Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
BaiXuePrincess
milvus
提交
d7e75bf9
milvus
项目概览
BaiXuePrincess
/
milvus
与 Fork 源项目一致
从无法访问的项目Fork
通知
7
Star
4
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
milvus
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
d7e75bf9
编写于
3月 22, 2019
作者:
J
jinhai
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Update for more function
上级
4787922d
变更
9
隐藏空白更改
内联
并排
Showing
9 changed file
with
247 addition
and
11 deletion
+247
-11
pyengine/engine/__init__.py
pyengine/engine/__init__.py
+3
-3
pyengine/engine/controller/group_handler.py
pyengine/engine/controller/group_handler.py
+29
-0
pyengine/engine/controller/index_file_handler.py
pyengine/engine/controller/index_file_handler.py
+2
-1
pyengine/engine/controller/index_manager.py
pyengine/engine/controller/index_manager.py
+8
-4
pyengine/engine/controller/raw_file_handler.py
pyengine/engine/controller/raw_file_handler.py
+18
-0
pyengine/engine/controller/vector_engine.py
pyengine/engine/controller/vector_engine.py
+176
-0
pyengine/engine/model/file_table.py
pyengine/engine/model/file_table.py
+4
-1
pyengine/engine/model/group_table.py
pyengine/engine/model/group_table.py
+6
-1
pyengine/engine/settings.py
pyengine/engine/settings.py
+1
-1
未找到文件。
pyengine/engine/__init__.py
浏览文件 @
d7e75bf9
...
...
@@ -10,7 +10,7 @@ app.config.from_object('engine.settings')
print
(
"Create database instance"
)
db
=
SQLAlchemy
(
app
)
from
engine.model.
GroupT
able
import
GroupTable
from
engine.model.
FileT
able
import
FileTable
from
engine.model.
group_t
able
import
GroupTable
from
engine.model.
file_t
able
import
FileTable
from
engine.controller
import
IndexManage
from
engine.controller
import
index_manager
pyengine/engine/controller/
GroupH
andler.py
→
pyengine/engine/controller/
group_h
andler.py
浏览文件 @
d7e75bf9
import
os
,
shutil
from
engine.settings
import
DATABASE_DIRECTORY
class
GroupHandler
(
object
):
@
staticmethod
def
CreateGroupDirectory
(
group_id
):
path
=
GetGroupDirectory
(
group_id
)
path
=
G
roupHandler
.
G
etGroupDirectory
(
group_id
)
path
=
path
.
strip
()
path
=
path
.
rstrip
(
"
\\
"
)
if
not
os
.
path
.
exists
():
if
not
os
.
path
.
exists
(
path
):
os
.
makedirs
(
path
)
print
(
"CreateGroupDirectory, Path: "
,
path
)
@
staticmethod
def
DeleteGroupDirectory
(
group_id
):
path
=
GetGroupDirectory
(
group_id
)
path
=
G
roupHandler
.
G
etGroupDirectory
(
group_id
)
path
=
path
.
strip
()
path
=
path
.
rstrip
(
"
\\
"
)
if
os
.
path
.
exists
():
if
os
.
path
.
exists
(
path
):
shutil
.
rmtree
(
path
)
print
(
"DeleteGroupDirectory, Path: "
,
path
)
@
staticmethod
def
GetGroupDirectory
(
group_id
):
return
DATABASE_DIRECTORY
+
'/'
+
group_id
\ No newline at end of file
print
(
"GetGroupDirectory, Path: "
,
DATABASE_DIRECTORY
+
'/'
+
group_id
)
return
DATABASE_DIRECTORY
+
'/'
+
group_id
pyengine/engine/controller/
RawFileH
andler.py
→
pyengine/engine/controller/
index_file_h
andler.py
浏览文件 @
d7e75bf9
class
RawFileHandler
(
object
):
class
IndexFileHandler
(
object
):
@
staticmethod
def
Create
(
filename
,
type
):
# type means: csv, parquet
...
...
pyengine/engine/controller/
IndexManage
.py
→
pyengine/engine/controller/
index_manager
.py
浏览文件 @
d7e75bf9
from
flask
import
Flask
,
jsonify
,
request
from
flask_restful
import
Resource
,
Api
from
engine
import
app
,
db
from
engine.model.
GroupT
able
import
GroupTable
from
engine.controller.
VectorE
ngine
import
VectorEngine
from
engine.model.
group_t
able
import
GroupTable
from
engine.controller.
vector_e
ngine
import
VectorEngine
# app = Flask(__name__)
api
=
Api
(
app
)
...
...
@@ -25,12 +25,13 @@ class VectorSearch(Resource):
def
__init__
(
self
):
self
.
__parser
=
reqparse
.
RequestParser
()
self
.
__parser
.
add_argument
(
'vector'
,
type
=
float
,
action
=
'append'
,
location
=
[
'json'
])
self
.
__parser
.
add_argument
(
'limit'
,
type
=
int
,
action
=
'append'
,
location
=
[
'json'
])
def
post
(
self
,
group_id
):
args
=
self
.
__parser
.
parse_args
()
print
(
'vector: '
,
args
[
'vector'
])
# go to search every thing
return
"vectorSearch post"
return
VectorEngine
.
SearchVector
(
group_id
,
args
[
'vector'
],
args
[
'limit'
])
class
Index
(
Resource
):
...
...
@@ -46,9 +47,12 @@ class Group(Resource):
def
__init__
(
self
):
self
.
__parser
=
reqparse
.
RequestParser
()
self
.
__parser
.
add_argument
(
'group_id'
,
type
=
str
)
self
.
__parser
.
add_argument
(
'dimension'
,
type
=
int
,
action
=
'append'
,
location
=
[
'json'
])
def
post
(
self
,
group_id
):
return
VectorEngine
.
AddGroup
(
group_id
)
args
=
self
.
__parser
.
parse_args
()
dimension
=
args
[
'dimension'
]
return
VectorEngine
.
AddGroup
(
group_id
,
dimension
)
def
get
(
self
,
group_id
):
return
VectorEngine
.
GetGroup
(
group_id
)
...
...
pyengine/engine/controller/raw_file_handler.py
0 → 100644
浏览文件 @
d7e75bf9
class
RawFileHandler
(
object
):
@
staticmethod
def
Create
(
filename
,
type
):
# type means: csv, parquet
pass
@
staticmethod
def
Read
(
filename
,
type
):
pass
@
staticmethod
def
Append
(
filename
,
type
,
record
):
pass
@
staticmethod
def
GetRawFilename
(
group_id
):
return
group_id
+
'.raw'
\ No newline at end of file
pyengine/engine/controller/
VectorE
ngine.py
→
pyengine/engine/controller/
vector_e
ngine.py
浏览文件 @
d7e75bf9
from
engine.model.GroupTable
import
GroupTable
from
engine.model.FileTable
import
FileTable
from
engine.controller.RawFileHandler
import
RawFileHandler
from
engine.controller.GroupHandler
import
GroupHandler
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
import
sys
,
os
class
VectorEngine
(
object
):
group_dict
=
None
@
staticmethod
def
AddGroup
(
group_id
):
def
AddGroup
(
group_id
,
dimension
):
group
=
GroupTable
.
query
.
filter
(
GroupTable
.
group_name
==
group_id
).
first
()
if
group
:
print
(
'Already create the group: '
,
group_id
)
return
jsonify
({
'code'
:
1
,
'group_name'
:
group_id
,
'file_number'
:
group
.
file_number
})
else
:
new_group
=
GroupTable
(
group_id
)
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
()
GroupHandler
.
CreateGroupDirectory
(
group_id
)
return
jsonify
({
'code'
:
0
,
'group_name'
:
group_id
,
'file_number'
:
0
})
@
staticmethod
def
GetGroup
(
group_id
):
group
=
GroupTable
.
query
.
filter
(
GroupTable
.
group_name
==
group_id
).
first
()
if
group
:
print
(
'Found the group: '
,
group_id
)
return
jsonify
({
'code'
:
0
,
'group_name'
:
group_id
,
'file_number'
:
group
.
file_number
})
else
:
print
(
'Not found the group: '
,
group_id
)
return
jsonify
({
'code'
:
1
,
'group_name'
:
group_id
,
'file_number'
:
0
})
# not found
@
staticmethod
def
DeleteGroup
(
group_id
):
group
=
GroupTable
.
query
.
filter
(
GroupTable
.
group_name
==
group_id
).
first
()
...
...
@@ -37,10 +47,18 @@ class VectorEngine(object):
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
()
return
jsonify
({
'code'
:
0
,
'group_name'
:
group_id
,
'file_number'
:
group
.
file_number
})
else
:
return
jsonify
({
'code'
:
0
,
'group_name'
:
group_id
,
'file_number'
:
0
})
@
staticmethod
def
GetGroupList
():
group
=
GroupTable
.
query
.
all
()
...
...
@@ -54,56 +72,95 @@ class VectorEngine(object):
print
(
group_list
)
return
jsonify
(
results
=
group_list
)
@
staticmethod
def
AddVector
(
group_id
,
vector
):
print
(
group_id
,
vector
)
file
=
FileTable
.
query
.
filter
(
and_
(
FileTable
.
group_name
==
group_id
,
FileTable
.
type
==
'raw'
)
).
first
()
file
=
FileTable
.
query
.
filter
(
FileTable
.
group_name
==
group_id
).
filter
(
FileTable
.
type
==
'raw'
).
first
()
if
file
:
if
file
.
row_number
>=
ROW_LIMIT
:
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
:
# read data from raw file
data
=
GetVectorsFromRawFile
()
# create index
index_filename
=
file
.
filename
+
"_index"
CreateIndex
(
group_id
,
index_filename
)
# create another raw file
raw_filename
=
file
.
seq_no
InsertVectorIntoRawFile
(
group_id
,
raw_filename
,
vector
)
# insert a record into database
db
.
session
.
add
(
FileTable
(
group_id
,
raw_filename
,
'raw'
,
1
))
db
.
session
.
commit
()
index_filename
=
file
.
filename
+
'_index'
CreateIndex
(
group_id
,
index_filename
,
data
)
# update record into database
FileTable
.
query
.
filter
(
FileTable
.
group_name
==
group_id
).
filter
(
FileTable
.
type
==
'raw'
).
update
({
'row_number'
:
file
.
row_number
+
1
,
'type'
:
'index'
})
pass
else
:
# we still can insert into exist raw file
InsertVectorIntoRawFile
(
file
.
filename
,
vector
)
# update database
# FileTable.query.filter_by(FileTable.group_name == group_id).filter_by(FileTable.type == 'raw').update('row_number':file.row_number + 1)
# 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
+
'
_0
'
raw_filename
=
group_id
+
'
.raw
'
# create and insert vector into raw file
InsertVectorIntoRawFile
(
raw_filename
,
vector
)
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
()
return
jsonify
({
'code'
:
0
})
@
staticmethod
def
SearchVector
(
group_id
,
vector
,
limit
):
# find all files
files
=
FileTable
.
query
.
filter
(
FileTable
.
group_name
==
group_id
).
all
()
for
file
in
files
:
if
(
file
.
type
==
'raw'
):
# create index
# add vector list
# train
# get topk
print
(
'search in raw file: '
,
file
.
filename
)
pass
else
:
# get topk
print
(
'search in index file: '
,
file
.
filename
)
data
=
IndexFileHandler
.
Read
(
file
.
filename
,
file
.
type
)
pass
# according to difference files get topk of each
# reduce the topk from them
# construct response and send back
return
jsonify
({
'code'
:
0
})
@
staticmethod
def
CreateIndex
(
group_id
,
filename
):
path
=
GroupHandler
.
GetGroupDirectory
(
group_id
)
+
'/'
+
filename
print
(
group_id
,
path
)
def
CreateIndex
(
group_id
):
# 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
)
return
jsonify
({
'code'
:
0
})
@
staticmethod
def
InsertVectorIntoRawFile
(
group_id
,
filename
,
vector
):
print
(
sys
.
_getframe
().
f_code
.
co_name
)
path
=
GroupHandler
.
GetGroupDirectory
(
group_id
)
+
'/'
+
filename
# 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
()
VectorEngine
.
group_dict
[
group_id
]
=
[]
VectorEngine
.
group_dict
[
group_id
].
append
(
vector
)
print
(
'InsertVectorIntoRawFile: '
,
VectorEngine
.
group_dict
[
group_id
])
# if filename exist
# append
...
...
@@ -112,3 +169,8 @@ class VectorEngine(object):
# append
return
filename
@
staticmethod
def
GetVectorListFromRawFile
(
group_id
,
filename
):
return
VectorEngine
.
group_dict
[
group_id
]
pyengine/engine/model/
FileT
able.py
→
pyengine/engine/model/
file_t
able.py
浏览文件 @
d7e75bf9
...
...
@@ -9,6 +9,7 @@ class FileTable(db.Model):
row_number
=
db
.
Column
(
db
.
Integer
)
seq_no
=
db
.
Column
(
db
.
Integer
)
def
__init__
(
self
,
group_name
,
filename
,
type
,
row_number
):
self
.
group_name
=
group_name
self
.
filename
=
filename
...
...
@@ -17,5 +18,7 @@ class FileTable(db.Model):
self
.
type
=
type
self
.
seq_no
=
0
def
__repr__
(
self
):
return
'<FileTable $r>'
%
self
.
tablename
\ No newline at end of file
return
'<FileTable $r>'
%
self
.
tablename
pyengine/engine/model/
GroupT
able.py
→
pyengine/engine/model/
group_t
able.py
浏览文件 @
d7e75bf9
...
...
@@ -5,10 +5,15 @@ class GroupTable(db.Model):
id
=
db
.
Column
(
db
.
Integer
,
primary_key
=
True
)
group_name
=
db
.
Column
(
db
.
String
(
100
))
file_number
=
db
.
Column
(
db
.
Integer
)
dimension
=
db
.
Column
(
db
.
Integer
)
def
__init__
(
self
,
group_name
):
def
__init__
(
self
,
group_name
,
dimension
):
self
.
group_name
=
group_name
self
.
dimension
=
dimension
self
.
file_number
=
0
self
.
dimension
=
0
def
__repr__
(
self
):
return
'<GroupTable $s>'
%
self
.
group_name
\ No newline at end of file
pyengine/engine/settings.py
浏览文件 @
d7e75bf9
...
...
@@ -6,4 +6,4 @@ SQLALCHEMY_TRACK_MODIFICATIONS = False
SQLALCHEMY_DATABASE_URI
=
"mysql+pymysql://vecwise@127.0.0.1:3306/vecdata"
ROW_LIMIT
=
10000000
DATABASE_DIRECTORY
=
'/home/jinhai/Document/development/vecwise_engine/db'
\ No newline at end of file
DATABASE_DIRECTORY
=
'/home/jinhai/disk0/vecwise/db'
\ No newline at end of file
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录