id, '')
                logging.debug(f'design_doc: {js_doc}')
                logging.debug(f'design_doc id: {id} at: {path}/{id}')

#.........这里部分代码省略.........