jumpserver/apps/terminal/backends/command/es.py

# -*- coding: utf-8 -*-
#
import pytz
import inspect

from datetime import datetime
from functools import reduce, partial
from itertools import groupby
from uuid import UUID

from django.utils.translation import gettext_lazy as _
from django.db.models import QuerySet as DJQuerySet
from elasticsearch import Elasticsearch
from elasticsearch.helpers import bulk
from elasticsearch.exceptions import RequestError, NotFoundError

from common.utils.common import lazyproperty
from common.utils import get_logger
from common.utils.timezone import local_now_date_display, utc_now
from common.exceptions import JMSException
from terminal.models import Command

logger = get_logger(__file__)


class InvalidElasticsearch(JMSException):
    default_code = 'invalid_elasticsearch'
    default_detail = _('Invalid elasticsearch config')


class NotSupportElasticsearch8(JMSException):
    default_code = 'not_support_elasticsearch8'
    default_detail = _('Not Support Elasticsearch8')


class CommandStore(object):
    def __init__(self, config):
        self.doc_type = config.get("DOC_TYPE") or '_doc'
        self.index_prefix = config.get('INDEX') or 'jumpserver'
        self.is_index_by_date = bool(config.get('INDEX_BY_DATE'))
        self.exact_fields = {}
        self.match_fields = {}
        hosts = config.get("HOSTS")
        kwargs = config.get("OTHER", {})

        ignore_verify_certs = kwargs.pop('IGNORE_VERIFY_CERTS', False)
        if ignore_verify_certs:
            kwargs['verify_certs'] = None
        self.es = Elasticsearch(hosts=hosts, max_retries=0, **kwargs)

        self.exact_fields = set()
        self.match_fields = {'input', 'risk_level', 'user', 'asset', 'system_user'}
        may_exact_fields = {'session', 'org_id'}

        if self.is_new_index_type():
            self.exact_fields.update(may_exact_fields)
            self.doc_type = '_doc'
        else:
            self.match_fields.update(may_exact_fields)

        self.init_index(config)

    def init_index(self, config):
        if self.is_index_by_date:
            date = local_now_date_display()
            self.index = '%s-%s' % (self.index_prefix, date)
            self.query_index = '%s-alias' % self.index_prefix
        else:
            self.index = config.get("INDEX") or 'jumpserver'
            self.query_index = config.get("INDEX") or 'jumpserver'

    def is_new_index_type(self):
        if not self.ping(timeout=3):
            return False

        info = self.es.info()
        version = info['version']['number'].split('.')[0]

        if version == '8':
            raise NotSupportElasticsearch8

        try:
            # 获取索引信息，如果没有定义，直接返回
            data = self.es.indices.get_mapping(self.index)
        except NotFoundError:
            return False

        try:
            if version == '6':
                # 检测索引是不是新的类型 es6
                properties = data[self.index]['mappings']['data']['properties']
            else:
                # 检测索引是不是新的类型 es7 default index type: _doc
                properties = data[self.index]['mappings']['properties']
            if properties['session']['type'] == 'keyword' \
                    and properties['org_id']['type'] == 'keyword':
                return True
        except KeyError:
            return False

    def pre_use_check(self):
        if not self.ping(timeout=3):
            raise InvalidElasticsearch
        self._ensure_index_exists()

    def _ensure_index_exists(self):
        properties = {
            "session": {
                "type": "keyword"
            },
            "org_id": {
                "type": "keyword"
            },
            "@timestamp": {
                "type": "date"
            },
            "timestamp": {
                "type": "long"
            }
        }
        info = self.es.info()
        version = info['version']['number'].split('.')[0]
        if version == '6':
            mappings = {'mappings': {'data': {'properties': properties}}}
        else:
            mappings = {'mappings': {'properties': properties}}

        if self.is_index_by_date:
            mappings['aliases'] = {
                self.query_index: {}
            }
        try:
            self.es.indices.create(self.index, body=mappings)
            return
        except RequestError as e:
            if e.error == 'resource_already_exists_exception':
                logger.warning(e)
            else:
                logger.exception(e)

    @staticmethod
    def make_data(command):
        data = dict(
            user=command["user"], asset=command["asset"],
            system_user=command["system_user"], input=command["input"],
            output=command["output"], risk_level=command["risk_level"],
            session=command["session"], timestamp=command["timestamp"],
            org_id=command["org_id"]
        )
        data["date"] = datetime.fromtimestamp(command['timestamp'], tz=pytz.UTC)
        return data

    def bulk_save(self, command_set, raise_on_error=True):
        actions = []
        for command in command_set:
            data = dict(
                _index=self.index,
                _type=self.doc_type,
                _source=self.make_data(command),
            )
            actions.append(data)
        return bulk(self.es, actions, index=self.index, raise_on_error=raise_on_error)

    def save(self, command):
        """
        保存命令到数据库
        """
        data = self.make_data(command)
        return self.es.index(index=self.index, doc_type=self.doc_type, body=data)

    def filter(self, query: dict, from_=None, size=None, sort=None):
        body = self.get_query_body(**query)

        data = self.es.search(
            index=self.query_index, doc_type=self.doc_type, body=body, from_=from_, size=size,
            sort=sort
        )
        source_data = []
        for item in data['hits']['hits']:
            if item:
                item['_source'].update({'id': item['_id']})
                source_data.append(item['_source'])

        return Command.from_multi_dict(source_data)

    def count(self, **query):
        body = self.get_query_body(**query)
        data = self.es.count(index=self.query_index, doc_type=self.doc_type, body=body)
        return data["count"]

    def __getattr__(self, item):
        return getattr(self.es, item)

    def all(self):
        """返回所有数据"""
        raise NotImplementedError("Not support")

    def ping(self, timeout=None):
        try:
            return self.es.ping(request_timeout=timeout)
        except Exception:
            return False

    def get_query_body(self, **kwargs):
        new_kwargs = {}
        for k, v in kwargs.items():
            new_kwargs[k] = str(v) if isinstance(v, UUID) else v
        kwargs = new_kwargs

        index_in_field = 'id__in'
        exact_fields = self.exact_fields
        match_fields = self.match_fields

        match = {}
        exact = {}
        index = {}

        if index_in_field in kwargs:
            index['values'] = kwargs[index_in_field]

        for k, v in kwargs.items():
            if k in exact_fields:
                exact[k] = v
            elif k in match_fields:
                match[k] = v

        # 处理时间
        timestamp__gte = kwargs.get('timestamp__gte')
        timestamp__lte = kwargs.get('timestamp__lte')
        timestamp_range = {}

        if timestamp__gte:
            timestamp_range['gte'] = timestamp__gte
        if timestamp__lte:
            timestamp_range['lte'] = timestamp__lte

        # 处理组织
        should = []
        org_id = match.get('org_id')

        real_default_org_id = '00000000-0000-0000-0000-000000000002'
        root_org_id = '00000000-0000-0000-0000-000000000000'

        if org_id == root_org_id:
            match.pop('org_id')
        elif org_id in (real_default_org_id, ''):
            match.pop('org_id')
            should.append({
                'bool': {
                    'must_not': [
                        {
                            'wildcard': {'org_id': '*'}
                        }
                    ]}
            })
            should.append({'match': {'org_id': real_default_org_id}})

        # 构建 body
        body = {
            'query': {
                'bool': {
                    'must': [
                        {'match': {k: v}} for k, v in match.items()
                    ],
                    'should': should,
                    'filter': [
                                  {
                                      'term': {k: v}
                                  } for k, v in exact.items()
                              ] + [
                                  {
                                      'range': {
                                          'timestamp': timestamp_range
                                      }
                                  }
                              ] + [
                                  {
                                      'ids': {k: v}
                                  } for k, v in index.items()
                              ]
                }
            },
        }
        return body


class QuerySet(DJQuerySet):
    _method_calls = None
    _storage = None
    _command_store_config = None
    _slice = None  # (from_, size)
    default_days_ago = 5
    max_result_window = 10000

    def __init__(self, command_store_config):
        self._method_calls = []
        self._command_store_config = command_store_config
        self._storage = CommandStore(command_store_config)

        # 命令列表模糊搜索时报错
        super().__init__()

    @lazyproperty
    def _grouped_method_calls(self):
        _method_calls = {k: list(v) for k, v in groupby(self._method_calls, lambda x: x[0])}
        return _method_calls

    @lazyproperty
    def _filter_kwargs(self):
        _method_calls = self._grouped_method_calls
        filter_calls = _method_calls.get('filter')
        if not filter_calls:
            return {}
        names, multi_args, multi_kwargs = zip(*filter_calls)
        kwargs = reduce(lambda x, y: {**x, **y}, multi_kwargs, {})

        striped_kwargs = {}
        for k, v in kwargs.items():
            k = k.replace('__exact', '')
            k = k.replace('__startswith', '')
            k = k.replace('__icontains', '')
            striped_kwargs[k] = v
        return striped_kwargs

    @lazyproperty
    def _sort(self):
        order_by = self._grouped_method_calls.get('order_by')
        if order_by:
            for call in reversed(order_by):
                fields = call[1]
                if fields:
                    field = fields[-1]

                    if field.startswith('-'):
                        direction = 'desc'
                    else:
                        direction = 'asc'
                    field = field.lstrip('-+')
                    sort = f'{field}:{direction}'
                    return sort

    def __execute(self):
        _filter_kwargs = self._filter_kwargs
        _sort = self._sort
        from_, size = self._slice or (None, None)
        data = self._storage.filter(_filter_kwargs, from_=from_, size=size, sort=_sort)
        return data

    def __stage_method_call(self, item, *args, **kwargs):
        _clone = self.__clone()
        _clone._method_calls.append((item, args, kwargs))
        return _clone

    def __clone(self):
        uqs = QuerySet(self._command_store_config)
        uqs._method_calls = self._method_calls.copy()
        uqs._slice = self._slice
        uqs.model = self.model
        return uqs

    def count(self, limit_to_max_result_window=True):
        filter_kwargs = self._filter_kwargs
        count = self._storage.count(**filter_kwargs)
        if limit_to_max_result_window:
            count = min(count, self.max_result_window)
        return count

    def __getattribute__(self, item):
        if any((
                item.startswith('__'),
                item in QuerySet.__dict__,
        )):
            return object.__getattribute__(self, item)

        origin_attr = object.__getattribute__(self, item)
        if not inspect.ismethod(origin_attr):
            return origin_attr

        attr = partial(self.__stage_method_call, item)
        return attr

    def __getitem__(self, item):
        max_window = self.max_result_window
        if isinstance(item, slice):
            if self._slice is None:
                clone = self.__clone()
                from_ = item.start or 0
                if item.stop is None:
                    size = self.max_result_window - from_
                else:
                    size = item.stop - from_

                if from_ + size > max_window:
                    if from_ >= max_window:
                        from_ = max_window
                        size = 0
                    else:
                        size = max_window - from_
                clone._slice = (from_, size)
                return clone
        return self.__execute()[item]

    def __repr__(self):
        return self.__execute().__repr__()

    def __iter__(self):
        return iter(self.__execute())

    def __len__(self):
        return self.count()