vector_engine.py 7.5 KB
Newer Older
J
jinhai 已提交
1 2 3 4 5 6 7 8
from engine.model.group_table import GroupTable
from engine.model.file_table import FileTable
from engine.controller.raw_file_handler import RawFileHandler
from engine.controller.group_handler import GroupHandler
from engine.controller.index_file_handler import IndexFileHandler
from engine.settings import ROW_LIMIT
from flask import jsonify
from engine import db
X
xj.lin 已提交
9 10
from engine.ingestion import build_index
from engine.controller.scheduler import Scheduler
X
xj.lin 已提交
11
from engine.ingestion import serialize
J
jinhai 已提交
12 13 14 15
import sys, os

class VectorEngine(object):
    group_dict = None
J
jinhai 已提交
16 17 18
    SUCCESS_CODE = 0
    FAULT_CODE = 1
    GROUP_NOT_EXIST = 2
J
jinhai 已提交
19 20 21 22 23 24

    @staticmethod
    def AddGroup(group_id, dimension):
        group = GroupTable.query.filter(GroupTable.group_name==group_id).first()
        if group:
            print('Already create the group: ', group_id)
J
jinhai 已提交
25 26
            return VectorEngine.FAULT_CODE, group_id, group.file_number
            # return jsonify({'code': 1, 'group_name': group_id, 'file_number': group.file_number})
J
jinhai 已提交
27 28 29 30 31 32 33 34
        else:
            print('To create the group: ', group_id)
            new_group = GroupTable(group_id, dimension)
            GroupHandler.CreateGroupDirectory(group_id)

            # add into database
            db.session.add(new_group)
            db.session.commit()
J
jinhai 已提交
35
            return VectorEngine.SUCCESS_CODE, group_id, 0
J
jinhai 已提交
36 37 38 39 40 41


    @staticmethod
    def GetGroup(group_id):
        group = GroupTable.query.filter(GroupTable.group_name==group_id).first()
        if group:
J
jinhai 已提交
42
            return VectorEngine.SUCCESS_CODE, group_id, group.file_number
J
jinhai 已提交
43
        else:
J
jinhai 已提交
44
            return VectorEngine.FAULT_CODE, group_id, 0
J
jinhai 已提交
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61


    @staticmethod
    def DeleteGroup(group_id):
        group = GroupTable.query.filter(GroupTable.group_name==group_id).first()
        if(group):
            # old_group = GroupTable(group_id)
            db.session.delete(group)
            db.session.commit()
            GroupHandler.DeleteGroupDirectory(group_id)

            records = FileTable.query.filter(FileTable.group_name == group_id).all()
            for record in records:
                print("record.group_name: ", record.group_name)
                db.session.delete(record)
            db.session.commit()

J
jinhai 已提交
62
            return VectorEngine.SUCCESS_CODE, group_id, group.file_number
J
jinhai 已提交
63
        else:
J
jinhai 已提交
64
            return VectorEngine.SUCCESS_CODE, group_id, 0
J
jinhai 已提交
65 66 67 68 69 70 71 72 73 74 75 76 77


    @staticmethod
    def GetGroupList():
        group = GroupTable.query.all()
        group_list = []
        for group_tuple in group:
            group_item = {}
            group_item['group_name'] = group_tuple.group_name
            group_item['file_number'] = group_tuple.file_number
            group_list.append(group_item)

        print(group_list)
J
jinhai 已提交
78
        return VectorEngine.SUCCESS_CODE, group_list
J
jinhai 已提交
79 80 81 82 83


    @staticmethod
    def AddVector(group_id, vector):
        print(group_id, vector)
J
jinhai 已提交
84 85 86 87
        code, _, _ = VectorEngine.GetGroup(group_id)
        if code == VectorEngine.FAULT_CODE:
            return VectorEngine.GROUP_NOT_EXIST

J
jinhai 已提交
88
        file = FileTable.query.filter(FileTable.group_name == group_id).filter(FileTable.type == 'raw').first()
X
xj.lin 已提交
89
        group = GroupTable.query.filter(GroupTable.group_name == group_id).first()
J
jinhai 已提交
90 91 92 93 94 95 96
        if file:
            print('insert into exist file')
            # insert into raw file
            VectorEngine.InsertVectorIntoRawFile(group_id, file.filename, vector)

            # check if the file can be indexed
            if file.row_number + 1 >= ROW_LIMIT:
X
xj.lin 已提交
97
                raw_data = VectorEngine.GetVectorListFromRawFile(group_id)
X
xj.lin 已提交
98
                d = group.dimension
J
jinhai 已提交
99 100

                # create index
X
xj.lin 已提交
101
                index_builder = build_index.FactoryIndex()
X
xj.lin 已提交
102
                index = index_builder().build(d, raw_data)
X
xj.lin 已提交
103 104

                # TODO(jinhai): store index into Cache
J
jinhai 已提交
105
                index_filename = file.filename + '_index'
X
xj.lin 已提交
106
                serialize.write_index(file_name=index_filename, index=index)
J
jinhai 已提交
107

X
xj.lin 已提交
108 109 110
                FileTable.query.filter(FileTable.group_name == group_id).filter(FileTable.type == 'raw').update({'row_number':file.row_number + 1,
                                                                                                                 'type': 'index',
                                                                                                                 'filename': index_filename})
J
jinhai 已提交
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
                pass

            else:
                # we still can insert into exist raw file, update database
                FileTable.query.filter(FileTable.group_name == group_id).filter(FileTable.type == 'raw').update({'row_number':file.row_number + 1})
                db.session.commit()
                print('Update db for raw file insertion')
                pass

        else:
            print('add a new raw file')
            # first raw file
            raw_filename = group_id + '.raw'
            # create and insert vector into raw file
            VectorEngine.InsertVectorIntoRawFile(group_id, raw_filename, vector)
            # insert a record into database
            db.session.add(FileTable(group_id, raw_filename, 'raw', 1))
            db.session.commit()

J
jinhai 已提交
130
        return VectorEngine.SUCCESS_CODE
J
jinhai 已提交
131 132 133 134


    @staticmethod
    def SearchVector(group_id, vector, limit):
J
jinhai 已提交
135 136 137 138 139
        # Check the group exist
        code, _, _ = VectorEngine.GetGroup(group_id)
        if code == VectorEngine.FAULT_CODE:
            return VectorEngine.GROUP_NOT_EXIST

X
xj.lin 已提交
140 141
        group = GroupTable.query.filter(GroupTable.group_name == group_id).first()

J
jinhai 已提交
142 143
        # find all files
        files = FileTable.query.filter(FileTable.group_name == group_id).all()
X
xj.lin 已提交
144
        index_keys = [ i.filename for i in files if i.type == 'index' ]
J
jinhai 已提交
145
        index_map = {}
X
xj.lin 已提交
146
        index_map['index'] = index_keys
X
xj.lin 已提交
147
        index_map['raw'] = VectorEngine.GetVectorListFromRawFile(group_id, "fakename") #TODO: pass by key, get from storage
X
xj.lin 已提交
148
        index_map['dimension'] = group.dimension
J
jinhai 已提交
149

J
jinhai 已提交
150
        scheduler_instance = Scheduler()
X
xj.lin 已提交
151 152 153
        vectors = []
        vectors.append(vector)
        result = scheduler_instance.Search(index_map, vectors, limit)
J
jinhai 已提交
154

J
jinhai 已提交
155 156 157
        vector_id = 0

        return VectorEngine.SUCCESS_CODE, vector_id
J
jinhai 已提交
158 159 160 161


    @staticmethod
    def CreateIndex(group_id):
J
jinhai 已提交
162 163 164 165 166
        # Check the group exist
        code, _, _ = VectorEngine.GetGroup(group_id)
        if code == VectorEngine.FAULT_CODE:
            return VectorEngine.GROUP_NOT_EXIST

J
jinhai 已提交
167 168 169 170
        # create index
        file = FileTable.query.filter(FileTable.group_name == group_id).filter(FileTable.type == 'raw').first()
        path = GroupHandler.GetGroupDirectory(group_id) + '/' + file.filename 
        print('Going to create index for: ', path)
J
jinhai 已提交
171
        return VectorEngine.SUCCESS_CODE
J
jinhai 已提交
172 173 174 175 176 177 178 179 180


    @staticmethod
    def InsertVectorIntoRawFile(group_id, filename, vector):
        # print(sys._getframe().f_code.co_name, group_id, vector)
        # path = GroupHandler.GetGroupDirectory(group_id) + '/' + filename
        if VectorEngine.group_dict is None:
            # print("VectorEngine.group_dict is None")
            VectorEngine.group_dict = dict()
J
jinhai 已提交
181 182

        if not (group_id in VectorEngine.group_dict):
J
jinhai 已提交
183 184 185 186 187 188 189 190 191
            VectorEngine.group_dict[group_id] = []

        VectorEngine.group_dict[group_id].append(vector)

        print('InsertVectorIntoRawFile: ', VectorEngine.group_dict[group_id])
        return filename


    @staticmethod
X
xj.lin 已提交
192
    def GetVectorListFromRawFile(group_id, filename="todo"):
X
xj.lin 已提交
193
        return serialize.to_array(VectorEngine.group_dict[group_id])
J
jinhai 已提交
194

J
jinhai 已提交
195 196 197 198 199 200
    @staticmethod
    def ClearRawFile(group_id):
        print("VectorEngine.group_dict: ", VectorEngine.group_dict)
        del VectorEngine.group_dict[group_id]
        return VectorEngine.SUCCESS_CODE