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
|
from sqlglot import exp, transforms
from sqlglot.dialects.dialect import (
Dialect,
format_time_lambda,
if_sql,
no_ilike_sql,
no_safe_divide_sql,
rename_func,
str_position_sql,
struct_extract_sql,
)
from sqlglot.dialects.mysql import MySQL
from sqlglot.generator import Generator
from sqlglot.helper import list_get
from sqlglot.parser import Parser
from sqlglot.tokens import Tokenizer, TokenType
def _approx_distinct_sql(self, expression):
accuracy = expression.args.get("accuracy")
accuracy = ", " + self.sql(accuracy) if accuracy else ""
return f"APPROX_DISTINCT({self.sql(expression, 'this')}{accuracy})"
def _concat_ws_sql(self, expression):
sep, *args = expression.expressions
sep = self.sql(sep)
if len(args) > 1:
return f"ARRAY_JOIN(ARRAY[{self.format_args(*args)}], {sep})"
return f"ARRAY_JOIN({self.sql(args[0])}, {sep})"
def _datatype_sql(self, expression):
sql = self.datatype_sql(expression)
if expression.this == exp.DataType.Type.TIMESTAMPTZ:
sql = f"{sql} WITH TIME ZONE"
return sql
def _date_parse_sql(self, expression):
return f"DATE_PARSE({self.sql(expression, 'this')}, '%Y-%m-%d %H:%i:%s')"
def _explode_to_unnest_sql(self, expression):
if isinstance(expression.this, (exp.Explode, exp.Posexplode)):
return self.sql(
exp.Join(
this=exp.Unnest(
expressions=[expression.this.this],
alias=expression.args.get("alias"),
ordinality=isinstance(expression.this, exp.Posexplode),
),
kind="cross",
)
)
return self.lateral_sql(expression)
def _initcap_sql(self, expression):
regex = r"(\w)(\w*)"
return f"REGEXP_REPLACE({self.sql(expression, 'this')}, '{regex}', x -> UPPER(x[1]) || LOWER(x[2]))"
def _no_sort_array(self, expression):
if expression.args.get("asc") == exp.FALSE:
comparator = "(a, b) -> CASE WHEN a < b THEN 1 WHEN a > b THEN -1 ELSE 0 END"
else:
comparator = None
args = self.format_args(expression.this, comparator)
return f"ARRAY_SORT({args})"
def _schema_sql(self, expression):
if isinstance(expression.parent, exp.Property):
columns = ", ".join(f"'{c.text('this')}'" for c in expression.expressions)
return f"ARRAY[{columns}]"
for schema in expression.parent.find_all(exp.Schema):
if isinstance(schema.parent, exp.Property):
expression = expression.copy()
expression.expressions.extend(schema.expressions)
return self.schema_sql(expression)
def _quantile_sql(self, expression):
self.unsupported("Presto does not support exact quantiles")
return f"APPROX_PERCENTILE({self.sql(expression, 'this')}, {self.sql(expression, 'quantile')})"
def _str_to_time_sql(self, expression):
return f"DATE_PARSE({self.sql(expression, 'this')}, {self.format_time(expression)})"
def _ts_or_ds_to_date_sql(self, expression):
time_format = self.format_time(expression)
if time_format and time_format not in (Presto.time_format, Presto.date_format):
return f"CAST({_str_to_time_sql(self, expression)} AS DATE)"
return f"CAST(SUBSTR(CAST({self.sql(expression, 'this')} AS VARCHAR), 1, 10) AS DATE)"
def _ts_or_ds_add_sql(self, expression):
this = self.sql(expression, "this")
e = self.sql(expression, "expression")
unit = self.sql(expression, "unit") or "'day'"
return f"DATE_ADD({unit}, {e}, DATE_PARSE(SUBSTR({this}, 1, 10), {Presto.date_format}))"
class Presto(Dialect):
index_offset = 1
null_ordering = "nulls_are_last"
time_format = "'%Y-%m-%d %H:%i:%S'"
time_mapping = MySQL.time_mapping
class Tokenizer(Tokenizer):
KEYWORDS = {
**Tokenizer.KEYWORDS,
"ROW": TokenType.STRUCT,
}
class Parser(Parser):
FUNCTIONS = {
**Parser.FUNCTIONS,
"APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list,
"CARDINALITY": exp.ArraySize.from_arg_list,
"CONTAINS": exp.ArrayContains.from_arg_list,
"DATE_ADD": lambda args: exp.DateAdd(
this=list_get(args, 2),
expression=list_get(args, 1),
unit=list_get(args, 0),
),
"DATE_DIFF": lambda args: exp.DateDiff(
this=list_get(args, 2),
expression=list_get(args, 1),
unit=list_get(args, 0),
),
"DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"),
"DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"),
"FROM_UNIXTIME": exp.UnixToTime.from_arg_list,
"STRPOS": exp.StrPosition.from_arg_list,
"TO_UNIXTIME": exp.TimeToUnix.from_arg_list,
"APPROX_PERCENTILE": exp.ApproxQuantile.from_arg_list,
}
class Generator(Generator):
STRUCT_DELIMITER = ("(", ")")
ROOT_PROPERTIES = {
exp.SchemaCommentProperty,
}
WITH_PROPERTIES = {
exp.PartitionedByProperty,
exp.FileFormatProperty,
exp.AnonymousProperty,
exp.TableFormatProperty,
}
TYPE_MAPPING = {
**Generator.TYPE_MAPPING,
exp.DataType.Type.INT: "INTEGER",
exp.DataType.Type.FLOAT: "REAL",
exp.DataType.Type.BINARY: "VARBINARY",
exp.DataType.Type.TEXT: "VARCHAR",
exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP",
exp.DataType.Type.STRUCT: "ROW",
}
TRANSFORMS = {
**Generator.TRANSFORMS,
**transforms.UNALIAS_GROUP,
exp.ApproxDistinct: _approx_distinct_sql,
exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]",
exp.ArrayConcat: rename_func("CONCAT"),
exp.ArrayContains: rename_func("CONTAINS"),
exp.ArraySize: rename_func("CARDINALITY"),
exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})",
exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
exp.ConcatWs: _concat_ws_sql,
exp.DataType: _datatype_sql,
exp.DateAdd: lambda self, e: f"""DATE_ADD({self.sql(e, 'unit') or "'day'"}, {self.sql(e, 'expression')}, {self.sql(e, 'this')})""",
exp.DateDiff: lambda self, e: f"""DATE_DIFF({self.sql(e, 'unit') or "'day'"}, {self.sql(e, 'expression')}, {self.sql(e, 'this')})""",
exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)",
exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)",
exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)",
exp.If: if_sql,
exp.ILike: no_ilike_sql,
exp.Initcap: _initcap_sql,
exp.Lateral: _explode_to_unnest_sql,
exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"),
exp.Quantile: _quantile_sql,
exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"),
exp.SafeDivide: no_safe_divide_sql,
exp.Schema: _schema_sql,
exp.SortArray: _no_sort_array,
exp.StrPosition: str_position_sql,
exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)",
exp.StrToTime: _str_to_time_sql,
exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))",
exp.StructExtract: struct_extract_sql,
exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT = '{e.text('value').upper()}'",
exp.TimeStrToDate: _date_parse_sql,
exp.TimeStrToTime: _date_parse_sql,
exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))",
exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})",
exp.TimeToUnix: rename_func("TO_UNIXTIME"),
exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)",
exp.TsOrDsAdd: _ts_or_ds_add_sql,
exp.TsOrDsToDate: _ts_or_ds_to_date_sql,
exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})",
exp.UnixToTime: rename_func("FROM_UNIXTIME"),
exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)",
}
|
