-
Notifications
You must be signed in to change notification settings - Fork 0
/
Encoder.py
268 lines (248 loc) · 11.7 KB
/
Encoder.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
from psqlparse import parse_dict
import numpy as np
from Connector import PGConnector
from config import Config
import torch
class Expr:
def __init__(self, expr,list_kind = 0):
self.expr = expr
self.list_kind = list_kind
self.isInt = False
self.isFloat = False
self.val = 0
def isCol(self,):
return isinstance(self.expr, dict) and "ColumnRef" in self.expr
def getValue(self, value_expr):
if "A_Const" in value_expr:
value = value_expr["A_Const"]["val"]
if "String" in value:
return "'" + value["String"]["str"].replace("'","''")+"\'"
elif "Integer" in value:
self.isInt = True
self.val = value["Integer"]["ival"]
return str(value["Integer"]["ival"])
elif "Float" in value:
self.isFloat = True
self.val = value["Float"]["str"]
return str(self.val)
elif "TypeCast" in value_expr:
if len(value_expr["TypeCast"]['typeName']['TypeName']['names'])==1:
return value_expr["TypeCast"]['typeName']['TypeName']['names'][0]['String']['str']+" '"+value_expr["TypeCast"]['arg']['A_Const']['val']['String']['str']+"'"
else:
if value_expr["TypeCast"]['typeName']['TypeName']['typmods'][0]['A_Const']['val']['Integer']['ival']==2:
return value_expr["TypeCast"]['typeName']['TypeName']['names'][1]['String']['str']+" '"+value_expr["TypeCast"]['arg']['A_Const']['val']['String']['str']+ "' month"
else:
return value_expr["TypeCast"]['typeName']['TypeName']['names'][1]['String']['str']+" '"+value_expr["TypeCast"]['arg']['A_Const']['val']['String']['str']+ "' year"
else:
print(value_expr.keys())
raise "unknown Value in Expr"
def getTableName(self,):
return self.expr["ColumnRef"]["fields"][0]["String"]["str"]
def getColumnName(self,):
return self.expr["ColumnRef"]["fields"][1]["String"]["str"]
def __str__(self,):
if self.isCol():
return self.getTableName()+"."+self.getColumnName()
elif isinstance(self.expr, dict) and "A_Const" in self.expr:
return self.getValue(self.expr)
elif isinstance(self.expr, dict) and "TypeCast" in self.expr:
return self.getValue(self.expr)
elif isinstance(self.expr, list):
if self.list_kind == 6:
return "("+",\n".join([self.getValue(x) for x in self.expr])+")"
elif self.list_kind == 10:
return " AND ".join([self.getValue(x) for x in self.expr])
else:
raise "list kind error"
class Comparison:
def __init__(self, comparison):
self.comparison = comparison
self.column_list = []
if "A_Expr" in self.comparison:
self.lexpr = Expr(comparison["A_Expr"]["lexpr"])
self.column = str(self.lexpr)
self.kind = comparison["A_Expr"]["kind"]
if not "A_Expr" in comparison["A_Expr"]["rexpr"]:
self.rexpr = Expr(comparison["A_Expr"]["rexpr"],self.kind)
else:
self.rexpr = Comparison(comparison["A_Expr"]["rexpr"])
self.aliasname_list = []
if self.lexpr.isCol():
self.aliasname_list.append(self.lexpr.getTableName())
self.column_list.append(self.lexpr.getColumnName())
if self.rexpr.isCol():
self.aliasname_list.append(self.rexpr.getTableName())
self.column_list.append(self.rexpr.getColumnName())
self.comp_kind = 0
elif "NullTest" in self.comparison:
self.lexpr = Expr(comparison["NullTest"]["arg"])
self.column = str(self.lexpr)
self.kind = comparison["NullTest"]["nulltesttype"]
self.aliasname_list = []
if self.lexpr.isCol():
self.aliasname_list.append(self.lexpr.getTableName())
self.column_list.append(self.lexpr.getColumnName())
self.comp_kind = 1
else:
# "boolop"
self.kind = comparison["BoolExpr"]["boolop"]
self.comp_list = [Comparison(x)
for x in comparison["BoolExpr"]["args"]]
self.aliasname_list = []
for comp in self.comp_list:
if comp.lexpr.isCol():
self.aliasname_list.append(comp.lexpr.getTableName())
self.lexpr = comp.lexpr
self.column = str(self.lexpr)
self.column_list.append(comp.lexpr.getColumnName())
break
self.comp_kind = 2
def isCol(self,):
return False
def __str__(self,):
if self.comp_kind == 0:
Op = ""
if self.kind == 0:
Op = self.comparison["A_Expr"]["name"][0]["String"]["str"]
elif self.kind == 7:
if self.comparison["A_Expr"]["name"][0]["String"]["str"]=="!~~":
Op = "not like"
else:
Op = "like"
elif self.kind == 8:
if self.comparison["A_Expr"]["name"][0]["String"]["str"]=="~~*":
Op = "ilike"
else:
raise
elif self.kind == 6:
Op = "IN"
elif self.kind == 10:
Op = "BETWEEN"
else:
import json
print(json.dumps(self.comparison, sort_keys=True, indent=4))
raise "Operation ERROR"
return str(self.lexpr)+" "+Op+" "+ str(self.rexpr)
elif self.comp_kind == 1:
if self.kind == 1:
return str(self.lexpr)+" IS NOT NULL"
else:
return str(self.lexpr)+" IS NULL"
else:
res = ""
for comp in self.comp_list:
if res == "":
res += "( "+str(comp)
else:
if self.kind == 1:
res += " OR "
else:
res += " AND "
res += str(comp)
res += ")"
return res
def get_join_tables(self):
return self.lexpr.getTableName(), self.rexpr.getTableName()
cfg = Config()
pgc = PGConnector(cfg.database, cfg.username, cfg.password, cfg.pghost, cfg.pgport)
JOIN_TYPES = ["Nested Loop", "Hash Join", "Merge Join"]
LEAF_TYPES = ["Seq Scan", "Index Scan", "Index Only Scan", "Bitmap Index Scan"]
ALL_TYPES = JOIN_TYPES + LEAF_TYPES
def normalize(cost):
return int(np.log(2+cost)/np.log(cfg.max_time_out)*200)/200.
class SqlEncoder:
def __init__(self, table2index):
self.column2index = {}
self.table2index = table2index
def getColumnIndex(self, column):
if column not in self.column2index.keys():
self.column2index[column] = len(self.column2index)
return self.column2index[column]
def to_vec(self, sql):
table_num = cfg.max_table_num
parse_result = parse_dict(sql)[0]["SelectStmt"]
self.from_tables = [x["RangeVar"] for x in parse_result["fromClause"]]
self.join_matrix = np.zeros((table_num, table_num), dtype = np.float)
if "BoolExpr" in parse_result["whereClause"]:
self.comparision_list = [Comparison(x) for x in parse_result["whereClause"]["BoolExpr"]["args"]]
else :
self.comparision_list = [Comparison(parse_result["whereClause"])]
self.predicates_selectivity = np.asarray([0]*cfg.max_column, dtype = float)
for table in self.from_tables:
table["relname"]
for comparision in self.comparision_list:
if len(comparision.column_list) == 2:
left_table, right_table = comparision.get_join_tables()
idx1 = self.table2index[left_table]
idx2 = self.table2index[right_table]
self.join_matrix[idx1][idx2] = 1
self.join_matrix[idx2][idx1] = 1
else:
table = comparision.aliasname_list[0]
col_index = self.getColumnIndex(comparision.column)
selectivity = pgc.getPGSelectivity(table, str(comparision))
self.predicates_selectivity[col_index] = self.predicates_selectivity[col_index] + selectivity
return np.concatenate((self.join_matrix.flatten(), self.predicates_selectivity))
class PlanEncoder:
def __init__(self, table2index):
self.table2index = table2index
def to_feature_cost(self, plan):
return [normalize(plan["Total Cost"]), normalize(plan["Plan Rows"])]
def to_table_id(self, plan):
return np.asarray([self.table2index[plan["Relation Name"]]])
def to_feature_join(self, plan):
# n为表的最大数量，left_table和right_table都是长度为2n的one-hot编码
feature_type = np.zeros(len(ALL_TYPES))
feature_type[ALL_TYPES.index(plan["Node Type"])] = 1
feature = np.concatenate((feature_type, self.to_feature_cost(plan)))
feature = torch.tensor(feature,device = cfg.device,dtype = torch.float32).reshape(-1,cfg.input_size)
return feature
def to_feature_scan(self,plan):
feature_type = np.zeros(len(ALL_TYPES))
rel_name = plan["Relation Name"]
feature_type[ALL_TYPES.index(plan["Node Type"])] = 1
feature = np.concatenate((feature_type, self.to_feature_cost(plan)))
feature = torch.tensor(feature,device = cfg.device,dtype = torch.float32).reshape(-1,cfg.input_size)
return (feature, torch.tensor(self.to_table_id(plan), device = cfg.device,dtype = torch.long))
def plan_to_feature_tree(self, plan):
if "Plan" in plan:
plan = plan["Plan"]
children = plan["Plan"] if "Plan" in plan else (plan["Plans"] if "Plans" in plan else [])
# len(children) == 1 表示node type为gather，是并行时产生的结果，没有实际意义
if len(children) == 1:
return self.plan_to_feature_tree(children[0])
if plan["Node Type"] in JOIN_TYPES:
left = self.plan_to_feature_tree(children[0])
right = self.plan_to_feature_tree(children[1])
pos= self.to_feature_join(plan)
return (pos,left,right)
if plan["Node Type"] in LEAF_TYPES:
pos= self.to_feature_scan(plan)
return pos
class ValueExtractor:
def __init__(self,max_value = 20):
self.max_value = max_value
# def encode(self,v):
# return np.log(self.offset+v)/np.log(2)/self.max_value
# def decode(self,v):
# # v=-(v*v<0)
# return np.exp(v*self.max_value*np.log(2))#-self.offset
def encode(self,v):
return torch.tensor([[int(np.log(2+v)/np.log(cfg.max_time_out)*200)/200.]], device = cfg.device, dtype = torch.float)
return int(np.log(self.offset+v)/np.log(config.max_time_out)*200)/200.
def decode(self,v):
# v=-(v*v<0)
# return np.exp(v/2*np.log(config.max_time_out))#-self.offset
return np.exp(v*np.log(cfg.max_time_out))#-self.offset
def cost_encode(self,v,min_cost,max_cost):
return (v-min_cost)/(max_cost-min_cost)
def cost_decode(self,v,min_cost,max_cost):
return (max_cost-min_cost)*v+min_cost
def latency_encode(self,v,min_latency,max_latency):
return (v-min_latency)/(max_latency-min_latency)
def latency_decode(self,v,min_latency,max_latency):
return (max_latency-min_latency)*v+min_latency
def rows_encode(self,v,min_cost,max_cost):
return (v-min_cost)/(max_cost-min_cost)
def rows_decode(self,v,min_cost,max_cost):
return (max_cost-min_cost)*v+min_cost